Medicare Advantage health care Data Brief #6 December 2016 utilization: observation stays KEY FINDINGS Trends in Medicare Advantage outpatient observation stays Use of observation stays in- In this data brief, the Health Care Cost Recently, the use of observation stays creased from 2010 through Institute (HCCI) reports on national within the Medicare FFS population rates of utilization and trends of out- has become an area of focus for policy 2014. patient observation stays in the Medi- makers, health care providers, and Similar increases occurred in overall use care Advantage (MA) population, aged patients because coverage and provid- and within 30 days of a hospitalization. 65 years and older, for the period er reimbursement differs for inpatient 2010 through 2014. This is the second and outpatient events. These differ- The percent of multiple day stays HCCI brief in an ongoing series on the ences potentially have cost and access utilization of health care services by implications for Medicare FFS benefi- and the average length of stay the MA population. The first brief re- ciaries. For example, Medicare Part A was constant over time. ported on hospital readmission rates.1 (hospital insurance) pays for inpatient Approximately 21% of observation stays MA, formerly known as Medicare Part hospital services; Medicare Part B spanned more than one day with an av- (medical insurance) covers, among C, is available to individuals eligible erage length of stay of 1.55 nights. other services, outpatient services. for fee-for-service (FFS) coverage. The MA program allows individuals to ob- Due to differences in Part A and B cov- 81.8% from 3.14 to 5.71 per 100 MA mem- erage, it is possible that patients will tain health insurance through com- bers. However, substantially larger annual face larger out-of-pocket payments for mercial health plans, in lieu of FFS increases occurred in 2013 and 2014. with an outpatient stay than for an inpa- coverage and provides similar cover- year over year percentage changes of 35.8% age but the structure of the benefit tient stay. Moreover, the FFS coverage rules have implications for whether and 21.0%, respectively. design may differ.2 As of 2015, over follow-up services such as skilled In Table 2, we report the percent of observa- 31% of Medicare beneficiaries were enrolled in an MA plan. 3 nursing facility care will be covered.5 tion stays with a length of stay (LOS) of one Within the MA population there are or more nights (i.e. the percent of observa- The HCCI data analyzed for this brief tions stays spanning 2 or more consecutive included health insurance member- also likely cost and utilization implica- tions to patients and providers based days). As noted above, observation stays are ship and claims from MA enrollees in billed as outpatient facility events, but it is on the MA plan benefit design. Howev- all 50 states and the District of Colum- possible that an observation stay spans mul- bia. On average, the HCCI data ac- er, there is generally less published research on the MA population com- tiple days and includes one or more over- counted for approximately 25% of the night stays in a hospital setting. The percent pared to Medicare FFS. Therefore, we national MA population over the study of multiple day observation stays was gener- report on the rate of observation stays period. ally consistent over the study period. The 5- within the MA population. This brief Observation stays contributes to the publicly available year average is 21.1% with the annual per- centage of total observation stays spanning Prior to or following outpatient facility data on the patterns of health care more than one day ranging from a low of services, such as a same day surgery, utilization in the MA population. Ex- amination of expenditures related to 20.3% in 2011 to a high of 22.5% in 2013. health care providers may request that a patient stay for evaluation or observation stays in the MA popula- Table 2 also shows the average LOS for mul- testing without admitting the patient tion is left for future reporting. tiple day stays in each year of the study peri- to the hospital as an inpatient. This The utilization of observation stays in od. LOS is calculated as the number of con- continued evaluation is considered an the MA population increased from secutive days minus one; thus, LOS, is equiv- observation stay and the services are 2010 through 2014 alent to the number of nights. The average classified as outpatient facility events, LOS is also consistent over time with an av- Table 1 shows the rate of observation erage of 1.55 nights for years 2011 through even if the observation stay lasts more stays per 100 MA members for each 2013, increasing to only 1.60 in 2014. than one day. Generally, observation year of the study period. The annual stays are expected to be used in situa- It is possible that a two day, one-night stay is change in observation stays is also tions where the patient is not ex- less than 24 hours. For example, a patient’s shown. Over the course of the study pected to stay more than two nights. 