Reprinted from THe Jounna or Panvenrrvé Mepicine . Vol. 6, No. &, September, 1£32 _ STUDIES OF THE DISTRIBUTION OF IMMUNITY TO YELLOW FEVER IN BRAZIL* I. POSTEPIDEMIC SURVEY OF Macf, Rio DE JANEIRO, BY COMPLE- MENT-FIXATION AND MONKEY-PROTECTION TESTS € FRED L. SOPER, M. FROBISHER, Jr., J. A. KERR anp N. C. DAVIS Received for publication April 21, 1932 Before the successful maintenance of the yellow fever virus in the laboratory and the development of the protection test with Macacus rhesus monkeys, which facilitates the postconvalescent diagnosis of yellow fever,’ individual suspected cases of this disease could be posi- tively diagnosed only after examination of tissues removed at autopsy. Conclusions regarding the distribution of the disease, therefore, de- pended on knowledge of the case-fatality rate and the factors which might cause such fatality rate to vary. But experience has shown that, in both epidemic and nonepidemic periods, even fatal cases of yellow fever often pass unnoted and that during epidemics deaths due to other causes are frequently attributed to yellow fever. The difficulties encountered in the diagnosis of nonfatal cases ate such that in the absence of epidemic conditions clinicians are unwilling to declare even moderately severe cases. The apparent casé-fatality rate varies, therefore, through errors in the clinical diagnosis of both fatal and nonfatal cases. . Furthermore, although the fatality of yellow fever is known to vary * The studies and observations herein reported were conducted with the support and under the auspices of the International Health Division of the Rockefeller Foundation, The authors are indebted to the directors of the official health services of the states of Sao Paulo and Rio de Janeiro for cooperation facilitating the studies. For assist- _ ance in the collection of specimens they are indebted to Drs. S. Ferreira Pinto, Mario Bio, A. Andrade, Waldemar Rocha, Frederico Motta, Carlos Del Negro and Domingos Bellizi. They are indebted to Dr. H. Aragao for timely assistance in preparing serum for transport to the laboratory. 1 Stokes, A.; Bauer, J. H.; and Hudson, N. P.: Amer. Jour. Trop. Med., 1928, 8: 103- 164 (March). 341 THE JOURNAL OF PREVENTIVE MEDICINE, VOL. 6, NO. 5 342 SOPER, FROBISHER, KERR AND DAVIS widely, the factors determining such variation have never been demon- strated. Carter? cites yellow fever mortality rates ranging from 4 to 25 per cent in extensive outbreaks, and up to 85 and even 98 per cent in certain “closed-group” epidemics. Variations in the virulence or dosage of the yellow fever virus might help to explain variations in the severity of individual cases or differences between one epidemic and another, but they cannot clarify variations in attack rate and severity in different groups living under identical conditions. Practically all authors are agreed that in outbreaks of yellow fever in endemic areas the highest attack and fatality rates are to be found among foreigners,’ especially those who have recently arrived. The disease would appear, from the descriptions given, to exemplify, par excellence, Cobbett’s‘ general statement: ‘‘Peoples are more resistant to the attack of the infectious diseases of their own countries and dis- tricts than to that of foreign diseases with which neither they nor their ancestors have been accustomed to come in contact.” The relative freedom of natives and long-time residents in endemic areas from the ravages of yellow fever has been generally attributed to a specific immunity acquired through mild attacks in early childhood or to a partial immunity built up from repeated hypothetical subin- fective doses of the virus. In regions having a large negro population native immunity has also been attributed to racial factors. But until recently no method for testing the distribution of immunity was available, and acquired immunity without diagnosed attack could not be distinguished from natural resistance to infection or from absence of exposure. Although it has long been recognized that recent arrivals from areas in which yellow fever is nonendemic form the group most highly sus- ceptible to the disease, the equally important observation that yellow fever in native populations of endemic areas may be a very elusive disease, even among white adults, has not been sufficiently emphasized. In a place where the disease is permanently endemic it has been al- * Carter, H. R.: In Byam & Archibald’s “The Practice of Medicine in the Tropics,” 1921-23, 2: 1228-1253. London (Frowde), 3 vols. * The term foreigners, as used in this article, refers to individuals born in areas not sub- ject to endemic or repeatedly epidemic yellow fever. , * Cobbett, L.: Tubercle, 1925, 6: 577-590 (Sept.). POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 343 most hopeless to expect diagnosis in natives of the place. The epi- demiologist engaged in tracing the distribution of yellow fever has therefore been forced largely to supplement with hypotheses the meager scientific information secured from such fatal cases as were followed by autopsy. How faulty such hypotheses, based for the most part on negative findings, may be in endemic areas with little movement of foreigners, is strikingly revealed by the Teappearance of yellow fever in the department of Santander, Colombia, in 1929 after an apparent absence from the country for five years and with no evidence pointing to recent reintroduction of the disease from with- out.’ Past experience in Brazil has also repeatedly demonstrated that the absence of declared suspected cases of yellow fever from an area does not necessarily mean that the disease itself is absent. Pe- tiods of many months without confirmed cases in all Brazil have fre- quently occurred, only to be followed by more or less extensive outbreaks in the absence, apparently, of any reintroduction from outside sources. Announcement of the infection of M. rhesus monkeys with yellow fever virus (March, 1928) and publication of details of the protection test occurred just as authorities in Brazil were beginning to discover isolated cases of yellow fever, which presaged its wide, known distri- bution in the country in 1928, 1929, 1930 and 1931. From April, 1927, to March, 1928, no cases of the disease had been confirmed in all Brazil. Early in March, 1928, however, a positive diagnosis was made at autopsy in Estancia, Sergipe (figure 1). In April a fatal case occurred at Timbatba, Pernambuco. And on May 16 an autopsy was performed in Rio de Janeiro on a soldier dead from the disease. No connection was ever established between any two of these three widely separated cases. Further Cases are not known to have occurred in either Estancia or Timbatiba during many succeeding weeks of observation when no control measures were being applied. In Rio de Janeiro, epidemiological studies? failed to reveal the original case ® Pefia Chavarria, A.; Serpa, R.; and Bevier, G.: Jour. Prev, Med., 1930, 4: 417-457 (Nov.), ® Confirmation during this period was Possible only by autopsy, as other laboratory methods were not yet available, * Barreto, J.: Archivos de Hygiene, 1929, 3: 93-193 (No. 1, May). 344 SOPER, FROBISHER, KERR AND DAVIS or even the original focus of the disease in the federal district. The facts collected demonstrated the existence of at least four distinct foci in Rio de Janeiro in the first half of May (Villa Militar, Sado Chris- tovdo, Safide and Catumby), but no relationship was ever established between any two of these foci. AMAZONAS nee, STATE OF RIO® DE JANEIRO ‘Atlantec Ocean @l- ESTANCIA @ 2+ TIMBACBA @ 3- RIO DE JANEIRO @ 4. MAGE @ S&S. PIRACICABA @ 6. STO. ALEIXO Fic. 1 It is impossible to state how long yellow fever had been present in the federal district of Rio de Janeiro before being diagnosed in May, 1928, although a superficial study of general and specific death rates indicates that the disease was probably not an important cause of death in Rio de Janeiro in January, February and March, but may POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 345 have been so as early as the first week in April. The impossibility of digging out the past ramifications of the disease in Rio was explained on the basis of the supposed occurrence of a large proportion of mild ' cases at the beginning of epidemics. After the recognition of the disease, great anxiety was felt for the safety of the city because of the large nonimmune element in the population of the federal district, composed of foreigners, Brazilians from nonendemic regions, and na- tives of the city born since the end of Oswaldo Cruz’s memorable campaign in 1908. Later events justified this anxiety regarding foreigners and Brazilians from nonendemic regions, but, contrary to expectation, there was not an appreciable number of cases among natives of Rio under 20 years of age. Of 125 officially recognized cases in 1928, ninety-seven, or 77.6 -£2.5 per cent, occurred in for- eigners, although according to the latest census (1920) only 20.8 per cent of the population are foreigners. Furthermore, 18 of 28 cases (64.3 +6.2 per cent) occurring in Brazilians were in persons who had resided in Rio de Janeiro less than five years. This failure of Rio de Janeiro to produce in two decades a highly susceptible native popula- tion was at first attributed to the fact that the disease predominated in sections of the city, SaGde and Catumby, largely filled with for- eigners. While it is true that, with the spread of the disease later to other parts of the city, a significant increase was noted in the