The Growth of Medical Information Systems in the United States Donald A.B. Lindberg Lexington Books The Growth of Medical Information Systems in the United States The Growth of Medical Information Systems in the United States Donald A.B. Lindberg University of Missouri LexingtonBooks D.C. Heath and Company Lexington, Massachusetts Toronto Library of Congress Cataloging in Publication Data Lindberg, Donald A B The growth of medical information systems in the United States. An outgrowth of the author’s The development and diffusion of a medical technology, 1977 prepared for the Committee on Technology and Health Care of the National Academy of Sciences. Includes bibliographies. 1. Information storage and retrieval systems—Medicine. 2. Medicine —United States—Data processing. 3. Medical innovations—United States. 4. Diffusion of innovations—United States. I. Title. R858.L56 029’.9’61 79-1555 ISBN 0-669-029114 Copyright © 1979 by D.C. Heath and Company All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage or retrieval system, without permission in writing from the publisher. Published simultaneously in Canada Printed in the United States of America International Standard Book Number: 0-669-029114 Library of Congress Catalog Card Number: 79-1555 Contents Preface vii Chapter 1 Introduction 1 Importance of Examining MIS Technology 1 Limitations of This Study 2 General Plan for Examining the Development of MISs in the United States 4 International Literature of MIS Technology 5 Summary Conclusions 6 Summary Recommendations 6 Chapter 2 Definition and Means of Comparing MISs 9 Summary 9 Working Definition 9 Implications 9 Systems Philosophies 10 Information Content of MISs 14 An Analytic Framework for Comparing MISs 15 Chapter 3 State of the Art of MISs 21 Summary 21 Feasibility of Specific Kinds of MISs 21 Tasks Thus Far Infeasible 30 Chapter 4 Description of MISs 37 Summary 37 Systems Designed for a Particular Institution 37 Commercially Offered Systems 47 Systems of Historical Interest 58 Chapter 5 Evaluating the Worth of MISs 69 Summary 69 The Need to Evaluate 69 For Whom and in What Sense Might MISs Be Worthwhile? 70 An Inventory of Methods at Hand for Evaluating the Worth of MISs 72 Results of Employing the Evaluation Methods 78 V VI Growth of MISs in the United States Chapter 6 Barriers to the Development and Diffusion of MIS Technology 105 Summary 105 The General Nature of Diffusion of Technological Innovation 105 Diffusion of Technological Innovation in Medicine 106 Barriers to MIS Technology 113 Chapter 7 Effects of Changes in Technology on Future MISs 129 Summary 129 Microcomputers 129 Laser-Etched Disk Storage Systems 135 Communications Technology 139 Artificial Intelligence Techniques 145 Chapter 8 Impact of Public Policy on the Development, Adoption, and Diffusion of MIS Technology 157 Summary 157 Effect of Nonmedical Federal Policies on MISs 157 Research Support for Computers in Medicine 158 Encouragement of MIS Development 160 Health Care Reimbursement Policies 163 Future Management of MIS Technology 165 Mechanism for Managing MIS Diffusion 167 Chapter 9 Conclusions 179 State of the Art of MISs 179 Evaluating the Worth of MISs 179 Barriers to Further Development and Dissemination of MIS Technology 180 Effects of Changes in Computing and Information Systems Technology on MISs 180 Potential Impact of Public Policy on the Development and Dissemination of MIS Technology 181 Name Index 183 Subject Index 189 About the Author 195 Preface This book grew out of a background paper, “The Development and Diffusion of a Medical Technology: Medical Information Systems,” that I prepared for the Committee on Technology and Health Care of the National Academy of Sciences in the spring and summer of 1977. My original June draft was revised according to the critique of the Committee and the August draft was accepted for inclusion as Appendix E of the final report from the Committee as issued by the Academy. The National Academy of Sciences agreed to permit inclusion of this core material in the present work, for which permission I am most grateful. The request that I describe the evolution of medical information systems (MISs) as an example of one form of medical technology seemed strange to me at first. On the one hand, MISs are intriguing productions of computer science and as humane an application as one could find. Furthermore, medicine very much needs help with its information processing. These considerations made the selection of MISs by the Academy Committee an excellent example for study. On the other hand, it seemed most strange that the request that I do the writing came from Jordan Baruch and Morris Collen, who surely know more about the evolution of MISs than anyone in the country. Nonetheless, they persuaded me that they were busy with other matters, including chairing and being a member of the Committee, respectively, and that I really ought to know enough after almost twenty years in the field to do the paper without undue effort. As it turned out, I had a great deal of learning and relearning to do. The field has expanded vastly in the eleven years since I published The Computer and Medical Care. The contributors to the field of MISs were then a small and fairly clubby group, mostly physicians, who saw MISs as an essential step toward improving medical care, either via research into patients and patient records or via improvements in hospital and health care systems management. Since then the federal Medicare program and its annual amendments have appeared, health care costs have outpaced general inflation, President Carter has committed himself to working toward a universal national health insurance program, and even political liberals are looking desperately for means to cap or reduce expenditures for medical care. In this context MISs are being viewed in a new way, as a potential means to an immediate moderate saving in direct health care costs and also to get a handle on the whole process of hospital management. This book assesses the realism of these expectations. With this broader question in mind, and also a touch of amazement in his eye, Professor Amitai Etzioni urged me to complete my study of MISs and put it in the context of his projects for the Milbank Memorial Fund. In retro- spect, it seems to me that his amazement was at the extent to which we medical systems builders have not recognized how many of our problems arise from VII VIII Growth of MISs in the United States the behavior of people and social systems rather than from naughty electronic circuits. In any event, Professor Etzioni has in no way urged this or any other conclusion on me. His substantial contribution has been to support the work through a grant from the Center for Policy Research and from time to time to ask some extremely penetrating questions concerning the issues and the presentation. I am grateful for his observations and hope for the sake of medical computing that his interest in the field is a permanent one. I wish to express appreciation to my colleagues and students in the Health Care Technology Center at the University of Missouri-Columbia for many hours of conversation concerning health affairs. These have greatly broadened my perception of the field. The following individuals shared relevant views with me: Jay Goldman, Samuel J. Dwyer, Paul Blackwell, Stuart Wesbury, Daryl Hobbs, Derek Gill, James Hedlund, Patricia Mullins, Gordon Sharp, Donald Brenner, A.E. Rikli, John Simpkins, Michael Leonard, and Larry Kings- land. Most helpful library assistance was provided by Beatrice Engley, Arjun Reddy, Bea Judd, and Don Foster, and editorial assistance by Normand DuBeau. For mechanical preparation of the manuscript and for a host of good offices I am grateful to Jean Gorges. Outside our group, the following friends and associates shared information and advice with me: Bruce Waxman, Octo Barnett, Homer Schmitz, Lawrence Weed, Richard DuBois, Marion Ball, Judith Wagner, Polly Ehrenhaft, Bernard Glueck, Reed Gardner, and William Spencer. The Growth of Medical Information Systems in the United States 1 Introduction Importance of Examining MIS Technology The term medical information system or MIS refers to a set of formal arrange- ments by which the facts concerning the health or health care of patients are stored and processed in computers. A great variety of arrangements can be used to achieve this objective. Indeed forms of such systems in limited numbers have existed for fifteen years. While one might easily concede that such matters are of concern to computer specialists in medicine, the reader may reasonably question why they are im- portant and interesting to persons outside the field itself. The United States health care system has become subject to increasing public criticism. At the same time, computing systems are ever more ubiquitous and successfully used in nonhealth fields to increase management capability and labor productivity. It is natural to wonder why health system management problems have not also benefited from dosing with the remedy of computer systems. It is this overriding simple question which makes the case of interest. Why has medicine not been able to use computer systems to solve its informa- tion processing problems? Typical of the general interplay between science and society, a multitude of small peculiarities come to light after a short inspection. These have included the usual underestimations of technical difficulties. There has also been evi- dence of the self-defeating nature of federal government policies. There has been the usual ebb and flow of technical developments and application concepts, first one ahead and then the other. These features will be described. We shall attempt, however, to focus on the more pressing main questions: What is the state of the art of MISs? Why is it not better? When and how might it get better? Those acquainted with the history of cybernetics and its evolution into modern computer science know the fundamental theoretical connection be- tween management of a field such as health services and detailed information concerning the services performed. Norbert Wiener postulated in 1950 that the connection between control of systems is inherently related to—indeed, in his sense, is identical to—information about the operation of the system.1 Exten- sions of this concept lead to recognition of feedback systems and the elaboration of what has come to be called control theory. Wiener did not write about hospitals. It is not presumptuous, however, to draw the analogy that manage- 1 2 Growth of MISs in the United States ment (control) of the U.S. health care system clearly is dependent on access to the facts of patient care (information about system operation). About a fourth of the operating cost of hospitals is expended on informa- tion handling.2 The information, however, is not ordinarily handled by a system that provides for management of even a single hospital, much less an aggregate system. One does not generally have ready access to the facts of patient care, as anyone knows who has waited hours or days for his own patient chart to be recovered from the hospital record room. Likewise, no one appears to have control of the health care system, as evidenced by grossly increasing medical costs and mounting personal dissatisfaction or doubts about quality and access to health care. The reader should be assured that the connection between computing systems and health care is far more than just a theoretical one. Hundreds of re- search projects have been dedicated to developing systems that join the two. These have had the participation of many of the most prominent medical research institutions and have had substantial federal funding. Millions of dollars have been invested in seeking successful implementation of MISs. Progress reports have been almost universally encouraging, but the problem somehow remains. Federal funds have been dispensed with scientific detachment and fairness, and never a hint of impropriety. Yet we will see that information systems developers have found inappropriate funding to have been one of their major obstacles. Development of computer technology in general has been rapid and regular. The pattern of application to medicine and diffusion into the health care system has been fitful and incomplete. The diffusion has not been managed and certain- ly there has been no historian. Yet even so, the trail is well marked. In retro- spect one can see the obstacles, the interplay of social forces and technical ad- vances, and the role of the federal government. Based on these retrospective analyses, one can understand the many problems and recommend corrective policies. Limitations of This Study A Domestic Study This book deals with the development of MISs in the United States. Such pro- vincialism is from the strictly scientific viewpoint inexcusable. That is to say, one does not care in what nation a phenomenon of physical or biologic science has been demonstrated. The universal nature of such events makes the science and the literature inherently international. This is also true in the field of MISs to the extent that one deals with questions of what is feasible or what can be expressed algorithmically. On the other hand, to the extent that success in Introduction 3 developing and implementing the MIS technology may be determined by socio- logical, political, and managerial factors, the relevant environment for study will be limited to one’s own country. The reader will doubtless guess that limiting our view to the United States still leaves ample scientific literature and applica- tion reports for study. Over 80 percent of such studies are published by U.S. researchers. In the period 1955-1968, more than 90 percent of these reports were from domestic studies. Furthermore, the preponderance of U.S. reports accurately reflects the extent to which U.S. software and hardware systems have dominated worldwide developments in MIS technology in the past. Even though this major lead is fast being lost, it is reasonable for us to examine the domestic experience as a guide to analyzing the strengths and weaknesses of our own public policy with respect to MIS technology. Further- more, it is reasonable for us to take this experience as a mirror of the present state of the art. It is definitely not reasonable to take the domestic experience as a guide to future advances in the technology itself, nor even as a mirror of all rational management strategies. There is much to be admired in certain public policies abroad relating to MIS developments. Yet study of foreign political arrangements is not directly relevant to our choosing a domestic strat- egy for MIS development. It is another piece of work entirely to compare and contrast international uses of computers in MISs, with their widely differing goals and their widely differing political and economic environments. In any event, I have not attempted in this book to analyze the worldwide experience, only the U.S. experience. For the reader who wishes to explore the international literature in the MIS field, a brief guide is presented in a later section. State of Knowledge Some limits to our consideration stem from an incomplete state of scientific understanding. Advances in these areas would, of course, completely re-define our problems with practical MIS applications. In the sphere of general science, we do not fully understand the following: The formation and utilization of language The formation of thought and its relationship to language How to measure semantic content The limits of physical media for storage of information In the medical sphere, we do not fully understand the following: The logic of diagnosis and treatment 4 Growth of MISs in the United States Much basic physiology (such as brain function or memory) How to organize and manage health care delivery in the context of the democratic process Acknowledging these areas of ignorance brings to our attention that MISs tend to be an automation of things as they are now. Some differ a bit, but in most cases, MISs and MIS builders are not asking why things are done—they are automating the information processing, storage, and retrieval aspects of current health care practices. They ask to be judged by the values of the day. Yet at the same time, advances in understanding fundamental processes in science or medicine determine the significance of the MISs accomplishments. General Plan for Examining the Development of MISs in the United States Chapter 2 elaborates on the definition of MISs. There are two major approaches to building the desired complete system: the holistic and the incremental. Chapter 2 offers a general analytic framework for comparing the existing sys- tems and those to come. The significance of this analytic method is not to arm the expert with another tool; it is to make clear the scope and dimensions of the domain in which MISs can be employed. Because each MIS tends to offer itself as “the answer,” it is essential to know the magnitude and specification of the problem domain before analyzing the state of the art. Chapter 3 depicts the current state of the art of MISs. I also list some desirable functions of MISs that have thus far been infeasible. Chapter 4 provides detailed descriptions of some MISs, both from major academic institutions and the commercial sector. There are some striking dif- ferences in the intellectual goals of the university and the commercial systems. The university systems at their best have more ambitious goals than the com- mercial systems, and some are distinguished too by fitting specialized institu- tional needs. Nonetheless, good commercial systems exist, span a substantial range of functions, and are offered under a variety of conditions and arrange- ments. Chapter 5 discusses the problem of evaluating academic and commercial MISs. Evaluative tools such as marketplace outcome, operations research, cost-effectiveness and cost-benefit analyses, technology assessments, and scien- tific impact studies are described. The results of applying these methods to specific systems are presented, as well as some of the difficulties encountered when these techniques are used. The chapter is a long one, reflecting the multiple and still generally incomplete approaches to evaluating MISs and other medical technologies. Introduction 5 Chapter 6 analyzes the barriers to further development and dissemination of MIS technology. We examine the published reports of system builders and those who studied them, recapitulating the obstacles they encountered and those they foresee. Barriers to diffusion have been sociological and behavioral as well as technological. Management problems are also described. A number of per- sonal opinions are offered in an attempt to generalize over the range of in- dividual views. Technical barriers specific to the medical environment are identi- fied. In addition, we consider barriers germane to any newly evolving tech- nology. These barriers are analyzed in a framework suggested by classical studies of diffusion of innovative practices in agriculture. Chapter 7 describes the effect of changes in the technology of computing and information systems per se on the field of MISs. Special attention is paid to the effects of microprocessors, video disk storage technology, computer com- munications networks, and artificial intelligence techniques. Some of the imme- diate effects will enhance MIS developments. The broader, more remote, effects, especially on the entire health care system, are subjected to speculation. Chapter 8 describes federal policy with respect to this field, both its state- ments of purpose and the working of the federal programs. Effects of formal science and biomedical research programs and also the programs dealing with patient care delivery and reimbursement are discussed. The first effects are direct. The indirect effects of health delivery system regulation are somewhat more conjectural. Chapter 9 presents conclusions concerning the nature and state of the art of MISs, their potential worth, and possible approaches to improved manage- ment of their full development and utilization. Each chapter includes a summary of the aspect of the problem on which it presents information. Chapter 9 draws together the conclusions from each of the preceding sections. For the hurried reader, it would be best to read the chapter summaries and then the last chapter with the overall conclusions. In addition, I offer summary conclusions and recommendations at the end of this chapter. International Literature of MIS Technology From the early period (1955-1968), publications from England and Sweden are worth scanning. Publications of the Nuffield Provincial Hospitals Trust are important, and SPRI (the Swedish Planning and Rationalization Institute of the Health and Social Services) in Stockholm published Swedish and some English versions of a series of highly relevant reports. In 1966 SJURA (Council for Rationalization of Hospital Operations) in Stockholm published the proceedings of a NATO (North American Treaty Organization) meeting on this subject that brought together in Elsinore an interesting sampling of U.S. and western European MIS groups. 6 Growth of MISs in the United States After 1968 one should also study MIS publications from West Germany and Japan. Much relevant work and some publications are sponsored by the Medical Information System Development Center (MEDIS/DC) in Tokyo. The proceedings of the MEDIS meetings on MISs in Osaka beginning in the early 1970s are worthwhile. A recent review volume covers a good sampling of med- ical computer applications throughout Japan.3 The International Federation for Information Processing (IFIP) contributes to this field by sponsoring meetings and publications that include MISs through TC4, their Technical Committee on Information Processing in Medicine. Inter- national meetings in Stockholm in 1974 (MEDINFO 74)4 and in Toronto in 1977 (MEDINFO 77)5 contain much relevant material on this topic. MEDINFO 80 will be held in and published in Tokyo. Another relevant IFIP meeting and proceedings were on the topic Information Systems for Patient Care, edited by Van Egmond, Robbe, and Levy, from a meeting held in Amsterdam in 1976.6 In September, 1978, the first congress was held of the New European Federa- tion for Medical Informatics in Cambridge, England. The proceedings were published.7 MISs were also the subject of an international meeting in Cape- town, South Africa in April, 1979, also sponsored by IFIP. Two journals covering international developments in MIS technology are Information Methods in Medicine/Methodik der Information in der Medizin, edited by Gustav Wagner;8 and Medical Informatics/Medecine et Informatique, edited by Francois Begon and John Anderson.9 In addition, a new publication series is planned by the World Association for Medical In- formatics (WAMI),10 and a continuing series Lecture Notes in Medical In- formatics is produced by Springer-Verlag.7 ,n’12 Summary Conclusions MISs based on computers have a potentially major contribution to make to health services research. Existing systems should be viewed as partial examples of the ultimate potential. Even so, operation of the current systems can be cost-justified. The explanation of why this technology has not diffused smooth- ly into health care systems is complicated. Part of the explanation includes federal policies that have inadvertently been in conflict with one another. Opportunities still exist to manage the development and diffusion of this tech- nology as a significant contribution to health systems problems in concert with national objectives. Summary Recommendations The major elements of a strategy for managing MIS technology development and transfer include the following: Introduction 7 1. Define federal responsibility for managing MIS technology. Choose an interagency strategy with a single lead agency. 2. Stabilize funding of this program over longer than five years, at whatever level is consonant with other national objectives. 3. Acknowledge the rightful contributions of universities, industry, and government by appropriate representation on an advisory structure. 4. Define the purpose of the management to be the successful transit of technologic developments, whether equipment-embodied or procedural, from early innovation and development through general commercial avail- ability, with the prime health care objective being the best possible balance between excellent care, access to it by all, and cost containment. 5. Report freely and openly on progress toward these goals, and of obstacles to achieving them. 6. Use federal health care reimbursement mechanisms when appropriate to encourage beneficial application of MIS technology. Identify information processing tasks that are desirable from the point of view of health care management and appropriately done by automated information systems. Make these reimbursable as individual services rather than as inclusions in institutional per diem reimbursements. Examples include institutional quality control procedures, drug information systems, diagnostic assistance functions, risk assessment, patient education, continuing medical education, and clinical competency testing. Notes 1. N. Wiener, The Human Use of Human Beings (New York: Avon Books, 1950). 2. R.A. Jydstrup and M.J. Gross, “Cost of Information Handling in Hospitals,” Health Services Research (1966), vol. 1, pp. 235-271. 3. M. Saito and T. Furukawa, comps., “Advanced Health Care Tech- nology,” Technocrat Technical Survey Books (Tokyo: Fuji Marketing Re- search Co., Ltd., 1978), pp. 1-230. 4. J. Anderson and J.M. Forsythe, eds., MEDINFO 74, Proceedings of the First World Conference on Medical Informatics, Stockholm, August 5-10, 1974. (New York: American Elsevier, 1974). 5. D.B. Shires and H.K. Wolf, eds., MEDINFO 77, Proceedings of the Second World Conference on Medical Informatics, (Amsterdam: North Hol- land, 1977). 6. J. Van Egmond, P.F. de V. Robbe, and A.H. Levy, eds., Information Systems for Patient Care (New York: North-Holland Publishing Co., 1976). 7. J. Anderson, ed., “Medical Informatics Europe 78, Proceedings, Cambridge, England,” Lecture Notes in Medical Information, vol. 1, D.A.B. 8 Growth of MISs in the United States Lindberg and P.L. Reichertz, eds. (New York: Springer-Verlag, 1978), pp. 1-822. 8. Methods of Information in Medicine/Methodik Der Information inDer Medizin. G. Wagner, ed. (West Germany: F.K. Schattauer Verlag). 9. Medical Informatics/Medicine et Informatique, J. Anderson and F. Begon, eds. (London: Taylor and Francis, Ltd., 1976.) 10. World Association for Medical Informatics, 74 Rue de la Colonie, 75013 Paris, France. 11. D. Fenna, S. Abrahamsson, S.O. Loow, and H. Peterson, “The Stock- holm County Medical Information System,” Lecture Notes in Medical Informa- tics, vol. 2, D.A.B. Lindberg and P.L. Reichertz, eds. (New York: Springer- Verlag, 1978). 12. U. Kelchammer and K. Uberla, “Long-Term Studies on Side-Effects of Contraception-State and Planning. Symposium, Munich, 1977,” Lecture Notes in Medical Informatics, vol. 3, D.A.B. Lindberg and P.L. Reichertz, eds. (New York: Springer-Verlag, 1978), pp. 1-240. 2 Definition and Means of Comparing Medical Information Systems Summary An MIS is a set of formal arrangements by which facts concerning the health or health care of individual patients are stored and processed in computers. Some believe that all elements of a system must be designed initially, and some believe that the MIS must inevitably result from an aggregation of func- tioning subsystems. All agree that the final result should be a system that im- proves patient care by improving the quality and flow of information, and the extent to which all sources of patient study are appropriately interrelated by the MIS. A small number of items of information about patients is essential for an MIS. A much larger number of constituents is optional. The state of the art of MISs may be judged by examining the extent to which systems have been demonstrated feasible along certain specific dimensions of an analytic frame- work. The dimensions are patient population, health care or institutional set- ting, medical service area, data elements collected, functions performed, use of output of the system, and financial basis of the system. Working Definition A reasonable—and noncontroversial—definition of a medical information system is: A set of formal arrangements by which facts concerning the health or health care of individual patients are stored and processed in com- puters. Implications There is an implication to most workers in the field that whatever information the MIS contains is organized relative to the person to whom it pertains. The information is not primarily about groups of persons but about individuals. Similarly, there is the strong implication that the system itself is meant to facilitate treatment or health maintenance of the individuals whose information it contains. 9 10 Growth of MISs in the United States Most people familiar with medical computing assume that a medical sys- tem contains information about patients that arises, or at least enters the system, from multiple sources. This circumstance has historically provided the major rationale for the development of such systems. Some workers, however, believe that even single-source data systems, such as those containing sets of observa- tions made in a private physician’s office, might be called MIS. Workers in the field also disagree about a number of other attributes of past and current MISs. For example, many workers stipulate that in an MIS the patient record is available to and used by a number of health professionals in addition to the person who initially entered the information. The general reader may feel that extensive discussions of definitions are irrelevant, especially if a brief working definition is generally acceptable. In the case of MISs, however, an exception may be justified because the differences in opinion concerning definitions reveal so clearly the problems the field has en- countered. The definitional elements over which workers in the field differ do not pertain to the ultimate appearance of the systems. The differences generally deal with the magnitude of the system, the range of uses to which the systems are put, and, more important, the manner in which system building should be approached. It may be helpful to recognize that “medical information system” is a term that is roughly analogous to “gasoline-propelled vehicle”, and not to the more restrictive term “automobile.” This analogy suggests an idea of the range of users, costs, sizes, and features that may be present. It does not give much appreciation for the intensity of the dispute concerning the appropriate way to build an MIS. This issue has, however, been so prominent in the history of this field that one must acknowledge it, even if only to reject the arguments. Systems Philosophies Holistic Systems versus Cumulative Systems Two major system-building philosophies have influenced the field. First is the concept that a reasonable system can only result from a holistic concept of the final product: the machine and procedures applied to the total problem area. One big system, according to this view, should be designed, built, and installed in a setting such as a hospital. The second philosophy is that the ultimate system should result from a sum of the successful subsystems that would have been directed toward parts of the overall problem of medical information processing. The parties to these disputes were not merely sophomoric debaters. Their beliefs derived from good and bad experiences with problems of receptivity of the medical settings, the sources and magnitude of funding available for the research, and good and bad experiences with large and small computers. On the accep- Definition and Means of Comparing MIS 11 tance or rejection of these definitional matters rested—and still rests—the pro- fessional success or failure of scores of individual investigators, industrial project directors, federal research offices, and sizable research groups and corporations. The debate may or may not have shed a great deal of light on the subject, but it has generated, and can still generate, substantial heat. The Holistic View. One of the strongest statements about the holistic view was made by Melville Hodge, then president of Technicon Information Systems:1 MIS is achievable only through an integrated approach and direct professional use. . .. An integrated approach implies a single inte- grated system serving the institution. It is necessarily characterized by the following: integrated patient files generalized terminals communications and terminal handling programs generalized file management programs open-ended central processing and storage architecture redundancy and recovery techniques sufficient to assure high reliability and absolute protection against data loss.. . . Hodge makes clear that even the inner details of the system should be conceptualized initially and should not evolve out of the solution of individual medical application problems. “Generalized” means application independent. That is, the sine qua non of the integrated approach is an applications independent terminal processing, file management and control program.2 There is considerable merit in this line of reasoning. The holistic approach has often seemed most natural to the industrial groups who directed attention to the MIS problem. International Business Machines Corporation (IBM) spoke of “total hospital information systems” in their initial offerings. Even their subsequent MISP (Medical Information Systems Programs) were offered as generalized packages, although it was understood by this time that substantial handtailoring would be performed by individual customers. The General Electric Company also approached the problem with an equally optimistic view, even if the systems-building philosophy was a bit less global. The Cumulative View. An equally strong argument has been made by others that MISs must be built up from a number of subsystems to be successful. They believe that the integrity and eventual maximal utility of the system 12 Growth of MISs in the United States arise from the conceptualization of the interrelationships between the parts. The individual parts or subsystems are, however, intended to be useful before a total system has been assembled, or even if it never emerges. The component systems in this view may be implemented one at a time in whatever sequence appears feasible, and are designed to be useful in varying combinations. The subsystems are presumed to be able to come and go over time, depending on their local utility. It is presumed that even multiple copies of subsystems may exist within one overall MIS. G. Octo Barnett has been a strong spokesman for this view. He advised: . . . developing a modular system is far better, for the present, than seeking to construct a so-called “total medical information system.” Although the long-term objectives of the two strategies are identical, the methods of procedure and the intermediate goals are very different, as is the prognosis for their relative utility and near-term success. To aim for a total all at once is to court several fatal dangers and to em- brace weaknesses, in our opinion. The modular approach is less ambitious in scope but more discriminat- ing and precise in net effect. It sees as its main task the identification of the functional information processing units of medical activity and the assessment of real problems existing in a day-to-day flow of medical care.3 Intermediate Views. It is tempting to dismiss both arguments by assuming that industrial views favor creation of a monolithic solution that could represent a series of big sales, whereas the medical/research/academic community natural- ly favors a personalized and fragmentary series of experimental solutions. There is some statistical validity in this generalization, but it is too simple. There are too many exceptions for us to accept it. It does real injustice to the integrity of the two approaches. One does not yet know that one philosophy is inherently better than the other. Morris Collen has been a pioneer and a major MIS system builder. His view attempts to transcend the two disputing sides and look on the essence of an MIS in pure performance terms. He states: A medical information system ... is one that utilizes electronic data processing and communications equipment to provide on-line process- ing with real time responses for patient data within one or more general medical centers, including both hospital and out-patient services.4 In his view, subcomponents of MISs may include a hospital information system, a laboratory data system, a hospital administrative information system, and presumably others. He provides further specifications for intermediate objectives and general functional requirements. Definition and Means of Comparing MIS 13 The important aspects of Collen’s definition are the emphasis on provision of communications capability within the system and the insistence that the medical setting in which the system is to be judged should include ambulatory as well as hospital patient care. This view of the field has long-term validity. Its usefulness as a definition has become limited by the changes in the tech- nology itself over time. The “on-line” specification is in some ways not as help- ful as the universal demand that the information be provided in a timely fashion with respect to the uses intended. In the present computing era most data- acquisition jobs are performed on-line as a practical matter. On the other hand, when computer terminals are actually polled, the fact that they are on-line does not guarantee that they will respond promptly. Indeed, they may be functionless for minutes at a time. What the response from such a computer terminal system might be considered to be in real time varies with the urgency of the medical sit- uation and the need and patience of the user. The requirement of on-line termi- nals is not so meaningful as it once was. In theory both the holistic and the subsystem philosophies could produce workable systems that might ultimately be indistinguishable from one another. I shall, for the purpose of our examination, reject Hodge’s argument and accept that MISs can be either holistic or the sum of the subsystems’ parts. There is yet another strong view of the definitional problem. This view essentially ignores the issue of how the systems were built and looks strictly at the scope of the undertaking. Such a view could appropriately be taken of a hospital system, a total system, or any part of a system. The view has been offered lucidly by Ronald R. Henley and Gio Wiederhold. They recently con- ducted a survey and evaluation of 206 sites of development of automated ambulatory medical record systems. These systems deal with the records of patient care in doctors’ offices, clinics, and out-patient departments. These represent important even if specialized examples of MISs. The authors found substantial variations among the systems studied, and made a statement of this dilemma along with a possible definitional solution. At the present time, no definition or classification for A AMR’s (Auto- mated Ambulatory Medical Record systems) has been established, and there is no ready means for categorizing systems.5 In their own analysis they used a classification which characterized the systems studied according to their extent and purpose. This implicit classifica- tion scheme was stated explicitly in a summary of the reports prepared by Allen Bender.6 Thus automated ambulatory medical record systems were character- ized as total, general, limited, or special systems. Total systems were intended to serve all of the information management requirements of normal operation, including most medical record functions, but excluding photographs, referral letters, and such items. General systems were intended to meet some of the same 14 Growth of MISs in the United States information management requirements at their sites. Limited systems were those which by design or otherwise, served only a single information management function. Special systems served a special purpose or purposes and were not intended to be placed in general use or for routine operation. Information Content of MISs Regardless of the philosophical biases that influence systems building, the com- puter systems analyst thinks in general terms about what information the system should contain. What data elements are being collected, and how are these elements being grouped? Essential Data Elements Listed here are the barest bones of the content of an MIS. That is, these are the minimal essential data elements that would need to be collected, disregarding for the moment considerations of purpose, usage, and setting. Patient identification: number, name, address Hospital or ambulatory care location Demographic information, including sex and age (sometimes occupation) Past hospitalizations Diagnosis/diagnoses Linkage information: This varies from one installation to another. Some- thing more than name and number is required to link new information transactions in this kind of file to the preexisting patient record. Some systems use items such as maiden name, mother’s name, check digits on patients numbers, transaction numbers, or source codes Time qualifiers So far as I know, there is no MIS in the United States that collects only these elements. The Danderyd System in Sweden, however, limits itself to not much more.7,8 Danderyd is one of the largest operating MISs in the world, even though limited in its depth. Optional Data Elements A host of other data elements could also be considered as candidates for inclu- sion. The following list presents an idea of the most frequently considered ones. Definition and Means of Comparing MIS 15 Billing information, insurance status Invariant physiological information such as blood type, leukocyte antigen type(s), and chromosome karyotype Elements of health status provided by the patient Results of measurements or observations performed on the patient Interpretive information An Analytic Framework for Comparing MISs A simple, common usage definition of a thing, (as, for instance of a gasoline powered vehicle) is not in general sufficient to permit one to compare samples of the thing. Whatever definition of MIS one accepts, even if it includes the philosophical underpinnings of the approach to system building, will be insuf- ficient to shape a comparison among different versions of MISs. The systems differ in internal construction, (just as do vehicles with two or four or six or eight cylinders). The situations under which the MISs are expected to function differ even more widely (as with road surface, terrain and parking require- ments for vehicles). The functions that MISs need to perform also differ (as between ambulances, tanks, family cars, and motor homes). To agree that vehicles carry things and have propulsion is no help whatsoever in assessing the competition between different kinds of vehicles. MISs require rigorous defini- tions, and a general conceptual framework against which one can measure in- dividual systems. The problem domain in which MISs must operate and in which the tech- nology must be assessed can be bounded by certain dimensions including all of the following: 1. Patient population 2. Health care or institutional setting 3. Medical service area (usually a hospital or ambulatory care department, division, or office) 4. Data elements collected 5. Functions being performed 6. Uses of the output of the system 7. Financial basis of the system Each of the dimensions has subdivisions or multiple values. For instance, there are many kinds of patient populations. The health care or institutional setting in which an MIS may prove useful or even essential differ greatly. The functions that might prove useful or vital are multitudinous. Who uses a given 16 Growth of MISs in the United States system, especially considering that many persons or groups can use a system si- multaneously for many purposes, will be an important determinant of the ulti- mate success or failure of the technology. Characterizing the particular health care application in which an MIS does or does not function helps considerably. Examining the possible applications emphasizes the differences between them and suggests that MISs might also differ fundamentally themselves. The data elements and the function undertaken by a MIS are interrelated. A function may not be possible unless a data item or group of data items has been collected. Conversely, collecting data elements with purpose A in mind does not assure that they will be used, either for purpose A or purpose B. Collecting data elements is a necessary but not sufficient condition for the func- tions to occur. For instance, an automated operating room log system may collect data concerning surgical procedures. Even if it has full information for all patients treated, all personnel participating, rooms, times, and diagnoses, this does not that the system will ever analyze and optimize operating room utilization. It may remain forever an automated operating room log system, offering a kind of clerical convenience. The content and the functions are virtually separate aspects of the development and assessment of MISs. Careful consideration of the dimensions shows that this interrelationship between the dimensions is not limited to data content and function. All of the dimensions are in some way interrelated, and, consequently, not independent mathematically. I make no pretense that it is clear how to assess this technology in any overall sense by numerical techniques. I assert that the potential con- tribution of MISs to progress in health care, and the assessment of the present state of the technology can only be done properly within the full problem domain characterized by the dimensions named. To underscore the substantial intellectual distance over which MISs must be measured, the following are offered as provisional specifications of the dimen- sions of the problem domain. MIS Dimensions and Values within Each Dimension Population Served. Values. Healthy patients and the worried well Patient versus panel versus incidental or survey population Acutely ill versus recuperative versus chronically ill Definition and Means of Comparing MIS 17 General population versus special population, (for example, American Indian, central city, private referral, or military) Type of health care or institutional setting. Values. Office versus institution Individual, panel, or group practice Public versus private Periodic health screening facility versus general clinic General hospital versus special hospital (such as mental health center, rehabilitation center, or dialysis center) Medical service areas. Values. Admissions office Ambulatory care facility Clinical laboratory Dietetics department Intensive care unit Mental health center Operating room Pharmacy Radiology department Data elements collected. Values. Essential data elements include: Patient identification (number, name, address) Hospital or ambulatory care location Demographic information (sex, age, occupation) 18 Growth of MISs in the United States Past hospitalizations Diagnosis/diagnoses Linkage information Time qualifiers Optional data elements include: Billing information Invariant physiological information Elements of health status provided by the patient such as complaints, history of present and past illnesses, health status as reflected by daily work and personal functions, and immunizations Results of measurements or observations performed on the patient such as height, weight, laboratory test results, electrocardiograms (EKGs), radio- logical studies, results of physical examinations including the traditional physician’s general examination and special examinations such as range of motion, proctoscopy, pelvic examination, and cystoscopy Interpretive information such as that generally provided by the physician, including problem list, provisional or working diagnoses, treatment plans, therapeutic and diagnostic orders, descriptions of surgical procedures, and progress notes Functions performed. Values. Retrieval of patient records for patient care Physician assistance functions such as differential diagnosis and therapy calculations Direct patient monitoring Retrospective analysis of quality of care Analysis for improving institutional management Analysis for fiscal control Analysis for refining medical practices or discovering new knowledge Definition and Means of Comparing MIS 19 Users of system output. Values. Patient’s physician or immediate health care team Medical consultant Research investigator Health administrators, local or remote Financial support of system costs. Values. Patient fee-for-service with individual insurance coverage Patient prepays Insurance carrier Government intermediary Social Security Administration State or municipality Federal research agency Mixed funding One may define a MIS by selecting one or more values from each of the seven dimensions. Together they define a valid MIS. Notes 1. M.H. Hodge, “Large Scale Medical Data Systems—The Integrated Ap- proach,” in Journees D’Informatique Medicate (Symposium on Medical Data Processing), (Toulouse: March 4-7, 1975), pp. 11-3, 11-4. (le Chesney, France: Iria, 1975) 2. Ibid., p. 11-4. 3. G.O. Barnett, “The Modular Hospital Information System,” in Com- puters in Biomedical Research, vol. 4, R.W. Stacy and B.D. Waxman, ed. (New York: Academic Press, 1974), pp. 245. Reprinted with permission. 20 Growth of MISs in the United States 4. Morris F. Collen, “General Requirements for a Medical Information System (MIS),” in U.S. Department of Health, Education and Welfare, Pro- ceedings of a Conference on Medical Information Systems, January 28-30, 1970, p.2. 5. R.R. Henley and G. Wiederhold, Technical Report No. 13 (1, 2) An Analysis of Automated Ambulatory Medical Record Systems, vol. 1 and vol. 2, (San Francisco, Calif.: University of California, San Francisco Medical Center, June 1975), NTIS# PB 254-235, PB 254-236. 6. Henley and Wiederhold, An Analysis of Automated Ambulatory Medical Record Systems, vol. 1, pp. 104-226. As quoted in A.E. Bender, Summary of An Analysis of Automated Ambulatory Record Systems, p. 13, Rockville, Maryland, 1976 (unpublished). 7. S. Abrahamsson, K. Larsson, “Danderyd Hospital Computer System,” Computers and Biomedical Research, vol. 3, 1970, pp. 30-46 and vol. 4, 1971, pp. 126-140. 8. D. Fenna, S. Abrahamsson, S.O. Lobw, et al.: “The Stockholm County Medical Information System,” Lecture Notes in Medical Informatics, vol. 2, D.A.B. Lindberg and P.L. Reichertz, eds. (New York: Springer-Verlag, 1978). 9. D.A.B. Lindberg. The Computer and Medical Care (Springfield, Illinois: Charles C Thomas, 1968) pp. 77-86. 3 State of the Art of MISs Summary The problem domain for MISs, that is, the conceptual space in which such sys- tems may operate, is bounded by certain natural dimensions. The state of the art is revealed by describing points or values along each dimension for which suc- cessful MISs have been demonstrated. Successful applications have proved the feasibility of the MIS concept in many circumstances. For example, the dimension “medical service area” has been explored by successful MISs in many hospital departments. These include admissions offices, ambulatory care facilities, business offices, clinical laboratories, dietetics depart- ments, EKG units; intensive care units, medical records departments, mental health centers, operating rooms, pharmacies, radiology departments, and radio- therapy units. Systems have been successfully applied in private or community hospitals, university and public institutions, solo medical practices, and group practices. A variety of funding schemes has been utilized. The entire problem domain has not yet been explored, but a large part of it has. References to successful demonstrations are provided, along with reference to reviews of the field. In addition, analysis allows one to make a list of apparently reasonable things that have so far been infeasible for MISs. This list is not based on theore- tical infeasibility, but rather on the lack of demonstrated accomplishment of these tasks. Feasibility of Specific Kinds of MISs Analytical Framework In the analytical framework presented in chapter 2, the dimensions of the problem domain indicated what kinds of things to look for in any MIS. Any MIS will attempt to provide certain functions, using certain data elements that it collects. The MIS will be operated in a particular medical service area or areas, will have a particular patient clientele or population, and will be financed according to a given scheme. Appraisal of the state of the art of these systems is achieved by stating the circumstances under which each successful system has operated with respect to each of the general dimensions. 21 22 Growth of MISs in the United States One cannot attempt to make a statement about each of the combinations of attributes that define a particular sort of MIS. No single system has ever been utilized in all, or even a large percentage, of these combinations. On the other hand, it is encouraging to realize that, taken together, systems have operated successfully in a large number of the possible circumstances. Each of these has been a successful feasibility experiment for part of the future, fully matured matrix of MISs. Data elements were discussed in chapter 2. As a further example of the use of this analytical framework, we shall select “medical service area” as the next dimension with which to array sample applications. What Ronald Henley and Gio Wiederhold term a limited system (see chap- ter 2)1’2 corresponds quite well with what Morris Collen,3 G. Octo Barnett,4 and others call subsystems or modules. Often such systems have been thought of primarily in terms of the medical service area in which they were installed. The resulting implementations are the smallest current examples of MISs still capable of performing as isolated components. Useful subsystems include pro- grams and systems relevant to any or all of the medical service areas. The list of systems set out here is classified only along the single dimension of medical service area with a brief indication of the medical tasks involved. There are other ways to classify the systems. In addition, one must recognize that each system has additional dimensions, such as the type of patient popula- tion and the kind of institutional setting. Dimension: Medical Service Area Admissions Office. Examples of systems that demonstrate the feasibility of MIS applications in the admissions office are given in notes 5-8. These demon- strate the feasibility of MIS function at the point represented by the admissions office, along the general dimension of Medical Service Areas. The following tasks can be performed: patient scheduling for hospital ad- missions or outpatient appointments and hospital census, including admissions, transfers, and discharges. Ambulatory Care Department. Examples of systems that demonstrate the feasibility of MIS applications in the ambulatory care department are given in notes 9-12. The following tasks can be performed: ambulatory patient care record keeping; multiphasic health testing; quality of care analysis; and physician assistance functions. Business Office. Examples of systems that demonstrate the feasibility of MIS applications in this area are given in notes 13-15. The following tasks can be performed: room rates and charge posting; patient billing; insurance statements; accounts receivable; aging of accounts; general ledgers; accounts payable;inven- tory; payroll; workload analysis; and productivity analysis. State of the Art of MISs 23 Clinical Laboratory. Examples of systems that demonstrate the feasibility of MIS application in this area are given in notes 16-21. The following tasks can be performed: physicians’ orders; data acquisition from instruments or technologists; report generation; quality control; and cost analysis. Dietetics Department. Examples of systems that demonstrate the feasibility of MIS applications in this area are given in notes 22-25. The following tasks can be performed: food inventory control; institutional menu standard- ization and planning; nutrient analysis; patient selective menu operation; menu item forecasting; and food service management functions. EKG Units. Examples of systems that demonstrate the feasibility of MIS applications in this area are given in notes 26-28. The following tasks can be performed: examination requests; acquisition of the physiological signals; processing and interpretation; report generation; preparation and management of physical records; and cardiac pacemaker surveillance. Intensive Care Unit. Examples of systems that demonstrate the feasibility of MIS applications in this area are given in notes 29-31. The following tasks can be performed: monitoring of physiological signals in coronary, surgical, or pulmonary intensive care units; alerting personnel to aberrant conditions; and on-line, closed-loop control of infusions and medications. Medical Records Department. Examples of systems that demonstrate the feasibility of MIS applications in this area are given in notes 32-35. The following tasks can be performed: diagnosis registry, numerically coded or otherwise, including any of the following: admission diagnosis, clinical diagnosis, interpretation of surgical pathology, cytology, or autopsy examinations, inter- pretation of radiological or EKG diagnoses. Mental Health Center. Examples of systems that demonstrate the feasibility of MIS applications in this area are given in notes 36-40. The following tasks can be performed: interpretation of psychological tests; hospital administrative and management functions; acquisition of nurses’ and physicians’ progress notes; and analysis and interpretation of hospital records with predictions of patient behavior. Operating Room. Examples of systems that demonstrate the feasibility of applications in this area are given in notes 41-43. The following tasks can be performed: creation and maintenance of an operating room log; administra- tive and medical analysis of utilization, including case mix, resident experience, duration of procedures, and blood usage; and optimization of case scheduling and scheduling of operating room use. 24 Growth of MISs in the United States Pharmacy. Examples of systems that demonstrate the feasibility of MIS ap- plications in this area are given in notes 44-48. The following tasks can be performed: acquisition and recording of prescriptions; preparation of labels; analysis of appropriateness of drug dosage with respect to patient age and usual dosage level; compatibility with other drugs; alerting to known potentially dangerous situations such as allergy or sensitivity; and preparation and main- tenance of drug profiles for individual patients. Radiology Department. Examples of systems that demonstrate the feasibility of MIS applications in this area are given in notes 49-51. The following tasks can be performed: examination requests; patient scheduling; physician assistance functions; acquisition of radiologists’ interpretations; radiation treat- ment planning; report generation; management of film library; cost analyses; and departmental management. Examples Classified with Respect to All Dimensions The analytic framework recommended for MISs presents certain problems if one attempts to present analyses of all existing and past systems. Such a pre- sentation would be overly lengthy as well as rather dull. More important, a presentation of developing systems would become quickly outdated. To illustrate the analytic method, a complete analysis of a few examples is presented. These systems are rather special, having been designed for par- ticular institutions. They are presented individually in more detail in chapter 4. Institute of Living, Hartford, Connecticut. Dimension 1: Population served Patient population Acute and chronic illnesses General population living in Northeast Dimension 2: Type of health care or institutional setting Special hospital for psychiatric care Dimension 3: Medical service areas Admission; medical record keeping; complete mental health care record Dimension 4: Data elements collected Essential data elements: All collected State of the Art of MISs 25 Optional data elements: Billing information Elements of health status provided by the patient Results of measurements or observations performed on the patient Dimension 5: Functions performed Retrieval of individual patient records for patient care Analysis for improving institutional management Analysis for refining medical practices Dimension 6: Users of system output Patient’s physician and immediate health care team Local health administrators Dimension 7: Financial support of system costs Patient fee-for-service with individual insurance coverage University of Utah, Salt Lake City, Utah. Dimension 1: Population served Patient population Acutely ill and recuperative General population Dimension 2: Type of health care or institutional setting Institution Individual and group practice setting Private and public General hospital Dimension 3: Medical service areas Admissions; clinical laboratory; surgical recovery room; extensive care units Dimension 4: Data elements collected Essential data elements: All collected 26 Growth of MISs in the United States Optional data elements: Elements of health status provided by the patient Results of measurements or observations performed on the patient Interpretive information Dimension 5: Functions performed Retrieval of patient records for patient care Direct patient monitoring Physician assistance functions Analysis for refining medical practices Dimension 6: Users of system output Patient’s physician and immediate health care team Medical consultants Research investigators Dimension 7: Financial support of system costs Patient fee-for-service with individual insurance coverage Federal research agency Harvard Community Health Plan, Cambridge, Massachusetts, COSTAR System. Dimension 1: Population served Panel population General population, in and about Cambridge, Massachusetts; more than 40,000 persons, including 3,000 low-income subscribers Population expects full range of care for acute and chronic illnesses Dimension 2: Type of health care or institutional setting Panel of physicians and participating hospitals Private care Patients are eligible for hospital as well as ambulatory care, but the COSTAR MIS is only required to store records for office and clinic care Dimension 3: Medical services areas State of the Art of MISs 27 Ambulatory care facility Dimension 4: Data elements collected Essential data elements: All collected Optional data elements: Elements of health status provided by the patient Results of measurements or observations performed on the patient Interpretive information Dimension 5: Functions performed Retrieval of individual records for patient care Retrospective analysis of quality of care Analysis for fiscal control Dimension 6: Users of system output Patient’s physician and immediate health care team Medical consultants Local health administrators Dimension 7: Financial support of system costs Mixed funding, including patient prepayment; insurance carriers; and government intermediary for Medicaid University of Vermont, Burlington, Vermont, PROMIS System. Dimension 1: Population served Special population admitted to a ward of the medical service of a university hospital Dimension 2: Type of health care or institutional setting General hospital Essentially public Essentially group or institutional practice setting Dimension 3: Medical service areas System provides information concerning all aspects of hospital care 28 Growth of MISs in the United States Dimension 4: Data elements collected Essential data elements: All collected Optional data elements: Elements of health status provided by the patient Results of measurements and observations performed on the patient Interpretive information Does not include billing information or insurance status Dimension 5: Functions performed Retrieval of individual records for patient care Physician assistance functions Retrospective analysis of quality of care Dimension 6: Users of system output Patient’s physician and immediate health care team Medical consultants Research investigators Dimension 7: Financial support of system costs Federal research agency Texas Institute for Rehabilitation and Research, Houston, Texas. Dimension 1: Population served Patient population Acute and chronic illnesses, predominantly chronic General population Dimension 2: Type of health care or institutional setting Institution Public setting Special hospital (rehabilitation center) Dimension 3: Medical service areas State of the Art of MISs 29 Medical, surgical, and rehabilitative services Dimension 4: Data elements collected Essential data elements: All collected Optional data elements: Results of measurements or observations performed upon the patient Interpretive information. Dimension 5: Functions performed Retrieval of patient records for patient care Retrospective analysis of quality of care Analysis for refining medical practices Dimension 6: Users of system output Patient’s physician and immediate health care team Medical consultant Research investigator Local health administrators Other Special Systems In addition to these five relatively well-known systems, there are many others of great interest. For example, in contrast to the mental health system within a single institution presented here, a statewide mental health information system is described elsewhere by James Hedlund,52 and a multi-state mental health information system is described by Eugene Laska.53 The CLINFO system was developed by RAND Corporation under contract to the Biotechnology Branch of the National Institutes of Health (NIH). It provides extensive patient data storage and retrieval capacity for the special case of patients whose records are the subject of research, and is typically used in clinical research centers. A his- tory of the program and analysis of usage of the systems at three test sites was provided by Howard Thompson and colleagues.54 A detailed description of the system is available from RAND.55 Many other excellent systems deal with special clinical situations or have more restricted operational goals. A number are brought to the reader’s atten- tion in the form of references cited under the section Dimension, Medical Service Areas as examples of individual parts of the whole MIS concept. 