4 observation stay may start at 6:00 PM and period, observation stays increased www.healthcostinstitute.org 1 last until 8:00 AM the next day. Howev- in observation stays following hospitali- tient readmission. Any additional obser- er, an average LOS of 1.60 implies that zations were observed in the last two vation stays within the 30-day window the average multiple day stay is more years of the study period. Moreover, the were excluded from the calculation of than 24 hours because a LOS greater annual percentage increases in observa- the rate. than 1.00 requires three or more con- tion stays per 100 hospital-wide admis- The 30-day rate of all-cause observation secutive days. A stay spanning three sions were similar in magnitude to the stays were calculated as the sum of eligi- days, for example beginning at 10:00 PM percentage increases in observation ble observation stays divided by the Tuesday and ending at 6:00 AM on stays per 100 members in 2013 and total number of index admissions. The Thursday, includes two overnight stays 2014. rate was then multiplied by 100 in order – Tuesday night, and Wednesday night. Conclusion to report in terms of the number of ob- The rate of observation stays following servation stays. The results presented in this brief show hospitalizations increased for all cate- that for a large, national MA population, Limitations gories of hospitalizations studied. aged 65 and older during the years 2010 Although, this report provides new sta- We report 30-day all-cause observation through 2014, observation stays in- tistics on the utilization of health care stays following a hospital admission in creased over time, both overall and within the MA population, the results Table 3. Because observation stays are within 30 days of a hospitalization. The may not be generalizable to all MA plans billed as outpatient facility events, they percent and length of overnight obser- or members. The HCCI data set is a con- are usually not included in counts of vation stays, however, remained con- venience sample from three national hospital admissions or hospital readmis- stant. insurers and may not reflect the utiliza- sions, even if the observation stay in- Data and Methods tion of the full MA population or the cludes an overnight stay in a hospital. The analysis sample was limited to indi- trends of other MA payers. Therefore, there is interest in observa- tion stays following admissions because viduals aged 65 years and older enrolled There were likely numerous factors in- they may be used instead of inpatient in an MA plan included in the HCCI data- fluencing the use of observation stays admissions, which in turn would reduce base. Inclusion in the post hospitaliza- during the study period. However, no readmission rates. Table 3 reports ob- tion observation stay rate calculation causal inferences can be drawn from the servation stays as either per 100 hospi- also required at least one inpatient hos- results presented in this brief and those tal-wide admissions or per 100 admis- pitalization with three or more months influencing factors. Moreover, the anal- sions for one of four specific conditions: of MA plan membership prior to a hospi- yses did not attempt to identify any acute myocardial infarction (AMI), heart talization and at least 30 days of mem- causes underlying the observed rates or failure, pneumonia, or chronic obstruc- bership following a hospitalization. trends; and the analyses did not evalu- tive pulmonary disease (COPD). Outpatient observation stays were iden- ate the impact of any particular federal, tified by revenue codes 0760 or 0762, state, or insurer policies or initiatives Observation stays within 30 days of a targeting observation stay utilization. prior admission approximately doubled signifying either general or observation Additionally, the scope of this brief was from 2010 to 2014 for all five categories hours, or an observation HCPCS codes, G0378 or G0379. These claims were limited to observation stay utilization of admissions analyzed. Observation stays following any hospital admission identified in claims with a hospital out- measures. There are many other (i.e. hospital-wide admissions) in- patient