percent- age of cases occurring among Brazilians (table 1), foreigners continued to suffer disproportionately to their numbers, and some observers suggested that yellow fever may have been more or less constantly present, previous to 1928, in mild or at least undiagnosed form, thus maintaining a high degree of immunity in the native population. But this hypothesis is not reasonable, because Rio de Janeiro has had a constantly shifting foreign population and has, moreover, long been the Mecca of the Portuguese immigrant; this highly susceptible material would undoubtedly have given rise to explosive outbreaks in the past had the virus been present in sufficient quantity to immunize the local population. . There was continued difficulty throughout 1928 and 1929 in tracing the source of individual infections, although the fatality of diagnosed cases was very high, 59.6 1.2 per cent. For many of the cases which occurred the only apparent explanation was the assumption 346 that a considerable number of undiagnosed yellow present even during the height of the epidemi profession and the public were both “yellow-fever-minde Distribution of cases of yellow fever treated between 1892 and 1899, and cases diagnosed TABLE 1* SOPER, FROBISHER, KERR AND DAVIS fever infections were c when the medical d . 2? in the Hospital Sdo Sebastido, Rio de Janeiro, in Rio de Janeiro in 1928, in 1929 and in 1928 and 1929 HOSPITAL, 1892-99 | _ RIO, 1928 RIO, 1929 RIO, 1928-29 yq Total cases.........0. cece eens 14,546 125 613 738 Males........0ceee eee e eee eees 13,042 100 461 561 Females...........-0eeeceenee 1,502 25 152 177 Per cent males..........-2+5-+- 89.7 +0.2) 80.0 +2.4) 75.2 41.2 76.0 +:1.1 Brazilians. ........202ee scene 701 28 203 231 Foreigners........002 see sees 13,845 97 335 432 Unclassified.........0.2 eee eens 75 75 Per cent foreigners classified. .... 95.2 40.1) 77.6 42.5] 62.3 41.4 65.2 +4:1.2 Whites... 0.0.2 cc cece eee eee 14,309 120 485 605 Mulattoes...........ecee eee ees 107 43 Blacks.......... ees eceeeeeees 126 5 4 53 Yellow... ..ccccnecece ee ereees 1 Unclassified...... reer 80 80° ' Per cent whites classified........ 08.4 +0.1} 96.0 +:1.2] 91.0 +0.8) 91.9 +0.7 In Rio less than 3 years........ 92 268 360 In Rio more than 3 years....... 33 65 98 Unclassified...........-eeeeeee 280 280 Per cent classified in Rio less than 3 years......... 62s eee 73.6 42.6 80.5 +1.5| 78.6 21.3 Age 20 to 40 years..........4+5 9 ,083 83 308 391 Other agea.........e: eee eens 5,463 42 220 262 Unclassified...........eeeeeeee 85 85 Per cent age 20-40 classified... . . 62.4 40.2) 66.4 +2.81 58.3 £1.4) 59.4 1.3 * Data from A. da S. Mello (Brasil Medico, 1928, 42: 1079-87, Sept.), J. Barreto (Archivos de Hygiene, 1081 and 1113, Sept. 27 and Oct. 4), and S. Vianna (in C, Fraga’s Brasil,” 1930, pp. 221-229). 1929, 3: 93-193, No. 1, May), C. Fraga (Brasil Medico, 1930, 44: *& Febre Amarella no The distribution of cases of yellow fever occurring during the Rio de Janeiro epidemic of 1928 and 1929 is compared in table 1 with the POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 347 distribution of cases treated in the isolation hospital in Rio from 1892 to 1899. These figures are probably not strictly comparable, as the 1892-99 figures refer only to hospitalized cases and would therefore probably tend to include a larger percentage of cases among foreigners than among Brazilians, as well as a larger percentage of cases amongmen than among women. Even so, an analysis of table 18 shows that the absence of yellow fever from the city for two decades was not entirely without effect. Significant differences are found in the percentages of cases among Brazilians, among females, and among whites in the two periods studied. No appreciable difference is found however in the percentage of total cases occurring between the ages of 20 and 40 years. Mello® calls attention to the fact that the observed distribution of diagnosed cases in the Rio de Janeiro epidemic of 1928 corresponded closely to the observed distribution of cases during the period when yellow fever was endemic in the city. He hypothecates a nonspecific immunity to infection in natives and old-timers which he attributes, following Gordon,!