30 Growth of MISs in the United States Generally the references are to successful examples. Ideally appropriate followup analysis would present all MISs or subsystems, both successful and unsuccessful. They would be characterized according to the six secondary dimensions, and one would wish for validation of written claims by statistical sampling of sites with systematic examination by a trained, on-site review team. Producing such a presentation would involve substantial effort, but it is quite possible. Following this analysis, one would be able to estimate what percentage of the problem domain contained feasible MIS applications. More important, one would know which applications were infeasible or unrealistic based on past experience. Exploration of as much of the potential problem domain of MISs as pos- sible, would reveal many valid versions. Their validity will be determined in part by the environment in which they operate (patient population, institutional characteristics, and users, and so on). Analysis of working systems tells us which aspects of the concept are feasible. They are numerous and valuable. Tasks Thus Far Infeasible Concerning the matter of what is feasible and what is not, two points should be emphasized. First, information concerning failures or infeasibilities is rare in scientific journals. Second, infeasibility is not established by a system’s failure to thrive as a commercial venture. Most scientific experiments are not com- mercial ventures to begin with. In very few fields of science—medical or other- wise—is there such confusion of feasibility and profitability. Profitability may be merely an artifact of public policy concerning reimbursement or telecommunica- tions pricing. The former question, feasibility, is primarily a scientific one. There can be some outside common sense economic bounds put upon feasibility. We must not forget, however, that even ignoring costs, some computer applications are strictly infeasible. Automatic translation of natural language is the classical example. Attempts to accomplish this feat with computers (or any other auto- matic system) have failed consistently for years in spite of talented investigators, ample and consistent funding, national need, and a downright eagerness world- wide. It would be improper to state that language translation is impossible;yet who would argue nowadays with the conclusion that it is infeasible? Even this matter is somewhat complicated by very recent developments. A semiautomatic commercial product has been announced in the language translation field after almost fifteen years of inactivity.56 The system redefines the problem by offering to be a kind of electronic aide-memoire or dictionary page turner to the human translator. The new system may turn out to be pro- fitable, but it does not alter the statement that automatic machine translation is infeasible. Other examples can bolster the claim that it is of the greatest importance State of the Art of MISs 31 in assessing the state of the art of MISs to establish which specific systems are feasible and which are not. The most important step is systematically to explore the problem domain. I offer the following list of MIS tasks that have thus far proven infeasible as a personal view. In some cases, one could cite reports of limited progress which leave one to conclude that the task is infeasible. In some cases, no comments have been recorded formally. Infeasible but Potentially Useful Tasks Associated with Developing and Operating MISs 1. Creation of a generally usable thesaurus with a standardized medical termi- nology 2. Processing of free-text medical entries such as progress notes and consulta- tion notes 3. Processing of text diagnoses, as contrasted with numerically coded diagnoses 4. A practical means of recording the general physical examination 5. Execution of a general diagnostic logic 6. Positive patient specimen identification 7. A clinical laboratory quality control system that contributes to the MIS a measure of statistical confidence for the individual laboratory report 8. Automatic analysis of cardiac arrhythmias 9. A means of recording in the MIS the circumstances of an individual’s birth: for example, an index of labor analogous to the Apgar score 10. A practical means of documenting changes in the patient’s weight while in the hospital 11. A computer programming language or even an MIS command language suitable for use by the professional medical staff directly, as contrasted with languages to be used by computer programmers 12. An MIS computer terminal or other human-machine interface which is agreeable to use in the clinical setting 13. Voice input to computer with something approaching a general vocabulary 14. Acceptable voice output from computer with a useful medical vocabulary 15. MIS capability to recognize a drug interaction involving more than two drugs 16. Automated certification of clinical medical competence of the physician or medical attendant based on performance, examination, or the MIS record of medical care given 17. Automation of even simple treatment plans 18. Automatic clustering of symptoms so as to recognize new clusters or disease states 32 Growth of MISs in the United States 19. Discovery of new medical principles or knowledge that can reasonably be attributed to MIS usage. One last caveat concerning infeasibility should be stated. There is not suf- ficient theoretical foundation for medical information science to permit one to assert that any or all of the tasks just listed are impossible. Indeed, a casual inspection of the list induces a kind of reflex irritation. Surely, one thinks, someone could do certain of these tasks! Perhaps this will prove to be true. For the moment, however sensible, desirable, and reasonable such tasks appear to be at first glance, they have not been shown to be feasible. If one wishes to see one or more of these tasks proven feasible, one would do well to consider such an effort as a research project. It would not be wise to consider that im- plementation of a hitherto infeasible task would make a sensible part of a com- mercial or a routine MIS installation. Notes 1. R.R. Henley and G. Wiederhold, Technical Report No. 13 (1), An Analysis of Automated Ambulatory Medical Record Systems vol. 1: Findings (San Francisco, Calif.: University of California, San Francisco Medical Center, June 1975), NTIS #PB 254-235. 2. A.E. Bender, “Draft Summary Report of an Analysis of Automated Ambulatory Medical Record Systems, University of California, June, 1975,” Rockville, Md., June 1976. 3. M.F. Collen, “Reasons for Failure and Factors Making for Success,” paper prepared for the Symposium on the Development of Hospital Computing Systems-Toulouse, June 28-July 2, 1971. WHO Regional Office for Europe, Document EURO 4304/10. Reprinted with permission. 4. G.O. Barnett, “Massachusetts General Hospital Computer System (Boston),” in Hospital Computer Systems, M.F. Collen, ed. (New York: Wiley and Sons, 1974), pp. 517-545. 5. L.W. Cronkhite, Jr., “Computer Brings Order to Clinic Scheduling System,” Hospitals, vol. 43,1969, pp. 55-57. 6. M.R. Heard and J.C. Thomas, “A Hospital Information System. Its Impact on Costs, Personnel and Patients in One Department,” Cost Contain- ment, Caps, and Consumerism within the Health Care Delivery System. Pro- ceedings of the Annual Joint Systems Conference 2 (Chicago, 111.: American Hospital Association, 1978). 7. A.G. Jessiman and K. Erat, “Automated Appointment System to Facilitate Medical Care Management,” Medical Care, vol. 8 (3), 1970, pp. 234- 246. 8. C.T. Wood and A. Lamontagne, “Computer Assists Advance Bed Book- ings,” Hospitals, vol. 43 (5), March 1,1969, pp. 67-69. State of the Art of MISs 33 9. G.O. Barnett, Computer-Stored Ambulatory Record (COSTAR) (Cam- bridge, Mass.: Laboratory of Computer Sciences, Massachusetts General Hos- pital, 1976). 10. M.F. Collen, L.S. Davis, and E.E. Van Brunt, “The Computer Medical Record in Health Screening,” Methods of Information in Medicine, vol. 10, 1971, pp. 138-142. 11. C.J. McDonald, “Protocol-Based Computer Reminders. The Quality of Care and the Non-Perfectability of Man,” New England Journal of Medicine, vol. 295, December 9, 1976, pp. 1351-1355. 12. W.E. Hammond, W.W. Stead, S.J. Feagin, et al.: “Data Base Manage- ment System for Ambulatory Care,” in Proceedings of the First Annual Sym- posium on Computer Applications in Medical Care, Washington, D.C., October 3-5, 1977, H. Orthner, ed. (New York: I.E.E.E., 1977), pp. 173-187. 13. T.T. Thibodaux, W.S. Russell, G.A. Gigliotti, et al: “Computer-Based Information System,”Hospitals, vol. 47, March 16,1973, pp. 51-56. 14. P.O. Allen, “Management Reporting System: Intermountain Health- care, Inc.,” Management Information Systems: A Collection of Case Studies (Chicago, 111.: American Hospital Association, 1978), pp. 31-40. 15. M.H. Hodge, Medical Information Systems: A Resource for Hospitals (Germantown, Md.: Aspen Systems Corp., 1977). 16. J.K. Aikawa. “The Cost-Effectiveness of the C.U. Computerized Clinic- al Laboratory System,” Biomedical Science Instrumentation, vol. 10, 1974, pp. 89-92. 17. D.A.B. Lindberg, “Collection, Evaluation and Transmission of Hos- pital Laboratory Data,” Proceedings of the 7th IBM Medical Symposium (White Plains, N.Y.: IBM, 1965). 18. W.F. Hamilton and S. Raymond. “ECLIPS: An Extended Clinical Laboratory Information Processing System,” Computers in Biology and Medicine vol. 3,1973, pp. 3-12. 19. G.P. Hicks, M.A. Evenson, M.M. Gieschen, et al: “On-Line Data Ac- quisition in the Clinical Laboratory,” in Computers in Biomedical Research III, R.W. Stacy and B.D. Waxman, eds. (New York: Academic Press, 1969), pp. 15-53. 20. M.N. Spraberry, P. Fretz, T. Gascho, et al: “A Computerized Data Management System for the Clinical Laboratory,” American Journal of Medical Technology, vol. 42, June 1976, pp. 33-40. 21. Mechanization, Automation and Increased Effectiveness of the Clinical Laboratory T.D. Kinney and R.S. Melville, eds. (Bethesda, Md.: U.S. Depart- ment of HEW, 1976). 22. J.T. Andrews and B. Tuthill, “Computer-Based Management of Dietary Departments,” Hospitals, vol. 42, July 16,1968, pp. 177-123. 23. J.L. Balinfy, “Menu Planning by Computer,” Communications of the ACM, vol. 7 (1), April 1964, pp. 255-259. 34 Growth of MISs in the United States 24. Computer-Assisted Food Management Systems A. Moore and B. Tuthill, eds., (Columbia, Mo.: Technical Education Services, University of Missouri, 1971). 25. A.M. Messersmith, A.M. Moore, and L.W. Hoover, “A Multi-Echelon Menu Item Forecasting System for Hospitals,” Journal of the American Diete- tic Association, vol. 72, (5), May 1978. 26. H.V. Pipberger, “Evaluating Computer ECG Programs,” Circulation, vol. 50, December 1974, pp. 1287-1288. 27. A.D. Little, \nc., Automated Electrocardiography in the United States (Cambridge, Ma.: Arthur D. Little, Inc., 1976). 28. M.E. Grant and J.S. Hanson, “A Totally Computerized Cardiac Pace- maker Surveillance System,” Computers in Cardiology. Proceedings of Con- ference October 7-9, 1976, St. Louis, Missouri, (Long Beach, Ca: I.E.E.E. 1976) pp. 13-17. 29. L.C. Sheppard and J.W. Kirklin, “Cardiac Surgical Intensive Care Com- puter System,” in Computers in Life Science Research, W.Siler and D.A.B. Lindberg eds., (Bethesda, Md.: Plenum Press, 1974) pp. 37-41. 30. J.J. Osborn, J.O. Beaumont, J.C.A. Raison, et al: “Computation for Quantitative On-Line Measurements In an Intensive Care Ward,” in Computers in Biomedical Research III, R.W. Stacy and B.D. Waxman eds., (New York: Academic Press, 1969), pp. 207-237. 31. F. Wiener and M.H. Weil, “Cardiovascular Monitoring in the Medical Intensive Care Unit,” Medical Instrumentation vol. 11, (5) September-October 1977, pp. 268-273. 32. PROMIS Laboratory, Automation of a Problem-Oriented Medical Record, (Burlington, Vt.: University of Vermont, 1976). 33. D.A.B. Lindberg, G.R. Reese, and C. Buck, “Computer Generated Hospital Diagnosis File,” Missouri Medicine, vol. 61, October 1964, pp. 851 - 852,858. 34. R.A. Cote, G.E. Gantner, R.S. Beckett, et al: “Systematized Nomen- clature of Medicine,” Pathologist, July 1977. 35. T.L. Lewis and G.C. Macks, “Adaptation of a General Hospital Com- puterized Medical Information System to the Research Environment,” in Pro- ceedings, First Annual Symposium on Computer Application in Medical Care, Washington, D.C. October 3-5, 1977, H. Orthner ed., (New York: I.E.E.E., 1977) pp. 111-115. 36. B.C. Glueck, in Progress in Mental Health Information Systems, J.L. Crawford ed., (Cambridge, Ma.: Ballinger Publishing Co., 1974), pp. 303-316. 37. I.W. Sletten, H. Altman, and G.A. Ulett, “Routine Diagnosis by Com- puter,” American Journal of Psychiatry, vol. 127, 1971, pp. 1147-1152. 38. Safeguarding Psychiatric Privacy E.M. Laska and R. Bank, eds., (New York: John Wiley and Sons, 1975). 39. J. Hedlund, I.W. Sletten, R.C. Evenson, et al: “Automated Psychiatric State of the Art of MISs 35 Information Systems, A Critical Review of Missouri’s Standard System of Psychiatry (SSOP),” Journal of Operational Psychiatry, vol. 8 (1), 1977, pp. 5-26. 40. J. Hedlund and C.V. Hickman, “Computers in Mental Health: A Na- tional Survey,” Journal of Mental Health Administration, vol. 6 (1) summer 1979, pp. 30-52. 41. D.A.B. Lindberg, The Computer and Medical Care (Springfield, HI.: Charles C Thomas, 1968), pp. 77-88. 42. R.J. Hanson and D.P. L’heureux, “CORMIS, a Computerized Manage- ment Information System for the Operating Room,” Cost Containment, Caps and Consumerism within the Health Care Delivery System 2 Proceedings of the Annual Joint Systems Conference (Chicago, HI.: American Hospital Associa- tion, 1978), pp. 411-435. 43. R. Bendix, V. Bhargava, W. Griffith, et al: “Computer Scheduling for the OR ''Modern Health Care, June 1976, pp. 16M-160. 44. R.K. Hulse, S.J. Clark, J.C. Jackson, et al: “Computerized Medication Monitoring System,” American Journal of Hospital Pharmacy, vol. 33, October 1976, pp. 1061-1064. 45. M.L. Braunstein and J.D. James, “A Computer-Based System for Screening Outpatient Drug Utilization,” Journal of the American Pharma- ceutical Association, vol. 16, February 1976, pp. 82-85. 46. S.N. Cohen, M.F. Armstrong, R.L. Briggs, et al: “A Computer-Based System for the Study and Control of Drug Interactions in Hospitalized Pa- tients,” in Drug Interactions, P.L. Morselli, S. Garaltini, and S.N. Cohen eds., (New York: Raven Press, 1974), pp. 363-374. 47. S. Garten, C.E. Mengel, W.B. Stewart, et al: “A Computer-Based Drug Information System,” Missouri Medicine, April 1974, pp. 183-186. 48. J. Nazzaro, Computerized Pharmacy System: Naval Regional Medical Center (Charlestown, S.C.: Naval Regional Medical Center, 1978). 49. G.S. Lodwick, R.J. Tully, C.R. Markivee, et al: “Missouri Automated Radiology System: A Dynamic, Interactive, Diagnostic and Management System for Radiant Images,” Journal of Medical Systems, vol. 1, 1977, pp. 237-250. 50. V. Smith and S. Clampitt, “A Comprehensive System for Radiotherapy Treatment Planning, Using the PDP-12 Computer,” Radiology, vol. 109, October 1973, pp. 187-192. 51. Proceedings. Fifth Conference on Computer Applications in Radiology, Albuquerque, N.M., June 8-12, 7977(Chicago, 111.: American College of Radiol- ogy, 1977). 52. J. Hedlund, I.W. Sletten, R. Evenson, et al., “Automated Psychiatric Information Systems, a Critical Review of Missouri’s Standard System of Psychiatry (SSOP),” Journal of Operational Psychiatry vol. 8 (1), 1977, pp. 5-26. 53. Safeguarding Psychiatric Privacy, E. Laska and R. Bank, eds. (New 36 Growth of MISs in the United States York: John Wiley and Sons, 1975). 54. H.K. Thompson, W.R. Baker, T.G. Christopher, et al: “CLINFO, A Research Data Management and Analysis System Acceptable to Clinical Users,” Proceedings of the 1st Annual Symposium on Computer Applications in Medical Care, October 3-5, 1977, Washington, D.C., F.H. Orthner, ed. (Long Beach, California: I.E.E.E., 1977), pp. 140-142. 55. G.F. Groner, N.A. Palley, W.L. Sibley, et al: CLINFO User’s Guide: Release Three (Santa Monica, California: Rand Corporation, September 1977). R-1543-3-NIH. 56. R.A. Shaffer, “California Firm to Unveil a Computer that Processes Words for Translators,” Wall Street Journal (New York: Wall Street Journal, October 24,1978). 4 Description of MISs Summary Systems built for particular medical institutions represent examples of the most advanced aspirations for MISs. In addition to relatively complete ranges of ser- vices, these systems contain features that yield non-monetary benefits by im- proving the quality of medical care and patient management and creating important research data bases. Fine systems are now available commercially. They are offered by compet- ing vendors in a substantial range of comprehensiveness. The sophistication of these systems in the medical sphere is distinctly below that of the best of the university-based systems. On the other hand, the range of hardware available is quite good, and the hardware and services are reliable. Commercial systems can be had under a variety of arrangements (on-site as well as off-site computers, evolutionary as well as holistic installation strategies, systems with direct physi- cian interaction, as well as systems with data intermediaries). A number of MISs, both commercial and re search-based, have come and gone. Often the losses were not attributable to technical inadequacies as much as inadequate management skills within the institutions and inconstancy and in- appropriateness of federal research policy. Systems Designed for a Particular Institution MISs have been used successfully in multiple, simultaneous applications by many kinds of health care professionals for many different purposes. By con- sidering a few of these, one can achieve a gross inspection of the considerable extent to which the total problem domain has been explored. The specific instances reviewed here represent an arbitrary selection. Institute of Living, Hartford, Connecticut Overall Description. This MIS serves an entire 400-bed private psychiatric institution. The system grew out of seventeen years of experience with com- puter-based information systems under the direction of Bernard Glueck. The institution has been a pioneer in such developments. The first real-time 37 38 Growth of MISs in the United States computerized psychiatric information system became operational in this institu- tion early in 1965.1 The primary purpose of the original as well as the current system is to facilitate patient care. The newer system, however, also attends to institutional management matters such as personnel files and patient billing. The Institute now utilizes two Digital Equipment Corporation (DEC) PDP-15 computers; the code was written in the MUMPS-15® language. MUMPS® is a Registered U.S. Trademark of the Massachusetts General Hos- pital. Detailed Description. The system provides an integrated patient record that includes essential data elements, plus results of psychological test instruments, nurses’ progress notes, a record of the general physical examination, medica- tions, and diagnoses. In the case of psychological tests such as the Minnesota Personality Inventory and the Minnesota Hartford Personality Array, the MIS itself scores as well as stores the tests results. Users of the system are techni- cians, nurses, physicians, administrators, and researchers. Special Features. This institution continues its early emphasis on nursing notes as an integral part of the automated information system. Graphical displays on computer terminals of all nursing factor scores are updated daily. In addi- tion, statistical analytical programs determine when significant changes occur in the scores. Information obtained from these programs and others are used in experimental programs to predict disruptive patient behavior, the potential for leaving the hospital without permission, and responses to various therapeutic regimens. A second special focus of the MIS has been the ability to support searches of the files for research and institutional management purposes. The files are arranged to give priority to patient care inquiries, but categorical searches are also supported across all admissions according to patient characteristics, test scores, therapy, and diagnoses. As a result, the MIS is also used to select specified and matched populations for testing scientific hypotheses. System costs were reported in 1974 to be $2.60 per patient per day.2 University of Utah, Salt Lake City, Clinical Computer Applications Overall Description. Many subsystems have been created over more than ten years of effort within the Department of Biophysics and Bioengineering of the University of Utah, under the direction of Homer Warner, Reed Gardner, and their associates, and implemented within the Latter Day Saints (LDS) Hospital.3 The systems generally operate on Control Data Corporation (CDC) computers under the Med Lab operating system. Currently, the central file management Description of MISs 39 and medical decision-making system operates on a CDC 3300 computer, with much of the data gathering and preprocessing performed on minicomputers in the laboratories and intensive care units.4 Detailed Description. Application areas and functions include a computerized integrated patient record with input from more than 100 “ports” in the hospi- tal.5 The data content includes demographic admitting data; diagnoses; screening data such as EKGs, pulmonary function tests, history, and vital signs; cardiac catheterization and blood gas measurements; clinical laboratory results; as well as the results of physiological monitoring and treatment records in the intensive care units.6 The basic principle of the physiological monitoring system is to measure arterial blood pressure by an in-dwelling arterial catheter and to com- pute cardiac stroke volume as a function of the pressure and the shape of the blood pressure-pulse wave. Extensions and elaborations to the system take into account the factors of treatment and disease states that alter the basic relationship between cardiac function and arterial wave form. The calculations are recalibrated against dye dilution curves as needed. At a minimum, the system provides an automatic source of vital signs measurements to the staff. The frequency of sampling is controlled by the computer programs.7 Output of the subsystems is used primarily by physicians, nurses, and pharmacists for direct patient care. Output of the integrated record system is also used for quality of care monitoring by hospital committees. Functions include many physician assistance features related to diagnosis and interpretation of patient measurements. Special Features. The HELP system is an advanced attempt to formalize the medical logic that interprets and integrates the numerous data elements incorporated within the automated patient record system. Historically the University of Utah group has always made special efforts to bring together and interrelate within their computer systems data arriving from various parts of the institution. The latest version of the system goes even further by in- corporating Data General minicomputers for special functions within the larger computer network at LDS and University of Utah hospitals. Creation of the entire system, as stated by Warner, was motivated by the belief that computers in medicine would make their most useful contribution by “uniting a knowledge base with a patient care data base to provide expert assistance to the physician in decision-making.”8 HELP is the computer programming system which represents knowledge concerning the meaning and interrelationships of the observations. Currently over 1,400 HELP sectors are operational at the LDS hospital. These deal with a host of circumstances, including interpretation of laboratory data, electro- cardiography, and physiological monitoring results, as well as suggestions for quality of care, clinical trials, adverse drug reactions, and other matters. Using 40 Growth of MISs in the United States the knowledge system, more than 60 acute care protocols (called alerts) are manifested by display of warning messages on printers and terminals and by visits and calls from laboratory and pharmacy personnel to the wards. The process is aimed at detecting and preventing life-threatening physiological crises in seriously ill patients. The current system performs well and has been accepted by clinical users.9,10 Because of the diversity of inputs to the system, and especially because of the clinical laboratory inputs, the system can serve all hospital patients to some extent. A measure of this general utility was given by Gardner et al.11 who reported on performance of the alerting system. In 1977 they experienced an average of fifty-four alerts per day in their 550-bed hospital. Of these, thirty (56 percent) were for patients in the general care sections of the hospital. The remaining twenty-four (44 percent) were for patients in intensive care units. The intensive care unit constitutes a special subpopulation that is extremely well served by the system. Fifty-seven patients in ten intensive care units in two hospitals are accommodated by the computerized system. The system plays an important function in providing continuity and expert advice during acute episodes. In pressing forward with these advances, Gardner and associates identify two primary problems.12 First, to achieve the most expert level of performance that medical knowledge permits, the computer system must be provided with physiological data in carefully prescribed time sequences, obtained according to strict rules. This requires tighter data acquisition procedures than is usual in the hospital setting. Second, they point to the current limits of medical understand- ing when they identify the need for establishing the “medicalware” to make wise and prudent decisions. Massachusetts General Hospital Computer Systems, Boston, Massachusetts Overall Description. A number of operational systems and subsystems were created by the Laboratory for Computer Science under the direction of G. Octo Barnett and associates. They all operate from terminals to DEC computers and were written in the MUMPS® computer language. COSTAR is a computer-based record system for ambulatory patients served by the prepaid Harvard Community Health Plan (HCHP).13 It has been operating since 1969,14,15 and provides the only record for about 50,000 patients.16 Users include technicians, nurses, physicians, and medical plan managers. Within the Massachusetts General Hospital other systems handle data Description of MISs 41 acquisition and recording for clinical chemistry, bacteriology, electroencepha- lography, hematology, portions of anatomic pathology, the blood bank, and radiology. Subsystems provide physician assistance functions in an anticoagula- tion clinic and in the pulmonary function testing laboratory. Others assist under- graduate medical education through automated testing and patient simulation. Detailed Description. COSTAR is based on an arrangement of batch data entry from paper data collection instruments (encounter forms), batch printing of selected portions of the patient records prior to scheduled visits, and also on- line, real-time, inquiry capability. Cathode ray tube (CRT) computer terminals are available to health care providers for retrieval of whatever further patient records are required for immediate patient care. In addition, there are appro- priate arrangements for administrative use, laboratory data input, and other functions. Barnett stresses that, while the system is a sophisticated information processor, its significance is not primarily as a “technological tour de force." Rather, it is a medical record system in which the human aspects are equal in importance to the computer elements.17 This approach has many secondary benefits. One is the ready linkage of information concerning a patient (or a disease) entered by many different providers. Structures, precoded encounter forms, and standardized vocabu- laries facilitate a team effort, even though free-text entry (linked to a coded item) is also accommodated. COSTAR permits implementation of a variety of quality control efforts directed to patient management. For example, Barnett’s group has reported successful use of the computer system to detect and induce correction of unacceptable management of streptococcal pharyngitis and a sep- arate corrective program directed toward proper management of patients with newly discovered hypertension. Such a system can adapt to the changing skills and needs of the medical users. The designers stress that practicing medicine with such an information system means a significant change from the patterns formed with manual sys- tems. Yet acceptance of the system has been good. The most obvious dangers in this system are erroneous transcription of information entered on the data sheets (estimated to occur on less than 1 per- cent of the data items), and constraint of the physicians by the fairly rigid forms. (The safety valve is the use of write-in free-form comments by the clini- cians.) Cost estimates by Barnett in 1976 showed COSTAR costs to be a total of $12.50 per member per year, as compared with an estimated cost of $14.00 for a similar manual system. These figures were calculated on the basis of a total 42 Growth of MISs in the United States enrollment of 40,000 persons; HCHP now has 60,000 members. Two aspects of the cost considerations should be noted, since the observations resemble those made for other MIS implementations. First, the estimated savings are a modest percentage (18 percent). Second, the estimates ignored the nonmonetary benefits that are a special virtue of MISs; namely, the rapid availability of medi- cal information and the use of the computer in quality assurance. Special Features. The most outstanding special feature of the programs them- selves is their ability to share and modify files using the MUMPS® language. The language is meant to provide flexibility and facilitate rapid system develop- ment and ease of program modification. Evidence of Dissemination. The COSTAR system was created specifically for use by HCHP, although it was built around general computer science practices. A subsequent system, called COSTAR V, has been built for use by similar centers. COSTAR V came into being because of the success of the original COSTAR system and the apparent need for such ambulatory patient record systems elsewhere. The COSTAR V system was created by the efforts of Bruce Waxman and collaborators in the National Center for Health Services Research, Barnett and collaborators at Harvard, and individuals at DEC and George Wash- ington University.18 COSTAR V is to be installed in about 20 sites according to the best infor- mation available to Waxman,19 as well as other sites for which the system is being provided by DEC. The system is designed so that customization for each location can be accomplished quickly by nonprogramming personnel. The authors report that only a few hours are required to establish a working dem- onstration version of COSTAR V. They caution, however, that considerably more effort is required to establish a fundamental system capable of ongoing medi- cal and financial record support to the clinical facility. Additional effort is needed to modify the program directories that align the system with the term- inology and procedures used in the clinical site. This generalized version of COSTAR was implemented in the standard MUMPS® language on DEC PDP-11 computers and runs under DECs standard MUMPS® operating system.20 The minimal equipment configuration costs about $75,000 and has 48K words of central memory, 28 million bytes of mass stor- age, one magnetic tape drive, five CRTs and a low-speed printer. This combina- tion is said to satisfy the needs of six physicians and a patient population of 10,000. The system has been designed to provide the following functions: medical records storage and retrieval, billing, scheduling of patients and physi- cians, registration, management reporting, data security and privacy assurance, and system backup. There are no published reports as yet concerning evaluation of COSTAR V in operation at user sites. Description of MISs 43 University of Vermont, Burlington, Vermont, Problem-Oriented Medical Information System (PROMIS) Overall Description. A series of MISs has emerged from this group since 1967. Various versions with various hardware configurations have been tested, eval- uated, redesigned, and retested. They have always been faithful to a unique unifying concept; namely, that it would be in the best interest of the patient if his or her medical record were organized according to the problems the patient has (and their management), rather than if the medical record were organized in the traditional fashion according to the hospital department or group that created the items in the record. Lawrence Weed’s concept of a problem-oriented record has been the center of substantial acclaim, messianic enthusiasm, rejection, debate, and gradual acceptance. Weed believes that only by adopting a problem orientation can the medical attendants keep focused on the patient’s needs, remember all of them including the social, and that only by this organization can the medical record be analyzed. The purposes of analysis were thought to include a reconstruction of why a particular test had been ordered or a therapy begun, that is, to understand the physician’s reason- ing. Weed stated, The objectives of the PROMIS system have not changed since the original 1967 formulation. They are to develop a system that will: 1. Facilitate good patient care (1) by making immediately available (in minutes) to the individual physician a complete, updated list of problems on any patient and (2) by providing simultan- eously, as a unit, all the data in sequence (narrative, laboratory, etc.) pertinent to any of the problems. 2. Make possible epidemiological studies and other research en- deavors in terms of problems, having all the data on any given problem immediately available. 3. Make possible a medical audit whereby the standards of care being provided for a given entity (e.g. hypertension) can be rapid- ly assessed because of the specific orientation of all the data. 4. Make possible a business audit to assess the physical, financial, and time resources that go into the solution or management of a given problem. The need for a more organized, efficient and economical approach to the management of common medical and surgical disorders may then be documented. 5. Provide problem-oriented records that can serve as the basis for clinical medical education and thereby enforce application of the principle that the goals of excellent teaching programs and excel- lent patient care are the same.21 In general, the PROMIS system is designed to store all pertinent medical 44 Growth of MISs in the United States information concerning any patient. It is not designed for a particular medical specialty setting. Furthermore, it is designed to preserve within its records of patient care the linkages necessary to relate all items to one or more explicitly stated problems of the patient in question. The system is computer-based, and has always utilized CRT terminals (generally with touch-sensitive screens) for acquiring information from patient and the health care team, and for redisplaying the information. The display of information, procedure options, or questions is presented as screen faces stored as frames within the system. Although the system can present printed versions of the record, its designers have made efforts to provide a variety of summary frame formats and special displays to encourage use of the computer and the terminal system for support of patient care, rather than to encourage reliance on paper records. During a four-year demonstration period there were no written records maintained in parallel. This and other demonstra- tions of the intact PROMIS system have been limited to a single ward unit of a hospital. At other institutions, the basic concepts of the system have been im- plemented in varying degrees.22,23 Detailed Description. The system is currently supported by minicomputers, the Varian Data Machines V76 with 256,000 16-bit words, designed to be networked to terminals by cables. Files are kept on four Century Data 215 disks. The system stores a single copy of the computer patient record which can be accessed, displayed, and amended by terminals at the patient care loca- tion, and in the radiology department, laboratory, pharmacy, and operating room. The system’s programs are based on the general software provided by the Varian Vortex system, but with substantial enhancements to accommodate the larger and more complex files demanded by the patient record application. The applications code is written in a special programming language created by the PROMIS group. The language embodies structural programming tech- niques and supports both the operation of the terminal interactions in the medical setting and the writing of the frames. All contents of the system are organized around the schema that patient management must be divided into four phases. These include acquisition of a data base (information concerning the patient); a statement of the problem list; development of a plan of action for each problem; and the keeping of progress notes on each problem. Achieving these objectives has involved some changes in the manner in which patients are cared for. The automated information system as Weed sees it is an active participant in medical management. In his terms, his system ... is not dependent upon the memories of individuals; preserves with- out ambiguity the logic of all actions; coordinates the efforts of all med- Description of MISs 45 ical persons who interact with the patient; and, provides feedback with which the performance of medical persons may be evaluated and improved, and with which the system itself, once performance within the system is adequate, can be evaluated and improved.24 Current costs of a complete PROMIS system are difficult to estimate. A gross cost for only an inpatient application has been estimated to be $5 per bed day.25 Special Features. This system provides medical coaching to the health care pro- fessionals. All medical computer systems prompt the user, and shape the data that the user enters. Usually this is a minor feature such as an insistence that dates accompany data, or that spelling be standardized. Sometimes the purpose of a system is to bolster quality control of a record service such as when labora- tory systems force the reporting of standard error terms. PROMIS actually volunteers to the user formalized medical knowledge whenever it feels he or she needs to know it. The system normally confronts the medical user with standardization of terminology, of management strate- gies, and with diagnostic criteria. For example, after the user selects “severe hypertension” as a statement of the patient problem, the PROMIS system volunteers: severe hypertension is defined as BP of 288-238/128-148 with symp- toms and signs of cardiac, renal, and/or cerebral dysfunction. /#HY/.26 By the designation “/#HY/,” the system says that it is prepared to display a citation to the published literature in which the diagnostic criteria have been validated. The system includes extensive provision for literature indexing and abstracting. One further example from the PROMIS system specifications conveys the extent to which this system is consistent with its own special aims. The following statement provides a Level 1 functional specification: A. Procedure information is forced upon users of certain classes (Ref: 2.5, TU5). Those for whom this information is not forced may choose to view this data or may choose not to. The radiology procedure data include the following: 1. test name; 2. method; 3. patient preparation; 4. cost; 46 Growth of MISs in the United States 5. target organ; 6. literature references B. The following data are necessary in ordering an IV i.e., an intra- venous infusion: 1. Specify replacement solution; 2. Specify bottle volume 3. Specify added ingredients, if any 4. Specify site (optional) 5. State with/without soluset 6. Specify total number of bottles C. A high-risk wall exists for some procedures which require that the user state that the procedure is one of the following: 1. recommended by consult; 2. ordered by consult; 3. ordered as part of a protocol. In addition the option “Decided against” is available for radiologic procedures. 7 It is beyond the scope of this work to dwell on the consequences of this systems strategy. PROMIS has its strong advocates, and its detractors as well. It seeks its ultimate justification in the production of benefits in patient out- come. Thus far it has not been possible unequivocably to prove better out- comes, according to the usual scientific requirements for control studies and matched settings. On the other hand, outcome measures in medicine are an exceedingly difficult criterion to work with.28 Reports examining process measures suggest that significant benefits flow from the PROMIS system. Texas Institute for Rehabilitation and Research Overall Description. This 81-bed special hospital has been developing and add- ing information processing and managing systems since 1957.29 William Spencer, Carlos Vallbona, and Allan Levy have continually added modules to handle patient testing and care information. First they used off-line techniques, then general time-sharing techniques, and more recently stand-alone processing techniques. Hardware has included unit record equipment and IBM 1401, IBM 1410, and IBM 360 and 370 computers and Four Phase Terminals. There has been a special focus on evaluating the technology and studying its effects on patient care. Detailed Description. Hospital management data derive from the data of in- dividual patient care. These are organized into modules that include the Description of MISs 47 customary items such as hospital census data, doctors’ orders, medication scheduling, vital signs, and clinical laboratory reports. Additional modules pro- vide for specific needs related to the rehabilitation mission of the institution. These include a grip strength test program, physical therapy, daily treatment modality report, personal independence reporting, fluid and electrolyte balance, pulmonary gas exchange, and arterial blood gas analysis. Special Features. The orientation of the investigators has been centered around the individual patient care process as opposed to the more common approach of emphasizing institutional organizational needs. The information system permits efficient scheduling of treatments. It facilitates evaluation of the extent to which the institution meets a patient’s needs and the progress of the patient toward rehabilitation. These medically advanced objectives have been achieved using comprehensive but conventional computer technology. The institution’s total approach results in moderate treatment costs, high quality care, and an increased number and complexity of patient problems seen per year. Over the years the productivity of the staff and the institution has increased and quality of care has been maintained or extended. The automated MIS is judged by the staff to be a key element in this accomplishment. Commercially Offered Systems Limitations All the systems referred to in this section operate at a level of medical sophistica- tion far below that seen in the noncommercial systems designed for particular institutions. Noncommercial systems are generally university-based, and often serve a research as well as a service function. The merit of the commercial sys- tems is that they are designed potentially at least, to be replicable in many institutions. An additional strong point of the commercial systems is that they have succeeded in implementation across a number of hospital departments. They are made up of a relatively large number of the subsystem components that are the building blocks of the ultimate mature MISs of the future. Along with the relative breadth of coverage offered by the commercial systems comes a tendency toward a shallow approach to individual medical areas. In no case do they accept narrative progress notes, the patient history, or the results of the general physical examination by the physician. Since these three types of information are central to the traditional medical record, it appears that the present-day commercial systems fall short of being total MISs. 48 Growth of MISs in the United States General Characteristics That commercial versions are available indicates that the state of the art of MISs has reached a certain maturity. It is especially encouraging to note that there are a number of vendors and approaches, and a span of competition and choice. The range of systems offered commercially is wide. The simplest are business office charge-reporting systems that have added communication capability. Other systems offer the additional ability for two-way communication so that information such as laboratory reports is carried back to the patient care areas. Still other systems provide for intercommunication between hospital or medical departments and offer some integration of patient information from the various sources. The advanced commercial systems provide desirable features such as the capability for customized sets of orders for each physician who uses the system. Some systems provide guidance with respect to hospital practices and policies for work-up and scheduling. A variety of terminals are offered, including CRTs (some black and white, some color), special keyboards and function keys, touch-sensitive panels, light pens, and a variety of printers. Often the applications programs are closely coupled with use of a particular terminal whose special features preserve the proprietary interests of the vendor. It is inadequate simply to state that communications or integrative func- tions are performed by commercial systems. There is variation as to how thoroughly and how well each function is performed. Some do quite a good job. All commercial systems offer a certain generality. That is, each must be fine-tuned to the particular institutions they serve. It is their virtue that they are meant to be transferable from one place to another, not uniquely suited to just one institution. Perhaps for this reason, perhaps because of avoiding the difficult or the cost, or because of a decision first to satisfy the well-delineated market demands—for whatever reasons—the commercial systems tend not to attempt services as sophisticated as the best of the systems built for individual institutions. As already noted, no commercial system provides for recording of the physical examination, the interrogative patient history, nor interval medical progress notes. In addition, most do not provide for permanent medi- cal record identity with continuity of the computer record throughout multiple hospital admissions and between hospital, clinic, and office. To identify these faults is not to imply that the vendors do not have the wit to create systems with better features. The importance of the faults is largely that the vendors have put together systems with the features they guess will meet and not exceed the requirements of their potential customers. For this reason the shape of commercial systems is a measure not only of the state of the art of MISs but also of the diffusion and maturation of the MIS concepts and their acceptance into practice. Description of MISs 49 One vendor executive, for example, told me with some indignation that his people knew quite well how to do a patient history program but that they could not identify a customer who wanted to pay for the use of it. Discriminating Features As noted in the previous section, commercial systems have a range of charac- teristics. The analytic framework I suggested (seven dimensions of the problem domain) seems appropriate for describing the state of the art of the MIS as a scientific form. This framework also seems to be a reasonable starting point for analysis of the commercial systems. The only problem in taking this ap- proach is that one would need to perform the analysis based on study of actual installations, not product brochures. This implies an amount of site visiting beyond the ability of one person. A team approach to such a future study will be described shortly. There are, however, alternative ways to view commercial MISs. Melville Hodge, who was president of Technicon Medical Information Systems (TMIS) during much of the development of the TMIS, classifies MISs into three levels of complexity, much as was noted in the section on General Characteristics of commercial systems.30 Such a view of system comprehensiveness is a com- posite of the analytic dimension of data element, functions, and medical ser- vices. Another way of looking at commercial offerings takes the point of view of the hospital administrator or health planner. From such a vantage point, it is important to focus on the potential problems that might be created by choosing one or another system. This consideration was well framed by a government health scientist Richard DuBois who noted what he considered . . . to be the key issues concerning the selection and implementation of an automated hospital information system. These issues include transferability of commercially available systems, distributed network versus large central processor designs, inhouse development versus commercially available systems, and what level of system compre- hensiveness should be sought by a hospital.31 He suggests that the health care planner evaluating MISs should question transferability. For example, numerous business office charge-reporting systems have been implemented easily in many different hospitals. In contrast, the more comprehensive systems have required considerable implementation effort in spite of having been commercially designed to be independent of their initial hospital implementation sites. Installation of the new Technicon system at the Clinical Center of NIH was programmed to include a substantial amount of 50 Growth of MISs in the United States implementation work.32 One year was required before the first nursing unit became operational, and three years work with this commercial system will be needed before the last of the twenty-seven nursing units is operational.33 The time required to adapt a commercial packaged system to a particular hos- pital will vary with the comprehensiveness of the system itself, the quality of the system’s design, and the complexity of the hospital environment. The Technicon system is reasonably complete (in the context of commercial develop- ments), and the Clinical Center at NIH is certainly a complex environment. In contrast, McDonnell Douglas states that their Hospital Financial Control System (which does not contain medical information concerning patients, out- side of census and billing information) has not required more than three months for installation in over 400 hospitals.34 A brisk installation timetable intermediate between these two was executed by Parkland Hospital in Dallas.35 This hospital received the Duke Hospital Information System software as a validation site before IBM subsequently marketed the system as the IBM Patient Care System. Programs were received in July, 1977. Radiology was in production by December, 1977; central services by March, 1978; pharmacy by June, 1978; and labor and delivery by September, 1978. A partial emergency room implementation of central services and phar- macy was in production in January of 1979, and EKG and preadmission orders had been implemented by February of 1979. Whether a commercial system is to be based on a computer network or a large central processor is highly relevant for any hospital administrator or health planner. Some commercial MIS systems (for instance, Technicon’s TMIS) are offered on the basis that computer services be provided by the vendor from its own remote service center. In this arrangement, proper backup, hardware redundancy, and systems discipline is the vendor’s responsibility. A hospital without its own computer center may consider this arrangement highly advan- tageous. In contrast, other systems (for instance, Medicus Corporation’s SPECTRA system) are offered on the basis that computer services are to be provided by hospital-owned computers housed within the hospital. Medicus Corporation provides software maintenance under contract. This firm will also undertake management of the computer center under a separate facility management contract. Alternatively, the hospital can undertake responsibility for operating its computers. There are advantages and disadvantages to this kind of arrange- ment and its flexibility that are independent of the state of the art of MISs. DuBois’ question of modular versus total implementation emphasizes the power of arguments concerning system-building philosophy to strike fire even after almost twenty years of discussion. There is a practical issue here, however, for a hospital or planner. Some commercial systems have concentrated exclusive- ly on business office and hospital management applications (for instance, those sold by Huff, Barrington, and Owens). Other companies offer systems aimed at Description of MISs 51 this market and additional systems aimed at more functionally medical markets (for instance, the Hospital Financial Control system of McDonnell Douglas and their Hospital Data Collection and Medical Record II systems). Still other com- panies recommend initial installation of their business office systems and offer incremental growth toward what are expected to be more complete hospital systems (for instance, Shared Medical Systems). In spite of the fear of disrup- ting hospital operation by implementing a total system, there is evidence in the history of commercial MIS systems that such an implementation is entirely feasible. Conversely, there is no evidence that a commercial hospital business office system has ever evolved into a comprehensive MIS. The probable reasons for this have nothing to do with the theory of systems or state of the art. The business officers who buy and use charge-reporting and fiscal management systems do not need the system to contain medical or patient information for business functions to be adequately done. They therefore have no motivation to extend the systems development. On the other hand, the medical staff derives no benefit related to patient care from the business systems. Consequently, they either ignore the business office system or start a separate system to con- tain medically oriented patient care data. Another difference between commercial systems seems to hinge on a funda- mental distinction in systems philosophy. This is the question of whether the system is premised on direct interaction with the clinical physician, or clerical or technical personnel acting as intermediates between the physician and the computer terminals. The reader may feel that such a question is superficial, since a terminal interaction can be executed by anyone. Furthermore, an incre- mental implementation could be imagined in which technical intermediaries were provided until physicians became accustomed to the system, or until the system had evolved to the point of requiring direct physician participation. In fact, this evolution has not occurred. Some systems were designed for direct physician use to begin with (for instance, Technicon TMIS and SPECTRA). Others provided for clerical or technical intermediaries (for instance, the various IBM MISP systems). The latter systems are still run by intermediaries. Some- times the differences between commercial systems at the user level appear a bit trivial, the advantages apparently easily copied, the deficits apparently easily remedied. Neither of these happen much in commerce. For nine years the Loyola Hospital in Chicago has had IBM information systems for receiving and processing physicians’ orders. The systems have always had clerks to transcribe the written orders to computer entries. The systems still require that each order be entered individually and that a code number for the order be looked up in a paper table or memorized by the clerk. This awkward situation could have evolved or improved, but it did not. Now a new MIS is being installed by Medicus. The new system is designed with English language order selection on display terminals and provides for physicians to establish (if they wish) a record of their own standard or customary order 52 Growth of MISs in the United States lists. No codes need be looked up or memorized, and in some cases it will be possible to enter a very small list of orders when the clinical circumstances are uncomplicated. In this hospital, physicians expressed their preference for elim- inating intermediate personnel and the inevitable errors of transcription in favor of direct physician order entry in the new system. Commercial systems do not evolve once an installation has been contracted. If a hospital wishes to pro- vide for physician entry, it must be an early decision in the process of system selection. DuBois also noted the difference between an MIS system philosophy of in- house development versus the packaged commercial system. Either one under- takes local systems development or not. This distinction is more apparent than real. The choice actually exists within the domain of commercial systems and vendor-supported software. In-house development does not mean the “do-it- yourself’ system more typical of the research system customized for an in- dividual hospital. Yet one can create a semicustomized system under the um- brella of