type of bill code with a first digit measures of health care services utiliza- creased to 2.44 in 2014 from 1.23 in of 1 and a second digit of 3 (i.e. hospital tion and expenditures, which may pro- outpatient). vide additional insight into the utiliza- 2010. Observation stays following a tion patterns of the MA population. COPD admission increased the most The inpatient admissions, i.e. index ad- over the study period from 1.03 to 2.21. missions, used to identify 30-day all- The smallest increase over the study cause post hospitalization observation period, 1.91 to 3.41, was for observation stays were defined by the same criteria stays following AMI admissions. Howev- used to identify index admissions in our er, the rate of 30-day all-cause observa- previous study of readmissions.6 Eligible tion stays following an AMI admission outpatient observation stays included was the highest in every year. an observation stay for any condition Annual percentage changes in observa- within 30 days of an index admission tion stays following hospitalization are regardless of the reason for the observa- shown in Table 4. Consistent with the tion stay. Only the first observation stay trend in observation stays per member following a hospitalization was counted. shown in Table 1, the largest increases This approach is consistent with the approach of only counting the first inpa- www.healthcostinstitute.org 2 Endnotes: from: https://www.cms.gov/ 1. Health Care Cost Institute. Medicare Regulations-and-Guidance/Guidance/ Advantage Health Care Utilization— Transmittals/downloads/R1760CP.pdf Hospital Readmissions. November 6. Health Care Cost Institute. Medicare 2016. http:// Advantage Health Care Utilization— www.healthcostinstitute.org/issue- Hospital Readmissions. November brief-MA-readmissions. 2016. http:// 2. Centers for Medicare and Medicaid www.healthcostinstitute.org/issue- Services. Medicare Advantage Plans. brief-MA-readmissions. https://www.medicare.gov/sign-up- change-plans/medicare-health-plans/ medicare-advantage-plans/medicare- advantage-plans.html 3. Kaiser Family Foundation. Medicare Advantage Enrollment: Total Medicare Private Health Plan Enrollment, 1999- 2015. June, 29, 2015. http://kff.org/ medicare/fact-sheet/medicare- advantage/ 4. Centers for Medicare and Medicaid Services. Are You a Hospital Inpaitent or Outpatient?. https:// www.medicare.gov/Pubs/ pdf/11435.pdf 5. For example, in the Medicare FFS population an observation stay is cov- ered by Part B (Medical Insurance) ra- ther than Part A (Hospital Insurance). Additionally, coverage of skilled nurs- ing facilities in Medicare requires an inpatient hospital stay. Even if the ob- servation stay is more than one day it does not count toward the overnight stays required to SNF coverage. Derived from CMS Manual System: Pub 100-04 Medicare Claims Processing. June 23, 2009. CMS. Available for download Authors Copyright 2016 Eric Barrette and Katharine McGraves-Lloyd Health Care Cost Institute, Inc. Unless explicitly noted, the content of this report is licensed Contact under a Creative Commons Attribution Non-Commercial No Eric Barrette Derivatives 4.0 License ebarrette@healthcostinstitute.org 571-257-1584 Health Care Cost Institute, Inc. 1100 G Street NW, Suite 600 Washington, DC 20005 202-803-5200 Acknowledgements This HCCI research product was independently initiated by HCCI and is part of the HCCI research agenda. www.healthcostinstitute.org 3 Table 1: Observation stays per 100 Medicare Advantage members Year Observation stays per 100 members Annual Percent Change 2010 3.14 - 2011 3.35 6.9% 2012 3.49 3.7% 2013 4.72 35.8% 2014 5.71 21.0% Source: HCCI, 2016. Table 2: Length of observation stays Percent of observations stays Standard deviation of length Year Average length of stay with a length of stay >1 day of stay 2010 21.0% 1.50 1.03 2011 20.3% 1.55 1.05 2012 20.8% 1.55 1.06 2013 22.5% 1.55 0.98 2014 21.3% 1.60 1.13 Source: HCCI, 2016. Table 3: Thirty-day, all-cause observation stays per 100 Medicare Advantage inpatient admissions Year Hospital-wide AMI Heart Failure COPD Pneumonia 2010 1.23 1.91 1.33 1.03 0.95 2011 1.39 2.07 1.50 1.30 1.04 2012 1.50 2.22 1.60 1.43 1.13 2013 2.03 2.88 2.09 1.75 1.60 2014 2.44 3.41 2.57 2.21 1.91 Source: HCCI, 2016. Table 4: Annual percentage change in thirty-day, all cause observation stays per 100 Medicare Advantage inpatient admissions Year Hospital-wide AMI Heart Failure COPD Pneumonia 2010 - - - - - 2011 12.4% 8.2% 12.5% 25.6% 10.2% 2012 8.2% 7.5% 6.4% 10.2% 8.3% 2013 35.6% 29.8% 30.9% 22.2% 41.8% 2014 20.0% 18.3% 22.8% 26.8% 18.8% Source: HCCI, 2016. www.healthcostinstitute.org 4