* to long-continued exposure to the bites of A. aegypti, with the production of a local tissue immunity in the skin. Mello points out the failure of the investigations of the health authorities to reveal evidence of any significant number of mild abor- ' tive cases in Rio to explain the selectivity of the disease and its unex- pected distribution, and refuses to accept specific acquired immunity due to invisible infection as the only explanation of the observed facts. While admitting that mild undiagnosable infections with the virus of yellow fever do occur, he insists on the validity of the ancient belief in an immunity of acclimatization, which can be progressively lost, by either natives or old-time residents in endemic areas, through con- tinued residence in cold nonendemic climates. He adduces this belief as an added argument for an unstable nonspecific immunity in persons long resident in endemic'regions, who have never suffered from attacks of the disease. Barreto’ notes the hypothesis that lowered basal metabolism ob- served in residents of tropical countries may contribute to greater * In considering table 1 the predominating influence of the classification “foreigners” on the other classifications presented must not be forgotten. ® Mello, A. da S.: Brasil Medico, 1928, 42: 1079-1087 (Sept.). 1@ Gordon, R. M.: Ann. Trop. Med. & Parasitol., 1922, 16: 229-234 (Oct.). 348 SOPER, FROBISHER, KERR AND DAVIS resistance to infection, but concludes (p. 93) that “the relative scarcity of cases in children depends really rather on the difficulties of the clinical diagnosis than on the paucity of patients.” In considering the attack and fatality rates of yellow fever in the recent outbreak in Rio de Janeiro it is interesting to know that Torres Homem and other early writers on the 1849-50 epidemic in that city, which occurred after an apparent absence of yellow fever from Brazil during many decades, reported that foreigners were far more suscep- tible to the disease than natives. A like observation, made at Bahia under similar conditions the same year, occurs in the report of the President of the Province of Bahia for 1850:" ‘The number of natives attacked between October, 1849, and the end of June, 1850, exceeded one hundred thousand, but among them the disease was much milder than among the foreigners. The mortality among the former was 3 to 4 per cent, but among the latter it was more than 30 per cent.” One of us (FLS) observed at close range the Rio de Janeiro yellow fever outbreak of 1928-29 and early came to the conclusion that in- formation of value might be secured if it were possible to carry out protection tests on a group of native-born residents living in known infected zones of the city. As reasonable certainty existed that yellow fever had not been endemic in Rio de Janeiro on a large scale since 1908, natives less than 20 years of age found to have a specific immu- nity would most probably have a recently acquired immunity. Plans were made for a series of protection tests, and in January, 1929, work was begun with Dr. Hugo Muench in Braz de Pinna, a suburb of Rio de Janeiro. The press of other services, however, made it necessary to discontinue this study. Before the project could be undertaken anew, yellow fever was reported in the nearby town of Magé, in the State of Rio de Janeiro, where, during February, March, and April, a small epidemic of 22 cases with 13 deaths occurred.2 Magé offered a much more concrete problem than did Braz de Pinna and was chosen as the site of the present study in the belief that the results might be Pereira, A. P.: “Conferencia sobre peste bubonica e febre amarella,” 1910, Bahia (Associacio Commercial da Bahia), 91 pp. 12 Lintz, A., and Parreiras, D.: ‘Notas e estudos epidemiologicos sobre a febre amarella (1928-1930). Nictheroy, 1930. POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 349 applied with profit to the interpretation of the observed distribution of yellow fever following the reintroduction of the virus into Rio de Janeiro, an old endemic center, after an apparent absence of twenty years. After the work in Magé had been begun (June, 1929), the first report of promising results with the complement-fixation test for yellow fever was published.* In view of the low cost of this test, it was decided to examine a much larger number of persons in Magé than would have been possible with the more expensive protection test, and provision was made ‘for a series of control tests with sera from a noninfected center. Piracicaba, Séo Paulo (figure 1), lying outside the endemic yellow fever zone, was selected for the collection of sera (1930) for the control series. Later (1931) an opportunity occurred to secure a control series of sera from another known postepidemic focus of yellow fever at Santo Aleixo, a small town near Magé. The data secured from these investigations are the basis of the Present report, which gives: (1) comparative results of complement- fixation tests in Magé and Santo Aleixo, both lying in a previously _ endemic area and both the seats of recent epidemics, and of similar tests in Piracicaba, Sao Paulo, believed to have been free of yellow fever for many years; (2) comparative results of complement-fixa- tion tests in Magé, four, sixteen and twenty-two months after a known epidemic of yellow fever; (3) comparative