vendor-supported overall systems code. This has been the approach of the IBM Corporation. Despite a spate of detractors, there are also happy users. Duke University Medical Center recently abandoned a vendor’s pack- aged system and undertook a semicustomized system under IBM software. The final discriminating consideration, the level of system comprehen- siveness appropriate for a given hospital, is the most important of all. One does not have a set of guidelines relating the attributes of a hospital or medi- cal practice (number of beds, types of patients, amount of surgery, size of emergency room practice, complement of specialty physicians, and house staff and so on) to an appropriate level of MIS comprehensiveness. Neither are there well-formulated rules that take hospital characteristics and specify relevant general MIS attributes, and certainly there are no guidelines or rules for narrow- ing the list of candidate systems. It would be useful to have such a guide. On the other hand, the value of such rules would derive from the integrity of the research into the analysis of actual system characteristics and implementation histories. A project has recently been initiated jointly between the American Hospital Association (AHA) and the Health Care Technology Center at the University of Missouri-Columbia that is aimed at satisfying these needs with respect to com- mercially offered systems.36 The overall plan includes coordinated activities and analyses on the parts of both institutions and involves the following steps: A national mail survey of more than 7,000 AHA member hospitals con- cerning their current use, experience with, and plans for computing systems A similar survey of the more than 250 vendors of hospital information sys- tems of subsystems More detailed information gathering from selected hospitals Description of MISs 53 On-site verification of systems at a smaller number of sites Production of a catalog of systems and sites Creation of a series of instructional institutes for demonstrating the use of the catalog as an aide to analysis of hospital needs and for MIS system selection Results from this undertaking are at least a year away. Even if successful, the long-run importance will be determined by whether or not it is possible to keep such a data base up to date. A recent vendor mail survey by another institution enumerated 158 firms in the business of purveying medical computer programming applications to hospitals and clinics.37 Most offer MIS subsystems or consulting assistance in the creation of subsystems. Nonetheless, no one has seen all these systems. Some of those not reviewed in this book may be excellent. In this case, my apologies are due both to the reader and to the vendor. The description of the commercial systems that follows is not presented as inclusive of all good and proper systems, but for the reader who desires an overview of the field with emphasis on its state of development, the systems reviewed here should present an adequate sampling. Specific Systems MISs are offered by Burroughs Corporation, Data Care, IBM, McDonnell Doug- las, National Data Communications, Inc., Shared Hospital Systems, Technicon Corporation, Medicus Medical Systems, NCR Corporation, and others. Historically it should be noted that the McDonnell Douglas Hospital Patient Care System was developed jointly with the Sisters of the Third Order of Saint Francis. Other McDonnell Douglas systems were developed by the company alone. Their Medical Record Information System MRII was purchased from California Health Data Corporation. The NDC System, originally developed by National Data Communications, Inc., was later marketed jointly with Honey- well (since it utilizes Honeywell computers), and later marketed solely by NDC. The Technicon MIS system was purchased from Lockheed Corporation and subsequently enhanced. Medicus Corporation developed its own business office system and purchased the SPECTRA Medical System from Spectra Corporation. There are currently nine installations of the Technicon system. There are four installations of the medical version of the Me Auto system, four installations of the NDC/Honeywell system, and six installations of the Medicus Corpora- tion’s SPECTRA System. Burroughs and Data Care each have about three 54 Growth of MJSs in the United States hospital installations of MISs. The same is true of the IBM old style MISP- based systems. There is a new IBM product that will be noted separately. Studying the hospital computer market in relation to the installed in- hospital computer hardware, Donald Huffmire interpreted recent survey data from 200 hospitals to suggest that 69 percent of the market was served by IBM, 16 percent by Honeywell, 12 percent by NCR, and 3 percent by Bur- roughs.38 These estimates did not differentiate among the uses to which the computers were put. Most were acknowledged to be business office functions. IBM has many hospital hardware installations. Since this company supplies overall systems software, computing equipment to users and to vendors of hospital services, and also proprietary software of old and new design, it is impossible to describe their market position in simple terms. It is probably sufficient to note that IBM has never been in the business of purveying turnkey systems to hospitals. During the 1960s IBM marketed to the medical field quite vigorously. Initially, hospital applications programs as well as systems software were distributed free to customers. Hospital programs include the early Program Application Library series (PAL packages) and the early Shared Hospital Ac- county System (SHAS). After unbundling software in connection with their consent agreement with the Department of Justice, application programs became a program product leased at a fee. These included later SHAS programs and MISP. The MISP systems provided overall systems support to medical and hos- pital customers who then created semicustomized systems under the Disk Operating System. Some are still running in one form or another, but nothing short of a countrywide survey would put accurate information concerning such matters into the public domain. After almost eight years of deemphasizing the MIS marketplace, IBM has recently returned to competition with new pro- prietary packages called the Health Care Support System (HCSS). There are so far three major installations. The system characteristics will be described shortly. The actual count on any given day of just how many commercial MIS systems are operational is subject to dispute. Installations come and go, and there is no central registry for recording new sales or failures. Companies com- peting in this field also come and go. Recently, for example, Shared Medical Systems of King of Prussia, Pennsylvania, announced its plans to offer exten- sions to its current hospital management systems so as to build an MIS incre- mentally. All vendors mentioned here offer application options that include the fol- lowing hospital areas: admissions office, medical records department, pharmacy, laboratory, radiology department, nursing stations, dietetics department, admin- istration, business office, emergency room, and outpatient departments. Three of the vendors also offer packages for heart stations, utilization review, and surgery. The extent to which these systems can support extensive and sophis- ticated functions at all of the hospital areas named is said by the vendors to be essentially unlimited. They do caution, however, that the systems must be Description of MISs 55 tailored to local hospital standards and procedures. This is a reasonable limita- tion in cases such as the standard battery of orders and treatments. Commercial companies such as Technicon, for example, take the position that they do not propose to offer standardized medical treatments as part of their systems. They insist that these items be undertaken by the hospital staffs. The Technicon system in El Camino offers option that medication orders be completely tailored to each physician. After identifying him or herself through a keyboard entry, the doctor can call up on the CRT his or her own set of commonly used orders and/or drug specifications. The system provides the physician with the capability of entering these to begin with, and of altering them at will. The SPECTRA System of Medicus Corporation also provides for individualization of physician orders. One of the consequences of the necessarily disciplined approach to data entry is enhanced completeness. Both systems can prompt the terminal users. Vendor systems differ in their data entry modalities. The Technicon system utilizes black and white CRT terminals and a light pen. The NDC VITAL system also utilizes a CRT terminal, but they include special function buttons and a badge reader. McAuto and IBM systems utilize a variety of standard terminals. The Medicus SPECTRA systems use a CRT with color display. The Technicon system and SPECTRA system are designed to encourage direct data entry by the physician; the other systems present this possibility, but do not normally oper- ate with physicians entering either data or orders. Some of the commercial installations are sizable. According to reports, the number of computer terminals employed, by company, is: Technicon Corporation, Maine Medical Center, Portland: 107 terminals Technicon Corporation, El Camino Hospital, Mountain View, California: 56 terminals Technicon Corporation, NIH Clinical Center, Washington, D.C.: 156 term- inals NDC, Deaconess Hospital, Evansville, Indiana: 83 terminals McAuto, Missouri Baptist Hospital, St. Louis, Missouri: 40 terminals IBM, Parkland Hospital, Dallas, Texas: 236 terminals.39-41 Patient charge reporting and financial systems include Burroughs BHIPS system; McAuto’s HFC, HFC/BASIC, and HPC systems; Tymshare’s Financial System; Huff, Barrington, and Owen’s Medpro system; Northrup Data System’s BDS series of systems, and Medical Information Technology’s packages. Some of these are not MISs, but rather primarily business office systems. They are noted because of the recommendation by some vendors that installa- tion of such a system precede or be the first step toward MIS installation. 56 Growth of MISs in the United States Some MIS systems concentrate on providing communication between medi- cal (most commonly, hospital) service areas. An example of such a function is the ability for a patient admission transaction to trigger housekeeping, dietary, and business office actions. These systems tend to cost $2 to $4 per patient per day. Savings are said to result from reduced clerical personnel through more efficient staffing and better utilization of facilities. Systems that also provide for reporting back to the hospital the location of required information and interdepartmental communication include: Shared Medical Systems’s ACTION NCR’s MEDICS McAuto’s Hospital Data Collection (HDC) Datacare’s PCIS NDC’s VITAL The most extensive client network is McAuto’s Hospital Financial Control (HFC) System, which has 475 installations. Their HDC System, which now includes a patient history and a nurse staffing/scheduling subsystem exists at 40 installations.42 Among the complete commercial systems, the most extensive range of MIS functions and systems is offered by three companies, Technicon Medical Systems, Medicus Medical Systems, and IBM. Technicon System. The TMIS is integrated into a single package operated on Technicon-owned computers at regional sites that amount to computer service bureaus. The Technicon approach centers about the concept of an interactive CRT computer terminal with which the physician enters orders. This is accom- plished via a branching display or menu of possible orders from which the physician selects using a computer light pen. Appropriate information is com- municated to hospital departments as a result of the orders by software provided by Technicon. Additional System Services. The physician is able to recall and alter the orders through the system and receive reports of laboratory and certain other services through the same system. A file or medical record is built for each patient with- in the system. It was a fault of the initial installation that the records were not kept after the patient’s discharge, and that a continuous patient record adding one hospitalization to another was not built. It is said that these deficiencies are being overcome by Technicon for future installations. SPECTRA System. Medicus Medical Systems Corporation’s SPECTRA System Description of MISs 57 is similar to the Technicon TMIS. This is a strongly competitive field and claims are made that, where the features are similar, the SPECTRA costs can be shown to be lower. It is beyond the scope of this work to make the kind of careful cost and detailed specification checking that a hospital would need to make before judging such a matter. For the general reader, it is fair to say that the two systems are grossly similar. Differences do exist. The SPECTRA System uses a special data station with a CRT, keyboard, light pen and printer, which they feel facilitates the interaction between health personnel and the computer. Color-coding conventions help the user discriminate between control functions, standing orders, and the next field for selection. Compatible Systems. A series of upward compatible systems is offered. The SPECTRA 500 provides for on-line admission, transfer, discharge census func- tions, in- and out-patients, bed control, and pharmacy. The SPECTRA 1000 system adds nursing station order entry and communication between depart- ments. The SPECTRA 2000 system also provides for medication scheduling, patient care planning and charting, patient record keeping, and customized physician orders. Additional features are planned for a future SPECTRA 3000 system to be used in teaching hospitals. Interface. The SPECTRA system provides medical logic and interaction and has been interfaced to a predecessor Medicus system that provides for business office functions. The Medicus Corporation takes a somewhat different approach to providing hardware facilities. They arrange for installation of the appropriate minicomputers within the hospital itself. Data General Eclipse computers are used; two are required at the hospital, providing for system backup. A five- year systems contract is required. The institution has the option of operating and maintaining the hardware itself or buying a facilities management contract from Medicus. The IBM Health Care Support System. HCS is a vigorous new entry into the MIS competition. Its general approach is compatible with the company’s past product line, including a focus on physicians’ order entry via light pen termi- nals. There is no compatibility with past software products. IBM still provides extensive software support for hospital customers to develop their own sys- tems. HCS goes further than past products by offering as proprietary program products a series of programs developed jointly by IBM and the Duke University Medical Center. The software includes 600 screens (text messages and formats for terminal displays) and 8,000 procedures (data collection lists for terminal display) as implemented at Duke. The screens and the data collection lists allow for forty-four application areas plus the nursing stations.43 The customer thus obtains “the Duke System,” and, in addition, receives full vendor support in making extensive changes in the displays and the manner in which the system 58 Growth ot MISs in the United States runs. Application segments are written in PL1 and Assembler language. The data base management uses the DL1 language and IMS (Information Management System) file structure. Modifications to adjust the screens and run commands are meant to be performed by nonprogrammer hospital personnel in each hospital. System Functions. The system incorporates patient care as well as administra- tive and resource management functions. It begins with registration and admis- sions, with extensions to serve central supply, pharmacy, laboratory, and other application areas. A second successful implementation was made at Parkland Hospital in Dallas, where David Mishelevich and colleagues adapted the system to their hospital’s requirements.44 Other installations are being made but have not been described publicly as yet. Control Data Corporation. Based on present accomplishments and published reports, CDC does not appear prominent in the MIS field, with the exception of the advanced systems at the University of Utah, which have been noted, and their ownership of the Med Lab commercial systems. Another CDC development warrants mention. CDC purchased the commercial rights to the PLATO com- puter-aided instruction system from the University of Illinois at Champaign- Urbana. Learning Centers. The company now has 55 learning centers utilizing their commercial version of PLATO through computer communications networks. No commercial MIS applications currently exist on PLATO, but experimental systems do. This possibility, especially relating to the ambulatory care health care system, is being systematically explored by CDC. All medically related activities have recently been concentrated into a health care services division. Because of the technical background of this company, their existing networks (PLATO, CYBERNET, and the former IBM Service Bureau system), their potential contribution to MIS development is worthy of note.45 Systems of Historical Interest General Electric MEDINET The GE MEDINET Department was established in June, 1966, and its com- pleted computer center and permanent offices dedicated on May 11, 1967. It proposed to offer a wide variety of computer services to health care institutions anywhere in the country via time-shared computers connected by dedicated telephone lines and transmission networks. The parent company, GE, already had the advantage of an extensive, nationwide telecommunications network. Description of MISs 59 The network of leased lines and message concentrators was used both for conduct of intracompany business and in connection with GEs time-sharing commercial computer business. In many ways this amounted to a nationwide private telephone company, GE had substantial experience and internal com- petence in business computer communications at the time it decided to enter the medical field. Its partner in the new business was Bolt, Beranek and New- man (BBN), a Boston engineering consulting firm. BBN had, since 1962, con- ducted a large research project at Massachusetts General Hospital in medical information processing based on computer time-sharing. The research project had been supported jointly by the federal government and the AHA. At its peak MEDINET employed 106 individuals as programmers and hospital or systems analysts. The system was demonstrated at the AHA meeting on August 21,1967. At that time a press release announced that “nationwide availability of the system to the medical community would not come until late 1968 or early 1969.46 A major administrative reorganization was announced publicly in December, 1967. Early in 1968 MEDINET ceased to offer the medical and patient care oriented computer services that had been aimed at a market of 20 to 50 institu- tions, and elected to offer business office and administrative services, aiming at what it considered to be a potential market of six hundred hospitals. At that time, MEDINET had spent about $16 million on systems development, mostly aimed at the MIS market.47 On May 20, 1970, GE announced that it would no longer engage directly in the manufacture and sale of general purpose digital computers. Its business computer equipment interests, both domestic and international, were merged with Honeywell in a new company in which GE was to own 18.5 percent. GE retained its time-sharing services, computer communications equipment, and process computer businesses.48 The MEDINET department was eliminated as an administrative entity in April, 1975. Detailed Description. The MEDINET prospectus proposed that “Hospital subscribers in a given geographic area [be] linked to a MEDINET center through communications lines. Each center, in turn, [will be] connected to other centers so that information from hospitals at one center can be made readily available at another.”49 Sharing data between hospitals was part of MEDINETs initial plan, and was already being done among GEs own commercial divisions. It is a sad commentary that, with the exception of hospitals owned by corporations, such a scheme has yet to be effected between hospitals by any group or com- pany even today. MEDINET recognized that major medical centers had already discovered the magnitude of their own data processing needs by 1966, and in some cases had already built partial computing services of their own. 60 Growth of MISs in the United States MEDINET proposed that “Large hospitals or special groups may have their own center which is connected to a MEDINET center for backup and network communication. ”50 The approach taken to terminals was also relatively flexible initially: Terminal equipment in the hospital itself may take various forms depending upon the needs of each specific user. A terminal device, connected to the computer center, may be a printer, a cathode ray screen, keyboards, and X-Y encoder, a badge reader—or a combination of these. Terminal devices will be located in the hospital at the points where they are needed most. Representative terminal locations might include: 1. Admission and Discharge Office 2. Patient Care Unit 3. Pathology Laboratory 4. Administrator’s Office (for management information) 5. Blood Bank 6. Clinical Laboratories 7. Accounting Offices 8. Chief Resident’s Office 9. Radiology 10. Dietician’s Office 11. Pharmacy 12. Research Laboratory51 The approach to implementation constituted a modified version of the holistic philosophy. The systems were to be imagined and programmed by GE as a whole, but the individual parts were to be implemented within the hospitals in subcomponents. The subcomponents would be adjusted to the local hospital environment through collaboration between hospital staff and programmers and MEDINET staff and programmers. The MEDINET offering at the time suggested: For hospitals, a phased approach into the MEDINET program is sug- gested. One function or application would be implemented at a time. For example, one hospital might introduce MEDINET services in this order: patient billing, payroll, admission and discharge, laboratory reports, research inventory, pharmacy and medication, doctor’s orders. Description of MISs 61 These and other applications will see continuing development by the MEDINET Company, working closely with selected hospitals, labora- tories, and medical libraries.52 In retrospect, it is clear that GE saw a great similarity between the in- formation processing needs of each hospital. They spoke of an emphasis on complete integrated information and automation systems. This perception, perfectly reasonable even now, was based on their industrial experiences. This is the reason that they could conceptualize the entire medical system on their own, and yet offer to each hospital the option to create its own routines for each service area. GE assumed that the differences from customer to customer would be rather slight. As it happened, the hospital staffs took considerable pleasure in focusing on the ways in which each hospital differed from other hospitals and in the extent to which its information processing requirements needed to be treated as unique. Special Features. Four elements of the MEDINET development stand out. First, a terminal was designed and built that incorporated keyboard, impact printer, rear-screen projector, X-Y coder, badge reader, and transmission modem. This was the MEDINET 661. The various internal coding and conversion con- ventions were unique in many ways. In addition, a new computer language, FILECOMP, was designed and written especially for hospital networking. This language was based on advanced concepts of a relatively high-level interpretive language with string-handling capability well suited to handling text passages and File Fields. The hardware also began as a rather special conFiguration, originally a GE 485 complexed to smaller Honeywell process control computers. Later the company elected to reconFigure the entire system to use the GE 600 series hardware exlcusively. The fourth special element of the GE development lay in its philosophy. The planners and managers saw health care delivery activities as a rather sim- plistic social phenomenon. They assumed that the arrangements among health practitioners and health care institutions were rational and explicit. They further assumed that decision makers would adapt themselves to the new computer capabilities in much the same way that banking and industry had. These Fields had already experienced the effect of integrated information systems in enhanc- ing the centripetal tendency already at work in industry to develop in a hierarchical pattern with the emergence of larger banks, holding companies, and conglomerates. Clinics, nursing homes, private physicians, laboratories, and other mem- bers of the medical community, will be able to join the MEDINET net- work as the hospitals in their area become members.53 62 Growth of MISs in the United States Centripetal forces were not at work in the hospital field. Historically, very little joining of health care elements has occurred. Kaiser-Permanente MIS By far the most advanced of all American general MISs was that at Kaiser- Permanente. The destruction of what had been built can be attributed to unique and unfortunate circumstances.54 The original multiphasic screening system had been built as a response to internal company needs and client demands. Twelve years of development were financed internally and successfully before any government research funding was accepted. Success with direct computer support of multiphasic screening, the demonstrated ability at Kaiser to analyze and compare these data with manage- ment data for the eleven northern California Kaiser Plan hospitals, and recogni- tion of information processing needs in their hospitals prompted Kaiser to com- mence major MIS development with partial support from federal funding. The scope of the project was large: a central medical computer system at one of the hospitals (with the ability to manage an integrated patient record including all clinically significant services) linked to a computer-supported communications system which was, in turn, capable of supporting ten medical centers. Both hospitalized and out-patients were to be served. The initial installation (and pilot version) was begun at Kaiser’s San Fran- cisco medical facility. Good progress was made at the subsystems level. The sub- systems included laboratory ordering and reporting, medication orders, phar- macy information management, and an innovative hospital and clinic patient record retrieval system for the emergency room. Progress on the project, both in the planning phases and the finished systems, had advanced further at Kaiser than anywhere in the United States. Final systems integration was planned but not completed. In the meantime, useful research studies were accomplished concerning health care practices, costs, and outcomes using the growing store of subsystems data. At a critical juncture, when there were very serious technical problems during expansion, the two major sources of federal funding were rather precipi- tiously withdrawn. These included federal funding to Kaiser as one of a dozen Health Services Research Centers, and cancellation of a major contract with the Food and Drug Administration (FDA) for important prospective drug reaction and toxicity studies. By then, the feasibility of the central record management system had been demonstrated at San Francisco. Justification of its costs was premised on bring- ing the other hospitals on-line. This process would have required another two to four years, a planning horizon not permitted by immediate withdrawal of the research support. Description of MISs 63 The public importance of the project was two-fold: first, to test whether this centralized, integrated computing strategy was the best model for other institutions to follow in MIS building, and, second, to create a working medical data base system that could serve the needs of the 1,110,000 Kaiser Plan mem- bers who could serve as a microcosm of the nation to answer many health services research questions relating to national health policy. The good of the public was not served by destroying this system after technical feasibility alone had been demonstrated. Nor were Kaiser Plan interests advanced by this experi- ment in resource sharing with the federal government. The Medical Methods Research Group was pulled off balance by the elimination of research support of developments that had been premised on a balance of service and research funding. This project exemplifies an inherent difficulty with major innovations in medicine. The difficulty is the long time required before benefits are mea- surable. The initial proposal for evaluating the benefits of the Kaiser periodic health examination and multiplasic screening recognized that five years of data collection would be necessary to detect improvement in morbidity. On the basis of statistical projection, ten years of data collection were known to be needed to detect benefits in reduced mortality. The five-year data did show reduced morbidity in the patients exposed to the innovation. Before the second half of the study, which might demonstrate reduced mortality, was completed, the institution had to give up the computers that made the MIS possible. University of Missouri-Columbia A number of MIS elements were developed at, or imported to, the University of Missouri during 1961 to 1972. These included computer systems for patient bed census; numerically coded clinical, radiological, and pathological diagnoses; dietary services; operating room log; radiology scheduling, reporting and film management; a radiological differential diagnosis physician assistance function; a drug information system; EKG interpretation; an integrated patient file; and various special registries. The laboratory system will be singled out as an ex- ample. The consistent success and diffusion of MIS technology in the clinical laboratory suggests that it is an example of a relatively simple innovation; it automates a current practice. The laboratory system based on on-line computer terminals at the Uni- versity of Missouri was an early one, developed in 1963.55 This system suc- ceeded an earlier off-line automated reporting system.56 The off-line system was developed with institutional funds and the on-line system with NIH re- search grant support. The on-line system used densely coded push button matrix keyboard terminal units for input of laboratory tests from all clinical 64 Growth of MISs in the United States laboratories. The special virtue of the system was the extensive editing, error checking, and medical quality control routines which were provided by the pro- grams that processed laboratory information. These resulted in trapping and cor- recting substantial numbers of erroneous laboratory reports, which would otherwise have been sent to the patient record. About half the errors were detected by the use of simple check digits that validated patient and specimen identification numbers. Detection of about half the errors required use of medical logic of varying degrees of complexity. The on-line system was successfully implemented within budget and on schedule in three years. Follow-up and internal evaluation studies were com- pleted.57 The host institution assumed the entire cost of subsequent operation of the system. In many ways the system achieved all of its objectives, as well as having succeeded in satisfying the then-current NIH standards for success. It worked, and was adopted by the hospital. The system continues with evolutionary improvement. It has had no federal research support for eleven years. In 1970 the system was inspected by visiting teams from forty-five medical institutions from the United States and abroad. Many copies of the system and the design documents were distributed. The success of such subsystem demon- strations contributes to further diffusion of this technology. On the negative side, however, is the fact that absence of research support for this application since 1967 has meant that no research has been done on the system per se for eleven years. Consequently, it is currently patched, inefficient, out-of-date, and incompatible with the many new advances and changes in laboratory technology that have arisen since then. The computer programs themselves are still doing their jobs. There is no parallel manual arrangement. Yet this system is no longer innovative. Consequently it is no longer able to exert a beneficial influence on the development of laboratory systems elsewhere, nor to enhance the diffusion of MIS technology more generally. Notes 1. B.C. Glueck and C.F. Stroebel, “Computers and Clinical Psychiatry,” Comprehensive Textbook of Psychiatry II, vol 1, A.M. Freedman, H.I. Kaplan, and B.J. Sadock, eds. (Baltimore, MD.: Williams and Wilkins, 1975), p. 414. 2. B.C. Glueck, R.P. Ericson, and C.F. Stroebel, “The Use of a Psychiatric Patient Record System,” in Computers in Life Science Research, W. Siler and D.A.B. Lindberg eds. (New York: Plenum Press, 1975), p. 173. 3. R.M. Gardner, D.P. Scoville, B.J. West, et al: “Integrated Computer Systems for Monitoring of the Critically 111, ” in Proceedings, First Annual Symposium on Computer Applications in Medical Care, Washington, D.C., October 3-5, 1977, H. Orthner, ed. (New York: I.E.E.E., 1977), pp. 301-307. Description of MISs 65 4. H.R. Warner, “Knowledge Sectors for Logical Processing of Patient Data in the HELP System,” Proceedings of the Second Annual Symposium on Computer Applications in Medical Care, Washington, D.C., November 5-9, 1978, F.H. Orthner, ed. (New York: I.E.E.E., November 1978), pp. 401-404. 5. Ibid., p. 401. 6. R.M. Gardner and T.P. Clemmer, “Computerized Protocols Applied to Acute Patient Care,” in Proceedings, 7th Technicon International Congress, New York, December 13-15, 1976. vol. 1 Clinical and Hospital Management Symposium (Tarrytown, N.Y.: Mediad, Inc. 1977) pp. 158-163. 7. A.D. Little, Inc., Evaluation of Computer-Based Patient Monitoring Systems: Final Report; Appendix D. A Review of the MEDLAB System in the Thoracic Surgery Intensive Care Unit at Latter Day Saints Hospital (Rockville, Md.: DHEW, March 1973) NTIS PB 247 421. 8. H.R. Warner, “Knowledge Sectors for Logical Processing of Patient Data in the HELP System,” Proceedings of the Second Annual Symposium on Computer Applications in Medical Care, Washington, D.C., November 5-9, 1978, F.H. Orthner, ed. (New York: I.E.E.E., November 1978), pp. 401-409. Reprinted with permission. 9. Gardner, et al: “Integrated Computer Systems for Monitoring of the Critically 111,” p. 305. 10. J.P. Barrett and R.N. Pesut, Physician Evaluations and Expectation for the HELP System. Final Report (Columbia, Ohio: Battelle Laboratories, April 1977) 11. R. Gardner et al: “Integrated Computer Systems for Monitoring of the Critically 111,” p. 306. 12. Gardner et al: “Integrated Computer Systems for Monitoring of the Critically 111,” p. 306. 13. G.O. Barnett, Computer Stored Ambulatory Record (COSTAR), NCHSR Research Digest Series (DHEW, 1976). 14. G.O. Barnett, N.S. Justice, M.E. Somand, et al: “COSTAR—A Com- puter-Based Medical Information System for Ambulatory Care,” Proceedings, Second Annual Symposium on Computer Applications in Medical Care, Wash- ington, D.C., November 5-9, 1978, F.H. Orthner, ed., (New York: I.E.E.E., 1978) pp. 486-487. 15. G.O. Barnett, N.S. Justice, M.E. Somand, et al: “COSTAR—A Computer-Based Medical Information System for Ambulatory Care,” I.E.E.E. Proceedings, (New York: I.E.E.E., September 1979). 16. G.O. Barnett, personal communication, March 1, 1977. Reprinted with permission. 17. Barnett, Computer-Stored Ambulatory Record (COSTAR), p. 2. 18. B.D. Waxman, P. Rowny, A. Zuckerman, et al: “An Approach to Medical Software Portability: The COSTAR V Project,” Proceedings of Hawaii International Conference on System Sciences, 1978. 66 Growth of MISs in the United States 19. B.D. Waxman, personal communication, January 9,1979. 20. National Bureau of Standards, MUMPS Language Standard NBS Hand- book 118 (Gaithersburg, Md.: National Bureau of Standards), January 1976. 21. PROMIS Laboratory, Functional Specification of a PROMIS Instance in Three Levels: Introductory Material (Burlington, Vt.: University of Vermont, February 1977). Reprinted with permission. 22. J.W. Hurst, “How to Implement the Weed System,” Archives of In- ternal Medicine, vol. 128, September 1971, pp. 456-462. 23. J.M. Aranda, ‘The Problem Oriented Medical Record: Exercises in a Community Military Hospital,” Journal American Medical Association, vol. 229 (5), July 29,1974, pp. 549-588. 24. PROMIS Laboratory, Functional Specification of a PROMIS Instance in Three Levels: Introductory Materials, p. 3. Reprinted with permission. 25. PROMIS Laboratory, Automation of the Problem-Oriented Medical Record, NCHSR Research Digest Series (U.S. Department HEW, April 1977). 26. PROMIS Laboratory, Automation of the Problem Oriented Record: Executive Summary, Fig. 3. Reprinted with permission. 27. PROMIS Laboratory, Functional Specification of a PROMIS Instance- Level I, p. 65. Reprinted with permission. 28. H.E. Klarman, “Application of Cost-Benefit Analysis to the Health Services and the Special Case of Technologic Innovation,” International Journal of Health Services, vol. 4, issue 2,1974, pp. 325-352. 29. Texas Institute of Rehabilitation and Research, Demonstration of a Hospital Data Management System, Five Year Summary Progress Report, January 1, 1967-June 30, 1972 (Houston, Tx., 1972). 30. M.H. Hodge, Medical Information Systems: A Resource for Hospitals (Germantown, Md.: Aspen Systems Corp., 1977). 31. R.M. DuBois, National Center for Health Services Research, personal communication, January 12, 1977. Reprinted with permission. 32. G.C. Macks, T.L. Lewis, R.E. Prior, et al: Study of the Impact of a Computerized Hospital Information System of the NIH Clinical Center, Final Report (Bethesda, Md.: Office of Clinical and Management Systems, NIH January 31,1974). 33. U.S. Congress, House, Committee on Science and Technology, Testi- mony of Arnold W. Pratt before the Subcommittee on Domestic and Interna- tional Scientific Planning, Analysis and Cooperation, 95th Congress, 2nd Session, May 9,1978. 34. McDonnell Douglas, McAuto Hospital Services Overview (St. Louis, Mo.: McDonnell Douglas, about 1977). 35. D.J. Mishelevich, M.D., Ph.D., personal communication, February 27,1979. 36. 1978 Survey of Automated Data Processing Systems. American Hos- pital Association and Health Care Technology Center, University of Missouri- Columbia, 1978. Description of MISs 67 37. D.K. Tao, Computer Applications in Medicine: A Survey of Vendors (St. Louis, Mo.: Washington University School of Medicine, February 1978). Monograph No. 328. 38. D.W. Huffmire, “The United States Medical EDP Market,” Medical Marketing and Media, August 1978, pp. 17-25. 39. Automated Hospital Information Systems, Case Study Report Pre- pared for Health Resources Administration, Rockville, Md. under Contract HRS-230-75-0063. (Systemedics, Inc., Princeton, N.J., April 29, 1976, reissued June 15,1977). 40. T.L. Lewis and G.C. Macks, “Adaptation of a General Hospital Com- puterized Information System to the Research Hospital Environment,” Pro- ceedings: First Annual Symposium on Computer Applications in Medical Care, October 3-5, 1977, Washington, D.C., (New York: I.E.E.E., 1977), pp. 111-115. 41. D.J. Mishelevich, M.D., Ph.D., personal communication, February 27, 1979. Reprinted with permission. 42. McDonnell Douglas, Hospital Data Collection System, 1975, (St. Louis, Mo.: McDonnell Douglas, 1975). 43. I.B.M., Health Care Support/DL/'/ Patient Care System, (White Plains, N.Y.: I.B.M., 1977). 44. D.J. Mishelevich and D. Cranfill, “Computer System (On-Line),” Guidelines for Determining Data Processing Needs for Texas Hospitals (Austin, Tx.: Texas Hospital Association), September 15,1978. 45. D. Walter, personal communication, February 19, 1979. Reprinted with permission. 46. American Hospital Association, MEDINET Announcement, Press Release (Chicago, HI.: AHA, August 21,1967). 47. J.J. Baruch, personal communication, March 7, 1978. Reprinted with permission. 48. General Electric. General Electric-Honey well Press Release (New York: General Electric, May 20, 1970). 49. General Electric. This is MEDINET (Watertown, Mass.: General Elec- tric, 1966), p. 4. 50. Ibid., p. 4. 51. Ibid., p. 5. 52. Ibid., p. 6. 53. Ibid., p. 6. 54. E.E. Van Brunt, L.S. Davis, and M.F. Collen, “Kaiser-Permanente Hospital Computer System (San Francisco),” in Hospital Computer Systems, M.F. Collen, ed. (New York: Wiley and Sons, 1974), p. 706. 55. D.A.B. Lindberg, “Collection, Evaluation and Transmission of Hos- pital Laboratory Data,” Methods of Information in Medicine, vol. 6, issue 3, July 1967, pp. 97-107. 68 Growth of MISs in the United States 56. D.A.B. Lindberg, “Electronic Processing and Transmission of Clinical Laboratory Data,” Missouri Medicine, vol. 62, April 1965, pp. 296-302. 57. D.A.B. Lindberg, J.J. Schroeder, Jr., L.R. Rowland, et al: “Experience with a Computer Laboratory Data System,” in Multiple Laboratory Screening (New York: Academic Press, 1969). 5 Evaluating the Worth of MISs Summary MISs represent a major hospital investment for which potential cost savings and other benefits are claimed. The relatively small number of experienced users of these systems tend to be enthusiasts. Others tend to be doubters. Hence, formal evaluation is necessary. The tools available for evaluation include marketplace mechanisms, opera- tions research, cost-effectiveness and cost-benefit analyses, technology assess- ments, and studies of scientific impact. The marketplace has selected predominantly hospital business office systems, not MISs. Operations research studies have identified improvements in hospital administration, health care provider and patient satisfaction, and additional objective process measures, including some oriented to immediate improvements in patient care. Economic analyses have revealed cost savings. Only limited technology assessments have been performed. These identified potential dangers to data privacy which necessitated legislative protection. The scientific impact of the MIS has virtually not been studied. Speculative observations suggest that there have been beneficial scientific impacts and that further potential scientific gains from use of MISs will be achieved. The Need To Evaluate MISs have been shown to benefit patients, directly and indirectly. Many reports of this beneficial effect have been published.1,2 In contrast, no reports of adverse effects on patients of MISs operating in a health care environment have been reported. One possible exception is the effects of MISs on individuals reported to have occurred when limited medical information was obtained in connection with applications for life insurance.3 Because there has been such enthusiasm for MISs on the part of individuals and institutions who have had direct experience with them, the need for formal evaluation of the technology has become evident. We noted in chapter 1 that a large part of all hospital operating costs are attributable to medical information processing, generally nonautomated. Likewise, it has been reported that MISs are costly to develop and operate, but that the net effect may be to save money and improve the health care system. Decisions to implement MISs, whether 69 70 Growth of MISs in the United States made by individual health care institutions or advocated by governmental agencies, cannot be based only on enthusiastic reports from system builders. More systematic and scientific evaluation is warranted. The present chapter analyzes the considerations that enter into the design of such evaluations, and describes the problems involved and the extent to which it has been pos- sible to estimate the worth of MIS technology. For Whom and in What Sense Might MISs Be Worthwhile? Worth must be estimated so as to emphasize medical benefits to the individual patient whose records are being processed. There are, however, other relevant parties. These include: The hospital department The health care institution The community Region or health service area The nation Benefits from an MIS might include all of the following: 1. Immediate improvement in the patient’s health care 2. Immediate improvement in the patient’s access to health care 3. Immediate reduction in costs of care 4. Immediate educational benefit (either to patient, health care personnel, or both) 5. Future educational benefits (to either group) 6. Future avoidance of cost to patient 7. Future research gains 8. Future improvement in patient care Such benefits might result directly or indirectly from operation of an MIS. For example, a drug information subsystem might detect a potential drug inter- action, delay administration of the second drug, and thus immediately and directly produce a benefit to a patient by preventing a dangerous drug-induced episode. Or, an MIS might produce a beneficial effect only in combination with a second system or technology. The synergism might be dependent on the existence of the MIS, such as indirect effects seen in the multiphasic health examinations demonstrated at Kaiser-Permanente.4’5 The benefit of reduced Evaluating the Worth of MISs 71 morbidity to patients who were members of the study group was dependent on periodic medical examinations and proper treatments. The existence of such a system, however, was dependent on the MIS as an infrastructure that provided efficient and reliable record keeping, reminding, appointment making, and data analysis. In a formal evaluation there must be agreement ahead of time as to what aspects of a system are to be measured. In the case of health systems, the follow- ing three aspects are proposed for observation and measurement: 1. Inherent value or content of the service under study. When an MIS contains standards, references, or didactic information, the evaluation instru- ment may test if the information within the MIS is equal in validity or superior to information available via a manual or alternative system. 2. The process of health care delivery. An MIS might be shown to result in an improvement in the manner in which services were provided, for example, if fewer venepunctures were required to obtain blood specimens; if reports of laboratory tests were reported more promptly than otherwise; or if the need for repeat x-ray studies were eliminated because the MIS prevented misplacement or loss of x-ray reports. 3. The outcome of the process. This measure examines the ultimate result of the system operation (some aspect of the health of the patient). Reduction in the number of days of hospitalization for a particular kind of illness, or even reductions averaged over all illnesses in a given hospital might be an outcome benefit attributable to an MIS, provided other relevant factors were constant over the time of the evaluation. While individual patients should be the prime beneficiary of an MIS or other health care innovation, there are other potential beneficiaries. Yet it is difficult to describe a general scheme for measuring benefits to remote beneficiaries such as a social group, region, or nation. Health service researchers often act as if they believed that process measures were grossly inferior to outcome measures in evaluating medical systems. In contrast, Avedis Donabedian, in a recent review dealing with the assessment of the quality of medical care, describes the coequal relationship between the two: It is not true that outcomes are a more valid measure of quality than is process, as it is fashionable to say. It is true that process measures are valid indicators of quality only to the extent that they relate to rele- vant outcomes. But it is equally true that outcomes are valid measures of quality only to the extent that they relate to the antecedent process of care.6 Many interesting factors tend to complicate the measurement of the worth of MISs and other medical technologies. Some of the factors and the problems created will be discussed in connection with the available evaluative methods. 72 Growth of MISs in the United States An Inventory of Methods at Hand for Evaluating the Worth of MISs Marketplace Outcome This is a default method that simply observes what remains of free enterprise market forces at work. The expectation is that what is preferred will prevail in the marketplace, that valuable systems and items will compete successfully with less valuable offerings. Use of this method is not suggested out of frivolity; it is essentially the method by which MISs are currently being evaluated. Operations Research This approach insists on a protocol for evaluation, with an explicit choice of variables to be examined, and an analytic methodology for processing the observations based on statistically valid procedures. Beyond these constraints, operations research methods are empiric; they may use economic measures, measures of personal convenience, systems efficiency, or other measures. The measurement strategy may use pre and post observations; it may discrimi- nate sharply between process and outcome measures; it may use matched sites. There will always be phases of quantitative observation, analysis, and con- clusion. The fundamental merit of the approach is that it insists that there is always something that can be measured, no matter how complex the process, and tends to assume that decisions will be wiser when based on some quanti- tative measure than on none. Cost-Effectiveness Analysis Cost-effectiveness analysis compares the economic efficiency of alternative systems directed at the same objective. Costs are calculated and compared for alternative ways of achieving a specific set of results. Medical methodologies are often compared in this way. The cost effectiveness of a method is traditionally measured by comparing it with an alternative method. Given that the tasks accomplished are roughly the same, it is not conceptually difficult to put a relative cost on the two op- erations. A familiar example is a manual procedure versus an automated pro- cedure. Frequently, automated procedure can be shown to be more cost effective. Evaluating the Worth of MISs 73 Many of the components or subsystems of the MIS can be compared in this way with some predecessor technology or manual method. Often this is an entirely fair comparison and an entirely sufficient basis for preferring the cost- effective method. Human decisions are based on many factors, however, and cost-effectiveness analysis is only intended to assist in formalizing the relative economic merits of the question to be decided. An example of a choice that could reasonably be based on cost-effectiveness analysis is that between automated clinical chemistry analyzers and manual methods. The automated system typically does more tests per hour at a small fraction of the unit cost for manual procedures. Assuming that observational data are available for both modalities, a number of reasonable routes exist along which to conclude the analysis. A reasonable decision criterion might be net present benefit, although return on investment might be chosen if this seemed more appropriate because of limited investment funds.7’8 These are rather general considerations and tend to be based on aggregate or averaged data. Local circumstances should be taken into account when a cost- effectiveness analysis is performed for a particular institution. Performance specifications should match local work loads and patterns. Similarly, local costs are more valuable for such an evaluation than nationally averaged figures. In the case of most currently available systems for laboratory automation, the analysis should acknowledge that the automated methods tend to fit better with batch testing. Single tests done by automated methods may be quite expensive. This should be reflected in the job mix that is the basis for com- parison. The laboratory may need to buy premixed reagents in surprisingly large quantities to eliminate compounding and mixing; these reagents require space and attention to safety. All this could be accounted for by adding to or subtracting from the costs included in the analysis, based on cost estimates for the given circumstances. Many subsystems of what we now think of as MIS can be evaluated by the cost-effectiveness method, whether in the general case based on averaged data, or with respect to a specific institution. This evaluation method is suitable for all subsystems of the MIS that have a manual counterpart. The comparisons must be contingent on some condition (such as patient acceptance or safety), but under these conditions, the evaluations and choices will still be reasonable from an economic point of view. The cost-effectiveness method is unsuitable for evaluating the entire MIS because there is no alternative methodology. The raison d’etre for MISs is to bring together all available parts of the patient record; to be more than the sum of the parts. Hence, there is not a valid basis for accepting a cost effectiveness comparison with respect to a complete MIS. 74 Growth of MISs in the United States Cost-Benefit A nalysis Cost-benefit analysis provides for calculation of the costs of an activity or sys- tem and the potential benefits to be derived from operating the system or activity. It is a method for identifying the relative economic merits of a ques- tion. Ordinarily it is used as an aid in choosing between two courses of action. Consequently, it is customary first to identify alternative actions or programs, and then make a quantitative assessment of all costs to be incurred and benefits to be gained. Last, one examines these alternative costs and benefits against some explicitly stated criterion, such as the ratio of benefits to costs, or more technical economic criteria such as rate of return on investment. Cost-benefit analysis is suitable, and cost-effectiveness analysis unsuitable, if one is attempting to answer the question, Should a new technology (such as an MIS) be employed? It is reasonable to question whether to expend re- sources on MISs versus some other purpose, even a non-health-related purpose. This is a valid application of cost-benefit analysis. The analysis will, however, require measurement of the costs and benefits of all the schemes being con- sidered, including the non-health-related ones. One MIS can be evaluated against another MIS if they are doing the same job. The evaluation measure, the basis for the comparison, could be cost, tho- roughness, reliability, precision, training requirements, space utilization, direct or indirect patient benefits, or societal gains, one or all of these, weighted or unweighted. For methodologies that have no counterpart or predecessor one is urged lately to prefer measures that compare the outcome of employing each method or the outcomes of employing the method in question versus investing for some other purpose. This strategy is useful partly because it tends to ask two ques- tions at once; namely, Which method tends to reach the goal best? and What is the validity of the goal in any case? It would be especially desirable to estimate health outcomes for such a comparison. On the other hand, it is difficult and costly to demonstrate rigorously that specific changes in health care delivery systems or mechanisms have a direct effect on health outcomes. The outcomes are remote, and the delivery systems do not remain static over time so as to permit study of isolated changes. In the case of MISs the outcomes or benefits to be measured may be speci- fied in terms centered on the patient, the institution, society, or all of these and others. The evaluator can be left to weigh the relative importance of each outcome measure either before or after the analysis has been made. One diffi- culty is that outcome measures that have been expressed in these various terms will often not naturally have the same unit of measure and hence are difficult to give weights to or to combine. An example of such benefits might be im- proved access to care, improved productivity, and patient satisfaction. The second difficulty is that some important outcomes may not have any readily Evaluating the Worth of MISs 75 apparent unit of measure at all. The classical example of such a benefit is im- proved quality of care. The third difficulty is that some outcomes are not measurable immediately after the methods have been tested nor within the same system under consideration. An example of the third difficulty, external- ities, arises in consideration of a universally desired benefit, cost containment. If the costs saving of a given procedure is accrued to some societal unit above that of the institution under study (a community or state, for example), the benefit will never be measured within the individual institution under study. Cost escalation presents a similar problem. The patient who is denied a computer tomography scan at Institution A may not have saved society’s health care dollars by traveling across town or across the state to get one at Institution B. Technology Assessment Technology assessment is a new, as yet not rigorously defined, approach to evaluation that appears especially well suited for decision making involving public policy. The approach was enunciated by Emilio Daddario in 1967 in studies preceding the establishment of the Office of Technology Assessment within the United States Congress. He stated: Technology assessment is a form of policy research which provides a balanced appraisal to the policy maker. It identifies policy issues, assesses the impact of alternative courses of action, and presents findings: it is a method of analysis that system- atically appraises the nature, significance, status, and merit of the technological program. . . [It] is designed to uncover three types of consequences—desirable, undesirable, and uncertain . . . The focus of Technology Assessment will be on those consequences that can be predicted with a useful degree of probability.9 As the technique evolved during the decade following this description, definitional scope remained extremely broad, though individual studies have fallen far short of the goals. One reviewer noted: Technology Assessment ... is a new commitment to develop and apply a comprehensive analytic strategy to achieve more effective public management of technological change. [This] would include, for example, effects of the proposed technology on the economy, the physical environment, institutions, culture, the social structure, mores, values, and the law. It provides a vehicle to couple the careful thought given by scientists 76 Growth of MISs in the United States and engineers about the technical efficacy of their projects with parallel thought by policymakers about the social efficacy of these endeavors.10 In spite of the breadth of this goal, more than one hundred such studies had been undertaken in various fields, generally nonmedical.11 Technology assessments typically involve multidisciplinary teams. Social impact analysis is the weakest and most vulnerable area of technology assess- ment, according to Sherry Arnstein.12 She has proposed a taxonomy for tech- nology assessment studies, in which three classes are recognized. These are: Macro or comprehensive assessments; Comprehensive in breadth and depth Requiring two or more years to complete Costing $300,000-$600,000 Mini assessments: Comprehensive in either breadth or depth, perhaps as pilot experi- ments Requiring less than a year Costing $30,000-$60,000 Micro assessments: This category is meant to include structured conferencing or brainstorming sessions, either exploratory in nature or aimed at defining the scope of a subsequent more ambitious assessment. Such exercises can be limited to hours or days Costs could fall under $5,000.13 Studies of Scientific Impact A scientific impact study has no formal body of techniques that can rival the gathering of forces of technology assessment experts. I mean to imply by this term merely that one can discover the contribution to scientific advances made by a technique, method, or system by looking specifically and diligently for the effects in question. Such studies involve the analysis of reports of experiments, published literature, literature citations, and sometimes personal interviews with scientists. Many historical innovations in science and medicine have had a well-known effect on subsequent discovery and scientific understanding. These include the importance of the light microscope to the subsequent understanding of anatomy, to development of the structure and function paradigm, and to the recognition of microbiological theory; the contribution of the colorimeter to clinical chemistry; of the statistics of distribution to clinical trials methodology; and of quantitative immunological methods to clinical epidemiology and rheumatology. It is possible to trace the stewardship of ideas in science, and often to know Evaluating the Worth of MISs 77 certainly from whom our techniques were learned and when they were first applied to clinical problems. Often the unit of currency in science is not a concept (such as the compound lens), but a packet of information, for instance, the knowledge that polio virus can propogate on tissue cultures of monkey kidney cells. Typically studies of the migration and diffusion of knowledge in science are retrospective, are performed by scientists, science writers, or historians, and are relatively inexpensive. Studies of scientific impact are typically induc- tive, reasoning from the particular to the general. They are also analytic, select- ing one effect out of many for study. MISs may contribute significantly to the development of new knowledge of health and disease and to a better understanding of health care processes. One would detect or document such a benefit by studies directed at the question of scientific impact. That is, one could trace the scientific impact of a new technology only by looking specifically for that impact. Some general purpose tools such as bibliometrics could be useful. It is most unlikely, however, that appreciation of scientific impact, that is, the power of a technique to stimulate the scientific imagination and to suggest new hypotheses for testing and new viewpoints for adoption, could arise through use of the standard inventory of technology evaluation tools. Operations research, cost-effectiveness, and cost- benefit analyses as usually practiced are not likely to reveal the scientific impact of technology. To see this process in terms relevant to MISs, one must imagine the “pac- kets” of information to be somewhat larger and more aggregate than in the case of the polio virus tissue culturing. MISs build systematic data bases of observations on the problems, care, and outcomes of health maintenance activities in defined populations. Much understanding of modern medicine has come from systematic observation of patients. Automated information systems, especially patient record systems, have the capacity greatly to facili- tate clinical research. This potential derives from the relative ease with which computer systems can examine records from many points of view, concatenat- ing and combining the observed and measured variables in all manner of ways. Of course, it is still the investigator who must recognize and evaluate the as- sociations. Automated information systems increase the ease with which he can test hypotheses, and greatly increase the reproducibility of such studies. It is commonplace to observe that computing systems allow the collection of large numbers of measurements and observations of an individual, and permit the analysis of these complexes. This medical data base building lies at the heart of MISs. If we ask, “To what end?” and “With what scientific impact?” it is apparent that the answer cannot emerge from studies of cost benefit, especially if the decision measure is oriented to individual patient outcome. The scientific impact of MISs is much more likely to derive from use within clinical medicine and epidemiology of the concepts of the quantitation of information and the 78 Growth of MISs in the United States challenge presented by working MISs as means of testing scientific speculations and hypotheses. The kinds of studies in which this impact is likely to be revealed include the following: 1. Long-term studies of the natural history of health and disease 2. Pooling of clinical patient records to study particular, especially rare, conditions and treatments 3. Examination of health care records aimed at identifying interacting factors, especially interactions between fixed biological factors, medical treatments, and environmental factors 4. Modeling of systems, either physiological or societal, validated against per- formance records 5. Systematic analysis of general health care schemes 6. Using the new vantage point of the information measurement to synthesize a new aim from previously unrelated information elements Results of Employing the Evaluation Methods Marketplace Outcome There is now no federal policy with respect to MISs. No encouragement is offered directly, except for significant but relatively modest support of individ- ual research projects, mostly in the past. Likewise, there is no overt discourage- ment to MISs except that reimbursement regulations mean, in effect, that the cost of such systems must be folded into the hospital operating expenses and absorbed into the per diem bed charges when they can be justified as essential. The result of the judgment of the market place is essentially the result of modified free enterprise. What is the result? Group medical practices in the United States use auto- mated systems for management purposes in 72 percent of the cases surveyed by Medical Group Management Association.14 Of these 31 percent had in-house computer capability and 41 percent used computer service bureaus. As for hospitals, one proprietary survey in 1976 reported that 82 percent of all U.S. hospitals use some form of computer systems for business office services, mostly billing. An independent survey of hospitals was performed in 1976 by the Hospital Financial Management Association who found that electronic data processing was used by over 90 percent of the respondents for at least one of their applications.15 More than one thousand hospitals used a computer in another hospital for data processing. Again the applications were preponderantly business rather than medical or paramedical. Many kinds of hospitals (and hospital needs) are represented in such aggregate data. This is reflected in the finding that 10 percent of the U.S. hospitals did not use any computing systems, Evaluating the Worth of MISs 79 while a different 10 percent used both their hospital-owned computer and also the services of an outside information processing vendor or service. A large number of vendors and service bureaus as well as in-hospital computer installa- tions provide these services. McDonnell Douglas Automation Company serves 475 client institutions in forty-three states with its Hospital Financial Controls system.16’17 It claims to be the nation’s leading supplier of hospital shared computer services.18 In contrast, not more than two hundred U.S. hospitals have MISs that include significant amounts of direct clinical services. Regret- tably the exact Figures are not known certainly. McDonnell Douglas and Shared Medical Systems offer features in addition to financial management. A more detailed description of the additional services is presented in chapter 4. For the present purpose it is probably sufficient to take note of the relative numbers of customers involved. Going beyond the Financial controls system, McDonnell-Douglas’s hospital Data Collection System provides for interdepartmental data communications. There are ap- proximately 40 user institutions.19 A more clinical system offered by this company is the Hospital Care System which now serves four hospitals. New systems are being written by McDonnell Douglas and others to extend the scope of the medical services offered. Some are said to have substantial market potential. Medical Information Technology Incorporated serves about fifty hospitals with their patient care package. They serve a lesser number of hospitals with their more extensive turnkey system that includes some medical records, pharmacy, and laboratory functions.20 Technicon, a commercial system with reasonably comprehensive medical services has nine installations.21 The SPECTRA Systems of Medicus Corporation also provide a relatively wide range of medical information processing services. Six systems have been sold, with three other likely possibilities.22 Many factors complicate interpretation of the current status of commercial vendors. These include purchases of completed and semicompleted systems by vendors, marketing difficulties, changes in systems due to changing computer technology, and changes in system priorities to mirror changes in federal re- porting requirements. None the less, it appears that the marketplace has valued and accepted computers for business office applications in hospitals and has not valued highly the medically oriented aspects of MISs. In summary, the marketplace test seems to have concluded that the worth of MISs is slim or undiscovered. A large market for this service has not de- veloped. Operations Research These systematized empiric measurements have been employed both by system builders and by government-sponsored teams to evaluate MISs. With 80 Growth of MISs in the United States a few exceptions, the studies have demonstrated substantial worth of the MISs examined. Homer Schmitz examined the impact of an automated system on the work pattern of hospital employees.23 He used a process measure with before and after comparisons. He found that deployment of the system caused a reduction in time spent on the telephone and with paperwork, thus relieving nursing time for patient-care duties. Furthermore, he and his team showed that the reduction in the nonmedical duties of registered nurses was a direct result of greater efficiency of processing hospital services by the automated system. Likewise, Robert Richart, in studies of the MIS at Kaiser-Permanente hospitals, documented a similar favorable impact on work patterns.24 At the component or subsystem level of MISs there are many favorable reports. Gen- erally the studies use empirical methods of observation, take process measures, and center on institutional or departmental interests. Benefits to patients are often most easily implied through cost reduction or containment or through demonstrated improved accuracy or timing of diagnostic services. Charles Flagle summarizes a number of studies of information systems using operations re- search methods in the clinical laboratory. He notes that the primary impact of the computer [seems] to be one of increasing productive capacity without increasing staff Studies of the speed and accuracy of reporting laboratory results show some improvement . . . [but] do not reflect the strong sub- jective preference of staff for the new procedures.25 Many other positive evaluations of MISs are based on operations research methods. The comments by Flagle, although well and fairly stated, are repre- sentative of one of the problems inherent in this method of measuring worth. The operations research method is suited to detecting dramatic changes. Yet MISs frequently yield a host of relatively small improvements in many dif- ferent aspects of the operation. Flagle noted two such effects of MISs on clinical laboratory questions: the ability for a laboratory to sustain a 200 percent in- crease in workload with the same number of staff, and the subjective satisfaction that the staff takes in neatly printed, more timely reports. In a system whose value is manifested in many ways, each way may be small. Under contract to the federal government, Ronald Henley and Gio Wieder- hold conducted an extensive review of automated medical record systems for ambulatory patients.26 They operated through a multidisciplinary team of observers, established formal data collection and evaluation protocols ahead of time, and made on-site visits. They looked at outcome as well as process mea- sures. In spite of the difficulties in quantitation, they concluded in nine of the seventeen systems visited that the quality of care had been improved. They were not convinced that MISs affected initial access to care, although they Evaluating the Worth of MISs 81 found that “improved secondary access was a major benefit at many sites.” They did not find evidence of cost savings. They found consistent evidence of contributions by MISs in improved institutional management. Patient satis- faction was difficult to measure. Few of the systems they visited were imple- mented with research as the primary goal. None did patient education as an MIS activity. The results of this survey show clearly identifiable benefits, no negative benefits, and calculable dollar costs. They did not undertake a formal economic analysis, balancing the benefits against the costs. The most extensive evaluation to date of an MIS installation was performed at El Camino Hospital in California. Two studies were involved: one made by the hospital management engineering staff, and an independent study by Battelle Columbus Laboratories. Both the hospital and the Battelle study were made under contract to, and monitored by, the federal government. Three and four years, respectively, were required for the studies. Their cost was in excess of $1,200,000. Even taking into account the length and size of the evaluation effort, the studies were not and did not propose to be technology assessment studies. They were strictly within the tradition of operations research methodology. The summary plan for each of the two studies states plainly: “The major ob- jective was to evaluate the impact of [the Technicon MIS] on the organiza- tion and administration of health care delivery at El Camino Hospital.”27 El Camino Hospital Staff Evaluation. The study by the hospital staff identified a number of favorable effects of the MIS implementation.28 The evaluations focused on hospital departments, the institution, and cost savings. Nurses strongly favored the system, reporting that it benefited them in many ways, including a major reduction in clerical tasks, greater availability of patient data, improved legibility of patient records, and improved patient care planning. The admitting, pharmacy, radiology, and laboratory departments demonstrated improvements in the timeliness, completeness, accuracy, and availability of information, and reductions in clerical work. Physicians as a group were the least positive; at the end of the initial four-year study 61 percent voted to retain and extend the system. Subsequently, the roster of physician supporters rose to 80 percent.29 The evaluation of the MIS at El Camino also concludes that there were significant cost savings. These are noted in the section Cost and Economic Analyses of MISs. Battelle Laboratories Evaluation. An evaluation of the Technicon MIS installa- tion at El Camino Hospital was conducted by Battelle Columbus Laboratories under contract to the National Center for Health Services Research.30 The evaluating group was wholly independent of either the system vendor or the hospital management engineering staff. The Battelle study had two parts or 82 Growth of MISs in the United States objectives. First, like the study by the hospital, it aimed to measure the impact of the new system on the organization and administration of health care de- livery in the hospital. Second it aimed to conduct a formal cost-benefit study. The first part of the study was published in December, 1975. The cost-benefit study, although promised by mid-1976, was not released as of June, 1979. We should examine the conclusions of the first part of the study, even if briefly. In the following section on economic analysis we give consideration to the problems encountered in such an evaluation, and why the second study has been so greatly delayed. Battelle concluded that there had been a demonstrably favorable impact on the organization and administration of the hospital. Like the hospital manage- ment team, Battelle reported that information used for administering patient care and for monitoring patient progress was more complete and accurate as compared to the time before the MIS installation. The same was reported for administration of tests and procedures and for ancillary and supportive services. Most departments concerned with direct patient care had been able to reduce their staff complement or avoid adding staff when additional services were required. According to Battelle, the MIS was associated with better communica- tion and coordination among doctors, nurses, and supporting hospital depart- ments. Nurses were consistently strongly in favor of the system; the physician group was by the time of the report 78 percent favorable. (The hospital study one year earlier had reported 61 percent of the physicians favorable.) Confidentiality of patient records was deemed to be suitably in compliance with California statute. The inherent conflict in objectives between restricted access to patient care records through the information terminals and use of the system to facilitate direct patient care was noted. The Battelle study looked primarily at process measures. Generally the study was focused on hospital staff, departments, or the institution. Direct patient benefits were not looked for; only patient benefits that could be implied by improvements in the processes by which the hospital rendered services to the patients were considered. The study specifically notes that no consistent change in nursing time spent in direct patient care was observed and that no significant change in hospital patterns of practice by the medical staff was ob- served. Patients and families were not interviewed or surveyed. University of Alberta Hospital Study. A similar set of studies was reported concerning the efficacy of a subsystem admitting department and census system installation in two Canadian hospitals.31 The goal of the studies was to install and evaluate information systems in hospitals on behalf of the Canadian govern- ment. The emphasis of the efforts was on medical, not business office functions. The expectation of the system builders was that this subsystem installation would be only the necessary first step in creating a medically useful information system. It came as a surprise to them to discover after installation that the Evaluating the Worth of MISs 83 subsystem had both subjective benefits and measurable cost savings. The latter are noted under economic studies. Subjective benefits of the admitting and census subsystem were documented in this study by a structured questionnaire technique, measurement being made before and after system installations. The opinions of patients as well as hospital staff were sampled. Both unstructured comments and a formal index of satisfac- tion were computed. In addition, objective system measures such as patient waiting time were made. Calculated indices of satisfaction showed improvement after installation of the systems for all categories of respondents. Patients and medical staff were most satisfied. The indices were accompanied by observations such as . .. fewer complaints from patients regarding the time taken from bed request to being admitted and the length of time patients were required to wait in the Admitting area.32 This subjective appraisal was confirmed by timing studies which documented that 25 percent of the former processing time had been eliminated through use of the MIS subsystem. Another expression of user satisfaction with the system was directed to its facilitative function: Less information was being lost or was difficult to trace because of careless filing or clerical errors.33 A related measurement tending to substantiate this belief was the finding of a reduction in the number of undesirable second charts issued to patients on admission when their previous hospital records could not be identified. In addition to these semiquantifiable improvements in the hospital process, the studies also recorded the pleasure and pride of the hospital staff in pre- siding over a more orderly system: .. . hospital management has better control waiting lists can be monitored in a way not previously feasible . . . reduced the effort required to correct statistics and billing records because of inaccurate data concerning patient movement.34 In summary, this study of the computer-based admitting/census subsystem substantiates improvements in the processes of hospital care of patients, as well as the subjective sense of satisfaction created for staff and patients. Quantita- tive measurements document some of the benefits. In addition, the installation was shown to reduce costs, as noted in the section on economic analyses. 84 Growth of MISs in the United States A.D. Little Study at Latter Day Saints Hospital. A systematic study of com- puter-based patient monitoring systems was conducted by the A.D. Little consulting firm under contract to the Department of Health, Education, and Welfare (HEW). One of the systems studied was the thoracic surgery intensive care information system of the Latter Day Saints Hospital in Salt Lake City.35 The full MIS of which this is a part was described in Chapter 4 under the head- ing, “University of Utah”. A.D. Little studied the pattern of care for all open heart surgery patients admitted to this unit over a one-month period. The unit is one of ten in which the information system operates. The study is now a bit dated, having been concluded in 1973. Nonetheless, operations research methodology did document that the system was functioning on a regular basis and was used as an integral part of the care process for these complex patient problems. A substantial portion of technician time (15 percent) was observed to be expended in operating the computer terminals. The designers use specially trained technicians rather than nurses or physicians for installing and adjusting the monitoring equipment and the computer attachments. According to this and other reports, the sub- stitution of this new personnel type works well. The study recorded the use of the computer terminals over a one-month period, and divided the usage into several categories. Entering and reviewing comments accounted for 34 percent of the usage, and reviewing of data another 28 percent. These figures support the contention of others that communication between members of the health care team does indeed constitute a major hospital activity, and one that can be mediated through an automated information system. Twenty-eight percent of usage was for reviewing alarms set by the system. A continuous development- al program adjusts the sensitivity and specificity of the alerting criteria. At the time of the study, about one-third of the alarms were judged to be caused by significant medical emergency conditions. The number of uses of the informa- tion system for each patient varied, as did the time distribution of the usage. This was generally highest in the first hour of monitoring. Subsequently, usage fell gradually to an average of two times per patient per hour. Certain imper- fections were noted, including a relatively high level of computer down time because of system reloading, and failure of the systems statistical trend analyses to satisfy the nurse or technician most of the time. In brief, the study documented a running system, with much interaction, many alarms, and a substantial percentage of valid alarms. It is a serious limita- tion of this form of evaluation that it does not seem to have any inherent mechanism for reaching significant conclusions. The A.D. Little report avoids a conclusion by noting that control studies have not been done at Latter Day Saints Hospital, so that hard data on the effects of the computer system on length of stay in the unit and patient outcome are not available. Evaluating the Worth of MISs 85 Cost and Economic Analyses of MISs El Camino Hospital Cost Evaluation by Hospital Staff. Probably the most carefully substantiated benefit reported from the study of the impact on the MIS was a reduction in hospital costs. The project reported that The net cost savings after paying for the system’s cost are conserva- tively estimated to range from $3 to $5 per patient day. These favor- able savings continue to increase with time.36 This amounted to $30,000 to $50,000 per month in savings to the hospital.37 Ninety-five percent of the saving was in labor, the major part of which was nursing salaries and wages. This amounts to 0.8949 hours per patient day. In addition, there was a reduction in the average length of stay of patients in the hospital during the three and one-half years following introduction of the information system. The relatively small percentage of reductions in patient stay, in nursing hours required per patient, and in nursing usage costs (2 to 4 percent) are significant when compared against the trend in comparable hospitals in the region and nationally. Elsewhere there were consistent comparable trends in the opposite direction. The hospital management felt that the information system helped to reverse the trend to higher costs. This was noted after a 1972 to 1973 study,38 and even more strongly after a 1975 study.39 A peculiarity of the contractual arrangement for this MIS lends credence to the conclusion of cost savings. According to the installation contract, all costs were to be borne by Technicon, the vendor, who was to be paid a fee only for savings generated by the MIS that were confirmed by all parties. The most revealing statement of the evaluation study relates to the diffi- culty of such appraisals. The features commented on by Flagle were also ob- served and disregarded in the El Camino study: reduced errors, improved time- liness, and enhanced availability of medical information. The El Camino report notes: Because those benefits are difficult to quantify, they are excluded here. Hence, the favorable economic conclusions are believed to be substantially understated.40 El Camino Hospital Cost Evaluation by Battelle. The cost and economic study was promised by mid-1976 but not ready three years later. Arrangements be- tween a federal agency and two large private institutions are complex. It would probably not be useful to review the exact details of this particular troika. It is, however, worth noting those problems at El Camino that are inherent in any such economic evaluation of a not-for-profit medical institution. 86 Growth of MISs in the United States First, the methods of costs analysis are inherently cumbersome; namely, to measure all costs and benefits, direct or indirect, small or large, and to express all in common terms, usually monetary. Any hospital like El Camino, an institution with 464 beds in service, is complex. It takes a lot of people to measure performance and systems impact. If the evaluators who make the measurements must be independent of the system so as to remove potential biases, all of them will be an extra cost. Hence, the second problem: formal evaluations are expensive. Last, and most critical, considerable time is required to accomplish the baseline measurement studies, the installation of the MIS itself, and a repetition of the measurement studies. During the four years required at El Camino the medical environment changed according to needs unrelated to the MIS experi- ment. Specifically, during the course of the Battelle study, the following changes occurred in the hospital: the addition of transitional care and intensive care beds, the initiation of an ambulatory surgery service, the addition of cardiac surgery, the introduction of the total hip replacement operation, and the institu- tion of kidney dialysis services. Roughly the same changes would be expected in any sizable community hospital during this time period. Indeed, the list is short. It could easily have included greatly increased use of emergency rooms for after-hours care, a general reduction in usage of obstetrical beds, increases in abortion and family planning services, a shifting of financing patterns for com- munity mental health services, the introduction of computer tomography machines in radiology, a newborn intensive care unit, increasing use of radio- nuclides in the clinical laboratories, and general inflation of all costs. Sorting out these effects is difficult. The basic task for the economics analyst in this situation is not to document increased sales (as it might be in an industrial setting), but rather to calculate expenditures that were not made but that would have been required were the MIS not in place. It is this formulation of the problem which explains both delays in such studies and the possibility of more than one interpretation of a set of measurements. University of Alberta Hospital Study. Economic savings were reported from the 1,200-bed Unversity of Alberta Hospital in connection with use of its computer- assisted admitting system.41 This part of the study was an informal cost- effectiveness measurement, comparing an automated method with a previous manual method. Since it was applied to an MIS subsystem, the method was reasonable. Savings produced by the system amounted to $36,200 per year. Of this $34,000 was in salaries of full and part-time personnel no longer re- quired. The job categories included admitting clerk, information clerk, key- punch operator, accounting clerk, and medical records personnel. The savings may be minimal, since no space, benefits, or overhead seem to have been in- cluded. Annual operating costs associated with use of the system totaled Evaluating the Worth of MISs 87 $29,000. This included $14,200 in hardware costs and $14,800 in personnel costs. The minimal net saving was therefore reported to be $7,200 per year. Some assumptions are necessary in such studies. All costs for admitting terminals and 20 percent of central computer costs were allocated to this sub- system on the basis that it consumed computing resources at this level. The researchers assumed that the remaining central computing costs would ulti- mately be totally allocated to other subsystems. It is also not clear how the evaluators handled the question of amortization of the computer investment, nor why savings in personnel overhead were not included. Another potentially interesting question is how to classify program maintenance costs (which were included in the $29,000). Assuming that the evaluators did successfully separate developmental from operating costs, as they stated they did, one could still consider the working system of programs itself as a thing of value. Equity in systems has been realized by a number of U.S. medical institutions which have sold their systems, as commercial products. Even short of this, a working system is inherently of substantial worth to its institution. In summary, this admitting/census system appears to be more than paying its own way at the University of Alberta Hospital, even without considering estimates of medical benefits. Analytical Services Study of Air Force Clinical Laboratory System (AFCLAS). Not all evaluations of clinical computing systems have yielded favorable re- sults. Analytic Services, Inc. performed a formal evaluation under contract to the Office of the Surgeon General of the Air Force on a commercial clinical laboratory system. The evaluation was preliminary to dissemination of the system throughout the service. Its designation, AFCLAS, was changed to TRILAB during the study as a token of consideration by the TriService Medical Information Systems (TRIMIS) office as a candidate for dissemination through all three armed services. It was a major outcome of the adverse evaluation that the TRIMIS office accepted the studies and immediately halted plans to pur- chase such systems. The studies themselves are fully described in formal reports approved for public release 42 _44 The computing system was installed at Wright-Patterson Air Force Base in Ohio and at Malcolm Grow USAF Medical Center at Andrews Air Force Base, Washington, D.C. The evaluation plan identified fifty-eight areas of potential impact of the new system when introduced in place of the existing manual in- formation systems. These areas included economic and non monetary aspects of clerical tasks in the laboratory and other parts of the institution, the complete- ness and quality of the medical records, the timeliness of laboratory reporting, and acceptance by and satisfaction of medical personnel and patients. Studies were performed before and after system installation. 88 Growth of MISs in the United States The cost-benefit analysis following studies at Wright-Patterson showed that the cost of operating the laboratory was $382,123 per year more with the automated system, to which had to be added $91,631 in initial purchase price. Some nonmonetary benefits were demonstrated in the areas of legibility and searchability of records. An amazing nonbeneficial result was the almost uni- formly greater time required for delivery of laboratory reports with the auto- mated system. Results at the Malcolm Grow installation were similar. There are a number of things to be learned from studying these reports. The most obvious is that honest and independent evaluations will identify bad information systems, and that the cost of such evaluation is small when it prevents the purchase of scores more of the system under evaluation. Further- more , the studies present a litany of virtually every possible mistake to be made in implementing such a system. Some of these errors derive from the inherent difficulties of the incredibly cumbersome process of procurement that federal offices seem legally compelled to use in' purchasing computing systems. Yet many of the errors and problems are similar to those noted in chapter 6 by investigators outside the federal government. The following observations are paraphrased from appendiceal material by Richard Brooks: 1. Implementation plans should include a means of preparing hospital staff for the changes inherent in using the automated system. 2. Ordinary laboratory operations must be susceptible to slight modification to utilize the new capabilities. 3. The new system should have a manager who knows both laboratory medi- cine and data processing. 4. Better vendor support of the system, especially for software changes, is needed. 5. The system should be used to perform more functions, including non- laboratory functions. 6. The MIS design should consider the impact on overall hospital operation. 7. Systems such as AFCLAS should be looked on as being in a developmental phase, not as if they were turnkey systems.45 8. Numerous other relevant observations resulted from the evaluation but relate very specifically to the local situation and system features. Deaconess Hospital. A number of economic studies have been reported for the MIS installation at Deaconess Hospital in St. Louis.46'49 In many respects the system does not go so far medically as others, concentrating on reporting of charges at the time of ordering and administrative and management reporting of various kinds. It does, however, provide for laboratory results reporting. One economic benefit is noted for this installation which we have not mentioned so far. Schmitz reported that the installation resulted in a reduction in the Evaluating the Worth of MISs 89 turnover of accounts receivable from 61.34 days to 38.69 days.50 The effect on patient care would naturally be negligible, but the effect on hospital cash flow represented a one-time inflow to the organization of $1,500,000. This figure represents short-term interest gained by investment or short-term in- terest avoided by not having to take short-term debt. Even though one-time in nature such a cash flow could easily pay in one year for purchase of the computing hardware required by the system for the next eight to ten years. Amortization of the equipment is needed to make available the capital to replace the computing equipment as needed in the future. A gain in cash flow of this sort is premised on two new viewpoints in hospi- tal administration. First, no such gain can be realized unless skilled and atten- tive managers convert the potential gain to a real gain by the appropriate financial operations. At Deaconess, for example, they must actually make the investments. Second, this businesslike way of looking at hospitals is in itself rather new. Communities and hospital boards of directors have always expected their med- ical institutions to be run “in a business-like way”, but they have not until recently expected the level of sophistication in management implied by this kind of economic analysis of MISs. Perhaps the moment the handwriting ap- peared most clearly on the wall was on March 1, 1973. The clear warning was contained in the testimony of Casper Weinberger, then Secretary of HEW, when he recommended termination of the government’s Hill-Burton program of financial assistance for hospital building.51 Mr. Weinberger announced to the Subcommittee on Interstate and Foreign Commerce of the United States House of Representatives that the Administration intended to eliminate the long-stand- ing Hill-Burton program in support of hospital and health facilities additions and construction. He offered two general sets of reasons for closing down Hill- Burton. First, that “the country now has a generally adequate supply of hospital beds. In fact, in some parts of the country, a definite surplus in hospital beds now exists. ..” Other groups have subsequently concurred with him in this view.52 Relevant to the information management problem in hospitals was the second category of reasons offered by the Secretary: . . . the rise of health financing systems, such as private and public health insurance plans, has made a fundamental change in the way that hospitals do or should do their business [italics added]. Nearly all types of health insurance, including Medicare and Medicaid, Blue Cross-Blue Shield, and other private plans, recognize depreciation as a valid component in the reimbursement of hospital and medical facilities. This aspect of current medical care financing should permit [italics added] hospitals and medical facilities to either set aside funds for facility improvements or pay back loans for construction over the useful life of the facility. ... A special Federal grant program for hospital construction is now unwarranted.53 90 Growth of MISs in the United States This attitude places an enormous accounting and management responsibility on hospital administrators. They must either include in each procedure charge the cost of equipment and facility obsolescence, or package into the per diem costs the cost of what Weinberger termed “depreciation” and “facility improve- ments.” These matters are to be attended to by all hospitals, including small community hospitals. Hospitals were told not only to behave responsibly, but to behave like the federal government. No wonder hospitals gladly embraced the prospect that commercial vendors could provide accounting and fiscal management through computer-based information systems. For a hospital administrator to be able to obtain cost savings through pur- chase or rental of a computer system is most important when we consider the limited education for administrative problems that current chief executive officers of hospitals have had. A national study in 1972 concluded that only 35 percent of hospital chief executive officers surveyed in a random sample held a master’s degree in health care administration.54 Furthermore, the per- centage of top administrators with master’s degrees had not increased in the previous eleven years. By 1976 a more encouraging picture was reported. Health services administrators holding positions in hospitals who held graduate degrees in some form of health services administration had increased to 40 percent.55 Even more encouraging was the estimate that of the new entries into hospital administration, approximately 95 percent have master’s degrees. The hospital administration field is now responding to the need for much greater management sophistication. The majority of hospitals, however, are directed by individuals trained or recruited in a previous era. They will continue to opt for information systems that give them relief in business office manage- ment problems. The information systems, in turn, will continue to produce potential savings requiring management skills to capture. Technology Assessments There have been no comprehensive technology assessments aimed specifically at MISs. No studies have attempted to encompass the effects of such systems on the immediate participants and the indirect and long-term social, legal, and ethical effects on other groups or on society as a whole. Limited technology assessments have examined the data privacy aspects of MIS development. Westin Privacy Studies. Some of the most relevant studies of broad societal effects of computer-based information systems in the health care system are those led by Alan F. Westin. He and his associates have investigated the im- plications of these systems with respect to citizens’ rights and focused rather strongly on actual and potential violations of the rights of privacy.56 The po- tential danger to medical privacy stemming from practices of the life insurance Evaluating the Worth of MISs 91 industry has been commented on by Alan Westin57 and others in legislative hearings.58 The Fair Credit Reporting Act of 1973 regulated credit reporting activities but exempted medical data. Subsequently, the Medical Information Bureau (MIB) representing 700 life insurance companies agreed to place itself in com- pliance with the act. MIB agreed to provide a written description of their information-sharing policies to clients in the future at the time such medical information is solicited. The Westin studies looked for actual and potential dangers from computing systems in all domains of society. Medicine and health care were included in the various investigations, but not as a major focus of the initial studies. Wes- tin’s team visited the Kaiser-Permanente Medical Group program in 1972. They found that there had been no known unauthorized access to the computer files, and that any irregularities in release of information had involved the manual files. In short, the MIS at Kaiser-Permanente had created no data privacy problems. Westin notes that, even in 1972: Compared to the Scandinavian countries, the U.S. has moved slowly both in the development of comprehensive preventive health care systems and in the computerization of medical records, laboratory testing, and patient monitoring.59 The report concluded rightly that very few medical institutions have moved as far as (or beyond) Kaiser. Secretary’s Committee on Personal Data Systems. Independent of the Westin studies, Secretary Richardson of HEW established an Advisory Committee on Automated Personal Data Systems in February, 1972. The Committee noted, in addition to the latest Westin report which had been conducted as a project by the National Academy of Sciences, Computer Science and Engineering Board,60 the recent publication of parallel studies and reports from Canada,61 and Sweden.62 The Committee was asked to analyze and make recommenda- tions concerning possible harmful consequences of using computer-based per- sonal data systems, means to prevent any potential harm, and means for redressing wrongs. In addition to these global considerations, a particular question was given to them; namely, What should be the policy of the Depart- ment of HEW concerning the use of the Social Security number as an identifier in such systems? The investigations of the Committee included testimony from more than one hundred witnesses from more than fifty different organizations, plus correspondence with 250 trade and professional associations and public interest groups. The Committee did not make site visits to computer installa- tions. In June, 1973 the final report was rendered to the Secretary (by this time, 92 Growth of MISs in the United States Mr. Weinberger).63 This document summarizes and indexes an enormous volume of information and efforts. The essence of the report was not to point to any damage to patients as a result of MISs or any other form of computer-based personal medical record keeping, but to emphasize some serious potential dangers. In their most general form the dangers were expressed in the follow- ing observation: . .. the net effect of computerization is that it is becoming much easier for record-keeping systems to affect people than for people to affect record-keeping systems.63 In addition, it was noted that Computerization creates a new class of record keepers whose functions are technical and whose contact with the suppliers and users of data are often remote.63 Potential benefits to the individual and society can arise from one’s par- ticipation in a computer-based MIS. To do so, as even with a manual system, involves voluntarily giving up some privacy. How else could one ever obtain medical treatment? The Committee concluded that while this might be a fair enough quid pro quo, in the case of the new computer technology additional legal protections ought to be provided. The Committee felt that “under pre- sent law, a person’s privacy is poorly protected against arbitrary or abusive record-keeping practices.” The Committee’s report recommended enactment of a federal Code of Fair Information Practice. The Privacy Act of 1974 turned out to follow the terms and even words of this report quite closely.64 The Fair Practices code was to be based on five principles that formed the basis for the subsequent law. These do not conflict in any way with the fundamental medical, scientific, or management aims for MISs. The principles are: 1. There must be no personal record-keeping system whose very existence is secret. 2. There must be a way for an individual to find out what information about him or her is in a record and how it is used. 3. There must be a way for an individual to prevent information about him or her that was obtained for one purpose from being used or made avail- able for other purposes without his or her consent. 4. There must be a way for an individual to correct or amend a record of identifiable information about him or herself. 5. Any organization creating, maintaining, using, or disseminating records of identifiable personal data must assure the reliability of the data for Evaluating the Worth of MISs 93 their intended purpose and take precautions to prevent misuse of the data. The Privacy Act of 1974, insofar as it applies to MISs, is enforced in federal systems. If this law were to be extended to oblige commercial MISs in the civilian community to include such facilities, a very new evaluation parameter would emerge. It is not the present practice to announce the existence of auto- mated medical record keeping systems to the public or to the patient involved. It is definitely not the present practice to offer the patient an opportunity to amend or correct the medical records that describe his or her health and ill- nesses. Such procedures are in no conflict with the fundamental aims of MISs, but they have not yet been implemented in civilian medicine. The implications and potentially desirable aspects of such a change have been discussed in a number of forums. Two examples do not adequately reflect the quality or extent of discussion but may give a sense of the problems involved. Budd Shen- kin has written of the inherent advantages to all parties of presenting to each patient a complete copy of his or her medical record. He expects that . . . giving patients their records might be more effective in monitoring the day-to-day quality of care that physicians give than all those sophisticated methods that have been devised. The sophisticated methods all depend on a centralized influence, and on formal measures to be taken. By contrast, giving patients their records would initiate an informal and decentralized way of quality control.65 Elsewhere Shenkin argues: Despite the fact that the Protocol promotes free communication and patient autonomy, it would not interfere with centralized reforms, such as Health Maintenance Organizations and Professional Standards Review Organizations.66 These arguments are premised on medical grounds, as are the various objec- tions raised to the suggestions. In any case, the use of MIS was not inherent in the discussions. Yet to give a patient his or her record would be compatible with the Privacy Act provisions directed toward the operation of automated record keeping systems. Another set of problems enters into consideration; namely, the added cost and complexity of accomplishing this feature once an automated system has been produced. In discussing the implications of the federal and state privacy bills for institutions, Edward Bloustein, the president of Rutgers University, endorsed the basic intent of privacy protection requirements and emphasized the diffi- culties that might be involved in complying. His university held both student and patient records in automated information systems. He notes: 94 Growth of MISs in the United States Next the Ware Report . .. suggests a way be provided for an individual to find out what information about him is in a record and how it is used. Here we are making very primitive attempts to do this, but I don’t think anyone should underestimate both the cost and the great sophis- tication of technique that you have to undertake to do this effectively. We’re all going to do it. My theory is that we are going to do it in form but not in substance because the cost of it may well be too high to do it effectively.67 The Ware Report dealt quite directly with the specific question addressed to the Advisory Committee to the Secretary, although the effect was much less enforceable than in the case of the Fair Practices recommendations. The report recommended that use of the Social Security number be limited to federal programs that have a specific federal legislative mandate to use the Social Security number, and that new legislation be enacted to give an individual the right to refuse to disclose his Social Security number under all other cir- cumstances. In summary, the Secretary’s Committee took a broad view of their charge. They emphasized potentially abusive practices and recommended legal protec- tion against them. The dangers and problems they identified partook in great measure of the effects on society of computer automation in general. They did not deal extensively with the question of positive medical benefits arising from proper medical use of such systems. Factors of economic cost were not much covered. The report essentially deals forthrightly with the potentially negative effects of the technology of MISs on societal groups. Presumably this effort would be considered a mini technology assessment, since it covered in depth a restricted aspect of the MIS technology impact; namely, its implications for data privacy. Scientific Impact General Studies. Formal studies of the scientific impact of MISs have not been conducted. Many interesting publications have been produced that deal with panoramic concepts such as the information explosion, the computer revolu- tion, the impact of computers on education, and on society generally. Technical volumes have presented composite views of usage in human affairs,68 medi- cine,69 life science research,70 hospitals,71 and libraries.72’73 It is impossible to extract or deduce conclusions concerning the scientific impact of MISs from these writings, although one can see in many of these reports a generally posi- tive and anticipatory view of the matter. It is clear that the existence of power- ful computing systems has been expected to make certain new kinds of clinical studies possible such as studies of large numbers of people, or studies involving large amounts of data on modest numbers of people. Computing power has been Evaluating the Worth of MISs 95 seen as making plausible the investigation of certain ideas such as multicausality of disease, genetic engineering, and artificial intelligence. Coupled to these concepts, MISs have been thought to create the possibility of carrying out cer- tain complex analyses including calculation of exact prognoses for individual patients, or execution of treatment algorithms. In the category of new vantage points achieved by use of MISs, one should mention an as yet incomplete piece of work. The Computer Based Examina- tion (CBX) project of the National Board of Medical Examiners and the Amer- ican Board of Internal Medicine is attempting to develop a computer-based system for certifying the clinical competence of candidates for certification in internal medicine. The project assumes that computer-accessible records of patient care will be available as the basis for determining the deficiencies and competencies of health practitioners of the future. Certifying instruments so far developed and field tested utilize a computer simulation of patient attributes and outcomes to treatment against which the candidate operates as if in an actual medical management setting. The funda- mental concept is of a large number of well-managed patients represented within the computer, and the user interacting with the system so as to relive the course of the treatments based on his or her own judgments and compe- tence. These judgments are then measured against the desirable standard of competence derived ahead of time from peer concensus. The advance of this idea over previous schemes for automated testing is primarily that the peer concensus is strongly determined by the actual outcome for the patients whose records form the basis of the simulations. In this sense the standard is deter- mined more by actual facts of patient care and less by opinions concerning process measures. The idea is not unique to medicine. Airline pilots are routine- ly trained on flight simulators which they “crash” and “save.” What makes the idea plausible in medicine is the possibility that MISs could provide the detailed information about numbers of patients, which would serve in place of the physical models of the airplane. Reports on this important project have been issued.74’76 The importance in the present context is that the MIS concepts have stimulated the synthesis of a new set of concepts. Finally, there has been a partially defined expectation that the computer- based MISs would make possible the discovery of some new and important conclusions concerning human bodily and mental behavior that would give an objective basis for patient management, at least for certain specific situations. Survey evidence indicates that one information system did create the perception that scientific benefits follow from use of the system in clinical research. RAND staff and clinical collaborators reported an evaluation of the CLINFO system previously noted in chapter 3.77 Thirty scientific users of CLINFO were surveyed before and after they employed the system in three clinical research centers. After using the system, 100 percent felt that the 96 Growth of MISs in the United States system provided greater insight into the relevant relationships than was possible without the automated system, and 89 percent felt the insight came earlier. Ninety-six percent felt that more questions could be answered from the data available, and 96 percent felt there were fewer data manipulation errors. In con- trast, most did not feel that the system made possible the use of fewer experi- mental subjects or fewer test samples per subject. Studies of the scientific impact of MISs could have been done on many system implementations. But they have not been done. One scientifically advanced medical application has been heavily influenced by the existence of MISs. The connection between the advanced clinical ideas and the underlying MIS concepts is so direct that one can see the effect of MISs in stimulating the scientific imagination to formulate a new viewpoint toward the problem of therapeutic decision making. Duke Prognostigram. The work of Eugene Stead and colleagues Andrew Wallace, Frank Starmer, and Robert Rosati at the Duke University Medical Center de- fined its goal as the production of an electronic textbook of medicine.78 By this they mean the provision to physicians at the time a patient management decision is to be made of information from the computer system relevant to the decision. Stead’s aim is not to structure current knowledge in a more convenient fashion, but to make available to practicing physicians new knowledge that could not come to them by any books, scientific journals, monographs, or tables now in existence. The most advanced “chapter” of the new electronic textbook deals with coronary arterial disease and myocardial infarction. The chapter allows physicians caring for patients with chronic heart disease to have the same kind of feedback via the computer concerning patient status that they have without the computer (that is, simply by observing) when they care for patients with acute illnesses. The Stead system presents the physician with carefully collected, validated, and analyzed information concerning the outcome, with either surgical or medical treatment, of the patients treated at the Duke Medical Center. The report is highly particularized to the patient under consideration. The records selected for display are those of previous patients similar to the new patient. There is a statistical basis for describing the precise ways in which the patients are similar which takes into account various attributes and measurements of the patients. The entire project hinges on a computer-based MIS that collects and con- tains the data of patient examinations, treatments, and outcomes. The carry- ing of the complete description of the patient and the linking of this description to known outcomes cannot be done by a paper and pencil system. Operation of the MIS to collect the information has been separate from the general hospital system in the past. Recently the Duke Medical Center abandoned use of a commercial MIS system, the Medidata System of Burroughs, Evaluating the Worth of MISs 97 and installed a hospital-developed system using IBM equipment. The special coronary artery disease MIS subsystem with its own minicomputers is now being integrated into the new general institutional MIS scheme. New program- ming language development has been undertaken so that both systems can utilize the same language, a superset of the PL/1 language. The rearrangement offers strong evidence that an MIS is central to this advanced concept and that compatibility with a general institutional MIS has been worth the substantial efforts involved. The operation of the computer-based information collection, its organiza- tion into an electronic textbook form, and the clinical interfacing produces for each patient a document called a Prognostigram. In one Prognostigram, a search of the files for patients with the following criteria produced a cohort of 153 similar patients: Criteria: No cardiomegaly by chest x-ray No history of congestive heart failure Three-vessel coronary artery disease Abnormal left ventricular contraction pattern Left ventricular contraction pattern not diffusely abnormal No left ventricular aneurysm Arteriovenous oxygen difference less than or equal to 5.5 No mitral insufficiency Of these, the Prognostigram identified sixty-four patients who had been man- aged with surgery and eighty-nine who had been managed by medicine alone. For both groups, the doctor (and presumably the patient as well) could see the relative numbers of patients who at the end of time periods up to five years after surgery were alive and those who were dead. In addition, the report showed the interval incidence of complete relief from pain in both groups of patients and the interval incidence of a myocardial infarction. The computer report does not tell the physician or the patient whether to pursue a surgical or medical treatment. It does present all the information available from the institution so as to make that decision a fully informed and intelligent one. The final outcome of using this approach is not yet certain. Discussion and evaluation of all the medical considerations of this approach and adjudication of the results of Duke treatment versus the approaches of other institutions is beyond the scope of this book. Briefly, the objective results of this decision- making scheme appear very desirable. The developers claim separation of 98 Growth of MISs in the United States questionable and seriously ill patients, avoidance of large numbers of unneces- sary surgical procedures, and reduction in stay and costs for hospitalization. Experiments are underway to incorporate into the system an excellent tertiary care institution in another region of the country. A final economic note should be sounded. The clinical studies that form the basis for the Prognostigram cost roughly $1,200 for each patient. Evidence from comparison with other institutions indicates that more than half the patients so studied would be operated on for relief of their angina had the Prognostigram studies not been available. The Prognostigram identifies only about 25 percent of the patients who will obtain benefit from surgery. Consequently, only about 25 percent of the persons actually receive a recommendation to undergo coronary bypass to treat the coronary artery problem. Each surgical procedure costs roughly $14,000. Recognizing the value of computer-based studies of this kind, the local Blue Cross health insurance plans have agreed to pay for the performance of the computer studies as a procedure. To summarize, this example offers evidence that the scientific impact of MISs can provide the technical infrastructure for scientifically advanced con- cepts in medicine. There is reason to be optimistic that the advanced concepts can produce cost savings in the health care delivery system and improved patient care. Notes 1. G.O. Barnett, “Massachusetts General Hospital Computer System (Boston)”, in M.F. Collen, ed. Hospital Computer Systems (New York: Wiley and Sons, 1974), p. 517. 2. D.A.B. Lindberg, The Computer and Medical Care (Springfield, 111.: Charles C. Thomas, 1968). 3. A.F. Westin, Computers, Health Records and Citizen Rights, National Bureau Standards Monograph 157 (Washington, D.C.: National Bureau Stand- ards, 1976). 4. M.F. Collen, L.G. Dales, G.D. Friedman, et al.: “Multiphasic Checkup Evaluation Study 4. Preliminary Cost Benefit Analysis for Middle-Aged Men,” Preventive Medicine, vol. 2,1973, p. 236-246. 5. S. Ramcharan, J.L. Cutler, R. Feldman, et al.: “Multiphasic Checkup Evaluation Study 2. Disability and Chronic Disease after Seven Years of Multi- phasic Health Checkups,” Preventive Medicine vol. 2, 1973, p. 207-220. 6. A. Donabedian, “Needed Research in the Assessment and Monitoring of the quality of Medical Care,” NCHSR Research Report Series, July 1978, pp. 1-37. 7. A.D. Little, Inc., Introduction to Use of Cost-Benefit Analysis in Planning for Emerging Health Care Technologies (Cambridge, Mass.: Arthur D. Little, Inc., February 1977). Evaluating the Worth of MISs 99 8. H.E. Klarman, “Application of Cost-Benefit Analysis to the Health Services and the Special Case of Technological Innovation,” International Journal of Health Services, vol. 4,1974, p. 325. 9. United States Congress, House. Committee on Science and Astronau- tics. Subcommittee on Science, Research and Development. Technology Assess- ment. Statement of Chairman, Emilio Q. Daddario (90th Congress, first session: 1967). 10. S.R. Arnstein, “Technology Assessment: Opportunities and Obstacles,” I.E.E.E. Transactions on Systems, Man and Cybernetics SMC-7, August 1977, pp. 572-573. Reprinted with permission. 11. Ibid., p. 572. 12. S.R. Arnstein and A.N. Christakis, “Summary of Assessors’ Perspec- tives,” in S.R. Arnstein and A.N. Christakis eds. Perspectives on Technology Assessment (Jerusalem, Israel: Science and Technology Publishers, 1975), p. 155. 13. S.R. Arnstein, “Technology Assessment: Opportunities and Obstacles,” p. 577. 14. F.P. Lawrence, personal communication (Denver, Colorado: Medical Group Management Association, February 21, 1979). Reprinted with per- mission. 15. Hospital Financial Management Association, The State of Information Processing in the Health Care Industry (Chicago: Hospital Financial Manage- ment Association, 1976). 16. P.G. Duffy, personal communication (St. Louis, Mo.: McDonnell Douglas Automation Co., February 25, 1978). Reprinted with permission. 17. P.G. Duffy, personal communication (St. Louis, Mo.: McDonnell Douglas Automation Co., January 26, 1979). Reprinted with permission. 18. McDonnell Douglas Automation Co., McAuto Hospital Services Over- view (St. Louis, Mo.: McDonnell Douglas Automation Co. MDC No. 826-077, 1977). 19. Duffy, personal communication, January 26, 1979. Reprinted with permission. 20. E. Pisinski, personal communication (Boston, Massachusetts: Medical Information Technology, Inc., February 15, 1979). Reprinted with permission. 21. M.R. Traska, “Methodist of Indiana Tailors Patient Computer System to Hospital Routine,” Modem Healthcare, October 1978. 22. Medicus Systems Corporation, Spectra 2000 Medical Information System (Chicago, 111.: Medicus Systems Corporation, 1977). 23. H.H. Schmitz, “Productivity Effectiveness: It Can Be Done in the Health Care Field,” Proceedings Ninth Annual Society of Management Informa- tion Systems (Chicago, 111.: Society for Management Information Systems, September 1977), p. 1. 24. R.H. Richart, “Evaluation of a Hospital Computer System,” in M.F. Collen ed. Hospital Computer Systems (New York: Wiley and Sons, 1974), p. 341. 100 Growth of MISs in the United States 25. C.D. Flagle, “Operations Research with Hospital Computer Systems,” in M.F. Collen ed. Hospital Computer Systems (New York: Wiley and Sons, 1974), p. 427. Copyright 1974 by John Wiley and Sons Inc. Reprinted by per- mission of John Wiley & Sons, Inc. 26. R.R. Henley and G. Wiederhold, An Analysis of Automated Record Systems: Vol. 1: Findings, Vol. II: Background Material (San Francisco, Calif.: University of California San Francisco Medical Center), June 1975, NTIS #PB 254-235, pp. 254-236. 27. J.P. Barrett, R.A. Barnum, B.B. Gordon, et al.: Evaluation of the Imple- mentation of a Medical Information System in a General Community Hospital. Final Report, December 19, 1975, Battelle Columbus Laboratories. Contract HSM 110-73-331, Department of Health, Education, and Welfare, Health Re- sources Administration, National Center for Health Services Research. NTIS Publication No. 248340. 28. J.E. Gall, D.D. Norwood, M. Cook, et al.: “Demonstration and Evalua- tion of a Total Hospital Information System,” Final Project Report, December, 1975. El Camino Hospital, Mountain View, California, Contract HSM 110-71- 128, Department of Health, Education and Welfare, Health Resources Adminis- tration, National Center for Health Services Research. 29. M.H. Hodge, personal communication, January 31, 1978. Reprinted with permission. 30. Barrett et al.: Evaluation of the Implementation of a Medical Informa- tion System in a General Community Hospital. 31. M.R. Heard and J.C. Thomas, “A Hospital Information System, Its Impact on Costs, Personnel and Patients in One Department,” in Cost Contain- ment, Caps and Consumerism within the Health Care Delivery System, vol. 2 (Chicago, 111.: Center for Hospital Management Engineering of the American Hospital Association, 1978). 32. Ibid., p. 31. Extracted with permission. 33. Ibid., p. 31. Extracted with permission. 34. Ibid., p. 33. Extracted with permission. 35. A.D. Little, Inc., Evaluation of Computer-Based Patient Monitoring Systems: Final Report, Appendix D. A Review of the MEDLAB System in Thoracic Surgery Intensive Care Unit at Latter Day Saints Hospital (Rockville, Md.: DHEW, March 1973, NTIS PB 247 421). 36. J.E. Gall, D.D. Norwood, M. Cook,et al.: “Demonstration and Evalua- tion of a Total Hospital Information System,” Final Project Report. (Mountain View, Ca.: El Camino Hospital, 1975. NTIS #PB-262 106). 37. J.E. Gall, Statement of El Camino Hospital, Mountain View, Cali- fornia, Before the Council on Wage and Price Stability (Mountain View, Calif.: El Camino Hospital, August 10,1976). 38. Ibid., figure 4. 39. Ibid., figure 5. Evaluating the Worth of MISs 101 40. J.E. Gall, D.D. Norwood, M. Cook, et al.: “Demonstration and Eval- uation of a Total Hospital Information System,” Final Project Report, Decem- ber, 1975. 41. Heard and Thomas, “A Hospital Information System, Its Impact on Costs, Personnel and Patients in One Department,” p. 32. 42. R.C. Brooks, R.G. Carlisle, I.J. Casey, et al.: Evaluation Plan for the Air Force Clinical Laboratory Automation System (AFCLAS) (Falls Church, Va. Analytic Services, March 1977), pp. 1-264. 43. R.C. Brooks, I.J. Casey, P.W. Blackmon, et al.: Evaluation of the Air Force Clinical Laboratory Automation System (AFCLAS) at Wright-Patterson USAF Medical Center Vol. 1: Summary (Falls Church, VA.: Analytic Services, March and May, 1977, pp. 1-44. 44. Richard C. Brooks, I.J. Casey, P.W. Blackmon, et al.: Evaluation of the Air Force Clinical Laboratory Automation System (AFCLAS) at Wright- Patterson USAF Medical Center Vol. II, Analysis (Falls Church, Va.: Analytic Services), May, 1977, pp. 1-356. 45. Ibid., Appendix K. 46. H.H. Schmitz, “Evaluation of the Health Information Systems,” CRC Critical Reviews in Bioengineering, vol. 2, January 1975, p. 209. 47. H.H. Schmitz, “An Evaluation of the Immediate Financial Impact of the Hospital Information System at Deaconess Hospital,” Health Information Systems Evaluation. Procceedings of the Symposium on Health Information Systems Evaluation (Boulder, Colorado: Association University Press, August 1973), p.265. 48. H.H. Schmitz, “An Evaluation of a Modular Hospital Information System ” Hospital Progress, June 1972, p. 70. 49. H.H. Schmitz, R.P. Ellerbrake, and T.M. Williams, “Study Evaluates Effects of New Communication System,” Hospitals, vol. 16, November 1976, p. 129. 50. H.H. Schmitz, “An Evaluation of the Immediate Financial Impact of the Hospital Information System at Deaconess Hospital,” Proceedings of the Symposium on Health Information Systems Evaluation, 1973, p. 265. 51. United States Congress, House, Committee on Interstate and Foreign Commerce, Subcommittee on Public Health and Environment, “Statement of Hon. Caspar W. Weinberger, Secretary, Department of Health, Education, and Welfarein Hearing on Oversight over Existing Health Programs Admin- istered by the Department of Health, Education, and Welfare (93rd Congress, 1st session, March 1,1973) p. 10. 52. United States Congress, House. Joint hearing before the Committee on Ways and Means; Subcommittee on Health and the Committee on Inter- state and Foreign Commerce, Subcommittee on Health and the Environment. Statement of Joseph A. Califano, Secretary of Health, Education and Welfare (95th Congress, 1st Session, May 11,1977). 102 Growth of MISs in the United States 53. United States Congress, House, Committee on Interstate and Foreign Commerce, Subcommittee on Public Health and Environment, “Statement of Hon. Caspar W. Weinberger, Secretary, Department of Health, Education, and Welfare,” p. 10. 54. S.A. Wesbury, “Career Patterns in Health and Hospital Administra- tion,” (Ph.D. dissertation, University of Florida, 1972). 55. W.R. Kirk, Your Future in Hospital and Health Services Administra- tion (New York: Richards Rosen Press, 1976). 56. A.F. Westin and M.A. Baker, Databanks in a Free Society: Computers, Record-Keeping and Privacy (New York: Quadrangle Books, 1972). Reprinted with permission. 57. A.F. Westin, Computers, Health Records and Citizen Rights, National Bureau Standards Monograph 157 (Washington, D.C.: National Bureau Stan- dards, 1976), pp. 723ff and 46Iff. 58. Hearings Before the Subcommittee on Consumer Credit of the Senate Committee on Banking, Housing and Urban Affairs on S.2360 The Fair Credit Reporting Act Amendments of 1973 (93rd Congress, 1st Session, 1973). 59. Westin and Baker, Databanks in a Free Society: Computers, Record- Keeping and Privacy. 60. Ibid. 61. Department of Communication/Department of Justice, Privacy and Computers, Report of a Task Force on Privacy and Computers (Ottawa: Infor- mation Canada, 1972). 62. Justice Department, Sweden. Data Och Integritet. Report of Swedish Committee on Automated Personal Systems (Almanna For Laget, 1972). 63. Secretary’s Advisory Committee on Automated Personal Data Systems, Records, Computers and the Rights of Citizens (Washington, D.C.: GPO, July 1973). 64. United States, Congress, Senate, Privacy Act of 1974, Public Law 93-579 (93rd Congress, 2nd Session, 1974). 65. B.N. Shenkin, “Medical Records in Patients Hands,” in Proceedings of Third Illinois Conference on Medical Information Systems, University of Illinois, November 1976 (Chicago, 111.: University of Illinois at Chicago Circle, 1977). Reprinted with permission. 66. B.N. Shenkin and D.C. Warner, “Giving the Patient his Medical Record, a Proposal to Improve the System,” New England Journal of Medicine, vol. 289 (13), September 27, 1973, p. 688. Reprinted by permission from New England Journal of Medicine 289:691, 1973. 67. E.P. Bloustein, “A University View of the Data Privacy Debate,” in Proceedings of Second Illinois Conference on Medical Information Systems, September 1975 (Urbana, 111.: Regional Health Resource Center, 1976). Re- printed with permission. 68. A. Mowshowitz, The Conquest of Will: Information Processing in Human Affairs (Reading, Mass.: Addison-Wesley Publishing Co., 1976). Evaluating the Worth of MISs 103 69. R.W. Stacy and B.D. Waxman, Computers in Biomedical Research, four vols. (New York: Academic Press, 1965-1974). 70. Computers in Life Science Research W. Siler and D.A.B. Lindberg, eds. (New York: Plenum Press, 1975). 71. M.F. Collen, Hospital Computer Systems (New York: Wiley and Sons, 1974). 72. INTREX, Report of a Planning Conference on Information Transfer Experiments C.F.J. Overhage and R.J. Harman, eds. (Cambridge, Mass.: MIT Press, 1965). 73. L.F. Carter, G. Cantley, J.T. Roundl, et al.: National Document- Handling System for Science and Technology (New York: Wiley and Sons, 1967). 74. J.R. Senior, Toward the Measurement of Competence in Medicine. (Philadelphia, Pa.: National Board of Medical Examiners, 1976). 75. D.A.B. Lindberg, “Computer Simulation for Testing Clinical Com- petence”, Simulation Councils Proceedings Series. Spanning the Applications of Simulation, vol. 4, (2), P. Brock, ed. (La Jolla, Calif.: Society for Computer Simulation, December 1974), pp. 51-56. 76. R.B. Friedman, “Computer-based Examination Project: Annual Report, April 1, 1977 - March 31, 1978.” (National Board of Medical Examiners and American Board of Internal Medicine: Philadelphia, Pennsylvania, March 31, 1978). 77. M.D. Hopwood, G.F. Groner, N.A. Palley, et al.: An Evaluation of the CLINFO Data Management and Analysis System (Santa Monica, Ca.: RAND Corporation, November 1977) R-2260-NIH. 78. R.A. Rosati, F. McNeer, F. Starmer, et al.: “A New Information System for Medical Practice.” Archives of Internal Medicine, vol. 135, August 1975. 6 Barriers to the Development and Diffusion of MIS Technology Summary Barriers to diffusion of MISs have been sociological and behavioral as well as technological. In addition, management difficulties spanned the two. The speed of technical change in this field has been far greater than the speed of social adaptation. Two kinds of technical barriers have been encountered: those typical of any newly evolving technological enterprise, and those more specific to the medical environment. With respect to the latter, medicine seems to suffer greatly from an administrative pattern of extreme balkanization. There are also major deficits in our understanding of basic physiological processes. Those faults have a direct influence on systems design and efficiency. Social barriers include slow adaptability by the health professions, even to educational requirements, and the seeking of short-term profits and earn- ings by both the computer and hospital industries. The General Nature of Diffusion of Technological Innovation Students of history, sociology, and engineering note a number of stages that technical innovations go through before becoming accepted as traditional mar- ketplace items or services. Frequently the sequence is represented to be research, development, demonstration, commercial prototyping, and production. Some writers show “invention” or “discovery” preceding this list. Some show “marketing” as following the list. Whatever terms best describe this process, it is clear that transitions between the stages occur at irregular inter- vals and that the total costs for each phase typically increase as one moves toward the market. Certainly it is clear that there is a major hiatus between a scientifically interesting and valid research system and anything which could be considered commercially practical. Doubtless there are interactions between the individuals and groups who function at each of these stages. Much has been written about these matters, especially about the two special relationships: that of basic to applied research, and the general effect or lack of effect of federal policy on the process of diffusion of technological innovation. To speak simply of the first question, 105 106 Growth of MISs in the United States earlier writers tended to look on a time sequential relationship between the discovery/basic research and the later stage of applied research/development. More recently, studies based on bibliometric measurements and sociological observations have cast doubt on this model. The new nonlinear model of technology transfer is depicted by a RAND Corporation commissioned policy analysis for federal biomedical research.1 RAND draws upon the work of W.H. Gruber and D.G. Marquis which distin- guished three mainstreams of activity; the development of a body of scientific information, the development of a state-of-the art technology, and the patterns of utilization of science and technology.2 The RAND transformation of this model to the biomedical setting presented the streams as follows: one, the basic research stream in the form of medical science and several of the fields of clinical medical science; and two, the stream of technological development in engineering, electronics and bioengineering. The third stream is represented in this picture only in general in the form of medical practice specialties and disease-specific subspecialties. This study correctly observes that the streams are largely independent, but also coupled in numerous ill-specified ways. In support of this depiction, RAND draws upon the work of Henry G. Small to document that advances in applied chemistry are virtually decoupled from the literature in basic chemistry3 and upon the work of Eugene Garfield who documents that the reverse is true in the case of strong coupling between cancer research and basic science literature.4 It seems remarkable to the present writer that the RAND analysts did not recognize that their third stream (medical specialists) cannot exist outside hospitals and within the framework of fairly extensive health care systems. There is considerable independence between the streams that constitute technological innovation in medicine; the independence is enhanced by the virtually self-propelled nature of the health care mission and of the hospital industry. Amitai Etzioni and Richard Remp summarize the current view by stating: “In principle, therefore, it seems more accurate to view applied research and basic research as two semi-autonomous activities which intereact with each other.”5 Diffusion of Technological Innovation in Medicine J.S. Coleman, E. Katz, and H. Menzel point out that: . . innovation in medicine takes place within a scientific community whose norms, at least officially, welcome change.”6 Yet the rate of adoption of innovative practices in U.S. hospitals varies greatly, depending on both the size of hospitals and possibly upon the characteristics of the innovation.7 For our purpose, it may be useful to consider that changes may take the Barriers to Development and Diffusion 107 form of new ideas, practices, and systems. The first two have been amply described; the diffusion of new systems within medicine has not been so well pictured. Diffusion of Ideas and Concepts in Medicine There is so large a published literature describing the dissemination of infor- mation per se in science that it is not appropriate to discuss it here at length. Complete exploration of ideas on this process can be read in works by A. DeReuck and J. Knight,8 C.F.J. Overhage and R.J. Harman,9 and the Report of the President’s Biomedical Research Panel.10 The general conclusion of these discussions is that information concerning scientific discoveries, experi- ments, and results of individual measurement or studies are communicated both by the formally published, indexed, and abstracted scientific literature and also by informal means whose patterns can be described. Ideas appear to move relatively rapidly through both the scientific and the medical communities. Writings on this subject tend to lament only that there are not means to speed up the process even more, and to facilitate the search through such a large corpus of published ideas. W.D. Garvey and Belver C. Griffith, for instance, object to the lag of months between the publication of a new idea in psychology and its subsequent appear- ance in a review journal.11 Martin M. Cummings12 and Derek Price13 emphasize the increasing size of the formal scientific literature, the problem of managing requests for information by individuals and institutions, and the possible utility of more discriminating publication and abstracting. It should be noted that these authors are directing attention to more or less raw information, embedded as it usually is in the formalisms of scientific writing. This is not quite the same thing as ideas and concepts, but is as close as one can find properly studied. The conclusion that communication of information within medicine and science needs only to be speeded up and made more perspicacious derives from studies of the function of medical scientists and medical practitioners. Recent studies of diffusion of medical technology have pointed out an additional problem. Information and ideas about scientific and technical ad- vances do not flow smoothly to all members of the health care network. That is, hospital-based paramedical specialists and health service administrators are not parties to the scientific information dissemination system, or, if they are, the system does not work equally well for them. In discussing diffusion of innovation within the health system, A.D. Kaluzny and colleagues emphasize the changes that have occurred since the Flexner report set the stage for the diffusion of medical practices in America.14 In Kaluzny’s view, the Flexner report was based in part on the following two assumptions: that technological 108 Growth of MISs in the United States expertise is necessary to ensure proper health care, and that physicians can and must be trained for competent use of most of the available technology. These assumptions provided the rationale for Flexner’s recommendation that medical education should occur close to universities, whose basic science faculties were at the time the source of much technological innovation in medicine. Kaluzny notes: Conditions have changed, however, since the Flexner Report was pub- lished. Perhaps the most crucial change has been the replacement of the physician’s office by the hospital as the central focus of non-routine medical care. Hospitals are now the dominant organizations in the health care system, and increasingly they act as the major repository of medical skills through which technology is presented to the popula- tion.15 Kaluzny finds this change to be most important to studies of diffusion of innovations in medicine. Most administrators are not medical experts ... On the other hand . . . he is under increasing pressure to work for greater cost efficiency and community responsiveness. Since these goals can conflict with the more technically oriented goals of the medical community, they will probably affect the speed of medical adoption. These facts must be considered in any study of the diffusion of medical innovations.16 A kind of global postscript to these considerations appeared in the 1976 Report of the President’s Biomedical Research Panel. In discussing the transfer of information in general and the NIH, in particular, the Report examines two questions: 1. In all the new mass of basic scientific data, are pieces of impor- tant information being set to one side, mislaid, or even lost, with- out being applied to disease problems? 2. How much of the new information, in fact, is applicable?17 The report concludes flatly: As to the first question, which concerns the “lag between bench and bedside,” we are unable to find evidence that this occurred . .. when a representative list of therapeutic and diagnostic advances of the past 25 years is reviewed, each based on preceding work in research labora- tories, the progress from discovery to application appears to have occurred in a reasonably timely and orderly fashion.18 The President’s Panel did not, however, fail to see problems in the process of transfer of information and ideas. They note: Barriers to Development and Diffusion 109 There is a third question, not so frequently asked: Are there occasions when application has occurred too soon, before the data were suf- ficiently certain, and is this a hazard for the future? Our impression is that it is a matter for concern.19 Diffusion of Practices in Medicine Along with modern communication systems, mass culture, and a general greater willingness to change, the speed of diffusion of new practices in medicine is far faster nowadays than in past times. Almost a century was required to achieve general use of percussion of the chest after its announcement by Auenbrugger in 1761. General use of the stethoscope was not complete for almost 80 years after Laennec’s description of the technique in 1808. General adoption of the electrocardiograph after Waller (1889) and Einthoven (1903) required not less than 20 years. In contrast, the methods of roentgenography took very little time to be adopted into medical practice in spite of being truly novel and requiring major capital expense for equipment. Indeed, x-ray sources were used to make photographs of medical abnormalities within months of Wilhelm Roentgen’s original publication of his technique in 1895. Study of the pattern of diffusion of a moden practice in modern times, namely, the use of new antibiotics, was done by J.S. Coleman et al. They con- cluded: . . there are truly ‘innovators’ and ‘conservatives’ among the doctors in those four communities—at least as far as the early introduction of relatively undramatic drug innovation is concerned.”20 Their observations were shaped by interest in the behavior of individual physicians and their relationship to other physicians and groups. These investigators accepted that for the individual there is a personal sequence of processes of awareness, interest, evaluation, trial, and adoption. Public media could serve only the first one or two steps. Peer judgments played the greatest role in the middle steps. Individual judgment entered into the last couple of steps of the sequence. The pattern of increasing use of new antibiotics was, however, not a general clue to the spread of innovative practices and ideas. Coleman et al. noted that the early adopters or nonadopters of new drugs ... are not necessarily innova- tors and conservatives in other respects.21 For instance, they do not necessarily also accept “. .. more diffuse modem trends, such as psychologic considerations in general medicine, or a more equal and active role for the patient . . .”22 Diffusion of Systems in Medicine While many ideas and practices have moved rapidly through modern medicine, the spread of systems has been irregular. The rate of diffusion of MIS technology 110 Growth of MISs in the United States has been in a single word, slow. The pattern of the diffusion appears to have been more like that characterizing technologic change in general in any field (that is, a gradual buildup) than it has been like “epidemic” models such as are seen with fashions (such as Beatlemania, the Twist, acupuncture, or Laetrile). The reasons for this may be the same as those offered in the past to explain the slow diffusion of scientifically sound agricultural systems.23'25 Classical studies of diffusion of innovation have been made with respect to agricultural practices and systems. An example is the concerted effort made by land grant universities and federal extension agents to encourage the use of “scientific farming.” Such studies may seem irrelevant because of being agri- cultural and because the innovations were described as “practices”. In reality a new system was being advocated. The first bag of fertilizer applied to a farm essentially represented a commitment to testing, analysis, correction of soil conditions ad infinitum, as well as a commitment to concommitant educational programs. Those who studied diffusion of innovation in agriculture made in- sightful observations that seem highly pertinent to our consideration of MIS acceptance. For this reason, I wish to present a speculative comparison of the two problems. In the case of agricultural systems, the emphasis of the studies was legiti- mately on individuals. Decisions concerning MISs are also made by individuals, although the adoption of an MIS is ultimately made by an institution rather than an individual. For this reason, we first review the problem in terms of the agricultural model, and then in terms of the corporate or institutional model under the section reserved for the MIS question. Comparison of Studies of Diffusion Processes: A Speculation Agricultural Model. H.F. Lionberger emphasized the relationship between the speed of diffusion of innovative practices, and the difficulty of the concept involved in the innovation as well as the difficulty in evaluating the benefits of adopting the innovation. An easily demonstrable practice may be more quickly adopted . . . than one which is difficult to understand, or one whose effects are indirect.26 Based on this observation, Lionberger constructed a taxonomy of innova- tions, which is paraphrased as follows: Grade One innovation: easily understood and easily demonstrated; use of pesticides, for example Barriers to Development and Diffusion 111 Grade Two innovation: somewhat more difficult to understand and demonstrable only over a minimum time of a crop cycle; use of improved fertilizers, for example Grade Three innovation: more difficult to understand and demonstrable only over a considerable period of time; improved genetic management practices, for example The increasing challenge of the innovations reflects the increasing complex- ity of the basic concepts and the indirectness or difficulty involved in the adapter coming to appreciate the benefit. Delay in adoption of the scientifically sound agricultural practices studied was proportional to these two factors as reflected in the taxonomy. Medical Model. Counterparts to these examples can be found in the field of MISs. The following examples of technologically innovative medical systems appear to fit the agricultural taxonomy quite well, and the speed of their adop- tion does indeed appear to be in accord with the taxonomy proposed. Grade 1 Innovation. Automation of a current simple practice such as patient accounting. This is easily understood, and the demonstration of feasibility and efficacy does not take long. This is especially true if it is possible to adopt almost entire- ly an existing “package” of programs and procedures. More than 85 percent of all U.S. hospitals use computer systems or ser- vices in connection with their patient billing, collections, and third party reim- bursements. The rate of adoption of computers in hospitals for such purposes has not been much slower than for industry in general. By 1976, 2.7 percent of all general-purpose conventional computers within the United States were owned by medical and hospital services. This compares with 2.9 percent owned by the transportation and carrier industry, and 2.7 percent owned by the printing and publishing industry.27,28 Grade 2 Innovation. Automation of a current, more complex procedure such as analysis of electrocardiographic signals. The general nature of the task required of the MIS is relatively easily under- stood. It is, to build an automatic system such that the diagnostic statements of the automation are identical to those that a skilled cardiographer would make in response to the same input record; namely, the words descriptive of the EKG tracing. The demonstration of successful operation of the automation is also rather easily shown over a relatively short time, indeed over minutes. Of course, the comparison between new fertilizers and automated EKGs is not adequate in all respects. Many additional considerations undoubtedly played roles in the medical case. 112 Growth of MISs in the United States In any event, the technology of automated EKG interpretation has diffused relatively more quickly and more widely than that of other components of the MIS concept. By August, 1976 more than 10 percent of all EKGs processed in the United States were interpreted by computer-based systems.29 Automated EKG interpretation systems are operating commercially quite independent of the original designers. Grade 3 Innovation. Automation of an entire system such as an automated patient medical record. This is a concept which is not at all easily explained or understood. It in- volves gathering data from a number of resources that have been conceptualized and recorded in quite different ways. The process demands coordination of actions of many persons. The execution must be timely. Furthermore, the beneficial results are quite indirect, relatively remote in time with respect to the actions of the subsystem components. Indeed there may be no benefits unless all of the subsystem components perform their jobs properly and in a timely fashion. In these respects the concept of the automated MIS is analogous to the con- cept of improved genetic practices in agriculture. In contrast to patient billing and EKG interpretation usage, no more than two or three dozen hospitals now use a full MIS. These institutions constitute less than half of one percent of U.S. hospitals. A large incremental innovative step might require for proper classification a Grade 4 category. This could take the following form: Grade 4 Innovation. Substitution of a new conceptual model of the activity in question and simultaneous automation of it. I do not have an example in agriculture per se, unless hydroponics might be satisfactory. In the field of MISs an example of this fundamental change is the single case of the PROMIS system of Lawrence Weed at the University of Vermont. (The system is described in detail in chapter 4.) Quite aside from whatever eventually turns out to be the benefits and costs of this system, the important thing to note for the moment is the fundamentally new conceptual model which is involved. In the Weed approach, the computer embodies in a finalized manner the known relationships between the various elements of human diseases and health. The system contains recommended procedures for medical study of the patient, as well as an implicit model of the proper medical handling of the disease processes themselves. The relationships are represented in this system by a thesaurus of terms, frames of statements and questions, and by rules that determine the sequence for display of the frames. The computer guides and assists the physician to match the attributes of a particular patient to the formalism of the profession contained in the com- Barriers to Development and Diffusion 113 puter system. The formalism may represent and organize the information in ways suited to the system and at variance with the individual physician’s tradi- tional work and thought patterns. This new procedure occurs in an automated mode. The physician is obliged physically to touch the computer terminal in special ways in order to operate the system. Touching the CRT screen face selects a chosen response from the lists displayed by the system. The benefits from adopting this new technology do not redound to the physician user (who may have to work harder), but to the patient or patients, and perhaps at a time quite remote from the initial encounter. The only obvious medical analogy historically is the diffusion of germ theory, with the consequent elaborate ceremony of aseptic techniques. This was based on a totally new model of disease. It did not involve automation. I do not say the problem-oriented record is an innovation of comparable magnitude. I only point out that the conversion of germ theory into aseptic techniques in medicine took a long time, and was not entirely pleasant for the innovators or early adopters. Insofar as systems such as PROMIS truly involve a new concep- tual model, namely, that there is a formal representation of good medical practice inside the computer, the diffusion of such a system must inevitably be slow. It is likewise inevitable that social acceptance must be a problem for such an innovation even greater than the inherent barriers constituted by the tech- nical difficulties of implementing such a complex system. Barriers to MIS Technology Operational Difficulties Throughout many system developments, implementations, and evaluations, certain difficulties have been reported with surprising consistency by the system builders and by the groups who studied the process. While it is difficult to classify each and every problem, it is clear that the barriers noted include tech- nical, social, and managerial issues. Many investigators have identified their problems in terms that echo Morris Collen’s formulation. After an analysis of reports of successful and unsuccessful attempts to produce operational MISs, Collen concluded that the reasons for the failure were: 1. A suboptimal mix of medical and computer specialists 2. Inadequate commitment of capital for long term investment 3. A suboptimal systems approach [that is, with either inability to fuse in- compatible subsystems or a too grand initial total design]. 4. Unacceptable keyboard terminals 5. Inadequate management organization 114 Growth of MISs in the United States In his view, factors tending toward success would require correction of all these deficiencies. He feels that, for success, ... an organization [a hospital or group of hospitals] of sufficient size is required, with effective management by technically sophisticated men who can make reliable decisions after considering technological alternatives.30 In a recent survey of automated ambulatory record systems, Ronald Henley and Gio Weiderhold summarized the problems encountered at fourteen of the systems site visited.31 Their list represents the most important problems as perceived by major systems builders. The list is remarkably similar to Collen’s. Like his, it is a combination of technologic difficulties, social, and management problems. Their origin listing was according to the site, reporting with no ap- parent further ordering. For the present analysis of barriers, the observations have been arranged into categories, while preserving the wording of the problem statements as reported to Henley and Wiederhold. Examples of technologic difficulties included: 1. Limited man-machine interaction 2. Diversity of agencies, record formats, and service requirements 3. Voluminous text Examples of social barriers included: 1. Lack of interest from management 2. City-funding intermediary has other priorities 3. Limited services provided 4. Unique environment 5. Usage pattern of individual practitioners is not the same 6. Mobile population Examples of management obstacles included: 1. Changes in university computer system support 2. Poor management interface between [university and health care facility] 3. Organizational structure of county, [the state university], and [the health care facility] is complex 4. No control over cost of operation 5. Insurance rate setting is based on overhead-to-payout ratio 6. Limited experience in automated ambulatory medical record systems. System developers frequently reported difficulty in managing complex Barriers to Development and Diffusion 115 systems. One experienced and competent project manager said that the systems finally became so complex that he could not implement programming changes and additions without intolerable delays.32 Another viewpoint on barriers and limitations is presented by Richard Friedman and David Gustafson.33 They deal with the more general question of the reason for slow success of computers in clinical medicine, but the observa- tions are also relevant to the special case of MISs. The impediments, they feel, have been: 1. Not perfecting the computer-to-man interface. They, like Collen, em- phasize the awkwardness of current computer terminals, their physical charac- teristics, low speed of communication, and irritating communication protocols. 2. Not succeeding in producing applications that exceed the capabilities of the physician without the computer. They draw attention to the rapidity with which technologies have been accepted in the past, despite costs and inconveniences, when they performed a service or a measurement that went beyond what an individual alone could do. 3. Not having proved a significant impact on patient care. 4. Not easily transferring computer systems to other institutions. 5. Not having developed computer systems in support of changes which might be needed in health care practices. They draw attention especially to the patient record systems, noting systems that collect data rather than improve decision making. 6. Not having learned from unsuccessful system implementations. These authors call for follow-up on new systems and reporting of failures. They describe a case in which reports detailing the reasons for withdrawal from an automated medical history application were not accepted by journals that had previously published the initially enthusiastic reports. Friedman and Gustafson surveyed thirty-two computer systems. They included all those reported over a Five-year period in four selected clinical medical journals. They found that half the systems had been abandoned or temporarily stalled. Nineteen percent were used routinely. Those still in use were mostly funded out of patient fees, had begun with limited, well-defined goals, and had been consistently cost effective. The writers tend to be critical of the medical system designers. They feel that the objectives of the systems were not sufficiently innovative. Yet, on the other hand, they seem to have documented that successful surviving systems had been cost effective, modest in scope, and essentially practical from the outset. The two desiderata, to be innovative and to be practical, are not necessarily mutually exclusive. None- theless, to satisfy both requirements simultaneously while using an automated technology is surely a difficult task. Before leaving the question of barriers to success of MIS development as perceived by the system builders, we should take up the excellent study by Gerald Giebink and Leonard Hurst. Their project included a formal and 116 Growth of MISs in the United States thorough review of twenty-eight computer projects in health care. Almost all the computer projects included, and generally centered on, MISs. The study and report was commissioned by the Robert Wood Johnson Foundation. Giebink and Hurst report on the developmental and operational problems encountered in the computer application of each of the projects. Once again, I have taken the liberty to rearrange the order to present these in the three categories into which it seems to me they fall naturally. In all cases the wording is unaltered. Examples of the technical problems reported by Giebink and Hurst include the following:34 1. Moving head disk hardware problems 2. No back-up computer system 3. Slow response time of terminals, and unreliable terminals 4. Major technical problems in computer representation of extensive medical logic 5. Overly optimistic expectations for speedy implementation 6. Irritations with terminals, printers, and communication lines 7. Software that required extensive development 8. Cost of computer services and lack of alternative suppliers 9. Space limitations in the clinic 10. Unreliable telephone lines, hardware breakdown, system unreliability following each major systems change 11. Power failures 12. Difficulty in adapting to institutions of different sizes 13. Badly designed telecommunications software 14. Difficulties in data collection from paper forms 15. Difficulty with hard copy format and printing fonts 16. Software developmental problems. The following were obstacles encountered which one might consider to have been managerial in nature: 1. Selection of hardware prior to design of the applications 2. Delays caused by large numbers of people involved in decision making 3. Poor acceptance of the system because of “top down” decision making in its creation 4. System became so complicated that long lead times were required for additions or modifications. Computer Hardware and Software Problems While there have been vast improvements in computer hardware during the past twenty years, there are still obvious major defects and obstacles. Awkward, Barriers to Development and Diffusion 117 slow, expensive computer terminals have been an impediment to all computer- system builders. Likewise, all users suffer from the malfunctioning of moving head disk information storage devices. In spite of the increases in disk storage capacity that have occurred in commercial equipment, large medical record files still frequently exceed the storage capacity of many systems. In a sense, this issue is a trade-off of convenience against costs because at a greater cost one can often obtain an increase in storage capacity. Yet the combination of costs and direct access memory capacity remains a serious general problem. Computer reliability has likewise improved enormously in recent years. Yet people expect more of computers now. Clinical systems must be reliable, just as many other critical control systems must be. For computers this generally requires redundant hardware, which often requires double or triple hardware cost. In the case of missile launches, the costs are acceptable. In the hospital, they are not currently acceptable. One example of this problem, and a workable solution, is the Technicon MIS at El Camino Hospital. Reliability of the system is enhanced by the availability of a back-up computer at a Technicon service center. During two study periods of six months each, the scheduled and unscheduled use of the back-up computer system was found to range from 4.7 to 80.5 hours per month.35 Purchase of this service on an hourly basis was not costly. In this situation Technicon was able to use the back-up computer profitably in its other business commitments. In contrast, had the hospital been obliged to rent the back-up machine totally, as Kaiser had to for their pharmacy and ward terminal system, the hardware costs of the system would have almost doubled. Reliability is still a major impediment to MIS implementation in most set- tings. Software development costs remain high and progress slower than in hard- ware systems. The software interface is still to the computer professional rather than to the physician user, or to the other subject matter experts. In other words, programming the computers is still sufficiently difficult technically that the medical or health systems logic cannot be programmed directly by experts in these fields. The subject matter expert must work with intermediaries such as systems analysts and computer programmers. The intermediaries then work directly with the computer to create the necessary software or programs. This awkward necessity tends to increase personnel costs and to make the accomplishment of system redesign difficult. Worse yet, it separates the health care professionals from direct participation in the creative aspects of the ap- plication development. Medical Barriers to MIS Development Nothing about the computer techniques used in MISs makes them fundamental- ly different from such systems used in nonmedical fields. There are, however, 118 Growth of M ISs in the United States two special nontechnical barriers inherent in the medical application. These are limitations on the state of medical knowledge about illness and health, and limitations on the state of medical systems management. Limitations of Medical Knowledge. There simply is still much art in medicine. Often this is inherent to our ignorance of basic bodily mechanisms and mental processes. We considered in chapter 4 examples of MISs that illustrate the limitations as well as the nonuniformity of our medical understanding. Contrast the functions of the fine mental health system described with those of the system alerting to intensive care units. The mental health system deals only with demographic variables, institutional management problems, and test scor- ing. The intensive care system is already at the stage of representing the knowl- edge about physiological variables and applying this knowledge to patients. Of course future years will doubtless see increases in understanding of both mental and bodily processes. At the moment, however, the different levels of sophistica- tion of the two systems accurately reflect the relative state of medical knowl- edge in the two fields. The highest aspirations of the MIS’s are limited by the profession’s understanding of the basic processes in the patient. Friedman and Gustafson warn that we must try to make computer systems that do things physicians cannot do.36 Yet at least one barrier to this happening is to identify the person with the innovative idea. There are ways to encourage more people, and even more imaginative people, to join the field. Still, the fundamental barrier is creating the idea for the significant new medical function that can be accomplished with the help of the computer. It is the occasional insights into useful new information systems tasks which create major benefits from computer-based systems. The alternative is merely automation of current practices. In MIS work, data base systems are used for patient medical records. This is the essence of the MIS. In this situation the limitation of medical knowl- edge becomes quite apparent. To state the issue in a somewhat extreme form, in using file systems the medical person’s problem is hypothesis generation, whereas the computer analyst’s problem is record retrieval. Smoothly working and efficient computer record systems are possible only when the content of the input records and the exact nature of the patterns of retrieval request are known in advance of the system design. In contrast, such record systems are maximally useful in medicine in the opposite circumstances. Thus in treating the individual patient, one tends to make diligent search of the old medical record when some current complaint is not well understood or not well treated. Consequently, the pattern of the search is difficult to predict. This discrepancy does not mean computer file systems cannot be used in clinical medicine; it simply emphasizes the difference in expectations of the medical and the com- puter system professionals. The most extreme and easily recognized divergence of such expectations occurs in the field of clinical or public health research. A striking example is the Thalidomide story. The association of ingestion Barriers to Development and Diffusion 119 of the sleeping medication Thalidomide by patients in the first trimester of pregnancy with the subsequent birth of malformed infants with phocomelia is widely known.37’38 Recognition of this association came rather late, indeed after some 10,000 malformed infants had been born: 5,900 in West Germany and many hundreds in Japan, Italy, Canada, and the United States. Recogni- tion did not come as a result of automatic scanning of a computer data base system, in spite of the fact that such systems existed at the time.39 The problem is easily understood. Once one has made the hypothesis (or knows or guesses) that the variables “first trimester pregnancy” and “soporific Thalidomide” are relevant to the diagnosis “phocomelia,” it is technically easy to construct the appropriate data system for containing the patient records and for making the search that tests the hypothesized association. Until one has such an hy- pothesis, however, the association will tend to be lost in the larger number of what seem to be more obviously important observations such as blood pressure, weight change, family history, infections, and morning sickness. Even though the Thalidomide/phocomelia cases became recognized as a major tragedy of international scope there is still no early warning computer record system to detect such an association in the future. To make such a system would tax the wisdom of computer scientists as well as medical scien- tists. On the other hand, computer systems do perform this kind of hypothesis- generation function in nonmedical settings. These include earth satellite scanning and tracking, sonar interpretation, and sentence understanding. The major barrier to accomplishing such a worthwhile function in the sphere of drug practices is the absence of explicit medical logic of any generality. Several other important statistical associations in the public health sphere warrant mention. In these cases, as in Thalidomide/phocomelia, the data base systems are technically easy to construct after the association of the pieces of the medical record has been recognized or hypothesized. A second example, Diethylstilbesterol (DES) therapy given to preparturient women to prevent spontaneous abortion appears to have caused an increase in the incidence of endometrial adenocarcinoma in the daughters born from such a pregnancy. Demonstration of the association required accurate records to have been kept for twenty years. In addition, it required an inspired hypothesis and considerable follow-up to find the daughters. An automated file system had not been designed to detect this complica- tion because the state of medical knowledge at the time of the treatments did not include awareness of such a complication. Examination of this failure in the past helps us to see how the fundamental limits of medical knowledge have prevented us from employing automatic information systems to recognize many obvious statistical relationships. As a guide to the future, this kind of experience must be interpreted broadly in terms of multiple potential benefits from pro- posed MISs. There will always be limits to medical knowledge. One cannot guarantee discoveries of associations such as the Thalidomide/phocomelia or 120 Growth of MISs in the United States the DES cancer. A discussion is presented in chapter 5 of the extent to which one might reasonably expect MISs to have such a scientific impact. A third example of MIS concepts in the public health domain should be noted. Ninety-six percent of individuals suffering from the form of arthritis known as ankylosing spondylitis have the special histocompatibility antigen HL-A27 on their white blood cells.40 In contrast, this antigen occurs in only 4 percent of the unaffected population. The association has great importance both to diagnosis and treatment of the condition, and also as a clue to better understanding of the hereditary aspects of the disease. Discovery of the asso- ciation was accomplished using computational and statistical techniques. The significance from the point of view of medical information systems is to take note that the histocompatibility typing is a variable or a record field which had no logical association whatever with the clinical diagnosis of arthritis. The association was not predicted by medical knowledge. Yet the hypothesis was made by medical personnel after the fact, using advanced information pro- cessing systems for testing the idea. The same kind of difficulty arises in data base systems needed to detect industrial carcinogen exposure and those needed to detect drug interactions. In the absence of such fundamental health knowledge, enormous complexity is required of the data base systems in order to “shoot in the dark” searching for relevancy among the patient record variables. Two additional undesirable general medical practices stem from the incom- plete state of our understanding of human biology. The practices include the use of extremely irregular terminology in medicine and our reliance on necessarily ad hoc patient identification systems. Both deficits present serious barriers to automation. The language of medicine is simply unstructured. The lack of a standard medical vocabulary is a serious obstacle to the creation of MISs as well as to the transplantation of a successful system to other locations. Most medical terms have not been given rigorous definitions. This is true for symptoms, findings including laboratory tests, and many diagnoses. There are substantial efforts to correct these weaknesses, but it is still true that a computer medical record file system would not be directly transferable from one hospital to another without a complete reexamination of its descriptors of the elements of the medical records. The second difficulty is comparable but not so easily solved. The problem of identification of individuals, their medical samples, and observations about both in a computer-based information system is a fundamental source of error and uncertainty in all MISs. Corn flake boxes and railroad cars are now made with “zebra stripes”, people are much more difficult to identify. As a conse- quence, multiple hospital records are frequently created for the same individual, and laboratories are confounded by lack of positive identification of speci- mens. Barriers to Development and Diffusion 121 Limitations of Management. The second barrier peculiar to MISs is the medical environment, or what is now more properly called the health care systems environment. The U.S. health care system is made up of thousands of relatively autonomous units, centering on large hospitals, which are themselves made up of relatively autonomous divisions and departments. There is no common ownership nor meaningful common accounting system. In addition there is an apparent shortage of individuals capable of managing—or even rearranging— complexity. To the extent that health care institutions do not work smoothly and sensibly with one another, the MIS cannot be shared or transplanted. To the extent that health care institutions are balkanized into small administrative parcels, the information systems must of necessity be small as well. It is clear why minicomputers are so popular in medicine, and why large data base systems are so rare. The minisystem matches the miniadministrative fiefdom. The large data base systems represent one of the many institutional goals to which the institutions often cannot manage their way. Even at the level of MIS subsystems relevant to hospital administration, management abilities may represent the primary barrier. In an excellent review of hospital information systems, Charles Austin and Barry Greene observe that information system developments in hospitals have not met their potential and have lagged behind information system developments outside the hospital field. They conclude that the main factor explaining the lag is that: Many hospital administrators simply do not possess the technical knowledge and skills required to understand control processes.41 Austin and Greene conclude that MIS components once installed may not be used by top managers simply because the managers do not know how to make use of the information the systems are capable of providing. Use is indeed the critical element in successful systems, rather than technical perfection in the design concepts. They go on to speculate that whatever initial flaws might be present in such a system would be corrected if only the systems were used to make important management decisions. Social Barriers to MIS Development Health Profession’s Response to Information Systems Technology. Technical changes can and generally do occur more rapidly than society’s adjustment to them.42’43 This has certainly been true of computers in general, medicine, in general, and the combination of the two fields in particular. Evidence for this can be seen in the shortage even today of skilled and experienced computer technologists and the even smaller number of medically trained individuals who have experience or cross-training in computer or information science. 122 Growth of MISs in the United States Medical school curricula have been notoriously slow to change. None currently provide formal courses in computing to students of medicine as part of their regular curricula. Likewise no testing for information processing skills is included on the medical certification examinations of the National Board of Medical Examiners in spite of the fact that the examination itself is totally processed and scored by computer.44 Postdoctoral training in medical computer work is provided at ten institu- tions.45 The oldest program is five years old. None of the medical specialty boards has requirements for or gives credit for computer and information systems experience as part of the specialty physician’s post graduate education. Hence, they strongly penalize those who invest postgraduate training years in obtaining computer skills. Problems with operational MISs also attest to the claim that social engineer- ing proceeds less rapidly than hardware engineering. Studies of barriers to MIS development repeatedly mention difficulties in communication between medical and computing personnel on the same research team, and in establishing com- munication and cooperation between health care institutions in the same city. Creativity and technological innovation, are prerequisites for the develop- ment of MISs. No one knows for sure how to manage the creative process, in science or elsewhere. The building of MISs requires teamwork by a multidisci- plinary group. This complicates matters by adding another substantial manage- ment problem. Furthermore, the activity is expected to proceed in the face of unstable financing, intermittent encouragement from government, and dis- incentives from medical specialty societies. All in all, there is a long way to go before social adaptation to the computer has caught up with the technical state of the art. Computer Industry’s Response to Medical Needs. There has at times been moderate enthusiasm in the computer industry for developing and marketing MISs. In 1973 Marian J. Ball reviewed fifteen commercial systems 46 By 1977, six of the companies were not in the MIS business. Major companies leaving the computer field altogether have included RCA and GE. IBM, an early enthusiast of MISs, virtually abandoned medicine as a high-priority marketplace for the better part of a decade. There are still large numbers of computers going into hospital business offices to do accounting jobs. One can only conclude that industry takes its mandate from stockholders to maximize profits over the long run, and that medical applications, especially MISs, are not judged to be the most profitable investment. Yet there are profits to be made elsewhere in the medical computing field. Witness the installation in the United States alone of 320 computer tomography units (CAT scanners) at about $500,000 each, with another 224 units approved and the orders on backlog as of July 19 7647 and over 500 installed by 1978 48 Barriers to Development and Diffusion 123 Witness eleven U.S. companies in the marketplace, and eight foreign competi- tors. Note that all this has transpired since Hounsfield’s first demonstration in England in 1971 and in the United States in 1972. Computer tomography and MIS developments have much in common. Like MIS work, the computer tomography problems are challenging and scientific. They involve multi- disciplinary teams to produce advances in the state of the art. Yet in the case of computer tomography, the state of the art has advanced rapidly. Third genera- tion equipment has been announced while there still is a backlog of orders for first generation scanners. Hospital Industry’s Response to MISs. Contrasting the problem of slow develop- ment of MISs with all too rapid diffusion of the technology of computer tomo- graphy presents the question: Why the difference? One can only conclude that industry does not respond to economic and scientific challenge. Industry re- sponds to opportunities to sell hardware at a profit. Customers, including hospitals, seem to buy more of those hardware items with which they in turn can increase earnings. Computer tomography units increase hospital and pro- fessional earnings as do automated laboratory instruments. Sometimes it seems that MISs cost everyone a great deal of money and anguish. When MISs are successful they reduce the cost of care and the amount of hospital billings and earnings. There is no escaping the conclusion that hospitals as corporate entities have not up until now been sufficiently strongly motivated to cut costs. MISs are just one of the ways to reduce costs, although they seem to offer a substantial cost reduction, provided good hospital administration is available to capture the savings. On the other hand, up until the current efforts of the Carter ad- ministration to cap costs, all hospital costs have been reimbursable. Only rare institutions have made serious efforts to implement MISs or other major cost- saving experiments. Outside of immediate cost reduction through administrative efficiencies, many of the apparently desirable applications of MISs are aimed at obtaining long-term benefits for groups outside the immediate hospital population. These benefits will naturally have their costs, so that the short-term effects of some MIS applications may temporarily or locally increase cost. This problem is discussed more fully in chapter 5 as a problem in evaluating MISs. For the moment we may note, if the desired outcome of the development of MISs is recognition of new diseases, or creation of community medical data bases, or prevention of future drug side effects, or more complete reporting to Pro- fessional Standards Review Organizations (PSROs) or Medicare intermediaries, or shorter hospitalizations, or recognition of occult disease in ambulatory popu- lations, then someone must pay for these outcomes. All the outcomes are desirable (and even cost justifiable) on the basis of society as a whole and within a time frame of decades. Yet each of them either decreases the revenues 124 Growth of MISs in the United States of an individual hospital or increases health care overhead in the short term, or both. No societal accounting system exists at the moment for rationalizing this problem of short-term or local cost versus long-term or general benefits. The absence of such machinery has been a serious obstacle to MIS diffusion. Notes 1. United States Department of Health, Education, and Welfare, “Ap- proaches to Policy Development for Biomedical Research: Strategy for Budget- ing and Movement from Invention to Clinical Application,” in Report of the President’s Biomedical Research Panel: Appendix B, April 30, 1976. (Wash- ington, D.C.: G.P.O.) 2. W.H. Gruber and D.G. Marquis, “Introduction,” in Factors in the Transfer of Technology (Cambridge, Mass.: MIT Press, 1969). 3. H.G. Small, Characteristics of Frequently Cited Papers in Chemistry, Final Report to the National Science Foundation (Philadelphia, Pa.: Institute for Scientific Information, 1974). 4. E. Garfield, “Journal Citation Studies 15: Cancer Journals and Articles,” Current Comments, vol. 17, October 16,1974, p. 5-12. 5. A. Etzioni and R. Remp, Technological Shortcuts to Social Change (New York: Russell Sage Foundation, 1973), p. 198. © Russell Sage Founda- tion. Reprinted with permission. 6. J.S. Coleman, E. Katz, and H. Menzel, Medical Innovation: A Dif- fusion Study (Indianapolis, Ind.: Bobbs Merrill Co., 1966), p. 8. Reprinted with permission. 7. L.B. Russell and C.S. Burke, Technological Diffusion in the Hospital Sector (Washington, D.C.: National Science Foundation, Office of National Research and Development Assessment, 1975). 8. A. DeReuck and J. Knight, eds., Communication in Science: Docu- mentation and Automation. A Ciba Foundation Symposium (Boston, Mass.: Little Brown and Co., 1967). 9. C.F.J. Overhage and R.J. Harman, eds., INTREX, Report of a Planning Conference on Information Transfer Experiments (Cambridge, Mass.: MIT Press, 1965). 10. U.S. Department of Health, Education and Welfare, Report of the President’s Biomedical Research Panel, April 30, 1976. (Washington, D.C.: GP.O.). 11. W.D. Garvey and B.C. Griffith, “Communication in Science: the System and its Modification,” in Communication in Science: Documentation and Automation. A Ciba Foundation Symposium (Boston, Mass.: Little Brown and Co., 1967), p. 16. Barriers to Development and Diffusion 125 12. M.M. Cummings, “The Biomedical Communications Problem,” in Communication in Science: Documentation and Automation. A Ciba Founda- tion Symposium (Boston, Mass.: Little Brown and Co., 1967), p. 110. 13. D.J. DeSolla Price, “Communication in Science: The Ends-Philosophy and Forecast,” in Communication in Science: Documentation and Automation. A Ciba Foundation Symposium (Boston, Mass.: Little Brown and Co., 1967), p.199. 14. A.D. Kaluzny, D.Y. Barhyte, and G.G. Reader, “Health Systems,” in G. Gordon and G.L. Fisher, eds. The Diffusion of Medical Technology (Cam- bridge, Mass.: Ballinger Publishing Co., 1975), p. 29. 15. Ibid., p. 32. 16. Ibid., p.32-33. 17. U.S. Department of Health, Education and Welfare, “The Place of Biomedical Science in Medicine and the State of the Science,” in Report of the President’s Biomedical Research Panel: Appendix A, April 30, 1976, p. 16. 18. Ibid., p. 16,17-19. 19. Ibid., p. 16-17. 20. J.S. Coleman, E. Katz, and H. Menzel, Medical Innovation: A Diffusion Study, p. 35. Reprinted with permission. 21. Ibid., p. 35. 22. Ibid., pp. 35-36. 23. H.F. Lionberger, Adoption of New Ideas and Practices (Ames, Iowa: The Iowa State University Press, 1960). Reprinted by permission. © 1961 by the Iowa State University Press, Ames, Iowa 50010. 24. Coleman, et al.: Medical Innovation: A Diffusion Study. 25. G. Gordon and G. Lawrence Fisher, The Diffusion of Medical Tech- nology (Cambridge, Mass.: Ballinger Publishing Co., 1975). 26. H.F. Lionberger, Adoption of New Ideas and Practices. Reprinted by permission. 27. R.M. Davis, “Evolution of Computers and Computing,” Science, vol. 195, March 18,1977, p. 1096. 28. Other surveys of the U.S. information processing industry give some- what differing but generally similar figures. For example, see P.S. Nyborg, P.M. McCarter, and W. Erickson, “Information Processing in the United States, A Quantitative Summary” (Montvale, N.J.: American Federation of Informa- tion Processing Societies, Inc., 1977), pp. 1-54. 29. A.D. Little, Automated Electrocardiography in the United States (Cambridge, Mass.: Arthur D. Little, Inc., 1976). 30. M.F. Collen, “Reasons for Failure and Factors Making for Success”, paper prepared for the Symposium on the Development of Hospital Comput- ing Systems, Toulouse, 28 June-2 July, 1971. WHO Regional Office for Europe. Document EURO 4304/10. Reprinted with permission. 126 Growth of MISs in the United States 31. R.R. Henley and G. Wiederhold, An Analysis of Automated Record Systems: Vol. I: Findings, Vol. II: Background Material (San Francisco, Calif.: University of California San Francisco Medical Center), June 1975. 32. G.A. Giebink and L.L. Hurst, Computer Projects in Health Care (Ann Arbor, Mich.: Health Administration Press, 1975), p. 172. Reprinted with permission. 33. R.B. Friedman and D.H. Gustafson, “Computers in Clinical Medicine, a Critical Review,” Computers and Biomedical Research, vol. 10, March 1977, pp. 199-204. 34. G.A. Giebink and L.L. Hurst, Computer Projects in Health Care. Re- printed with permission. 35. J.P. Barrett, R.A. Bamum, B.B. Gordon, and R.N. Pesut, Final Report on Evaluation of the Implementation of a Medical Information System in a General Community Hospital, (Columbus, Ohio: Battelle Memorial Institute, 1975) NTIS#PB-248 340, pp. 10.9-10.15. 36. Friedman and Gustafson, “Computers in Clinical Medicine, a Critical Review,” p. 200. 37. M. Mintz, The Therapeutic Nightmare (Boston, Mass.: Houghton Mifflin Co., 1965). 38. H.B. Taussig, “The Evils of Camouflage as Illustrated by Thalidomide,” New England Journal of Medicine, vol. 269, November 7,1963, pp. 92-94. 39. G.W. Mellin, “The Fetal Life Study of the Columbia-Presbyterian Medical Center: a Prospective Epidemiological Study of Prenatal Influences on Fetal Development and Survival,” in Research Methodology and Needs in Perinatal Studies (Springfield, 111.: Charles C. Thomas, 1963). 40. D.A. Brewerton and D.C. James, “The Histocompatibility Antigen (HL-A27) and Disease,” Seminars in Arthritis and Rheumatism, vol. 4, February 1975,p.191. 41. C.J. Austin and B.R. Greene, “Hospital Information Systems: A Cur- rent Perspective,” Inquiry, vol. 15(2), June 1978, p. 95. Reprinted with per- mission. 42. A. Mowshowitz, The Conquest of Will: Information Processing in Human Affairs (Reading, Mass.: Addison-Wesley Publishing Co., 1976). 43. A. Toffler, Future Shock (New York: Random House, 1970). 44. J.P. Hubbard, Measuring Medical Education: The Tests and Test Pro- cedures of the National Board of Medical Examiners (Philadelphia, Pa.: Lea and Febiger, 1971). 45. National Library of Medicine, Health Sciences and Computer Tech- nology. Report of Training Directors (National Library of Medicine, May 1976). 46. M.J. Ball, “Fifteen Hospital Systems Available,” in How to Select a Computerized Hospital Information System (Basel: S Karger, 1973). Barriers to Development and Diffusion 127 47. American Hospital Association. CT Scanners: A Technical Report (Chicago, 111.: American Hospital Association, 1977). 48. E. Kennedy. Congressional Record, vol. 124 (124), (Washington, D.C., August 9, 1978) p. 12931. 7 Effects of Changes in Technology on Future Medical Information Systems Summary Two major advances in computer and information systems hardware have re- cently occurred. These are the computer-on-a-chip microprocessors and the laser-etched video disk information storage systems. These new technologies will eliminate or bypass a number of current MIS problems but will require a significant rethinking of many MIS concepts both with respect to system objectives and evaluation criteria. Developments in communications networking hardware and practices will also profoundly affect MISs. The way in which current communications technology is deployed for broad social purposes will be the critical factor for MISs. At the same time, computer software difficulties will become more of a problem and hindrance to MIS development. One software technique is an exception. The increasing use of artificial intelligence techniques to support the creation of knowledge-based clinical systems can have a major beneficial impact on the kinds of MISs that are developed. Growth of this software technology is synergistic with the growth of the three hardware-based tech- nologies. Its effect may be to support the development of independent informa- tion resources at the periphery of the health care system. The major obstacles to full development of the ultimate computer-based information systems in medicine will continue to be rigidity in the health care systems, persisting difficulty in the human-machine interface at the level of the interactive terminal devices, and the difficulty of integrating national social planning with the management of technological development. Microcomputers Description of the Technology A new class of systems now exists called microcomputers. Such systems are marketed by more than 120 companies. They are based upon hardware devices called the “computer-on-a-card” or “computer-on-a-chip.” The basic hardware devices, or the microprocessors themselves, are manufactured by about twenty U.S. companies in about thirty different designs. The basic rights to the most popular general-purpose microprocessor are held by such U.S. firms as Texas 129 130 Growth of MlSs in the United States Instruments, Motorola, Hewlett-Packard, Fairchild, Digital Equipment, National Semiconductor, and IBM. Some of these companies permit manufacture of their chips by other “second course” companies under license. In all cases, a vast number of electronic elements of the computer circuits have been repre- sented in solid-state form on tiny wafers of a silicon mixture. R.N. Noyce of Fairchild compared current systems with an early com- puter: Today’s microcomputer, at a cost of perhaps $300, has more comput- ing capacity than the first large electronic computer, ENIAC. It is 20 times faster, has a larger memory, is thousands of times more reliable, consumes the power of a light bulb rather than that of a locomotive, occupies 1/30,000 the volume and costs 1/10,000 as much.1 From the user’s point of view, a much smaller and much cheaper version of a traditional digital computer has been created. The reduction in size of the circuits themselves is somewhat illusory. A single four-by-six-inch card, or even a silicon chip less than an inch square, can contain all the logical elements and capability that required 10 to 20 cubic feet of transistor and magnetic core technology or a building full of vacuum tube technology. On the other hand, since power supplies, typewriters, printers, display screens, and the nonlogical computer elements have not become appreciably smaller, the real reduction in size of a complete system is less spectacular, although still substantial. What used to require a computer center can now be housed in a small room. What used to require a small room can now be kept on a desk top. Costs have fallen comparably, with reductions ranging from 10:1 to 30:1 if one compares 1979 microprocessor systems with 1970 minicomputer systems. These changes result solely from advances in engineering and manufactur- ing methods. No fundamental changes in computer architecture per se have yet been involved. The microcomputers are simply much smaller, cheaper versions of the old-style computers. The same kind of programs will execute on the new machines. This is a mixed blessing. There is great convenience in being able to transfer work more or less directly to newer, cheaper computers. On the other hand, the level of sophistication of the computer languages has been frozen. The microprocessors now offer the individual user a few thousand dollars computer power roughly equivalent to what a half million dollar computer center offered in the mid and late 1960s. Consequently the computer languages and software problems of microprocessors are likewise about the same as they were in the late 1960s. There are some additional implications for the user from hardware con- ventions in microprocessor systems that are still evolving. Most important, these systems tend to utilize solid-state memories which, unlike the older, large, and more expensive magnetic core memories, do not retain their informa- tion store once the electrical power is shut off. Program instructions and data Effects of Changes in Technology on Future MISs 131 must be retained if the systems are to be workable. Consequently, microproces- sor systems tend to include another new device, the floppy disk. This is a rela- tively inexpensive form of permanent magnetic store based on a revolving (incidentally flexible) plastic disk. From the user’s point of view the importance of this convention is: first, the amount of information that can be stored on-line at present is limited to between one quarter and about a million words, and secondly, that the system is highly dependent on the new inexpensive disks and their drives in order to operate reliably. Major Uses of Microprocessors in MISs The most obvious use to which these new systems surely will be put is to per- form individual medical applications as stand-alone units. That is, one can easily imagine single medical functions of the MIS being done successfully by the microprocessors. Examples might be hospital admission scheduling, nurse scheduling, menu planning, automated interrogative patient history-taking, certain physician assistance functions such as differential diagnosis, electro- cardiogram and spirogram interpretation, and a number of elements of the automated clinical laboratory and patient monitoring functions. These areas are certain to expand. The question concerning patient record storage is much more in doubt. At issue is not whether the microprocessor systems can store records, but whether secondary systems will evolve to take care of the problem of getting the records in and out in ways which will appear reasonable to the people involved. In short, as the MIS tasks can be described in ever-greater detail in ever-smaller pieces, the likelihood of microprocessors being successful tends to increase. Problems in these developments center on the obvious question of capacity of the new devices. When their storage capacity is exceeded by the needs of a medical application, be it a ward, a laboratory instrument, or the records for a single patient, can the microprocessor be linked to a larger machine in some kind of hierarchy? The same issue arises if one raises the question, Do the records of a single patient or instrument need to be added to or related to those of other patients or instruments? When such conceptual relationships need to be established, then electronic interconnection between microproces- sor systems will also be required. These matters are receiving considerable attention from computer manufacturers. The technical terms that reference much of this development are network architecture2,3 and distributed pro- cessing.4'6 The ability to interconnect computing systems, whether micro- computers or large computing complexes, constitutes a new technology in itself. This is described in the third section of this chapter. Certain operational difficulties associated with traditional large computers have been observed to be barriers to diffusion of MIS technology (see chapter 132 Growth of MISs in the United States 6). These include high costs of telephonic connections between user locations and the computer center, problems with the management conventions adopted by multiuser computer centers, and awkwardness at the interface between the user and the computer terminal. It is too early in the development of microprocessors to allege that all of these difficulties will disappear. Certainly the new systems, even when they offer advantages, will bring with them some unanticipated problems of their own. Yet on balance, it seems likely that the microprocessor-based MIS subsystems will circumvent many of the past difficulties arising out of telecommunication costs and awkwardness. For the applications in which microprocessors can perform as stand-alone systems or subsystems, it will not be necessary to main- tain telephone lines to a remote computer center, or to deal with the center’s management conventions. Furthermore, microprocessor technology creates the potential for sufficient computer logic local to the terminal that better human engineering appears within reach. More remote effects are difficult to judge at the moment. The effect on data privacy, for example, would also appear to be favorable. Patient records or even sensitive portions of records could be kept at the patient care site on the small disks and not sent to central file storage. This prospect is too remote to judge now because it is not clear what forms the human organiza- tions will take that will employ the new hardware. The main effect of this change in the technology will be centrifugal: it will move intelligence further toward the periphery of the health care system. The low cost of microprocessors will make it possible for quite small health care units to purchase and operate the systems, and also because those sys- tems that can be operated “stand-alone” obviate the costs and the communica- tion problems associated with the older systems. Where large central computers are used today, the operating units of the medical system are connected by cable or by telephone lines. Numerous concommitant costs and inconveniences of time-sharing or multiprocessing will be bypassed as microprocessors simply do the job at the level of the service unit without recourse to central facilities or to central management. On the other hand, there are potential centripetal effects as well. Some MIS subsystem applications that currently use central computers solely could use microprocessors for performing services at the patient site. At the same time such subsystems could still serve as sources of valid input in hierarchical relationship to central computer record storage systems. Synergistic relation- ships between peripheral microprocessors and central larger computers are quite possible. The central machines could even dispatch computational and mechanical jobs to the peripheral units. The interpretation of electrocardiographic records by computers is a prac- tical example of a subsystem task that is currently centralized. In 1976 four Effects of Changes in Technology on Future MISs 133 million such patient records were processed in this way.7 The examinations were made at the patient site, recorded electronically, signals sent over telephone lines to a central computing facility, interpreted, and results sent back elec- tronically and/or verbally to the site by telephone lines. Since inexpensive microprocessing systems now have greater computer power than the older central computers, transmissions to the center need no longer occur. At least, there is no longer any technical reason for central interpretation of the single electrocardiographic record. Likewise the various difficulties associated with correct patient and sensor lead identification are better handled with computer logic at the patient site rather than centrally. Such considerations encourage one to believe that the microprocessor technology will support advanced information processing close to the patient. This is an important goal, especially for sparsely populated rural areas. There are, however, functions that may prove to be best done centrally. These include storage of the records and/or interpretations of previous electrocardiographic records and registry functions such as the recording and surveillance of pace- maker implants.8 All such information systems are most efficient when heavily used. The capital-intensive nature of the investment in even small computers may cause such services to be excluded from the sparsely populated areas most in need of automated assistance. A survey of automated EKG systems has recently been completed.9 Among nine factors that tended to encourage the diffusion of this technology was decreasing costs of equipment for low- volume users. Among six factors observed to mitigate against diffusion was increasing cost of transmission over telephone lines. In spite of the fact that the computational power of microprocessors permits jobs such as EKG interpretation to be performed locally, many other factors such as profitability, maintenance, and sales problems work against this possibility. For example, in 1967 the Caceres EKG interpretation programs were first demonstrated in a statewide service, concentrating on rural areas.10 There were no major technical or scientific problems. The limiting factor was cost, which derived from three elements. The cost of the central computer (a CDC 8090) was about $90,000. The cost of the data acquisition and trans- mission cart necessary in each physician’s office or clinic was about $9,000. The cost of the telephone calls varied with distance, but was always governed by high intrastate tariffs. In contrast, twelve years of progress in computer engineering have brought changes and improvements of all kinds. So one is told. In fact, the newest of the electrocardiographic interpretation systems (and certainly an excellent system) is the Hewlett-Packard ECG Management System. This system now provides all the program features which were needed twelve years ago in a machine with ten times the throughput capacity of the older computer. On the other hand, the cost of the new device is about $150,000, the cost of the data carts needed on-site is about $14,000, and intra- 134 Growth of MISs in the United States state telephone call charges are similarly inflated in price. The net effect of this progress is to make such services profitable in metropolitan centers, and to leave rural areas even more underserved than in the past. A contrasting situation exists with respect to another special computer service, the planning of radiotherapy for neoplasms. Computers can optimize detailed plans of this therapy in ways that match the expert and exceed the ability of the nonexpert. Consequently, radiotherapy planning services have been provided by computer centers.11 In this application, the new stand-alone machines probably will be sufficiently inexpensive duplicates of this capability to permit the services to be performed in a less centralized fashion. At least one radiation planning system has been reported that is operated on a micro- processor-based graphic display.12 The discrepancy between the old, rather modest computing and patient management ideas and the substantial computing power available locally with microprocessors may determine the eventual effect. The new computer tech- nology may decentralize decision making because complex decision making can now be executed on inexpensive stand-alone machines close to the patient. The effect of the technological change represented by microprocessors will be to permit information processing intelligence to move to the periphery of the health care system. Whether the new engineering possibilities of microprocessors will be enough to overcome the long-run tendency to cluster technical capa- bility in metropolitan areas as opposed to rural areas is not clear. Within an institution, where the periphery is represented by individual departments and clinics, the change to local microprocessor-based medical computing support will surely occur. Changes in MISs caused by technological progress will have consequences for research as well. Microprocessors make it possible to reexam- ine patient care applications that were not practical when they required con- nections to remote, large time-sharing computers. Estimation of Outcome The net effect of the microprocessor technology is inherently centrifugal. This tendency presents the problem within single institutions such as hospitals that, unless the developments are managed, the effect might be further to compart- mentalize and fragment the institutional systems. On the contrary, the technical possibilities presented by microprocessors for information processing in rural areas might not come to fruition without additional strong nontechnical sup- port. Microprocessors are but devices. They could be used to distribute medical knowledge, to strengthen medical capability at remote locations, and to facil- itate appropriate communication in an orderly overall system. The configuration of the overall system need not be strictly hierarchical or even star-shaped. Effects of Changes in Technology on Future MISs 135 It could be an organizational and management network of any form appro- priate to the nation’s needs. It is clear, however, that the microprocessor tech- nology does not demand and will not itself create a management network. Quite the reverse. Only a substantial management effort on the part of society can shape the technology toward an integrated national health care compe- tence. Laser-Etched Disk Storage Systems Description of the Technology The common digital computer of 1979 is virtually identical in architecture to the early so-called von Neumann stored-program computer of 1950. The central processing unit (CPU) contains an arithmetic-and-logical unit that oper- ates on data or instructions read from computer memory. It does one opera- tion at a time. The memory itself has taken various forms. Main memory for most computers in the past has been ferrite magnetizable cores. Microprocessors use other sold-state information storage devices. In either case, the large mem- ory stores of the computer are one level below main memory, usually taking the form of rotating magnetizable disks onto which information is written and read in binary (on-off) form. Magnetic disk memory is still relatively ex- pensive. Typical 1978 commercial disk memory costs about one cent per 1,000 bits of on-line storage. Microprocessors do not reduce this storage cost; they reduce only the CPU cost. Consequently, disk storage still constitutes as much as 40 to 70 percent of system costs for the hardware of an MIS system, and could constitute 90 percent of the hardware costs of a microprocessor-based system. Thus disk storage limits the capacity of many systems. Laser-etched optical storage is a new development with the potential to change those considerations and the technical mode of operation of MISs in a radical way. Commercial disk storage units have been announced by N.V. Philips in Holland and MCA in the United States. These devices are sometimes called video disks because one planned consumer application is likely to be the retail sale of disks that contain commercial television programs or motion pictures. In this form, the disks are about the size of 33 1/3 rpm phonograph records. The physical recording medium in the case of video disks is a thin metallic film in which holes or pits have been burned by a laser. The record- ing is made in concentric tracks, as are the traditional magnetic disks. The laser-etched holes or pits represent the equivalent of binary on or one bits. Reading is likewise done via a laser under computer control. The information capacity of the disks is great: 2 X 1011 bits per side. This is sufficient for digital storage of about 30 minutes of commercial television. It is as great in 136 Growth of MISs in the United States terms of data elements as the storage available on current large commercial computers, and much larger than that for most hospital subsystems. Costs of the new storage medium are estimated to be one cent per 100,000 bits of on-line storage.13 Two peculiarities are inherent in this technology. First, the disks are “write once.” That is, once the holes or pits have been burned, they cannot be changed. Second, writing the original or master disk is more expensive than making copies. Major Uses of Laser Disk Storage in MISs No one is absolutely sure of the eventual uses of this technology in MISs. Units are just now being produced commercially. The range of applications for ordi- nary digital storage of digital information will depend on costs, complexity of writing the disks, and price changes of competitive storage media. The most common use will likely be in data storage. The result will be to remove virtually all limitations on the size of the storable data file. Certainly from the point of view of any personal file, storage would become essentially infinite. Laser disk storage offers an even more interesting use in medical record work in the storage as images of those parts of the patient chart which are not digital to begin with. It is possible that some portions of the medical record can be treated forever merely as images. That is, since the systems are used to store motion picture and television images, they may also find a major use in the storage and retrieval of medical images. What might the medical images be? The most obvious initial candidates are the radiant images in radiology: the roentgenograms, fluoroscopic views, and radioisotopic views now stored on film or paper. In addition, radiology departments are faced with the problem of management of the information produced through a variety of new techniques. These include computer tomo- graphic examinations, ultrasound studies, and, in a few cases, images of patients made with positron cameras and electromagnetic resonance image studies. There is more than one reasonable way to deal with each of these studies. Some, such as computer tomography, result directly in digital images. In such studies the information resides “in the numbers.” The technique of the medical examination itself results in absorption densities, not in images. The image the radiologist sees is calculated or derived from the digital numbers. At present, a few selected images are stored in the medical record in the form of paper photographs of the computer screen display. Yet a large number of possible images can be calculated from any given computer tomography scan. An as yet unresolved question is whether it is best to store digitally the unambiguous and unique set of images the radiologist inspected, or whether it will be better to store the much more voluminous numbers representing the raw measure- ments from which the images are calculated. In either case, the new disk Effects of Changes in Technology on Future MISs 137 technology may offer the only practical system for long-term storage of either representation of this portion of the medical record. In contradistinction to this situation, the digital representation of commer- cial television frames is simply one-for-one. In addition to the potential to store radiological images, the most obvious extension of the laser disk technology to MISs is for storage of images of the pages of the current handwritten medical record. Handwritten records have up until now been an anathema to the com- puter systems analyst because there was no way to process them. Yet most of the medical record is a handwritten document, for the most part created by physicians and nurses. Great efforts have been expended by MIS builders to devise systems and apparati so as to cajole physicians and nurses to enter this information into computer systems via keyboards. The purpose of entry into the system has generally been so that the computers could store the information. The major purpose of the storage was to make possible redisplay of the same information later to the same or different physicians and nurses. This had been termed the data acquisition problem, and has been a major obstacle in all MIS implementations. Laser disk storage could obviate the difficulty of the data acquisition problem. Even with the new means of image storage, some digital identifica- tion data would be required for computer management of the image. At a minimum, this would include patient number, date, and a form identifier. These items could easily be precoded so that no keyboard entry would be needed. The page images could be transmitted or accessed by whatever hospital or office locations require them. The more data that were keyed or precoded in overtly digital form, the more the system could evaluate their meaning. Yet, retrieval of the page image itself, or better yet, selected portions of many pages for the medical attendants and appropriate hospital departments, would completely satisfy the initial major objective of the MIS. This ability to store away with an identifier and a few index terms the image of one’s private papers was imagined by Vannevar Bush in 1945.14 He foresaw by a decade the coming of the general digital computer and by three decades the arrival of the personal computer. His imaginary MEMEX system for document-image management comes very close to what might evolve from the current MIS systems combined with laser disk storage technology. Major Effects of the Laser Disk on MISs A major obstacle to MIS implementation has been the unacceptability to physi- cians and nurses of the teletype-style keyboard as a means of keeping a medical record. There have been other constraints of course. No system has been able to handle unstructured, free text. Keyboard and display screen systems have heavily structured the input data with respect both to content and to format. 138 Growth of MISs in the United States Insofar as the new technology could usefully operate on a simple page image, it could completely bypass the major social impediment to MISs. The laser disk storage scheme also requires a rethinking of our approach to medical information. Which items, for example, have been assiduously key- stroked into computers simply to display back to users? If such data elements are not used by the computer logic to relate to other data elements, they need not be treated as overtly digital information. Only where the computer logically uses information for some purpose beyond storage and retrieval need we actually acquire the information in a digital form. It is not profitable to speculate further at this early stage in the evolution of the laser disk technology. In addition to script, a host of potentially useful medical images are candidates for storage, depending on eventual costs and clearer estimates of the benefits. Images of abnormal blood cells, chromosome karyotypes, patient gait and visage, cytology-histology microscopic fields, and pages of relevant scientific literature are examples. Images of all these elements are sometimes needed in the patient chart. We have no practical way to include them in the medical record now; and hence, have no clear data on benefits. Again, the general effect is to oblige a re-evaluation of the elements of the present medical record. Estimation of Outcome from Changing Technology Successful use of the laser disk for storage and retrieval of page images of the medical record would represent a distinct move toward use of the computer system as a partner in the human-machine synergism, as opposed to a develop- ment further toward automating the record. A system in which the machine stores and fetches the page image would still be very much dependent on the human being who does the reading. This relationship is more explicit than with present systems in which the human must deal with the information transaction more or less on the machine’s terms. On the other hand, the less the computing system does to guide and structure the input, the less its con- tribution. An entirely new balance would need to be struck. Such a new system would present the absolute need for doing two things which many investigators and research administrators have resisted: 1. To complete basic research on the human-machine interface with respect to computing terminals, and deal with the dual problems of acceptability of these systems to the individual users and acceptability to new groups of users, including public users. 2. To direct the development of the new system contingent on the results of research concerning acceptability to individual and to group. Effects of Changes in Technology on Future MISs 139 The reader may reasonably fear there is some exaggeration in urging the need to manage the development of such a technology with social and technical goals equally in mind. The alternative is to permit the new technology to be shaped by market forces and ad hoc engineering decisions. Commercial tele- vision is a technology which has grown in just this manner. Current U.S. television sets scan at 525 horizontal lines, an enforced U.S. standard. We frequently hear that this prosperous country has more television sets than bathtubs, and that, at the same time, television’s potential for public education is great. The scan standard, however, is so coarse that text cannot easily be read from a television screen. Consequently book pages in effect cannot be displayed. Computers, which can easily be attached to television sets for display, cannot display significant amounts of text words. They are currently displaying electronic ping pong games or their equivalents. The lesson is clear. The technical standards that emerge from MIS displays will determine absolutely the purposes to which they can be addressed. It is important that the standards for display screens for the new laser disk image systems be chosen so that the technical standards do not exclude appropriate scientific and educa- tional information display. Furthermore, since the technology is strongly driven by a consumer market, that is, potential retail sales of video storage and replay machines, the potential benefits from computer coupling to these systems is of broad societal importance, not merely a matter concerning medical informa- tion automation. Communications Technology Description of the Technology The technology of electronic communications is essential to full deployment of information networks and computing systems. The range of devices and techniques is broad, running from orbiting earth satellites to digital dial tele- phones. Much of interest has been written about this field, indeed far too much for proper summary in this work. Yet the relevance to MIS development can be stated simply. Health care providers may be able to join or use the emerging telecom- munications networks. Such use may facilitate the creation of interconnec- tions between hospitals and their satellite locations, or other local or distant health care institutions. The health-related uses of these networks will not significantly influence the nature of the network. Rather health activities will benefit from systems built for other purposes. What is new and relevant to MISs in this rapidly changing field is the progress that has been made toward standardizing the means to interconnect computer networks. The technology of course takes the form of hardware 140 Growth of MISs in the United States devices, e.g., new transmission equipment such as concentrators and switches, new terminals and interfaces, or modem devices. More important, the hardware that emerges, whose engineering is already well known, will be determined by the agreements and standards for usage that are being discussed. Ira Cotton, in a review of computer interconnection problems and prospects, notes: The data communications world has begun to undergo a major change in recent years, from an ad hoc array of “home brew” systems that worked, more in spite of than due to, publicly available facilities, to a more planned and integrated set of facilities designed both for the communication of data and for public use.15 The major elements of the technology that have brought about the im- minent change are: the introduction of computers into the communications system as computer costs fell steadily and rapidly compared with communica- tion costs; and the consequent advent of packet-switched networks that use computer-based intelligence to decrease the amount of communications band- width cost required. Savings are achieved by utilizing the communications link only during transmission of the highly compressed packets of information, rather than with traditional circuit-switched telephonic systems in which the communications circuit is tied up for the duration of the conversation or mes- sage exchange. Many other features are available with the new systems, such as alternate routing, enhanced error detection and correction, and adaptation to the proto- cols and speeds of many kinds of terminals. These additional features form the basis for a new type of business, the value-added communication network companies. Such services are provided by firms that lease the basic telephonic connections from common carriers, and then add on both services and charges to the computer network customers. Permission for such companies to do business, and also for microwave communication companies to operate in competition with the regulated utilities, has been given by the Federal Com- munications Commission (FCC) only since 1971. The new communications network technology makes it possible for any user, medical or otherwise, to use his computer terminal to connect either to his local computer within the hospital, or to connect to a regional service bureau computer, or to send to or receive data from a computer terminal or network anywhere in this country (and others with whom bilateral arrangements have been made). The communication charges with some arrangements are essen- tially independent of the number of miles between the user and the host com- puter, and are based solely on volume of data transmitted. Likewise the user is unaware of the transmission conventions and modalities employed. The efficiency of the communication is such that the costs are frequently less for special or rarely used functions than if they had been maintained locally. Some computer networks use proprietary, obsolete, or deliberately obscure protocols Effects of Changes in Technology on Future MISs 141 and hence are difficult to access. Under the best of circumstances, however, the terminal user can get in and out of many remote networks with no more trouble than in using a local computer. The engineering principles employed are presumably universal, but the competitive situation that stimulates these advances is unique to the United States. A number of companies are actively developing, proposing, and oper- ating advanced communication systems of various kinds. These include Tymnet, Telenet, Satellite Business Systems, Microwave Communication, Inc., and International Telephone and Telegraph. In July and November of 1978, the American Telephone and Telegraph Company requested permission of the FCC to establish their Advanced Com- munications Service (ACS) using existing Bell System digital facilities.16,17,18 ACS is designed to allow computers to be linked to a distributed communica- tions network without regard to proprietary hardware and communications compatibility. According to one analysis, “Initially, ACS will support about 450 different terminal models from one hundred vendors.”19 A computer manufacturer, in endorsing the ACS proposal, stated that the development . . . will significantly broaden the market for small computers and terminals, and allow a greater number of users to take advantage of computer networking. Users will be able to select computer hardware based on the individual features of the equipment rather than being forced to make decisions based primarily on the need to assure com- patibility with existing systems.20 In February of 1979, ATT reaffirmed its intention to offer ACS, but announced a delay in its filing of tariffs because of unforeseen software work to be done.21 Major Uses of Communication Technology in MISs MISs developed for particular institutions usually use cabled terminals and relatively short-distance traditional teleprocessing. Commercial MIS services already make extensive use of traditional teleprocessing communication ser- vices, sometimes over relatively long distances. McDonnell-Automation, Shared Medical Services, Technicon Medical Systems, and Medical Information Tech- nology are examples. Where the distances between the application site and the host computer are substantial, smaller minicomputers are often placed at the hospital sites to do data acquisition and preprocessing. So far as published reports reveal, none yet uses the new packet-switching communication network technology. Undoubtedly, the major information system vendors will in the future use whatever public and private packet-network services can increase their customer roster and reduce communication costs. The advantage to the Growth of MISs in the United States 142 vendor will be reduced overall costs and hence pricing more competitive with manual methods, along with the potential for competition through the pro- vision of special communication features. The advantage of such communication networks to medical users will be the ease with which they can deal with multi- ple vendors. One medical advance of potential importance to MISs, the knowledge-based consultant systems to be described next, is being developed through the use of the advanced networking technology. These artificial intelligence projects at a number of institutions are dependent on connections to either or both of the special facilities at Stanford and at Rutgers.22 A number of examples of specific MIS functions appear to be candidates for which a hospital might in the future consider a remote vendor of services, using an appropriate computer networking scheme. These include: Monitoring of drug-drug interactions. Updating the data base is a kind of national responsibility. Massachusetts has already established a statewide Consortium Poison Control System and a centralized Information Center to maintain its data base.23 Transferring and abstracting records of previous hospital admissions. Trans- fers from other hospitals such as the Veterans Administration is an ex- ample. Performance of ordinary hospital information functions not yet developed locally. This might include hospital bed advance booking, scheduling of the surgical operating room, dietary menu planning, and electrocardiography interpretation. Operation of these systems on local microprocessors will be an alternative to access via communication network. Appraisal of hospital and medical functions. HAS and PAS are remote batch services that currently serve this need, but there is room for plenty of competition. Special clinical consultations. An example is advice concerning diagnosis and management of unfamiliar or difficult clinical problems. Examples of prototype knowledge-based consulting systems are described in the fol- lowing section. Major Effects of Communication Technology on MISs The effect of advanced communications capability on MISs cannot be a direct, dramatic one. Medicine is too small a part of the computing business, and our ideas for MISs are yet too far from complete. The ultimate effect of communica- tions technology on the stand-alone hospital MIS, however, will be as gradual Effects of Changes in Technology on Future MISs 143 and as profound as the effect which the automobile and rural electrification had on the one-room schoolhouse. The most advanced MIS systems are very much like excellent one-room schoolhouses. Users of the Utah HELP system, for example, cannot connect to and use logic in the PROMIS system, or vice versa. Even COSTAR and the Institute of Living systems are totally separated, in spite of being written in the same computer language. Currently no MIS system can use any part of the capability of any other MIS system, even if it is relevant, without the necessity to purchase and install a complete duplicate of the system with the desired feature. Computer communications networking will not cure all of these prob- lems, but it has the capability to obviate those obstacles which stem from electronic and software intercommunications incompatibilities. These are a major obstacle in themselves. Consequently one would expect to see more hospitals considering the possibility of obtaining one information processing service from remote source A, and a second service from remote source B. Their decision to use a remote rather than a local source could then be based solely on selection of the best logic available for the job. On the commercial side, better communications would enlarge the mar- ketplace for each vendor and increase competition within any locality among vendors. The effect on the hospital consumer of a greater range of choices among remote and local vendors and service bureaus might be confusing at first, but would soon be a happy dilemma. In the case of communications technology, the effects on MIS development and diffusion do not appear to be contingent primarily on the emergence of new devices. Rather the critical issue is how the present capabilities can be deployed for more general social purposes. The technical ability already exists to interconnect computers through telephone lines, cables, and communica- tion satellites. There are many difficulties with such schemes, including serious technical, social, and organizational problems. Networking merely provides technical interconnections between different computer systems. Stephen Kimbleton, in a report from the National Bureau of Standards, cautions that networking . . . does not provide any means to alleviate the effects of differences across systems. As a result, network users are forced to resolve such differences; the resulting learning and programming burden inhibits effective utilization of networking capabilities to achieve active re- source sharing.24 He reviews the concept of Network Operating Systems as a potential means to mask system differences from the computer network user. Access to service via a computer network is viewed as proceeding through the stages of acquisi- tion, initialization, utilization, and termination by Watkins and Kimbleton.25 They make the assertion that of these stages, utilization is not amenable to 144 Growth of MISs in the United States centralized network level support. Functions for the other three stages can be provided for by the network. The implication for a hospital or medical network user is that he or she must have the capability of dealing at least with the in- dividual peculiarities of each of the application programs themselves, even though the interconnection between user and program were provided through network auspices. These increasingly refined differentiations reflect a growing experience with working scientific and commercial networks, as well as theoretical exploration of the developmental tools that may be necessary. An example of the latter is the concept of the National Software Works.26 Such discussions of operational problems contrast with what in retrospect were farsighted but somewhat naive expectations for networking in earlier years. In 1965, for instance, the Massa- chusetts Institute of Technology (MIT) Planning Conference for the INTREX Project predicted that the library of each university would be the hub through which faculty and students would access documents, data, and knowledge via a network extending throughout the academic community. The report of the conference suggested that: The users of the network will communicate with each other as well as with the library; data just obtained in the laboratory and com- ments made by observers will be as easily available as the text of books in the library or documents in the departmental files. The information traffic will be controlled by means of the university’s time-shared computer utility. . .27 A recent RAND Corporation position paper describes the feasibility of developing local, regional, and nationwide utilities for information storage and retrieval.28 Individual hospitals could potentially obtain access to special data bases or to special services. The RAND paper points out, however, that the practicability of such systems will be determined by economic questions. Gabriel Groner and col- leagues identify three economic issues as needing special study. These are: 1. The costs of computer networks using broad band width coaxial cable, as well as other methods of data transmission such as com- munications satellites and telephone lines. 2. The importance of interconnection costs in comparison with the many other costs associated with computer use. 3. Two regulatory issues: (a) problems associated with allocating common costs among multiple users, and (b) implications of the recent FCC decision to permit new specialized common carriers to compete with the Bell System in supplying interconnection for computer uses and for many other purposes.25 Effects of Changes in Technology on Future MISs 145 Artificial Intelligence Techniques Microprocessors, video disks, and telecommunications networks are primarily hardware developments. Each requires some programming and new concepts, but the actual development resides in a device or system of devices. It is a truism in the computer field to note that hardware advances tend to outpace the growth of programming of software techniques. Indeed there are in the software sphere no new concepts, developments, or languages that stand out as comparable in magnitude or importance to the hardware growth factors. Yet one trend should be noted because of its potential to support major medical and scientific advances in computing, including especially MISs. The trend is the increasing use nowadays of the computer science techniques known as artificial intelligence as a means to support the development of medical knowl- edge-based systems. To elevate the importance of this new usage in medical applications of known computer science technique to a level comparable with microprocessors, is not merely a personal judgment. Others share this expectation. Yet even so, the reader should be aware that we are now discussing an extremely specula- tive area. If predicting the future is hubris, predicting medical software develop- ments is the rankest hubris. Description of the Technology Edward Feigenbaum presents a brief description of the general field of artificial intelligence. Artificial intelligence research is the part of Computer Science that is concerned with the symbol-manipulation processes that produce intelligent action. By “intelligent action” is meant an act or decision that is goal-oriented, arrived at by an understandable chain of symbolic analysis and reasoning steps, and is one in which knowledge of the world informs and guides the reasoning.30 In this broad guise, artificial intelligence technology is employed in many nonmedical areas and has goals much broader than health care per se. These include fundamental investigations aimed at understanding heuristic search, creation of general systems for semantic information processing, and develop- ment of precise models of human psychology in symbol-processing terms. Early research in the field employed game playing and robotics as experimental test systems. Pursuit of these general goals has come to include recently a focus on 146 Growth of MISs in the United States methods of representing knowledge and the subsequent creation of high- performance systems that perform at or near the level of human experts in specialized fields. Feigenbaum points out that the problem-solving power ex- hibited by such systems (i.e., agents) ... is primarily a consequence of the specialist’s knowledge employed by the agent, and only very secondarily related to the generality and power of the inference method employed. Our agents must be knowl- edge-rich, even if they are methods poor.31 The spill-over into biomedicine of concepts stemming from analysis of human reasoning in information processing terms is claimed to have had a substantial scientific impact, especially in psychology and psychiatry. An ex- ample is the concept of paranoia as a data input transformation. In chapter 4 we examined scientific impact as an aspect of evaluation. Regrettably, no rigorous studies of scientific impact have been reported in the case either of MISs or of artificial intelligence techniques. The medical applications of artificial intelligence techniques which are relevant to MIS building have so far been limited to the creation of knowledge- based programs that act as clinical consultants over a specified range of medical problems. In such systems, medical knowledge of the special area is represented within the computer program. The knowledge is built into the program through explicit interaction with one or more medical subject matter experts. The struc- turing of the knowledge and the interlocution between computer and expert is performed by a new kind of science professional called a knowledge engineer. The system is developed by the usual train-and-test iterations. After the program has been completed, it is typically used to provide consultation. The findings concerning a particular patient, real or imaginary, are transmitted to the program by the human medical user. Since the user is a qualified medical observer, but a nonexpert in the field in question, he or she desires to obtain the reasoning of the program, which functions as a surrogate for the expert or group of experts whose reasoning it has come to embody. The output of the program is com- municated through a terminal system and takes the form of diagnosis, recom- mended therapy, a request for further clinical testing, or whatever form is appropriate to the circumstances of usage. A number of programs of this general type have performed with amazing facility. One author described the output of such a chemical consultant pro- gram as in “world’s champion class” for certain specialized families of molecules. Some purely clinical programs also give outstanding performances. Nonetheless, such programs are still generally being used in a research mode. Effects of Changes in Technology on Future MISs 147 Major Uses of Artificial Intelligence Techniques in MISs We will consider evidence that artificial intelligence programming and system building techniques have succeeded in creating computer programs that perform in medical settings with information processing skill comparable to or exceeding that of the human expert. If the reader will grant for the moment that the feasibility of this development has been demonstrated, it is then necessary to deal with the question, why is this so important? Why in other words would such a development be expected to have a general effect upon many MISs or upon many parts of the MIS as we know it? The answer has three parts. First, the artificial intelligence approach has the capability to enlist the enthusiasm and participation of the biomedical community because it focuses on the essence of processes (the knowledge structures) rather than the superficial appearances of the processes such as tend to be addressed in ordinary programming techniques. Second, the artificial intelligence approach urges on the user advanced and relatively comfortable programming and system building styles. For many knowledge bases, English language is used because it frequently permits the most natural formulation of present knowledge. Similarly, high-level program- ming languages such as LISP and INTERLISP interpreters are often selected because of the relative ease with which such programs can be modified or redone during the developmental stages of systems building. This approach, which may be inefficient with respect to computer re- sources, banks on the assumption that computer resources will always be orders of magnitude cheaper, whereas human input costs will always be more ex- pensive. Consequently developments that increase human input efficiency and reduce operating costs at the expense of computer capital expenses will almost certainly be worthwhile investments. Third, the artificial intelligence approach thrusts on its users a new set of goals for their work. The goals include, in addition to the fundamental step of producing a formal representation of medical knowledge for a given problem area, the self-imposed requirement that the final knowledge system be able to justify its conclusions (for example, its decisions or recommendations) to the user. Merely to present the justification is not as difficult as it sounds. One of the systems we will cite helps to choose an antibiotic for treating a bacterial infection. It justifies its recommendation by indicating that it was given ob- servations A, B, and C, and hence invoked rules 1,2, and 3, which inevitably produced its recommendation. The technique, however, is powerful and well- suited to the medical context. Such programs attempt to understand the prob- lems presented to them, stand ready to reveal the heuristic logic and specific knowledge by which they analyze the problem situation and ask to be judged according to the standards of human subject matter experts. 148 Growth of MISs in the United States There is great strength in this approach. Also there are some serious dis- advantages. The largest one is that the functions selected for the artificial intelli- gence problem solving must be understood sufficiently well that the knowledge can be made explicit and formal. Yet the medical functions must be significant. The degree of success would be measured in part by whether the problem- solving circumstances was of sufficient importance to warrant the effort to create the system. Like all good research, it must be feasible but important. Not all such applications have proved to be feasible and not all the finished programs here addressed major problems. There are, however, enough signi- ficant successes to suggest that this new technology will radically change the way MISs are built in the future. Selected Current Medical Applications of Artificial Intelligence. A major pio- neering venture in artificial intelligence work was the DENDRAL system of Edward Feigenbaum, Bruce Buchanan, and Joshua Lederberg for analysis and interpretation of mass spectral chemical data.32 This system deals with input data somewhat analogous to physiological signals. Fortunately the DENDRAL system has the benefit of having sound chemical models of the data generating process on which to commence its reasoning. The major obstacle to employing such techniques directly in clinical information systems is the selection of problem areas sufficiently constrained so that good theoretical constructs comparable to the atom-molecule models are available as the basis of the reason- ing systems. Based on understanding and techniques derived from the mass spectrometer work, another knowledge-based system called MYCIN was built by Edward Shortliffe and others.33 This system acquires information concerning patient and laboratory data and offers a recommendation concerning the optimal antibiotic or antibiotics for treating the infection in question. The logic takes the form of production rules, which are English language statements of experts which have the form “if. . . then . .. .” About four hundred such statements were sufficient for septicemia. There are many reservations that can be brought against such a system. There is still ample room for improvement, especially if the system were to aspire to be a true infectious disease consultant. On the other hand, the major practical difficulty encountered in actually using the system has been the excessive amount of time required for the user to provide to the system the facts about the patient and the laboratory findings. Were MYCIN to become a integral part of a working MIS, virtually all the necessary observations would already be part of the computer record. Its ques- tions to the user physician could then be limited to a practical, small number of high-level interpretive issues. The same limitation on the need for facts about the patient will hold true for all artificial intelligence consultant programs. All would be far more effective as parts of working MISs, and the MISs would be far more powerful with such a Effects of Changes in Technology on Future MISs 149 staff of internal expert consultants. At least one of the advanced MISs, PROMIS (described in chapter 4), is founded on Weed’s determination to keep the auto- mated information system whole and in possession of the entire patient record. INTERNIST is an artificial intelligence system whose task is differential diagnosis in internal medicine.34’35 It is the result of collaboration between the expert diagnostician Jack Myers and the excellent computer scientist Harry Pople. The knowledge is represented as a semantic network. It is difficult to quantify the performance of such programs, but the reader will understand something of the sophistication of the logic by knowing that INTERNIST can operate correctly on Clinical Pathological Conference cases from the New England Journal of Medicine. Again there are difficulties in entering all the necessary patient findings into the computer interaction. Again such a system would be a wonderful and synergistic addition to an MIS. CASNET and its successor program XPERT are artificial intelligence pro- gramming systems created by Casimir Kulikowski, Saul Amarel, Sholom Weiss, and colleagues at Rutgers Computer Resource and Aran Safir at Mt. Sinai Medical School.36 The program deals with the management of glaucoma. The logic is a combination of rules and network structure. Insofar as possible, the system utilizes the conceptual model of the anatomic structures of the eye, and necessarily the current level of understanding of the physiology of the normal and diseased eye. The CASNET knowledge base was built through interaction with a number of centers for treatment of glaucoma. Information for building the knowledge base was collected through a formal computer network for data sharing. Operation of the program yields advice at the level of human experts. Specific Future Uses of Artificial Intelligence Techniques in MISs It is not likely that artificial intelligence approaches will be employed in the fore- seeable future as the overall logic for entire MISs. Rather it seems more likely that this approach will permit many individual parts of the MIS to exhibit intel- ligent, goal-directed activity. There are several functional areas of MISs in which goal-directed activity would be obviously appropriate. These are noted below. Editing and Evaluating Input Data. Many current systems do not edit data at all. When it is done, the checking is most often performed against COBOL-type masks which look for alphabetic characters in numeric fields, field length restric- tions, arithmetical range restrictions and such considerations. Essentially no checking is done which takes into account semantic meaning. The general avail- ability of techniques that facilitate editing according to actual meaning would greatly enhance the ability of MISs to record patient history, nursing notes, physical findings, interval progress notes, problem statements, and treatment effectiveness. 150 Growth of MISs in the United States Laboratory Quality Control. In spite of the fact that clinical laboratories often deal in numerical quantities, the logic by which quality is controlled is rich in highly content-specific knowledge and the evaluation of uncertainties. To be brief, there is currently no suitable automated laboratory quality control system. Control where it exists derives from close human supervision by an expert. Traditional programming systems have proven cumbersome and inflexible when tested against the extremely dynamic new methodologies and procedures. Again a new approach would do a great deal of good. Interpreting Laboratory and Physiologic Information. This potential applica- tion is exemplified by the work of Homer Warner, Reed Gardner, and colleagues in the Utah HELP system.37 Expert Clinical Consultation. This area is clearly a fruitful and needful one, especially as American medicine is pushed headlong into an emphasis on pro- duction of generalists.38 Working examples of prototype clinical consulting systems have been cited. This function will tend to include diagnostic systems and systems that guide patient management according to treatment protocols. File Management. The most difficult MIS area is the management of very large files. Video disk systems, for example, solve only the physical storage problem. It is clear that an entirely new approach to the logic of retrieving records is needed in order to use the future storage devices. Work on the use of artificial intelligence techniques for searching large files has just begun.39 The technique and the problem have a kind of natural good match. The search trees of pos- sible solutions in artificial intelligence problems are typically too bushy to permit an exhaustive search. Consequently it is customary in this work for the heart of the system to be an algorithm that prunes the tree or generates only the most likely branches to be searched for solutions. The analogy is relevant to files or even indices that may in the future be far too large to permit ex- haustive searching. Major Effects of Artificial Intelligence Techniques on MISs The major effect of increased use of artificial intelligence techniques will be to enhance the development of the most advanced MISs. The difference between the systems built for a particular institution and those commercially available general systems will increase. The institutionally-specific systems will continue to lead the way. Artificial intelligence techniques will be synergistic with further development of all three major new hardware technologies. MISs such as the Utah HELP system and the Vermont PROMIS system Effects of Changes in Technology on Future MISs 151 already incorporate substantial amounts of medical knowledge, albeit with essentially traditional computing techniques. The Utah system contains over 1,400 representations in which incoming patient care data can be evaluated or interpreted by the system itself according to its coded instructions.40 As noted previously, these representations have already resulted in the system’s ability in 60 kinds of circumstances to alert medical users to potentially dangerous situations which require human intervention.41 Considering that the system originally was designed to test and process physiological signals, it has evolved a long way in the direction of a primary knowledge-based system. As the general techniques for knowledge representation become better known, advanced sys- tems such as the Utah one will be found ideally situated to move forward fast. In a similar way the PROMIS system has a commitment to contain internal- ly a kind of standard of good medical practice. It is according to such an im- plicit standard that PROMIS currently guides and monitors the actions of its users. The system is already fundamentally knowledge-based. The state of the art of knowledge engineering is not sufficiently advanced or widely known yet to have influenced this system directly. Weed and his associates are, however, not unaware of the problem. It is because of the complexity of writing, working with, and maintaining the large number of frames within their system, that they went to the effort to create their own programming language. It seems likely that the future will hold even more powerful aids for this work in the form of the newer programming systems addressed directly at the knowledge representation problem. With respect to the hardware developments, artificial intelligence tech- nology is also synergistic. The new microprocessor hardware is approaching in size and capability the older central computing systems on which the knowl- edge systems originally were developed. It is certainly reasonable in the future to put intelligence as far as possible out toward the system periphery, that is, to operate the program on microprocessors, close to the user and to the patient observations. The predictable advent of larger, cheaper video disk storage mem- ory at the periphery enhances this capability by providing for local storage of the knowledge bases. The relationship of knowledge-based systems to net- working is also interestingly synergistic. Here the situation is somewhat differ- ent. It is desirable that knowledge-based systems reside at the local site of use once a final system has been created. It is often necessary, however, to create systems by using a centralized special facility that can provide the computa- tional ease and capacity that enhances experimentation in the developmental stages. Access to such a developmental facility (for example, the Stanford SUMEX facility or the Rutgers Artificial Intelligence Resource) has been pro- vided to current collaborating institutions through communications networking arrangements (for example, the existing commercial networks TELENET and TYMNET and the federal ARPANET). Experience with successful development of knowledge-based clinical 152 Growth of MISs in the United States consultant programs via remote use of these two special facilities suggests that the present networking is adequate for development of the new programs. Cur- rently the interconnections are for the most part between remote terminals and the two central computers, with only infrequent interconnection of the two large computers. Improvement in the speed and efficiency of the future com- munications networks—especially improvements in the facility to attach com- puters and intelligent terminals to the networks—will greatly enhance the devel- opment of future knowledge-based computer programs. Yet another aspect of these developments is synergistic with communica- tions network technology. The ambition of developers of knowledge-based systems to explore fully machine representations of human reasoning in medical and other domains will doubtless continue to exceed the capacity of local processors of whatever future great capacity. The most advanced experiments, in other words, will continue to require access to the most capacious computing systems available anywhere. The new networks technology means that such access can be arranged (at least technically) at a regional, national, or interna- tional level. Which arrangement emerges as practical will depend on future economies of communication costs and on public policy in the area of federal research management. The second major effect of artificial intelligence technology on MISs will likely be to reduce the art and to increase the science in medicine. The effect thus constitutes an enhancement to an already obvious general trend. Computer- based information systems have not thus far had a marked influence on making the medical process explicitly logical. This is true even in Sweden, where SPRI was constituted by name as a commission to rationalize medicine. Ordinary computer-based MISs have been focused on automating things as they are, without explicitly questioning why things are done or exposing even the com- puter logic. The PROMIS system is an exception. Artificial intelligence techniques fix on the human reasoning steps and the conceptual models at the root of human understanding of medical circum- stances. Insofar as such approaches become generally known and facilities become available for their use, they have the potential to give powerful support to the current efforts to strengthen the scientific component of clinical medicine. The MIS field has gotten the attention of excellent computer scientists interested in data base building. It has not gotten so full a measure of help from the clinical side. Now that MIS development has moved beyond its primary concern with technical problems and is ready to focus on representation of medical knowledge, one can imagine a greater interest and willingness to partici- pate from the clinical experts. Favoring this prediction is the increasing ease with which clinical logic can be expressed and represented by the new systems; empiric rules, for example, are more easily stated by clinical experts than the numerical and statistical statements that used to be required for traditional Effects of Changes in Technology on Future MISs 153 medical computer programming. Advances in medicine have historically arisen out of attempts to state publicly a model of a process, and to change these publicly as the model is fitted to actual clinical circumstances. In the past many models such as the concept of renal clearance have been numerical rather than semantic. The new technology offers a chance to develop models of all parts of the logic of medicine dealing with any circumstances in which symbols can be used to represent the processes. There are some considerations that argue against the desirability and hence the likelihood that artificial intelligence technology will have a major impact on MISs and other medical computing applications. The first is the possibility that physicians will resist committing their medical knowledge to computing systems out of fear of losing work or prestige. Some eminent workers say they have encountered this attitude, and that it is widespread.42 I have not personally had this experience in almost twenty years of work on many medical computing applications in many sites. If this attitude exists, it likely will be limited to metropolitan areas in which too many physicians already compete for patients. Since the more general problem occurs where expert medical personnel is the limiting circumstance, abstention out of fear does not seem to me to be a likely general case. A more serious possible factor limiting the pentration of artificial intelli- gence technology into medicine is the sincere protest from some quarters that such a thing should not be done for ethical reasons. A notable advocate of this consideration is Joseph Weizenbaum, whose qualifications as a pioneer com- puter scientist himself are quite impeccable. He argues in his recent book that there are some tasks that simply should not be committed to a computing system, and further, that analysis of human decision making in terms suitable for computer rendering will inevitably degrade and debase the human func- tions.43 One cannot do justice to such a consideration in the short space which our present study justifies. The reader should be aware that such arguments have been put forward rather insistently by legitimate disputants and could become a major obstacle to limit the use of artificial intelligence techniques. Notes 1. R.N. Noyce, “Microelectronics,” Scientific American. September 1977, p. 65. Reprinted with permission. 2. J.H. McFadyen, “System Network Architecture: An Overview,” l.B.M. Systems Journal, vol. 15, 1976, p. 4-23. 3. National Cash Register Corp., NCR Distributed Architecture (Dayton, Ohio; National Cash Register Corp., 1977). 4. R.H. Eckhouse, Jr., J.A. Stankovic, and A. Van Dam, “Issues in Dis- tributed Processing—An Overview of Two Workshops,” Computer, vol. 2, issue 1, January 1978, pp. 22-26. 154 Growth of MISs in the United States 5. P.H. Enslow, Jr., “What is a ‘Distributed’ Data Processing System?” Computer, vol. 2, issue 1, January 1978, pp. 13-21. 6. R. Peebles and E. Manning, “System Architecture for Distributed Data Management,” Computer, vol. 2, issue 1, January 1978, pp. 40-47. 7. A.D. Little, Automated Electrocardiography in the United States (Rockville, Md.: Health Resources Administration, August 1976). 8. M.E. Grant and J.S. Hanson, “A Totally Computerized Cardiac Pace- maker Surveillance System,” Computers in Cardiology. Proceedings of Con- ference October 7-9, 1976, St. Louis, Missouri, H.G. Ostrow, ed. (Long Beach, Calif.: I.E.E.E., 1976), pp. 13-17. 9. A.D. Little, Automated Electrocardiography in the United States, pp.8-9. 10. R.J. Dobrow, A. Fieldman, W.P.C. Clason, et al: “Transmission of Electrocardiograms from a Community Hospital for Remote Computer Analysis,” The American Journal of Cardiology, vol. 21, May 1968, pp. 687-689. 11. J.W. Johnson (Chairperson), J. Cedarlund, F.W. George, III, et al: “Panel Discussion: Radiotherapy and Computer Information Systems,” Com- puter Applications in Radiation Oncology. Proceedings of the 5th International Conference on the Use of Computers in Radiation Therapy (Hanover, N.H.: University Press of New England, 1976), pp. 49-53. 12. D.D. Leavitt, “Comprehensive Radiation Therapy Treatment Planning with a New Low-Cost Basic Graphic Computing System,” Proceedings of the Fifth Conference on Computer Applications in Radiology, Albuquerque, N.M., June 8-12, 1977. (Chicago, 111.: American College Radiology, 1977). 13. G.C. Kenney, “Special Purpose Applications of the Optical Videodisc System,” I.E.E.E. Transactions on Consumer Electronics, November 1976, pp. 327-337. 14. V. Bush, “As We May Think,” A tlantic Monthly, July 1945, p. 101. 15. I.W. Cotton, Computer Network Interconnection: Problems and Prospects, (Gaithersburg, Md.: National Bureau of Standards, April 1977), pp.1-73. 16. American Telephone and Telegraph Company. Advanced Communica- tions Service (ACS): A News Release, July 10,1978, pp. 1-2. 17. American Telephone and Telegraph Company. Advanced Communica- tions Service (ACS): A News Release, November 15, 1978, pp. 1-2. 18. American Telephone and Telegraph Company. Advanced Communica- tions Service: Technical Overview. New York: November 1978, pp. 1-44. 19. “AT&T Specialists Supply Technical Details on ACS Net,” Data Com- munications, August 1978, pp. 15-24. 20. Digital Equipment Corporation. Interim Report to Shareholders, (Maynard, Ma.: Digital Equipment Corporation, January 1979). Reprinted with permission. Effects of Changes in Technology on Future MISs 155 21. American Telephone and Telegraph Company. Advanced Communica- tions Service (ACS): A News Release, February 16, 1979. New York, pp. 1-2. 22. S. Amarel, C.A. Kulikowski, S. Levy, et al: Proceedings of the Third Annual Artificial Intelligence in Medicine Workshop, Rutgers University, New Brunswick, New Jersey, July 5-8,1977. 23. F.H. Lovejoy, D.L. Caplan, T. Rowland, et al: “A Statewide Plan for Care of the Poisoned Patient: The Massachusetts Poison Control System,”New England Journal of Medicine, vol. 300(7), February 15, 1975, pp. 363-365. 24. S.R. Kimbleton, H.M. Wood, and M.L. Fitzgerald “Network Operating Systems—An Implementation Approach,” National Computer Conference- AFIPS Conference Proceedings, June 5-8, 1978, Anaheim, California, vol. 47, S.P. Ghosh and L.Y. Liu, eds., (Montvale, N.J.: AFIPS Press, 1978), pp. 773- 782. 25. S. Ward Watkins and S.R. Kimbleton, “Network Access Technology— A Perspective,” National Computer Conference AFIPS Conference Proceedings, June 5-8, 1978, Anaheim, California, vol. 47, S.P. Ghosh and L.Y. Liu, eds. (Montvale, N.J.: AFIPS Press, 1978), pp. 495-503. 26. D.P. Geller, “The National Software Works—Access To Distributed Files and Tools,” ACM 1977: Proceedings of the Annual Conference, October 17-19, 1977, Seattle, Wash., (New York: Association for Computing Machinery, 1977), pp. 39-43. 27. INTREX-Report of a Planning Conference on Information Transfer Experiments, C.F.J. Overhage and R.J. Harman, eds. (Cambridge, Mass.: MIT Press, 1965). Reproduced by permission of the MIT Press; copyright 1968 by The Massachusetts Institute of Technology. 28. G.F. Groner, N.A. Palley, M.A. Rockwell, et al: Applications of Com- puters in Health Care Delivery: An Overview and Research Agenda (Santa Monica, Calif.: RAND Corp., p. 5185, February 1974), pp. 1-50. 29. Ibid., p. 36. Reprinted with permission. 30. E.A. Feigenbaum, “Artificial Intelligence Research: What is it? What has it Achieved? Where is it going?” Invited paper, Symposium on Artificial Intelligence, Canberra, Australia, 1974. Reprinted with permission. 31. E.A. Feigenbaum, The Art of Artificial Intelligence: 1. Themes and Case Studies of Knowledge Engineering, (Stanford, Calif.: Stanford University, June 1977), Computer Science Department Report No. STAN-CS-77-621. 32. E.A. Feigenbaum, B.G. Buchanan, and J. Lederberg, “On Generality and Problem Solving: A Case Study Using the DENDRAL Program,” Machine Intelligence, vol. 6 (New York: Edinburgh University Press, American Elsevier, 1971), pp. 165-190. 33. E. Shortliffe, Computer-Based Medical Consultations: MYCIN (New York: Elsevier, 1976). 34. H.E. Pople, Jr., J.D. Myers, and R.A. Miller, “DIALOG: A Model of Diagnostic Logic for Internal Medicine,” Proceedings of the 4th IJCAI, vol. 2, 156 Growth of MISs in the United States September 1975, pp. 849-855. 35. S.V. Lawrence, “INTERNIST: Computer Program Expressing Clinical Experience and Judgement of a Master Internist Constitutes a Unique Re- source,” Forum on Medicine, April 1978, pp. AAAI. 36. S.Weiss, C.C. Kulikowski, and A. Safir, “Glaucoma Consultation by Computer,” Computers in Biology and Medicine, vol. 8, 1978, pp. 25-40. 37. H.R. Warner, “Knowledge Sectors for Logical Processing of Patient Data in the HELP System,” Proceedings of the Second Annual Symposium on Computer Applications in Medical Care. November 5-9, 1978, Washington, D.C., F.H. Orthner, ed. (New York: I.E.E.E., 1978), pp. 401-404. 38. S.O. Thier and R.W. Berliner, “Manpower Policy: Base It On Facts, Not Opinions,” New England Journal of Medicine, vol. 299 (23), December 7,1978, pp. 1305-1307. 39. G. Wiederhold, “Management of Semantic Information for Data Bases,” Proceedings of the Third U.S.A.-Japan Computer Conference. American Federa- tion of Information Processing Societies, Inc. (AFIPS), Montvale, New Jersey, September 1978, pp. 192-197. 40. H.R. Warner, “Knowledge Sectors for Logical Processing of Patient Data in the HELP System,” p. 404. 41. R.M. Gardner, D.P. Scoville, B.J. West, et al: “Integrated Computer Systems for Monitoring of the Critically 111,” in Proceedings First Annual Symposium on Computer Applications in Medical Care. Washington, D.C., October 3-5,1977, (New York: I.E.E.E., 1977), p. 306. 42. W.B. Schwartz, “Decision Analysis: A Look at the Chief Complaints,” New England Journal of Medicine, vol. 300(10), March 8,1979, pp. 556-559. 43. J. Weizenbaum, Computer Power and Human Reason: From Judgment to Calculation San Francisco, California: (W.H. Freeman, 1976.) 8 Impact of Public Policy on the Development, Adoption, and Diffusion of MIS Technology Summary The medical computer market is not large enough to shape the major market forces. Consequently, MIS developments will merely share in the effects of hardware price reductions as a result of mass production for other users, the status of public telecommunication pricing structures, labor cost trends, and other large-scale economic effects. Some federal policies and practices bear directly on MIS diffusion. In the past, federal policies have supported development and exploration of MISs in the context of biomedical research. Federal policy has not provided for public support of the technology transfer processes that would permit the research systems to become disseminated throughout the health care community as com- mercially supported systems. One highly relevant issue is the manner in which health care cost reimbursement policies can reinforce and encourage beneficial uses of MISs. Possible government stances vis-a-vis MIS development are described. The federal government should take an active role in managing the diffusion of this technology. There are a number of strong federal agencies with talents relevant to this mission. An interagency strategy may be the appropriate initial step. Effect of Nonmedical Federal Policies on MISs Since computing in health-related activities constitutes less than 3 percent of the information processing marketplace, its needs cannot determine major developments in the computer hardware industry.1 Computing hardware costs have fallen by at least two orders of magnitude over the past twenty years. Health applications, including MISs, will continue to be the beneficiary of such savings in hardware costs and reflect the corresponding increase in the per- centage of costs contributed by technical personnel’s salaries. Insofar as these trends are influenced by federal policies with respect to write off of capital expenditures and minimum wage requirements, the medical computing field will continue to share the experience of other technical fields. In the past, even developmental computing center budgets allocated no more than 50 percent to personnel costs. Now it is not uncommon to find that labor and personal service costs make up as much as 70 percent of such budgets. The trend clearly is for 157 158 Growth of MISs in the United States computing expenses to be made up of an even higher percentage of personnel and software costs. MIS systems are not likely to be exceptions. Federal policies with respect to common carrier and private communica- tions systems have a similarly strong but nonspecific effect on MIS develop- ments. Telecommunication and terminal costs often constitute more than half the hardware costs of the complex of a computer and its groupings of user termi- nals. Consequently, cheaper communication rates enhance the development of large-scale centralized information systems. High rates enhance the develop- ment of stand-alone, decentralized information systems. Speculation concerning the effects of advanced public computer communications networks were pre- sented in chapter 7. Such developments are absolutely dependent on federal policy. Thus far the United States is unique in encouraging innovative and competitive private communication systems. Once again the effects on MIS developments could be salubrious but nonspecific. Research Support for Computers in Medicine The creation of hospital accounting and business office computer-based in- formation systems has proceeded on the basis of local funding and commercial development and sales. This application area for computers has been recognized for many years as more or less analogous to business office and accounting functions in other kinds of institutions. Hence it has always been an opportunity for entrepreneurial development, the success of which could be measured by cost reductions, labor savings, or at least some kind of suitable cost displace- ment. In contrast, attempts to develop research applications of computer-based information systems in medical areas has necessarily had to be contingent on funding combined with research. Since the end of World War II, the federal government has become the largest funder of medical research. Consequently, it was of great importance to the development of MISs that such efforts were formally recognized as legitimate research. This recognition came in 1960 with the establishment at the NIH of the Advisory Committee on Computers in Research, which was charged with defining general areas of biomedical com- puting and stimulating interest in them.2 This group was established as the Study Section on Computer Research in 1964. It became the Computer and Biomathematical Sciences Study Section in 1970. The Study Section was abolished on June 30,1977. With the creation of the Health Services and Mental Health Administration and its National Center for Health Services Research and Development in 1969, additional study sections were created which had the ability to support some aspects of the development, diffusion, and evaluation of the technology of MISs. Certain special aspects of such systems have also been supported by the National Impact of Public Policy 159 Library of Medicine (NLM) through the Biomedical Library Review Committee. The NLM has also been the main support for training programs to provide the special education and experience, both pre- and postdoctoral, to individuals entering this field from medicine, as well as from the computer-related disci- plines. It should be emphasized that, regardless of whether one considers the priorities and funding policies of these institutions to have been wise and con- sistent, they did give legitimacy to attempts to explore and define research uses of MISs. Government-Sponsored Computer Centers The initial research grants from NIH in this area took the form of facility sup- port awards. These were made to encourage and subsidize the creation and operation of computing facilities in selected major medical centers. The purposes of the facilities were by no means specific to MIS development. Rather, they were to provide appropriate and convenient computational support to bio- medical investigators in their institution or region. Even so, efforts to develop patient data base systems came relatively early. The most general rationale was reasonable enough, namely, that the systems were needed in order to render the records of patient care suitable to be the subject of research. NIH-supported computer centers were roughly comparable to the general university computer centers which were obtaining financial support at the same time from the National Science Foundation. Both agencies soon found that the country’s appetite for such funds was large, indeed, beyond the agencies’ means. Both began encouraging the computer centers to shift to fee-for-service operation, so as to be self-sustaining after the initial federal subsidies were withdrawn. University and medical facilities managers were en- couraged to take a strong administrative hand, so as to shift the operational costs to the individual institutions. Similarly, individual investigators were obliged to budget for, request, and defend computer use charges as an integral part of their individual grant applications. The various scientific study sections were instructed to honor these requests (when the research proposals them- selves were meritorious) because computing was no longer free. The net effect of these moves was to encourage—almost to compel—the development of large central computing facilities at major medical centers. This has turned out to be a mixed blessing. At least, however, it was accom- panied by a legitimization of attempts to utilize information systems in support of medical research. This is to say, the developments supported the asking of new questions about human health and disease, and supported attempts to do new things in the health care by using the emerging computer-based informa- tion processing technology. 160 Growth of MISs in the United States Effects on MIS Development Biomedical research support has been provided through relatively short-term funding of generally modest size. Regrettably, this pattern has never been suitable for computer projects. L.B. Lusted said that even in 1960 it was appar- ent that computer grant applications differed from other kinds of grant requests by being for larger sums of money and by including the purchase of computers (certainly a long term commitment) within the request.3 Development of infor- mation systems has never been compatible with year-to-year funding mechan- isms. The fact that the professional literature is not full of such pleadings is an artifact of the policies of refereed journals and the chameleon nature of informa- tion system developers. After conducting a survey of health computing projects, G.A. Giebink and L.L. Hurst conclude that funding policies placed a premium on quick transitions from applied research to operational demonstration.4 They identified two serious undesirable consequences of this policy. First, the systems were declared operational before they had been fully developed. Second, much research essential to subsequent development was never performed. Their implication is that there was not time within the original grant award period to complete all essential research, and that it could never subsequently be justified for funding because the investigator was obliged to “go operational.” Encouragement of MIS Development In the light of ever-clear hindsight, one must conclude that there has not been a systematic federal government plan for deployment of the MIS technology. Several issues require special notice. These are: (1) the problem of technology transfer, (2) the time frame for such accomplishments, and (3) the magnitude of developmental costs. Technology Transfer For MISs the hiatus between scientific success and successful demonstration in a practical environment is mirrored geographically by the distance from the NIH campus at Bethesda and the HSMHA (Health Services and Mental Health Ad- ministration, later the Health Resources Administration) in Rockville, Maryland (later moved to Hyattsville). Projects have existed in which demonstrations of information systems and other health-care innovations required support. The Health Resources Adminis- tration (HRA) has been authorized—although not always sufficiently well funded—to support such demonstration and evaluation projects. The transition Impact of Public Policy 161 from research support by NIH to HRA support for hospital demonstration can be rough, but some projects have survived it. The transition from HRA support to commercial viability is unheard of for anything as large as an MIS. Officials of funding agencies and the peer review groups that undertake review of research grant proposals have been aware of the problems of transi- tion. It has always been considered a wise plan to propose withdrawal of federal funding and substitution of hospital or institutional support for systems once they have achieved their research objectives. On the other hand, very frequently the hospital support has not been forthcoming. The results of scientific research, that is, new knowledge, do not always save money for the institution in which they were developed. In brief, there is no clear path for a research system to follow that can provide transition to a practice setting unless that system makes money for its original host institu- tion. In the absence of a strong national policy for managing technical innova- tion, the development must add to health costs or it cannot succeed. That this is so is quite understandable when one considers that the technological developments typically occur in tertiary care facilities, often university hos- pitals. These institutions fulfill their function partly by participating in research and development. Yet the developments may be aimed at a broad class of users, for instance, the far more numerous community hospitals. In the case of many MIS subsystems, the greatest economic benefits are available not to the originators of the systems but to potential user institutions with a smaller complement of health professionals. An example is the automated interrogative patient history system.5 Other examples include the physician assistance com- puter diagnosis system for bone tumors,6 drug information systems,7,8 patient medical record systems,9 and automated laboratory systems.10 Such useful systems are necessarily created within institutions that already have an abun- dance of competent health personnel in the area of the system’s function. Their participation in developmental efforts is as a public service, and often at federal expense. Consequently to expect that the same institutions will be the prime economic beneficiaries of that finished technology is unrealistic. They can sometimes support the final systems by adding an equivalent amount of cost to their patient care charges. They cannot realize a savings from that particular piece of automation because they cannot and should not displace the expert personnel who created the development in the first place. The outcome that is most in the public good is for the development to move out of the innovating institution and into community service where cost savings can be realized. It is unrealistic to expect the tertiary care institution to finance such a commercialization of technology transfer. Such a transfer should be managed and financed by the federal government so that the resultant final system can be beneficial to national health care needs. An example of such a scenario dealing with a noncomputer related field, is the polio vaccine technology transfer. The rapid transfer of basic research 162 Growth of MISs in the United States knowledge to an experimental vaccine and subsequently to a major field trial followed by large-scale production of the vaccine was funded from public sources. Under the circumstances it would have been feasible economically to have left the development entirely to private companies. To do so would have delayed the production of the vaccine and might not have resulted in the low cost, effective product we now enjoy. Medical applications of information systems do not present a picture quite so emergent, nor can one be so certain of immediate benefits as with the famous Salk vaccine. Nonetheless, a number of aspects of the development problem are quite similar. A more coordinated and supportive management attitude within the federal government toward MIS’s would even now assure a beneficial product which would serve the whole public. Time Frame for Accomplishments In the simplest possible terms, it has been the mission of NIH to support re- search, and it chose to include MIS research in this mission. The support has always terminated once the scientific success of the project could be declared (or earlier in the case of a failure). Research grants have been typically one to three years, never more than five years. This time frame may be consistent with the conduct of the research phase of a project. It clearly is not consistent with the time frame needed to bring a MIS to the stage of even a prototype commercial system. Such systems have been shown to require up to ten years for development.11 Completing a study “on time” in the terminal year of a grant has presented a serious problem: how to provide for transition of the system to a self-sustaining basis? Often this translates, how to get the hospital administration to pick up the costs? The Magnitude of Developmental Costs Since no real example exists of a full MIS in a practical environment, it may be illusory to speak definitively of developmental costs. Nonetheless one can reason from the substantial developmental costs reported for existing partial systems. Ronald Henley and Gio Wiederhold report on development costs for nine operational systems designed to be full MISs for ambulatory patients.12 Costs for the nine ranged from $230,000 to $10,000,000. Five of the nine had devel- opmental costs greater than $1,000,000. The five had annual costs for continu- ing development ranging from $154,000 to $539,000. This study did not include hospital MISs. In this category, the NDC/Honeywell system cost $12,000,000 to develop.13 Developmental costs for the Technicon MIS were $25,000,000.14 No operational unit of the Department of HEW has grant budgets sufficient Impact of Public Policy 163 to support the big systems. Research grants typically run $30 to $50 thousand per year. Research grants for computer work typically are somewhat larger, perhaps $25 to $150 thousand a year, with really large ones reaching $400 thousand per year. In the aggregate the Department of HEW spends large sums for support of biomedical research. The amounts available to single functional offices, however, are typically small compared with the magnitude of commit- ments which appear to be required to see through the development of an MIS. Health Care Reimbursement Policies The most profound effect on MISs was the initiation of the Medicare system under PL 89-97 (42 USCA 1395). This program, with its series of entitlements of ever larger numbers of individuals, provides for reimbursement to hospitals (and under Medicaid to physicians and other providers) of actual costs of care for citizens age 65 and older (plus other entitled groups). The enumeration of charges, the justification of costs, the certification of entitlement, and the huge cash-flow problems associated with delayed and partial reimbursements have forced all hospitals to devote greatly increased resources to these business office matters. The expenses are reimbursable and as such are folded into the rising per diem bed charges. With space at a premium in most hospitals, and trained clerical personnel never plentiful, most hospitals have been quite willing to shift these costs to support of computer installations or services, and to support of the administrative portions of MISs. Indeed these federal programs have created whole industries within the computer field that prosper mightily in computing, printing, and even reading the documentation of the health care services required for third party reimbursement. An example of the new computer services (outside the field of medical information systems) is Electronic Data Systems Corporation in Dallas. Com- puter processing of Medicare claims by this company constituted a $133 mil- lion business in 1976.15 With respect to the effect of the reimbursement legislation on hospitals, figures are a bit hard to come by concerning the increases in business office personnel associated with Medicare. Everyone agrees that they are substantial. One isolated example may be offered. A 450-bed hospital found that as a direct effect of Medicare, it had to increase its business office claims process- ing staff from twenty-five to thirty-two and increase the record room staff from twenty-two to twenty-six.16 In 1976, 82 percent of hospitals with one hundred beds or more were said to have computers or to use computer service bureaus.17 This strong increase in demand for accounting-oriented information systems for hospitals and clinics has had absolutely no effect on the development of the medical components of MISs. 164 Growth of MISs in the United States A system does not need to contain medical information to print a valid and correct bill, any more than the Texaco Company needs to understand one’s vacation plans in order to forward gasoline charge tickets. The hospital accounting systems are not really simple, but they are well within the state of the art. No research is needed. Their purpose is to collect bills, either from the patient (who used to pay far smaller and simpler statements than he now receives) or from the federal government’s Medicare intermediaries. Adding medical information to the accounting systems would be costly. It is not re- quired and will not be done under the present system. Indeed the cost of adding medical information is generally not considered to be reimbursable or “allow- able” by the same intermediaries who currently wonder at increasing hospital costs. Federal policy and practice in reimbursement of Medicare/Medicaid ex- penses relate directly to MISs. We can encourage or discourage deployment of this technology (and quite effectively) simply by regulation. The requirements for certification for Medicare would be a possible means, although there are many others. Implementation of such systems could be encouraged and made attractive by making specified services reimbursable costs. It would be espe- cially encouraging if these services (e.g., physician assistance functions, quality assurance analyses, screening studies, risk estimates, prognoses, treatment plans, or patient educational services) were made billable as services outside the nego- tiated per diem costs. Hospitals and practitioners have traditionally implemented services for which there was a market demand. Such a policy would create this demand and strongly influence the diffusion of the technology. The reader may accept as a general proposition that federal reimburse- ment policies must somehow have a significant effect on MISs, like all other aspects of health care technology, and yet may find it difficult to imagine just how policy actually has its effect. In fact, because there is no federal policy with respect to MISs, other essentially unrelated federal objectives can have major effects outside their intended field. The principle is more or less that regulations tend to expand to fill the vacuum. Let us consider for a moment a recent proposed federal rule which will serve as an example. Public Law 95-142, the Medicare-Medicaid Anti-Fraud and Abuse Amend- ments of 1977, established scores of new procedures aimed at monitoring and eliminating illegal practices in payment of benefits under these programs. One part of the bill, Section 19, required the establishment of uniform reporting systems for providers participating in the programs. These included all hos- pitals, skilled nursing facilities, and intermediary care facilites by 1978, and included by 1979 the addition of home health agencies, health maintenance organizations, and other types of health services facilities and organizations. The uniform reporting systems must provide information on costs, volume of services, rates, capital assets, discharge data, and billing data. Such requirements are burdensome, but monitoring so large a federal Impact of Public Policy 165 program is probably necessary. As noted previously (chapter 5), the govern- ment has long ago made the assumption that hospitals and such health organ- izations are managed and administered along the lines that it employs. PL 95- 142 now states this requirement. The educational level of hospital administrators and their understandable interest in automated information services that can satisfy the government’s taste for reports have already been noted. The problem for MISs or MIS providers does not arise from the general principles of the law, which probably in the long-run will encourage further development, at least of business-oriented MISs. The problem arises from one section of regulations issued pursuant to the law itself. The Health Care Financ- ing Administration (HCFA) went to considerable effort to produce a proposed System for Hospital Uniform Reporting (SHUR), which it announced publicly on January 23, 1979 in the form of proposed rule making.18 One small part of this sizable proposal impinges on MISs. This is a standard calling for reporting of computer usage in terms of CPU minutes. It is a relatively minor issue to most people, but a critical matter to those who purvey computer services to hospitals. This particular measure does not distinguish between the minutes provided by a million-dollar CPU and those provided by a five-thousand-dollar microprocessor CPU. Similarly, the measure does not reflect whether files were used. It is possible to argue that the sophis- tication of the services provided by the computers could be reflected in the CPU-minutes, but even this is not consonant with modem practices of firmware and would not encourage the use of the most efficient algorithms that reduce CPU use. The biggest problem with the regulation as proposed is that it focuses on the financial auditing issues which are the natural concern of the agency, and inadvertently does a certain rough injustice to the issue of the quality of the computer-based information services employed. The financial aspects of medical computing are now to be accounted for; this is certainly the proper role of HCFA. The extent to which the systems are used to improve care is apparently not in the domain of any federal agency and is not currently being addressed. Since there is already a sizable commercial medical information processing industry, there doubtless will be comments on the regulations. Perhaps it will all end right. The outside observer is left, however, with the un- easy feeling that everyone cares what the medical computer systems cost, and no one seems to care what they do. Future Management of MIS Technology MIS technology can be managed under a variety of governmental attitudes. The consequences are estimated here for three separate paradigmatic views of the field. These are essentially the judgmental, the observational, and the man- agerial. 166 Growth of MISs in the United States Consequences of the Judgmental Paradigm: Examine, Evaluate, Appraise, and Permit The present de facto federal policy for managing MIS development is to shut off research support for the further development of MISs or the creation of new ones, and to emphasize evaluation of existing systems. Examination is proceeding, sometimes with favorable appraisals. The evaluation methods used have been formal, slow, and costly. Some secondary gains have appeared, mostly in the form of well written documentations and analyses that are somewhat helpful to scientists in this and related fields. The anticipated decision to permit or not to permit will unfortunately be based on considerations which may be quite relevant to some aspects of the health planning process but which may be quite unrelated to the priorities and poten- tials of the MISs themselves. Often the analyses must lean heavily on financial measures. Nearly all authors of the financial analyses admit that there are sub- stantial unmeasured benefits in other areas. If survival of the various forms of MISs must be dependent on the best we can do now by way of formal cost-benefit analyses, then there is a clear danger. Systems of borderline merit, which concentrate primarily upon business office and institutional management functions will be permitted. Some systems which are of great merit because of their clinical features (for example, physician assistance, education, prospective community database building) will not be susceptible to evaluation based on dollar accounting, and will be harmed or destroyed. If one believes that the correct paradigm is to examine, evaluate, and per- mit, one must surely also include the need for further research in the metho- dology of evaluation. Benefit analysis in health fields merits further research. Consequences of the Observational Paradigm: Observe and Predict A laissez-faire strategy of management of these developments (that is, not much management) will mean that ideas for innovative research based potentially on MISs will either hibernate or seek support as traditional, small, discipline- specific research projects funded over short time periods. As such they will, even if successful, not be capable of supporting the development of true full MISs. A research program cannot be funded by a collection of research projects. An alternative for some research ideas is to emerge repackaged as unneces- sarily large clinical trials. MISs that are primarily oriented to hospital administration will pay their own ways by producing small savings for individual institutions. There is no known case in which a business office system has ever evolved into a MIS. The Impact of Public Policy 167 concept of the full MIS under a laissez-faire paradigm will simply not be developed. Consequences of the Managerial Paradigm: MIS as a National Intent This paradigm implies management of the development and maturation of the MIS concept. It is clear that the MIS concept, like many other innovations, has to find its proper place in its problem space. That is, one must determine by experimental exploration of the problem domain just which areas are feasi- ble and fruitful. Next, at least some sample of the various kinds of MISs must be selected for transfer as far as possible through the known sequence of phases of matura- tion. One must be prepared to see systems moved from research, to develop- ment, to demonstration, to clinical trial, and to full implementation in the market. Each of the stages will need evaluation, but each will require quite a different set of measures. Each phase should have distinct functional limits, but there should not be a predetermined time for each phase. Costs will increase as a project moves through such a sequence. Different phases could be supported by separate branches of government, but it should be government’s responsi- bility to provide for a smooth transition between phases. There must, of course, be strict criteria for evaluation, and unsuccessful projects must be selected out. Transition or selecting out should be accompanied by formal reporting from the project so as to document the experiment and its results. The process of managing the development and documentation of the growth and changes in the field should rest with a permanent government body. The consequences of this paradigm would be to complete maturation of the concept, explore appropriate problem areas, and move to commercial availability of those systems for which such a metamorphosis is appropriate. In short, this paradigm sees MISs as a concept of potentially great national value. It sees government as having the opportunity and responsibility to man- age the development of this technology and to assure that society gains the benefit. Mechanism for Managing MIS Diffusion The question naturally arises, “What government body, office, or agency could manage the MIS technology diffusion?” The following strategies appear plaus- ible. 168 Growth of MISs in the United States Special Assistant Special assistant in the office of the Secretary of HEW or in the office of the Assistant Secretary for Health could be appointed. An individual in such a position could accomplish a great deal if equipped with a budget and a small but technically competent staff for monitoring contract studies and projects. The problem would be one of continuity, since the function would inevitably become associated with the political tenure of the office holder. This plan runs the risk of on-again-off-again funding over time frames which might be far too short to permit such an office to address itself to managing the technology transfer. Existing Agency Sole responsibility could be assigned to an existing federal agency or unit. Potential candidates in this category are relatively numerous. One should em- phasize in such a discussion that excellent computer and health scientists exist in many units of the federal government. The task with respect to managing MISs and related developments is not to search anxiously for competence, but rather to give consideration to what factors of office will enhance the manage- ment role. Division of Research Grants of NIH. This unit found itself quite able to manage traditional research grants for development of many early MISs as well as much closely related research in the theory and method of medical computing. Management was in the hands of a Study Section (now called an Initial Scientific Review Body). The Study Section was appropriately not attached to an individual Institute. Its recommendations went to a number of Councils. There has never been an identifiable program fund for medical computing work. Individual projects when approved were paid for by individual or com- posite funds from categorical Institutes and their programs. The relevant Study Section, that is, Computers and Biomathematics, was abolished in 1977. Disadvantages from the past would likely be present in the future. The process was strictly reactive to proposals from the academic community. One exception was the deliberate stimulation by one Study Section of interest within the scientific community in efforts to improve the quality of research into diagnostic decision-making systems.19’20 A second disadvantage has been the policy throughout NIH to drop support once scientific and technical feasibility has been demonstrated. This demands an extremely abrupt transition to commercial viability, more or less like throwing the child into the water to teach it to swim. The only alternative strategy outside the MIS area so far demonstrated Impact of Public Policy 169 by NIH seems to be to support extremely expensive clinical trials such as those for cancer chemotherapy. An additional current disadvantage, although not necessarily a permanent one, is the lack within NIH of a tradition of good working relationships with the industrial and hospital groups who now seem to be an essential component in the diffusion and utilization of medical computing technology. Biotechnology Resources Branch of NIH. This office has demonstrated a steady interest in technology transfer. It has responsibility for a spectrum of special facilities for support of biomedical research. Applications are heavily weighted to physics and chemistry equipment, but also include support for one relevant computer-based information system for medical records. This is the CLINFO system produced by RAND Corporation under government contract. It is not designed for general hospital use, but rather is carefully tuned and tested for use by Clinical Research Centers. Each system manages the records of a maxi- mum of three hundred patients. It is planned that the limited number of ex- perimental systems will be replaced next year by larger numbers of commercial- ly available systems. CLINFO itself is not directly relevant to the MIS problem, but on the other hand the Biotechnology Resources Branch’s careful management of the creation, testing, development, and diffusion of the CLINFO system may be a good example of the technology transfer which is needed for larger systems. Division of Computer Research and Technology (DCRT) of NIH. This unit (DCRT) provides computer services to scientists and administrators at NIH. It is one of the most technically advanced and well run medical computer facilities in the world. It is apparent that there is technical as well as substantial scientific management expertise within the organization. It does not have an extramural grants program and has not been permitted to offer its services to individuals or organizations outside the federal health community. DCRT could be a strong partner of a team effort, but is not constituted to assume prime responsibility in managing the diffusion of MISs or computers in medicine more generally. The National Center for Health Services Research. From many points of view and in a wholly rational world, this office would indeed be the logical organiza- tion to accept responsibility for fostering the development, evaluation, and diffusion of computers into the health care system. Certainly it has not done so. Why this has happened is not completely clear. The following factors may have played significant roles in perpetuating this problem. 1. HSMHA, the predecessor agency to HRA and HSA and partly of HCFA, had serious problems of a new agency with too numerous and sometimes con- flicting programs to administer. The National Center for Health Services 170 Growth of MISs in the United States Research was administered first by HSMHA, later by HRA, and recently by another unit. 2. It has undergone numerous radical executive reorganizations. For this reason and perhaps because of its applied mission, and perhaps because of its singularly graceless operating style, it has never attracted either to government itself nor to the outside constituency the very best medical and scientific investi- gators and thinkers in the numbers it needed. 3. The National Center for Health Services Research is the successor to a previous unit called the National Center for Health Services Research and Development. In both its former and present lives the National Center ad- ministers a modest extramural program of grant-supported research centers in universities. Dropping of the designation “and Development” is sharply reflected in the programs supported, which now have a strong evaluative, judg- mental flavor with respect to technological matters. Indeed, in spite of the fact that the legislation (PL 93-353) clearly directs one sponsored health services research center “to encourage the development and use of technology ... to increase the effectiveness of the therapeutic process,” there is a complete unwillingness on the part of the National Center to en- courage work supporting technological developments. The most recent re- organization within this area of HEW shifted the National Center for Health Services Research out of the Health Resources Administration and into the Office of the Assistant Secretary for Health.21 It is too soon to tell yet, but such a change ordinarily means a shift to a policy role rather than a research role. In addition, a separate National Center for Health Care Technology has been authorized, but during fiscal year 1979-1980 no funds were appropriated for its operation.22 4. There is a strong imperative within government to reduce health care costs immediately. This injunction seems to be felt especially strongly in HRA, HCFA, and in the NCHSR. While a reasonable goal, immediate cost reduction cannot be permitted to divert attention and support from programs of long-term importance such as those concerning MISs. Information systems development may require an initial governmental investment to achieve final systems that serve the general good rather than the good of an institution or corporation. Even beyond this, such programs, including the greater use of computers in health care, offer the hope of major reductions in overall societal costs of health, regardless of their immediate effect on hospital per diem charges. The National Library of Medicine. This institution accepts prime responsibility within the federal government for the collection, organization, and distribution of knowledge in biomedical science. It is the only supporter of training programs aimed primarily at producing scientific personnel in the area of information systems and the scientific aspects of computing in medicine. Its extramural research grants program has been relatively small, and must be divided so as to Impact of Public Policy 171 support basic library work, the cross-disciplinary training programs, limited amounts of fundamental information science research (such as linguistics, thesaurus building, and information and citation storage and retrieval), and lastly the relatively large Regional Medical Libraries network. It is the last that forms the information distribution network through which NLM provides the nationwide biomedical scientific community with library services. The per- centage of its extramural funding spent on the Regional Medical Library net- work has varied from 25 percent to 45 percent between 1970 and 1977. Over the same period, the percentage available for research projects has ranged from 10 percent to 18 percent. In dollar terms, these figures translate to relatively small research grants, with the residual flexibility for startup of only the high- est priority new research. Recently the NLM formulated a plan in response to external recommenda- tions that would expand both the base of support for new research projects in the field of information-based systems in medicine and permit the creation of a program for training advanced research directors in the field of computers in medicine.23’24 Support of these expanded programs in a government institution of im- peccably high scholarship would seem of top priority in terms of national needs. Expansion of NLMs research-support program and formal entry into a computers in medicine program are natural evolutionary developments. There is every reason to believe that this growth would be accomplished without com- promising their scientific and administrative credibility. The new developments are planned to include a more pro-active program to achieve a knowledge of and rationalization with research activities supported by other government agencies, especially HRA. For NLM to go beyond its current plans and be considered for sole re- sponsibility to administer a program to manage the diffusion of MIS technology might be a mistake. Such a major move would draw on known good and stable management and excellent rapport with the biomedical scientific community. Such a move could not draw on a substantial industrial constituency nor a strong bioengineering constituency. NLM ought certainly to be encouraged to expand its development of the research and training aspects of computers in medicine, as well as to continue its ambitious medical library network program and the Lister Hill Center. NLM could well be the mainstay of an interagency strategy but should not be asked to take the sole responsibility for managing MIS diffusion because of its existing heavy commitments to important old and new progarms. The National Bureau of Standards (NBS). The Bureau, located within the Department of Commerce, has at times represented one of the finest scientific laboratories in the world. It is currently in a rebuilding phase. Contributions of the NBS to computing have been substantial, including participation in 172 Growth of MISs in the United States developing the standards for the COBOL language. The Computer Science Institute of NBS has evidenced a strong interest and helpful participation in health affairs where a standard could be imagined to be involved. Examples of positive contributions are the numerous meetings on data privacy and pub- lication of a major report on Computers, Health Records, and Citizens Rights by NBS.25 In addition, NBS has fostered the adoption of a particular cipher as a national standard for data security and privacy protection.26 NBS seems to have excellent rapport with the industrial, engineering, and scientific communities. It probably has relatively underdeveloped rapport with the medical and the health care community. NBS is serving as convener and advisor to the armed services in their management of the development of TRIMIS, the TriService Medical Information System. This system is an un- usually large undertaking for any office. The NBS is relatively a newcomer to the project. In many ways TRIMIS is a bride with entirely too long a history for anyone to expect a fairy tale romance. In this sense, the results of their initial scientific and management assistance to TRIMIS need not be taken as the ultimate measure of the NBS product. There are aspects of the MIS problem, and by extension to related aspects of computers in medicine, that could easily be construed as problems in standards or the lack of them. With respect to all such issues NBS could be a strong partner in managing the diffusion of computer technology in health care. Standards-related issues include formulation of the following: 1. Standards for medical nomenclature and terminology 2. Format standards for computer-based military medical records 3. Standards for new medical image display systems 4. A rational medical identifier system viz, the Swedish system Social Security numbers bar-coding technology, etc. 5. Evaluation criteria, or even a useful taxonomy of appropriate evaluation methods 6. Standard descriptors for computer services to patients. This is a critically important need, especially if one wishes to use Medicare/Medicaid and the medical insurance industry as a means of inducing provision of advanced automated diagnostic assistance, patient access, and quality assurance services. While one must be careful not to impose standards prematurely or restric- tively, the issues noted appear to be susceptible to a program of standardization both as a solution and a constructive goal toward refining the problems. It should also be acknowledged that NBS is not primarily a health agency. Yet health and medical problems, especially those involving information system standards, are one of the areas for which NBS has already demonstrated in- terest and leadership. In any event, in the areas noted the NBS appears fully Impact of Public Policy 173 able to provide stable leadership in managing these scientific aspects of the problem solving. New Federal Body Vesting prime responsibility for MIS development could also be in a new federal office whose broad mandate would be to manage technology transfer in health care. Two legislative proposals deal in a general way with this issue, although both failed to pass Congress in 1978. They reveal a serious concern on the part of Congress with technology and health. They could be reintroduced or modi- fied in future years, and in any case, provide an example of the problems created for a particular technology such as MISs vis-a-vis a potential new federal body. The Kennedy Bill in the Senate27 and the companion Maguire Bill in the House28 proposed to establish the National Institutes of Health Care Research. The Institutes would have included a National Institute for Health Policy Re- search and a National Center for the Evaluation of Medical Technology. In addition, they proposed a National Institute for Health Statistics and Epid- emiology. The new institutes were to be charged with conducting and supporting research, demonstrations, evaluations, and statistical and epidemiological activities for the purpose of improving the effectiveness, efficiency, and quality of health care services in the United States. Among the numerous areas in which the institutes would operate were the safety, efficacy, effectiveness, cost effectiveness, and social, economic, and ethical impacts of medical tech- nologies, and alternative methods disseminating knowledge concerning health and health related activities. The actions implicit in this legislation fall short of a plan for active manage- ment of technological developments in medicine via federal policy. The implicit plan is cast in the old mold of “evaluate and permit” rather than “manage and facilitate.” Evidence for this negative appraisal comes from the bills them- selves. MISs and, more generally, computers in medicine were not readily identi- fiable in the mass of responsibilities assigned to the institutes. Where alluded to at all, there was the implication that evaluating MISs should be reserved to the Policy Institute. Such a focus would enhance the likelihood that the problems and opportunities presented by this technology would be utterly lost in the wealth of policies under urgent consideration. Not the least of these would be current policies with respect to health care costs and future potential policies for national health insurance. It is possible that such an institute could assist in information systems development, but the assistance would be extremely indirect. Technology is mentioned in association with the proposed Center for Evaluation of Medical Technology. Even here it is not clear at what conceptual 174 Growth of MISs in the United States level the activity was meant to exist. Information systems, for example, may have been excluded from the Technology Center. HR 10839, the Maguire Bill, defined medical technology as “any discrete and identifiable medical or surgical regimen or modality used to diagnose or treat illness, prevent disease, support life, or maintain patient well-being.” This area could, by default, be made up largely of issues of instrumentation and surgical procedures. If the instrumenta- tion were to be conceived of as separate from information systems, for which responsibility was assigned to the Policy Institute, the instruments would necessarily be conventional electromechanical devices. In contrast to this poten- tially unrealistic model presented by the legislation, the real-world develop- ment of instrumentation for medicine and industry is almost exclusively cen- tered on built-in computers and interconnections to even larger information systems. It is apparent from the administratively lower level assigned to the Tech- nology Center as contrasted with the institute status assigned to statistics, that a lower priority and sense of relative importance would be accorded information systems and related technologies. Participation by industry in the activities of the institutes would have been limited and conventional. It would have occurred via contracts administered by DHEW. This has not been satisfactory in the past. Coordination of the institutes would be the statutory responsibility of the Secretary assisted by a council. The constituents of the Council make it clear that the Division of Research Resources and the Biotechnology Resources Branch of the NIH Institutes, the NLM, and the NBS would not be represented. A solution that assigned primary responsibility for management and facilita- tion of computers in medicine, MISs specifically, to such institutes would not be wise. This legislation would have created a new configuration of the same persons and offices we now have. Their orientation would have been “evaluate and permit.” There would have been no support for technological development and technology transfer unless the legislation had been amended to tell them plainly to “manage and facilitate.” In contrast to the earlier Kennedy and Maguire bills, PL 93-623 was en- acted, which created a new federal organization called the National Center for Health Care Technology.29 Mr. Kennedy also introduced in the Senate the successful modified legislation. In the Senate version, the organization was to be known as the Office of Technology Evaluation. Fifteen million dollars was authorized for operating the new center in its first year when the final act emerged from the Senate-House Conference Committee. No funds were actually appropriated in 1979-1980. The Center does, however, have an acting director and deputy director. Among other things, the authorizing legislation calls for the appointment of a National Council of Health Care Technology which doubt- less will be accomplished during 1979. Under full operation of the Center, the Council will be an important and Impact of Public Policy 175 busy group, with responsibility for setting priorities for the work of the center and also for itself developing and disseminating exemplary standards, norms, and criteria for assessed health care technologies. According to the legislation, the technologies will be evaluated according to their safety, effectiveness, cost effectiveness, and their social, ethical, and economic impact. The Council is required to be made up of persons including a wide range of specified back- grounds, consistent with its broad charge. It is required that two of the eighteen individuals be representatives of businesses engaged in developing or producing health care technology. With the center in an early start-up mode, it is impossible to forecast its emphases and effects. The language of the legislation is somewhat ominous. “Medical technology” is redefined to mean “any discrete and identifiable medical or surgical regimen or modality used to diagnose or treat illness, pre- vent disease, or maintain patient well-being.” The bill’s stress on assessment and standard setting seems to suggest the adoption once again of the “evaluate and permit” paradigm. On the other hand, funds and authorization are provided to create nongovernmental technology research centers. It would be entirely possible for the new Center to pursue an active and positive role in identifying technology needs and managing the maturation and dissemination of appropriate technologies. Even if the management paradigm is selected, how much attention any particular technology such as MISs might receive is uncertain. However, information systems pervade and support all high technology including instru- mentation, biomaterials, robotics, and new clinical procedures such as coronary bypass surgery and breast cancer screening. Consequently, the enhanced final development of medical information management techniques ought to be a serious concern of the Center. The authorizing legislation does not provide specific guidance as to whether the activities of the Center are to be directed toward the long-term greatest national gains from optimal growth of beneficial technologies, or whether the activities are to be aimed at short-term control of emergent problems in the context of certificate-of-need cost control legislation and capping exercises. Interagency Strategy An interagency strategy would assign responsibility for managing MIS diffusion and perhaps, more generally, computers in medicine to a federal interagency committee. Typically, such committees meet regularly and have authority to coordinate the efforts toward their common goal through actions by individual agencies in domains for which they already have legislative authorization. For this reason, budgetary control of the entire technology transfer program should rest with the Committee. It ought not to be put in a position of merely presid- ing over existing activities. Industry and universities should be represented along with government in an appropriate management structure. 176 Growth of MISs in the United States Such an interagency committee usually has the authority to hold con- ferences, publish reports and studies, and draw upon nongovernmental partici- pants as invited guests or consultants. The committee is responsible for setting objectives and milestones for its work and reporting annually on its accomplish- ments, including identification of the problems it encounters. In this framework, the NLM should take the role of lead agency in orches- trating these activities. Other federal agencies with relevant competencies and missions include the Division of Research Resources of NIH Biotechnology Resources Branch of NIH Division of Computer Research and Technology of NIH Computer Science Institute of NBS National Center for Health Services Research Health Resources Administration Alcohol, Drug Abuse and Mental Health Administration (ADAMHA) Veterans Administration, Office of Health Services Research The disadvantages of an interagency strategy are well known. Such a com- mittee could encounter the traditional problem of each agency acknowledging validity of the committee’s goal in general but finding its own goals assuming a higher priority for action or funding. On the other hand, such an interagency committee would combine very extensive pockets of technical competence within the federal government. It would also provide the centralized point for yearly reporting of progress on the problem. This strategy would focus on the problems and accomplishments of computers in medicine and prevent the field from becoming lost in more global concerns. Notes 1. R.M. Davis, “Evolution of Computers and Computing,” Science, vol. 195, March 18, 1977, pp. 1096-1102. 2. L.B. Lusted, “Computers in Medicine—a Personal Perspective,” Journal of Chronic Diseases, vol. 19,1966, pp. 365-372. 3. Lusted, “Computers in Medicine—a Personal Perspective,” p. 368. 4. G.A. Giebink and L.L. Hurst, Computer Projects in Health Care (Ann Arbor, Mich.: Health Administration Press, 1975). Impact of Public Policy 177 5. W.V. Slack and L.J. Van Cura, “Patient Reaction to Computer-Based Medical Interviewing,” Computers and Biomedical Research, vol. 1, 1968, pp. 527-531. 6. G.S. Lodwick, “Radiographic Diagnosis and Grading of Bone Tumors, with Comments on Computer Grading,” Proceedings of Fifth National Cancer Conference (Philadelphia, Pa.: J.B. Lippincott Co., September 1964), pp. 369- 380. 7. J. Morrell, M. Podlone, and S.H. Cohen, “Receptivity of Physicians in a Teaching Hospital to a Computerized Drug Interaction Monitoring and Reporting System,”Medical Care, vol. 15(1), 1977, pp. 68-78. 8. S. Garten, C.E. Mengel, W.B. Stewart, et al: “A Computer-Based Drug Information System,” Missouri Medicine, April 1974, pp. 183-186. 9. PROMIS Laboratory, “Automation of the Problem-Oriented Medical Record,” NCHSR Research Digest Series. DHEW, April 1977. 10. D.A.B. Lindberg, The Computer and Medical Care (Springfield, 111.: Charles C. Thomas, 1968). 11. R.R. Henley and G. Wiederhold, An Analysis of Automated Ambula- tory Record Systems, vol. 1: Findings, vol. II: Background Materials (San Francisco, Calif.: University of California San Francisco Medical Center, June 1975). 12. Ibid.,vol. l,p. 157. 13. Systemedics, Inc., “Automated Hospital Information Systems, Case Study and Report,” Princeton, April 29, 1976. 14. E.C. Whitehead, Technology Impact on Health Care Costs Statement to United States House of Representatives Committee on Science and Technology, September 27,1978. 15. C. Matthews, “Texas Firm Moving into Position to Control U.S. Health Care,” St. Louis Post Dispatch, January 30, 1977, p. 1. 16. University of Missouri, personal communications. 17. Systemedics, Inc., “Automated Hospital Information Systems, Case Study and Report.” 18. Health Care Financing Administration, H.E.W., Uniform Reporting Systems for Health Services Facilities and Organizations. Federal Register Vol. 44(16), January 23,1979, pp. 4741 -4744. 19. Computer Diagnosis and Diagnostic Methods, J.A. Jacquez, ed. The Proceedings of a Conference Sponsored by the Biomedical Data Processing Training Program of the University of Michigan. University of Michigan, April 1964. 20. Computer Diagnosis and Diagnostic Methods, J.A. Jacquez, ed. The Proceedings of the Second Conference on the Diagnostic Process, University of Michigan, May 1970. (Charles C Thomas, Springfield, Dlinois, 1972). 21. Office of the Assistant Secretary for Health, Public Health Service Regional Offices, Health Resources Administration, and Health Services 178 Growth of MlSs in the United States Administration, Statement of Organization, Functions, and Delegations of Authority, Federal Register vol. 42(232), December 2,1977, pp. 61317-61318. 22. United States Congress. 95th Congress. Second session. Public Law 95-623, Health Services Research, Health Statistics and Health Care Technology Act of 1978, November 9, 1978. 23. Research in Biomedical Information Systems, Program Project Support Announcement, National Library of Medicine, November 1978, Bethesda, Maryland. 24. New Investigator Research Grants, Extra-mural Programs Announce- ment. National Library of Medicine, November 28, 1978, Bethesda, Maryland. 25. A.F. Westin, Computers, Health Records and Citizens Rights (Wash- ington, D.C.: National Bureau of Standards, 1976). 26. “Data Encryption Standard,” Federal Information Processing Standards Publication 46, National Bureau of Standards, January 15,1977. 27. United States Congress, Senate, The National Institutes of Health Care Research Act of 1978, S. 2466,95th Congress, 2nd Session. 28. United States Congress, House, National Institutes of Health Care Research Act of 1978, H.R. 10839, 95th Congress, 2nd Session. 29. United States Congress, Public Law 95-623. Health Services Research, Health Statistics and Health Care Technology Act of 1978. 95th Congress. 2nd Session, November 9, 1978. 9 Conclusions State of the Art of MISs 1. Systems have been demonstrated that exhibit the feasibility of MISs to perform in a number of medical service areas, in many different institution settings, doing many functions, and serving many purposes under many finan- cial structures. No system has been used in all combinations of circumstances, nor have all combinations been explored systematically. Even though most useful MIS tasks have been feasible, some have not. Nonetheless, much of the problem domain has been explored successfully. 2. Systems built to suit particular institutions, especially university re- search centers, represent important demonstrations of feasiblity and of medical- ly advanced expectations. Several of these groups use MISs effectively to press forward the state of the art of medical practice. These sought-after advances include improvements in quality of patient care and medical quality-control procedures, improvement in access to secondary care, and in the education and certification of health care professionals. Advanced systems use extensive formal representations of medical knowledge within the computer, total re- placement of the patient’s written medical record, and extensive interaction between MIS and the clinical physician concerning matters of medical judgment. 3. Excellent commercial systems are available from competing vendors using computer hardware of many manufacturers. These systems have a general- ity whose purpose is to facilitate transferability from one institution to another. For this and other reasons, the commercial systems lag behind the individual university systems in their most advanced medical functions. None provides for patient history, physical examination, or progress notes. Many do not provide for continuity of patient records between hospital admissions and between hospital and ambulatory care visits. 4. The most serious limitation of all general MIS implementations in the United States, commercial or research based, is that they have not yet reached the stage of interinstitutional communication and data pooling. Evaluating the Worth of MISs 1. Marketplace mechanisms have clearly demonstrated the worth of automated information systems for business office and hospital administrative functions. 179 180 Growth of MISs in the United States 2. Operations research methods as well as cost and economic analyses have demonstrated immediate modest dollar savings directly attributable to MIS operation, as well as nonmonetary benefits. The latter include improved institu- tional management, health care planning, and quality of care. The dollar savings alone can more than pay for the routine operation of MISs. 3. Technology assessments have been limited. They have raised questions of potential dangers to personal data privacy in the future, and have resulted in legislation to prevent or punish abuses. 4. Scientific impact studies are virtually absent. Partial analyses suggest that there are substantial positive contributions to be made by MISs to science and medical research. These benefits would stem from the more complex analyses of patient findings that computers permit, but, more important from the increased availability of reliable patient measurements and observations that would come from extensive routine use of MISs. Barriers to Further Development and Dissemination of MIS Technology 1. Obstacles described by system builders and students of the field fall into three general categories: technical, social, and managerial. The last two have been seriously underestimated in the past. 2. MIS development shares some of its problems with other emergent technologies. Ideas and practices, for instance, diffuse more easily than entire systems. Systems with immediate benefits are more readily accepted than those whose benefits are remote in time or place. The MIS innovations expected to have the slowest rate of diffusion and acceptance will be those which substitute a new conceptual model of the activity in question and simultaneously attempt to automate it. 3. There are obstacles inherent in the medical setting in which the MIS technology must make its way. The most notable barriers are the extreme administrative balkanization of medical affairs and its rather slow social adapta- tion to changing technology. A notable example of the slow social change is the fact that no American medical school offers instruction in computing or in- formation systems as a regular part of its curriculum. These features are not absolute barriers, nor even deliberate obstructions. They do, however, slow the diffusion of MIS technology and keep the emerging successful systems rather small. Effects of Changes in Computing and Information Systems Technology on MISs 1. Microprocessors and the laser-etched disk memory will have major favorable impacts on MIS development. Both require considerable rethinking of Conclusions 181 traditional MIS practices and assumptions in order for the improvements to be optimal. 2. Communications technology also can have a major impact on MIS developments. The advances needed are not devices but more wise general utilization of existing capabilities for broad social purposes including health care systems. 3. The effect of microprocessors, new memories, and artificial intelligence techniques will be centrifugal, adding capability at the periphery of the already decentralized health care system. In contrast, the long-range effect of advanced communication techniques will be centripetal, adding capability (at least poten- tial capability) for increasing the integrity of a centralized national information network for health care delivery. Potential Impact of Public Policy on the Development and Dissemination of MIS Technology 1. In the past the federal government recognized MIS development as a proper biomedical research topic and supported this research. More recently funds for research and development of medical technology in general and MIS technology specifically have been almost totally denied and a limited number of evaluations of existing systems have been emphasized. 2. In neither case has significant attention been paid to the process of technology transfer by which such innovations ought to evolve into commercial systems that would be generally available to the hospital community. 3. Federal policy ought to facilitate and manage the further development and diffusion of MIS technology. This should be done through a plan that defines goals for MIS technology, provides more long-term funding for tech- nology development, and allocates responsibility for portions of the effort to the appropriate federal units. 4. Considering the realities of the talent and problem mix within the federal government, the best means for accomplishing this plan will be an interagency strategy with the NLM the lead agency. Real budgetary control of such a program must rest with such an interagency governing body, and not be merely cosmetic administrative approval of old policies. Industry and universities should both be represented in such a body, reflecting the seriousness of the technology transfer questions. Name Index Abbott, R.P., 23, 34 (30) Abrahamsson, S., 6, 8 (11); 14, 20 (7, 8) Adams, J.B., 40, 65 (14, 15) Aikawa, J.K., 23, 33 (16) Allen, P.O., 22, 33 (14) Altman, H., 23, 34 (37, 39); 29, 35 (52) Amarel, S., 142, 155 (22) American College of Radiology, 24, 35 (51) American Hospital Association, 52, 66 (36); 59,67(46); 122, 127 (47) American Telephone & Telegraph Co., 141, 154 (16-19), 155 (21) Anderson, J., 6, 7 (4, 7), 8 (9) Andrews, J.T., 23, 33 (22) Aranda, J.M., 44, 66 (23) Armstrong, M.F., 24, 35 (46) Arnstein, S., 76, 99 (10-13) Arthur D. Little, Inc., 23, 34 (27); 39, 65 (7); 73, 98 (7); 84, 100 (35); 112, 125 (29); 133, 154 (7, 9) Austin C.J., 121, 126 (41) Baker, M.A., 90, 102 (56); 91, 102 (60) Baker, W.R., 29, 36 (54) Balinfy, J.L., 23, 33 (23) Ball, M.J., 122, 126 (46) Bank, R., 23, 34 (38); 29, 35 (53) Barrett, J.P., 40, 65 (10); 81, 100 (27, 30); 117, 126 (35) Barhyte, D.Y., 107, 125 (14); 108, 125 (15, 16) Barnett, G.O., 12, 19 (3); 22, 32 (4), 33 (9); 40, 65 (13-16); 41, 65 (17); 42, 65 (18): 69, 98 (1) Baruch, J.J., 59, 67 (47) Barnum, R.A., 81, 100 (27, 30); 117, 126 (35) Bateman, B., 38, 64 (3); 40, 65 (9, 11, 12); 151, 156 (41) Beaman, P.D., 40, 65 (14, 15) Beaumont, J.O., 23, 34 (30) Beckett, R.S., 23, 34 (34) Begon, F., 6, 8 (9) Bender, A.E., 13, 20 (6); 22, 32 (2) Bendix, R., 23, 35 (43) Berland, T„ 23, 35 (43) Berliner, R.W., 150, 156 (38) Bhargava, V., 23, 35 (43) Blackmon, P.W., 87, 101 (42-44); 88, 101 (45) Bloustein, E., 94, 102 (67) Brantley, B.A., 22, 33 (12) Braunstein, M.J., 24, 35 (45) Brewerton, D.A., 120, 35 (40) Briggs, R.L., 24, 35 (46) Brooks, R.C., 87, 101 (42-44); 88, 101 (45) Buchanan, B.G., 148, 155 (32) Buck, C., 23, 34 (33) Burke, C.S., 106, 124 (7) Bush, V., 137, 154 (14) Caceres, C., 133, 154 (10) Campbell, B., 70, 98 (5) Cantley, G., 94, 103 (73) Caplan, D.L., 142, 155 (23) Carlisle, R.G., 87, 101 (42) Carter, L.F., 94, 103 (73) Casey, I.J., 87, 101 (42-44); 88, 101 (45) Cedarlund, J., 134, 154 (11) Cho, D.W., 23, 34 (39); 29, 35 (52) Christakis, A.H., 76, 99 (12) Christopher, T.G., 29, 36 (54); 95, 103 (77) Ciesielski, V., 142, 155 (22) Clampitt, S„ 24, 35 (50) Clark, S.J., 24, 35 (44) Clason, W.P.C., 133, 154 (10) Clemmer, T.P., 38, 64 (3); 39, 65 (6); 40, 65 (9, 11, 12); 151, 156 (41) Cohen, S.N., 24, 35 (46); 161, 177 (7) Coleman, J.S., 106, 124 (6); 109, 125 (20- 22); 110, 125 (24) Note: Text page numbers follow the author’s name. Numbered notes at the end of each chapter present full citations and are listed in this index in parentheses. 183 184 Growth of MISs in the United States Collen, M.F., 12, 20 (4); 22, 32 (3), 33 (10); 62, 67 (53); 70, 98 (4, 5); 94, 103 (71); 114, 125 (30) Cook, M., 81, 100 (28); 85, 100 (36), 101 (40) Cote, R.A., 23, 34 (34) Cotton, I., 140, 154 (15) Cranfill, D„ 58, 67 (44) Cronkhite, L.W., 22, 32 (5) Cummings, M.M., 107, 124 (12) Cundick, P.M., 38, 64 (3); 40, 65 (9, 11, 12); 151, 156(41) Cutler, J.L., 70, 98 (5) Daddario, E., 75, 99 (9) Dales, L.G., 70, 98 (4, 5) Davis, L.S., 22, 33 (10); 62, 67 (53) Davis, R.M., 111, 125 (27); 157, 176 (1) DeReuck, A., 107, 124 (8) Dept, of Communication/Dept, of Justice, Canada, 91, 102 (61) Digital Equipment Corp., 141, 154 (20) Dittrich, F.J., 24, 35 (49) Dobrow, R.J., 133, 154 (10) Donabedian, A., 71, 98 (6) Donnelly, W.H., 23, 34 (34) DuBois, R., 49, 66 (31) Duffy, P„ 79, 99 (16, 17, 19) Eckhouse, R.H., 131, 153 (4) EUerbrake, R.P., 88, 101 (49) Emlet, H.E., 87, 101 (42-44); 88, 101 (45) Enslow, P.H., 131, 154 (5) Erat, K., 22, 32 (7) Erickson, W., Ill, 125 (28) Ericson, R.P., 38, 64 (2) Etzioni, A., 106, 124 (5) Evenson, M.A., 23, 33 (19) Evenson, R.C., 23, 34 (39); 29, 35 (52) Fazen, L., 142, 154 (23) Feagin, S.J., 22,33 (12) Feigenbaum, E„ 145, 155 (30); 146, 155 (31); 148, 155 (32) Feinburg, L.S., 24, 35 (46) Feldman, R„ 70, 98 (4, 5) Fenna, D„ 6, 8 (11); 14, 20(8) Fieldman, A., 133, 154 (10) Fisher, G.L., 110, 125 (25) Fishman, B„ 29, 36 (55) Fitzgerald, M.L., 143, 155 (24) Flagle, C.D., 70, 98 (4); 80, 98 (25) Fleming, J., 81, 100 (28); 85, 100 (36); 85, 101 (40) Foft, J.W., 23, 33 (20) Forsythe, J.M., 6, 7 (4) Fretz, P„ 23, 33 (20) Friedman, G.D., 70, 98 (4, 5) Friedman, R.B., 95, 103 (76); 115, 126 (33); 118, 126 (36) Furukawa, T., 6, 7 (3) Gall, J.E., 81, 100 (28); 85, 100 (36-39), 101 (40) Gantner, G.E., 23, 34 (34) Gardner, R.M., 24, 35 (44); 38, 64 (3); 39, 65 (6); 40, 65 (9, 11, 12); 151, 156 (41) Garfield, E., 106, 124 (4) Garten, S., 24, 35 (47); 161, 177 (8) Garvey, W.D., 107, 124 (11) Gascho, T., 23, 33 (20) Geller, D.P., 144, 155 (26) General Electric, 59, 101 (48, 49); 60, 101 (50, 51); 61, 101 (52) George, F.W., 134, 154 (11) Giebink, G., 115, 127 (32); 116, 127 (34); 160, 176 (4) Gieschen, M.M., 23, 33 (19) Gigliotti, G.A., 22, 33 (13) Glicksman, A.S., 134, 154 (11) Glueck, B., 23, 34 (36); 38, 64 (1, 2) Gordon, B.B., 81, 100 (27, 30)f 117, 126 (35) Gordon, G., 110, 125 (25) Gorman, P.A., 133, 154 (10) Graepel, P., 23, 34 (34) Grant, M.E., 23, 34 (28); 133, 154 (8) Greene, B.R., 121, 126 (41) Greenlie, J.K., 40, 65 (14, 15) Griffith, B.C., 107, 124 (11) Griffith, W„ 23, 35 (43) Groner, G., 29, 36 (54, 55); 95, 103 (77); 144, 155 (28, 29) Gross, M.J., 2, 7 (2) Name Index 185 Gruber, W.H., 106, 124 (2) Gustafson, D.H., 115, 126 (33); 118, 126 (36) Hakimi, B.R., 24, 35 (49) Hamilton, W., 23, 33 (18) Hammond, W.E., 22, 33 (12) Hannigan, J.F., 24, 35 (46) Hanson, J.S., 23, 34 (28); 133, 154 (8) Hanson, R.J., 23, 35 (42) Hansten, P.D., 24, 35 (46) Harman, R.J., 94, 103 (72); 107, 124 (9); 144, 155 (27) Health Care Financing Administration, HEW, 165, 177 (18) Health Care Technology Center, University of Missouri, 52, 66 (36) Health Education and Welfare, Department of, 106, 124 (1); 107, 124 (10); 108, 125 (17, 18); 109, 125 (19) Health Resources Administration, 170, 177 (21) Health Services Administration, 170, 177 (21) Heard, M.R., 22, 32 (6); 82, 100 (31); 83, 100 (32-34); 86, 101 (41) Hedlund, J., 23, 34 (39), 35 (40); 29, 35 (52) Henley, R.R., 13, 20 (5, 6); 22, 32 (1); 80, 100(26); 114, 126 (31); 162, 177 (11, 12) Henson, D., 23, 34 (34) Hickman, C.V., 23, 35 (40) Hicks, G.P., 23, 33 (19) Hodge, M., 11, 19(1, 2); 22, 33 (15); 49, 66 (30); 81, 100(29) Hoover, L.W., 23, 34 (25) Hopwood, M.D., 29, 36 (55); 95, 103 (77) Hospital Financial Management Association, 78,99 (15) Hubbard, J.P., 122, 126 (44) Huffmire, D., 54, 67 (38) Hulse, R.K., 24, 35 (44) Hunn, G.S., 24, 35 (46) Hurst, J.W., 44, 66 (22) Hurst, L.L., 115, 126 (32); 116, 126 (34); 160, 176 (4) International Business Machines (IBM), 57,67 (43) Jackson, J.C., 24, 35 (44) Jacquez, J.A., 168, 177 (19, 20) James, D.C., 120, 126 (40) James, J.D., 24, 35 (45) Jessiman, A.G., 22, 32 (7) Johnson, J.W., 134 (11) Justice Dept/Sweden, 91, 102 (62) Justice, N.S., 40, 65 (14, 15) Jydstrup, R.A., 2, 7 (2) Kaluzny, A.D., 107, 125 (14); 108, 125 (15, 16) Katz, E., 106, 124 (6); 109, 125 (20-22); 110, 125 (24) Kelchammer, U., 6, 8 (12) Kennedy, E., 122, 127 (48) Kenney, G.C., 136, 154 (13) Kimbleton, S., 143, 155 (24, 25) Kinney, T.D., 23, 33 (21) Kirk, W.R., 90, 102 (55) Kirklin, J.W., 23, 34 (29) Klarman, H.E., 46, 66 (28); 73, 99 (8) Knight, J., 107, 124 (8) Kulikowski, C.A., 142, 155 (22); 149, 156 (36) Lacy, W„ 29, 36 (54); 95, 103 (77) Lamontagne, A., 22, 32 (8) Lamson, B.G., 22, 33 (13) Larson, F.C., 22, 33 (19) Larsson, K., 14, 20 (7) Laska, E.M., 23, 34 (38); 29, 35 (53) Lawrence, F., 78, 99 (14) Lawrence, S.V., 149, 156 (35) Leavitt, D.D., 134, 154 (12) Lederberg, J., 148, 155 (32) Levy, A.H., 6, 7 (6) Levy, S., 142, 155 (22) Lewis, T.L., 23, 34 (35); 50, 66 (32); 55, 67 (40) L’Heureux, D.P., 23, 35 (42) Lindberg, D.A.B., 16, 20 (9); 23, 33 (17), 34 (33), 35 (41); 24, 35 (47); 63, 67 (54, 55); 69, 98 f2); 94, 103 (70); 95, 103 (75); 161, 177 (8, 10) 186 Growth of MISs in the United States Ling, R.O., 95, 103 (77) Lionberger, H.F., 110, 125 (23, 26) Lodwick, G.S., 24, 35 (49); 161 (6) Loow, S.O., 6, 8 (11); 14, 20(8) Lovejoy, F.H., 142, 155 (23) Lusted, L.B., 158, 176 (2); 160, 176 (3) Mabry, J.C., 95, 103 (77) Macks, G.C., 23, 34 (35); 50, 66 (32); 55, 67 (40) Manning, E., 131, 154 (6) Markivee, C.R., 24, 35 (49) Marquis, D.G., 106, 124 (2) Matthews, C., 163, 177 (15) McCarter, P.M., 111, 125 (28) McDonald, C.J., 22, 33 (11) McDonnell Douglas, 50, 66 (34); 56, 67 (42); 79, 99(18) McFadyen, J.H., 131, 153(2) McNeer, F„ 96, 103 (78) Medicus Systems Corp., 79, 99 (22) Mellin, G.W., 119, 126 (39) Melville, R.S., 23, 33 (21) Mengel, C.E., 24, 35 (47); 161, 177 (8) Menzel, H., 106, 124 (6); 109, 125 (20-22); 110, 125 (24) Messersmith, A.M., 23, 34 (25) Miller, R.A., 149, 155 (34) Mintz, M., 119, 126 (37) Mishelevich, D.J., 50, 66 (35); 55, 67 (41); 58, 67 (44) Mittler, B.S., 96, 103 (78) Moore, A., 23, 34 (24, 25) Moore, T.N., 24, 35 (46) Morrell, J„ 161 (7) Morris, J.J., 96, 103 (78) Mowshowitz, A., 94, 102 (68); 121, 126 (42) Myers, J.D., 149, 155 (34) National Bureau of Standards, 42, 66 (20); 172, 178 (26) National Cash Register Corp., 131, 153 (3) National Library of Medicine, 122, 126 (45); 171, 178 (23, 24) Nazzaro, J., 24, 35 (48) Nishimura, T.G., 24, 35 (46) Norwood, D.D., 81, 100 (28); 85, 100 (36) 101 (40) Noyce, R.N., 130, 153 (1) Nyborg, P.S., 111, 155 (28) O’Connor, J.L., 95, 103 (77) Osborn, J.J., 23, 34 (30) Overhage, C.F.J., 94, 103 (72); 107, 124 (9); 144,155 (27) Palley, N.A., 29, 36 (55); 95, 103 (77); 144, 155 (28, 29) Parent, M.S., 40, 65 (14, 15) Peebles, R., 131, 154 (6) Penchas, S., 40, 65 (14) Perez, C., 23, 66 (34) Pesut, R.N., 40, 65 (10); 81, 100 (27, 30); 117, 126 (35) Peterson, H., 6, 8 (11); 14, 20 (8) Pipberger, H.V., 23, 34 (26) Pisinski, E., 79, 99 (20) Podlone, M., 24, 35 (46); 161, 177 (7) Pople, H.E., 149, 155 (34) Price, D., 107, 125 (13) Prior, R.E., 50, 66 (32) PROMIS Laboratory, 23, 34 (32); 43, 66 (21); 45, 66 (24-26); 46, 66 (27); 161, 177(9) Raison, J.C.A., 23, 34 (30) Ramcharan, S., 70, 98 (5) Raymond, S., 23, 33 (18) Reader, G.G., 107, 125 (14); 108, 125 (15, 16) Reese, G.R., 23, 34 (33) Reinfrank, R.F., 133, 154 (10) Remp, R„ 106, 124 (5) Rexroad, M.G., 50, 66 (32) Richart, R., 80, 99 (24) Robbe, P.F., 6, 7 (6) Rockwell, M.A., 144, 155 (28, 29) Rosati, R.A., 96, 103 (78) Roundl, J.T., 94, 103 (73) Rowland, L.R., 64, 68 (56) Rowland, T., 142, 155 (23) Rowny, P., 42, 65 (18) Russell, L.B., 106, 124 (7) Name Index 187 Russell, W.S., 22, 33 (13) Rydell, R., 81, 100 (28); 85, 100 (36), 101 (40) Saathoff, J., 64, 68 (56) Safir, A., 149, 156 (36) Saito, M„ 6, 7 (3) Schmitz, H.H., 80, 99 (23); 88, 101 (46- 49); 89, 101 (50) Schroeder, J., 64, 68 (56) Schultz, L., 94, 103 (73) Schwartz, W., 153, 156 (42) Scoville, D.P., 38, 64 (3); 40, 65 (9, 11, 12); 151, 156 (41) Secretary’s Advisory Committee on Auto- mated Personal Data Systems, 92, 102 (63) Seiden, H.R., 94, 103 (73) Senior, J.R., 95, 103 (74) Shaffer, R.A., 30, 36 (56) Sheehan, J.M., 50, 66 (32) Shenkin, B., 93, 102 (65, 66) Sheppard, L.C., 23, 34 (29) Shires, D.B., 6, 7 (5) Shortliffe, E.H., 24, 35 (46); 148, 155 (33) Sibley, W.L., 29, 36 (55); 95, 103 (77) Siegelaub, A.B., 70, 9 (4, 5) Siler, W„ 94, 103 (70) Slack, W.V., 161, 177 (5) Sletten, J.W., 23, 34 (37, 39); 29, 35 (52) Small, H.G., 106, 124 (3) Smith, L.A., 24, 35 (46) Smith, R.L., 142, 155 (22) Smith, V., 24, 35 (50) Snowden, J.I., 50, 66 (32) Somand, M.E., 40, 65 (14, 15) Spraberry, M.N., 23, 33 (20) Sridharan, D., 142, 155 (22) Stacy, R.W., 94, 103 (69) Stankovic, J.A., 131, 153 (4) Starmer, F., 96, 103 (78) Stead, W.W., 22,33 (12) Stewart, D.H., 144, 155 (28, 29) Stewart, W.B., 24, 35 (47); 161, 177 (8) Straube, M.J., 22, 33 (12) Stroebel, C.F., 38, 64 (1, 2) Suntharalingam, N., 134, 154 (11) Swedlos, D.B., 50, 66 (32) Systemedics Inc., 55, 67 (39); 162, 177 (13); 163, 177 (17) Tao, D.K., 53, 67 (37) Taussig, H.B., 119, 126 (38) Texas Institute of Rehabilitation & Re- search, 46, 66 (29) Thibodaux, T.T., 22, 33 (13) Thier, S.O., 150, 156 (38) Thomas, J.C., 22, 32 (6); 82, 100 (31); 83, 100 (32-34); 86, 101 (41) Thomas, L.B., 23, 34 (34) Thompson, H„ 29, 36 (54); 95, 103 (77) Toffler, A., 121, 126 (43) Traska, M„ 79, 99 (21) Tully, R.J., 24, 35 (49) Tuthill, B., 23, 33 (22), 34 (24) Uberia, K., 6, 8 (12) U.S. Congress, Congress, 92, 102 (64); 170, 178 (22); 174, 178 (29); House, 173,178 (28); House Committee on Interstate and Foreign Commerce, 89, 101 (51), 102 (53); House Committee on Science and Astronautics, 75, 99 (9); House Commit- tee on Science and Technology, 50, 66 (33); House Committee on Ways and Means, Joint Hearing, 89, 101 (52); Senate, 173,178 (27); Senate Committee on Banking, Housing and Urban Affairs, Hearings, 91, 102 (58) Ulett, G.A., 23, 34 (37) University of Missouri, 163, 177 (16) Van Brunt, E.E., 22, 33 (10); 62, 67 (53) Van Cura, L.J., 161, 177 (5) Van Dam, A., 131, 153 (4) Van Egmond, J., 6, 7 (6) Wagner, G., 6, 8 (8) Wallace, A.G., 96, 103 (78) Wallace, E„ 94, 103 (73) Walsh, C., 23, 35 (43) Walter, D., 58, 67 (45) Warner, D.C., 93, 102 (66) Warner, H.R., 24, 35 (44); 39, 65 (4, 5, 8); 150, 156 (37); 151, 156(40) 188 Growth of MISs in the United States Watkins, S., 143, 155 (25) Watson, R.J., 81, 100 (28); 85, 100 (36), 101 (40); 94, 103 (73) Waxman, B„ 40, 65 (15); 42, 65 (18), 66 (19); 94, 103 (69) Weed, L., 43, 66 (21); 45, 66 (24) Weil, M.H., 23, 34 (31) Weiss, S., 149, 156 (36) Weizenbaum, J., 153, 156 (43) Wesbury, S.A., 90, 102 (54) West, B.J., 38, 64 (3); 40, 65 (9, 11, 12); 151, 156 (41) Westin, A.F., 69, 98 (3); 90, 102 (56); 91, 102 (57, 59, 60); 172, 178 (25) Whitehead, E„ 162, 177 (14) Wiederhold, G., 13, 20 (5, 6); 22, 32 (1); 80, 100 (26); 114, 126 (31); 150, 156 (39); 162, 177 (11, 12) Wiener, F., 23, 34 (31) Wiener, N„ 1,7(1) Williams, T.M., 88, 101 (49) Wolf, H., 6, 7 (5) Wood, C.T., 22, 32 (8) Wood, H.M., 143, 155 (24) Wyllys, R.E., 94, 103 (73) Yeh, L., 42, 65 (18) Yosten, L., 24, 35 (46) Zuckerman, A., 42, 65 (18) Subject Index Accounting systems, as example of innova- tion, 111 Administrators, hospital: educational back- ground, 90; requirement for cost savings, 89-90; technical skills as a barrier to MIS, 121 Admissions office: cost-effectiveness study of MIS, 86-87; microprocessors and, 131; tasks performed by MISs, 22 Advanced Communications Service (ACS), 141 AFCLAS evaluation, 87 Agricultural model of diffusion, 110 Amarel, Saul, cited, 149 Ambulatory Care Department, tasks per- formed by MIS, 22 American Telephone and Telegraph Co. (ATT), 141 Analytic framework, for comparing MISs, 15-19 Analytic Services Corp., study of clinical laboratory system, 87-88 Ankylosing spondylitis, genetic association as MIS problem, 120 Arnstein, Sherry, cited, 76 ARPANET, 151 Arthritis: HL-A27 Genetic Marker, 120 Artificial Intelligence technology: descrip- tion of the technology, 145-146; effects on MISs, 129, 150-153; future uses in MISs, 149-150;usesin MIS, 142,147-149 Auenbrugger, Leopold, cited, 109 Austin, Charles, cited, 121 Balkanization of medicine, barrier to MIS development, 105, 121 Ball, M.J., cited, 122 Barnett, G. Octo, cited, 12, 22, 40 Barriers to MIS diffusion: generally to com- puters in medicine, 113-117; hardware and software, 116-117; medical barriers, 117-121; operational difficulties, 113-116; social barriers, 121-124; with central computers, 131-132 Baruch, Jordan, cited, vii Battelle Laboratories, evaluation of MIS, 81-82, 85-86 Bender, Allen, cited, 13 Benefits, potential, from MISs, 70 Bibliometric studies of innovation, 106 Biomedical Research Panel, President’s Report of, 108-109 Biotechnology Branch of NIH, 29, 169 Bloustein, Edward, cited, 93 Bolt, Beranek and Newman, and GE MEDI- NET system, 59 Buchanon, Bruce, cited, 148 Burroughs Corp., hardware supplier, 54; MIS supplier, 53, 96 Bush, Vannevar, cited, 137 Business office: example of diffusion of innovation, 111; requirements for MIS, 89-90; tasks performed by MIS, 22 Caceres, Cesar, cited, 133 Canada: data privacy study, 91; University of Alberta, 82 CASNET, artificial intelligence clinical consultant program, 149 CBX. See Competency testing for clinical, 95 Census System, hospital, MIS applications in, 22, 83 Central processor vs. network architecture of MIS, 49, 50 CLINFO System: scientific impact, 95; specialized MIS, 29 Coleman, J.S., cited, 106, 109 Collen, M.F., cited, 12, 22, 113-114 Commercially offered MISs: discriminating features, 49-53; general characteristics, 37, 48; limitations, 47; specific systems, 53-58 Communications technology: description of the technology, 139-141; effects on MISs, 142-144; federal policy, 144; uses in MISs, 141-142 Comparison of MISs, analytic framework, 15 189 190 Growth of MISs in the United States Competency, testing for clinical, 95 Comprehensiveness of MISs: need to match hospital characteristics, 52-53; potential problems, 49 Computer tomography, rapid diffusion of, 122-123; storage of records of, 136 Control Data Corporation: Learning Cen- ters, 58; potential role in MIS, 58 COSTAR System, MIS application 40-42 COSTAR V MIS for outpatient practices, 42 Cost-benefit of MISs: as evaluation method, 74-75; detailed problems in analysis, 85-86; results of analyses, 85-90 Cost-effectiveness of MISs: as evaluation method, 72-73; results of analyses, 41- 42,85-90 Costs of health care: hospital, motivation for containment, 123; long vs. short term, 123; reduction by MISs, 41-42, 69, 81, 83, 85, 89 Costs of MIS development, 162-163 Costs of MIS operation: charge reporting systems, 56; Harvard Community Health Plan, 41-42; Institute of Living, 38; PROMIS, 45 Cotton, I., cited, 140 Cummings, Martin M., cited, 107 Cybernetics, 1 Daddario, Emilio, cited, 75 Danderyd MIS, 14 DES. See Diethylstilbesterol Deaconess Hospital, MIS evaluation, 88-90 Definition: of artificial intelligence, 145; of MIS, 9 DENDRAL, artificial intelligence chemistry consultant program, 148 Diagnosis: artificial intelligence and, 149; microprocessors and, 131 Dietetics: microprocessors and, 131; tasks performed by MISs, 23 Diethylstilbesterol: toxicity discovery as an MIS problem, 119-120 Diffusion: of scientific ideas, 77; of tech- nology innovation, 105-106; of tech- nologic innovation in medicine, 106-113 Dimensions for comparing MISs, 15 Distributed processing and microprocessors, 131 Donabedian, Avedis, cited, 71 Drug information systems, 142 DuBois, Richard, cited, 49, 40, 52 Duke University: MIS application, 52, 57; prognostigram system, 96-98 Education: of hospital administrators, 90;of physicians and specialists, 121-122, 159 Einthoven, Willem, 109 El Camino Hospital: computer hardware back-up, 117; customization of MIS, 56; evaluation of MIS by staff, 81, 85; studied by Battelle, 81-82, 85-86 Electrocardiography: computer interpreta- tion of, 111, 132-133; diffusion of auto- mated technology, 111-112, 132-133; microprocessors and, 132-133; tasks performed by MISs, 23 England: computer tomography, 123; MIS publications, 6 Epidemiological studies: and PROMIS system, 43; DES, 119-120; MISs in, 77-78; phocomelia, 118-119 Etzioni, Amitai, cited, 106 Evaluation of MISs: methods for, 72-78; need for, 69-71; results of, 78-98 Externalities, in cost-benefit analyses, 75 Fair Credit Reporting Act of 1973, 91 Fair Information Practice, code of, 92 Federal Communications Commission (FCC), 140, 144 Feigenbaum, Edward, cited, 145, 146, 148 Flagle, Charles, cited, 80 Flexner Report, 107-108 Food. See Dietetics Friedman, Richard, cited, 115 Gardner, Reed, cited, 38, 40, 150 Garfield, Eugene, cited, 106 Garvey, W.D., cited, 107 General Electric Company: detailed descrip- tion of MEDINET, 59-61 ;early position, 11; medical market, 122; MEDINET system, 58-62; special features of MEDINET system, 61-62 Subject Index 191 Germany, West: and phocomelia, 119; MIS publications, 6 Giebink, G.A., cited, 115-116, 160 Glaucoma. See CASNET Glueck, Bernard, cited, 37 Greene, Barry, cited, 121 Griffith, Belver C., cited, 107 Gruber, W.H., cited, 106 Gustafson, David, cited, 115 Hardware, computer, percentages supplied to hospital market, 54 Harvard Community Health Plan, MIS application, 26-27, 40-42 Health Resources Administration, role in management of MIS diffusion, 176 Hedlund, James, cited, 29 HELP system: as knowledge-based system, 150-151; in MIS application, 39; inter- connections, 143 Henley, R., cited, 13, 22, 80, 114, 162 Hewlett-Packard ECG system, 133 Hill-Burton program, 89 History: interrogative patient, 131; of early MIS systems, 58-64 Hodge, Melville, cited, 11, 13 Honeywell Corporation: hardware supplier, 54; MIS supplier, 53 Huffmire, Donald, cited, 54 Hurst, L.L., cited, 116, 160 Hypothesis generation, problem in MISs, 118 IBM: early position, 11, 54, 122; Health Care Support System, 54, 57; MIS ap- plications, 50-52, 54, 55; hardware supplier, 54 Ideas: diffusion of in medicine, 107-109; stewardship of, in science, 76 Identification of patients: as data privacy issue, 91; as obstacle to MISs, 64, 120 IFIP, meetings and publications on MISs, 6 Images: medical, storage of, 138; storage of radiological, 136; written medical record, 137 Infeasible tasks, in MISs, 30-32 Information: content of MIS, 14-15; scientific, patterns of use, 106 In-house development of MISs: potential problems, 49; under manufacturer- supplied software, 52 Innovations: grades of in agriculture, 110- 111; grades of in medicine, 111-113; technical, stages of, 105 Institute of Living, MIS application, 24-25, 37-38 Intensive Care Units: alerts in Utah MIS, 40; evaluation of Utah MIS, 84; tasks performed by MISs, 23 Interaction, direct, of MIS with physician. See Philosophy of MIS system building Interagency strategy, 175-176 INTERLISP, Computer programming language, 147 International: developments in MIS, 2-3; MIS literature, 5-6; scope of Thalidomide tragedy, 119 International Telephone and Telegraph (ITT), 141 INTERNIST, artificial intelligence clinical consultant program, 149 INTREX Project, 144 Japan: and phocomelia, 119; Medical In- formation Systems Development Center, 6; MEDINFO, 6; MIS publications, 6 Kaiser-Permanente: federal policy, 62; indirect effects of MIS, 70-71; MIS application, 62-63, 70, 80, 117; privacy of data, 91 Kaluzny, A.D., cited, 107, 108 Katz, E., cited, 106 Kimbleton, S., cited, 143 Knowledge based systems. See Artificial Intelligency technology Knowledge engineer, 146 Knowledge, limited state of, 3 Kulikowski, Casimir, cited, 149 Laboratory, clinical: AFCLAS evaluation, 87; artificial intelligence, 150; cost- effectiveness evaluation, 73; tasks per- formed by MISs, 23; University of Missouri system, 23, 63-64; University of Utah system, 39-40 192 Growth of MISs in the United States Laennec, Rene, cited, 109 Language translation, 30 Laser-etched disk storage: description of the technology, 135-136; effects on MISs, 137-139; uses in MISs, 136-137 Laska, Eugene, cited, 29 Latter Day Saints Hospital, MIS application, 38-40, 84 Lederberg, Joshua, cited, 148 Legislation, federal: Fair Credit Reporting Act, 91; Medicare, 163; Medicare-Medi- caid Anti-Fraud and Abuse, 164; Nation- al Center for Health Care Technology, PL93-623, 174; Privacy Act, 92 Levy, Allen, cited, 46 Linguistics and language translation, 30 Lionberger, H.F., cited, 110-111 Literature, indexing of scientific: as means of studying diffusion of ideas, 106; in PROMIS system, 45 Little, A.D., MIS hospital study, 84 Lockheed Corporation, 53 Loyola Hospital, MIS experience, 51 Lusted, L.B., cited, 160 Malcolm Grow, USAF Medical Center, MIS evaluation, 87 Management: obstacles to MIS, 121; of dif- fusion of MIS technology, 6-7, 165-176; of hospitals for MIS cost savings, 89-90 Marketplace of MISs: as evaluation method, 72; reflecting diffusion of MIS tech- nology, 48; results of, 78-79 Marquis, D.G., cited, 106 Massachusetts General Hospital, MIS ap- plication, 40-42 McDonnell-Douglas Automation, MIS ap- plications, 51, 53, 55, 141 Medical audit: and PROMIS, 43; and PSRO, 123 Medical Education: curricula of medical students, 122; medical computer train- ing, 122 Medical Group Management Association, cited, 78 Medical Information System: definition, 9; comparison of, dimensions, 15-19; im- plications of definition, 9-10 Medical Information Technology, Inc., 79, 141 Medical knowledge: limitations of, 118 Medicare, reporting requirement, 163 Medicus Medical Systems: MIS applications, 53; SPECTRA system, 55, 56 MEDINET. See General Electric Company MEMEX system, 137 Memory: central computer, 135; secondary, disk, 135 Mental Health: Institute of Living MIS, 37-38; interstate MIS, 29; statewide MIS, 29; tasks performed by MISs, 23 Menzel, H., cited, 106 Microcomputers: description of the tech- nology, 129-131; effects on MISs, 134-135; outcome, 134-135; radiation treatment planning and, 134; uses in MISs, 131-134 Microwave Communication, Inc. (MCI), 141 Multiphasic screening: MIS applications in, 70; indirect effect of MIS upon, 70-71 MUMPS language: in COSTAR MIS, 40; in MIS in Institute of Living, 38 MYCIN, artificial intelligence clinical consultant program, 148 Myers, Jack, cited, 149 NCR Corporation, 53, 54 NDC Systems, MIS applications, 53, 55 National Bureau of Standards, potential to manage MIS diffusion, 171-172 National Center for Evaluation of Medical Technologies, 173-174 National Center for Health Care Tech- nology, 174 National Center for Health Services Re- search : potential to manage MIS diffu- sion, 169-170; sponsorship of MIS evaluations, 80, 81-82, 114 National Institute of Health Care Research, 173 National Institutes of Health, Division of Computer Research and Technology, 169 National Institutes of Health, Division of Research Grants: potential to manage MIS diffusion, 168-169; support of com- puter research, 158, 159 Subject Index 193 National Library of Medicine: potential lead agency for MIS diffusion, 175-176; potential to manage MIS diffusion, 170- 171; training programs for medical computing, 122 National Science Foundation, 159 National Software Works, 144 Network: architecture, 131; operating systems, 143. See also Communications technology Noyce, R.N., cited, 130 Nuffield Provincial Hospitals Trust, 5 Nutrition. See Dietetics Operating Room, surgical, tasks performed by MISs, 23 Operations research: as evaluation method for MIS, 72; results of, 79-84 Outcome measures, patient: as evaluation of PROMIS, 46; as method of evaluation generally, 71; dfficulties in evaluating, 74 Packet switching, 141 Paradigms for managing MIS diffusion: judgmental, 166; managerial, 167; observational, 166-167 Parkland Hospital: MIS application, 58; MIS installation schedule, 50; MIS terminals, 55 Pharmacy, tasks performed by MISs, 24 Philips, N.V., Company, 135 Philosophy of MIS system building: ambu- latory system taxonomy, 13; and prac- tical hospital planning, 50-51; cumula- tive view, 11-12; direct interaction of MIS with physician, 51,52; holistic view, 10-11; intermediate view, 12-13 Phocomelia. See Thalidomide Physician data entry, handwritten data entry, 137. See also Philosophy of MIS system building Poison control information, 142 Policy, federal: encouragement of MISs, 78, 160; future management of MIS tech- nology, 165-176; health care reim- bursement, 163-165;mechanism for managing MIS diffusion, 167-176; non- medical, effect on MISs, 157; research support for computers in medicine, 158— 160; technology transfer, 160-163; television transmission, 139 Pople, Harry, cited, 149 Practices, diffusion of in medicine, 109 Price, Derek, cited, 107 Privacy, rights to: Act of 1974, 92-93; El Camino Hospital, 82; NBS cipher, 172; patient medical record, 93; Secretary’s Committee, 91-92; Westin studies, 90-91 Problem domain. See Analytic framework Problem-oriented medical records. See PROMIS Process measures: as method of evaluation generally, 71; in TIRR MIS system, 47 Profitability, vs. feasibility, 30 PROGNOSTIGRAM. See Duke University PROMIS: as example of innovation, 112- 113; as knowledge-based system, ISO- 152; interconnections, 143; MIS applica- tion, 27-28,43-46 PSRO, as potential user of MIS, 123 Quality: control in laboratory, 150; of patient care, 41, 93 Radiology: image storage, 136; tasks per- formed by MISs, 24; radiation therapy planning, 134 RAND Corporation: CLINFO System, 29; hospital communications networks, 144; model of technology transfer, 106 Recommendations for managing MIS tech- nology, 6-7 Records, medical: department, tasks per- formed by MISs, 23; example of diffu- sion of innovation, 112; hand-written, 137; laser disk storage of, 137; micro- computers and, 131; patient-owned, 93 Remp, Richard, cited, 106 Richardson, Elliot, cited, 91 Richart, Robert, cited, 80 Roentgen, Wilhelm K., cited, 109 Rosati, Robert, cited, 96 Rutgers University, 142, 151 Safir, Aran, cited, 149 Satellite Business Systems, 141 194 Growth of MISs in the United States Scandinavia, and MISs, compared to U.S., 91 Schmitz, Homer, cited, 80, 88 Scientific impact of MISs: as evaluation method, 76-78; competency testing, 95; CLINFO evaluation, 95; general studies, 94; PROGNOSTIGRAM, 96 Shared Medical Systems, MIS applications, 53, 54, 141 Shenkin, Budd, cited, 93 Simulation of patient, by computer, 95 Sisters of the Third Order of St. Francis, 53 SJURA. See Sweden Small, Henry G., cited, 106 Social Security number, as privacy issue, 91 SPECTRA. See Medicus Medical Systems Spencer, William, cited, 46 SPRI. See Sweden Standards, setting of: as part of PROMIS system, 45, 112; for reporting computer usage, 165; list, related issues in MIS development, 172; need in MIS develop- ment, 120 Stanford University, 142, 151 Starmer, Frank, cited, 96 Stead, Eugene, cited, 96 Sweden: Danderyd MIS system, 14; Report on Data Privacy, 91; publications on MISs, 6; SJURA, 6; SPRI, 6 System for Hospital Uniform Reporting (SHUR), 165 Systems, diffusion of in medicine, 109-110 Technicon Medical Information Systems Corp.: MIS applications, 53, 56, 117, 141; TMIS, 56 Technology transfer, 160-162 Technology assessment of Miss: as evalua- tion method, 75-76; results of, 90-94 Telenet, 141 Television: and education, 139; and video storage disks, 139; U.S. engineering standards, 139 Terminology, medical, 120 Texas Institute of Rehabilitation and Re- search, MIS application, 28-29, 46-47 Thalidomide, toxicity discovery as an MIS problem, 118-119 Thompson, Howard, cited, 29 Transferability of MISs, 49-50 TRILAB System, 87-88 Tymnet, 141 Understanding, limitations of medical knowledge, 118-120; limitations of state of knowledge, 3 University of Alberta: MIS application, 82; MIS evaluation, 86-87 University of Missouri: hospital survey, 52- 53;MIS application, 63-64 University of Utah: future potential, 151; MIS application, 25-26, 38- 40 University of Vermont, MIS application, 27-28,43-46 Valibona, Carlos, cited, 46 VARIAN computers, PROMIS system, 44 Veterans Administration, role in manage- ment of MIS diffusion, 176 Video disk storage systems: description of the technology, 135-136; effects on MISs, 137-138; outcome, 138; uses in MISs, 136-137 Vocabulary of medicine, standardization as an MIS need, 120 VORTEX operating system, 44 Wallace, Andrew, cited, 96 Waller, Augustus D., cited, 109 Ware Report, to Secretary of HEW, 91-92, 94 Warner, Homer, cited, 38, 39, 150 Waxman, Bruce, cited, 42 Weed, Lawrence, cited, 43, 44-45, 112-113 Weinberger, Caspar: privacy report to Secretary of HEW, 91-92; statement on Hill-Burton program, 89 Weiss, Sholom, cited, 149 Weizenbaum, Joseph, cited, 153 Westin, Alan, cited, 90, 91 Wiederhold, G„ cited, 13, 22, 80, 114, 162 Wiener, Norbert, cited, 1 XPERT, artificial intelligence consultant program, 149 X-ray. See Radiology About the Author Donald A.B. Lindberg is professor of pathology and director of the Information Science Group in the School of Medicine at the University of Missouri-Columbia, and director of the campus Health Care Technology Center. He received the M.D. in 1958 from the College of Physicians and Surgeons of Columbia University, and his specialty training in anatomic and clinical pathol- ogy at Columbia-Presbyterian Medical Center, New York. Dr. Lindberg also holds the D.Sc. from Amherst College, honoris causa. Dr. Lindberg did research in experimental pathology, dealing with pneu- monia, modes of antibiotic action, and extra-corporeal oxygenation. Subsequent- ly he began a long term investigation of the use of computers in medicine, founding one of the first medical computer centers, in 1963 at the University of Missouri. Early systems built by this group included admissions, diagnostic registry, clinical laboratory, educational systems, and a statewide EKG inter- pretation system. He is an editor of four journals and author of The Computer and Medical Care, Charles C Thomas, 1968, and editor with William Siler of Computers and Life Science Research, Plenum Press, 1974. He has published over 100 scientific articles and book chapters. He was a member of the Computer Science and Engineering Board of the National Academy of Sciences and is the U.S. representative to the Interna- tional Medical Informatics Association of the International Federation for Information Processing.