results of complement-fixa- tion and M. rhesus-protection tests in Magé; (4) an analysis of yellow fever immunity distribution in the population of Magé in 1929, as indicated by complement-fixation and W. rhesus-protection tests. YELLOW FEVER IN MAGE AND SANTO ALEIXO Magé, the county seat of Magé county, is situated at the junction of two railways, by one of which Rio de Janeiro may be reached in about an hour and a half (figure 1). Contact with Rio is also main- tained by small freight and fishing sailboats. The life of Magé de- pends largely on the operation of the local cotton mill, which is re- ported to have some five hundred people in its employ. The popula- tion, roughly estimated at 2500 for 1929, is, on the'whole, a remark- "Frobisher, M., Jr.: Proc. Soc, Exper. Biol, & Med., 1929, 26: 846-848 (June). 350 SOPER, FROBISHER, KERR AND DAVIS TABLE 2 . Results of unofficial census of Magé, January, 1925* AGE (YEARS) — TOTAL RACE. OR COLOR cane Orie [soma] woo] 9 | | BP Payer Indian... 6... cece eee eee 0 0 0 2 1 1 0 4, 0.2 Negro... ....c cee e eee e cere eee 12 | 36 | 48 | 61 | 39 [138 | 32 366] 16.2 Mulatto............ecen eee 20 | 65 | 80 | 76 | 58 |234 | 52 585{ 25.9 White......... cece eee 36 j153 |178 [154 [151 j507 |121 | 1,300) 57.7 Total... .. cece eee eens 68 |254 1306 {293 j249 (880 |205 | 2,255 Per cent........0..0ee cease 3.0) 11.2] 13.6] 13.0} 11.0) 39.0} 9.1 * From Boyd, M. F., et al.: Amer. Jour. Hyg., Monograph Series 5, 1926 (May). ably stable one, with distribution by race and age as shown in table 2. That part of the population studied for the present report showed the following distribution by age groups, corresponding closely to that of table 2: AGE (YEARS) NUMBER PER CENT Under 4 209 13.8 5 to > 189 12.4 10to14 | 212 14.0 15 to 19 178 11.7 20 to 49 594 39.0 50 and over 139 9.1 Total... eee eee 1,521 100.0 Santo Aleixo and the contiguous village of Andorinhas, forming one nucleus of population, comprise about five hundred houses lying largely between the two cotton mills which justify their existence. This center, which will be referred to as Santo Aleixo in this report, lies within the county of Magé, some eighteen kilometers inland from the town of Magé, with which it is connected by a small steam rail line operated for the convenience of the cotton mills. The history of yellow fever in Magé and Santo Aleixo is believed to have depended in the past on the history of the disease in Rio de Janeiro. These towns, like Rio, are not believed to have harbored yellow fever between 1908 and 1928. POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 351 Although Magé is close to, and in daily contact with, Rio de Janeiro, there were no rumors of yellow fever in the former town before Decem- ber, 1928, and no proof of its presence there until February 13, 1929, when the first positive autopsy findings were made. Control measures, including antilarval work, medical inspection, isolation of the sick and fumigation, were instituted on March 21 by the Yellow Fever (7, Non~fatal Case A- Andorinhas Fatal Casc sk see Aleixo ‘o se se 3 a gz 20} x i s s S 8 uM . ” § & 161 3 2 a & 2 a 3 8 » z R Official Cases Yellow Fever » 8 Collection Sera, Sto. Alexio, and Srd Colleation Sera, A fea an TER SGTi aa eel FAH Ta[iaaa Fic. 2, Disramuzron, ny Monvas, or Orrictarzry REcocnizep Cases or YELLow Fever In Macé anp Sanvo ALeIxo, wire Rererence 70 COLLECTION oF -Bioop Saweres yor Txsts Service of the State of Rio de Janeiro. The last known case in Magé occurred April 8, 1929. The distribution of officially recognized cases in Magé and Santo Aleixo is shown in figures 2 and 3. The official - duration of the epidemic was nine weeks. The last cases occurred eighteen days after control measures were begun, and the fatality of cases recognized before the health authorities arrived was much higher than that of later cases. Both these facts may be interpreted to mean that many more cases probably occurred in Magé than were diagnosed. 352 SOPER, FROBISHER, KERR AND DAVIS The two physicians practicing in Magé differed regarding the extent of the yellow fever outbreak. One believed that many cases had occurred in the town during January, February and March, with comparatively few deaths, while the other, although admitting that yellow fever had been present in the town and that he had seen two cases with suspicious symptoms, believed that most of the fever cases seen by him had been due to malaria. The monthly number of medical calls and consultations among families of employees of the cotton mill made by one physician, be- Official Notification Yellow Fever <———— Non-fatal Case Patal Case 6 MAGE °o STO. ALEIXO ANDORINHAS | m n Official Cases °o » xt a » » 2 » p a é a Ot Epidemic Week Measured from First Local Official Case (June 1929 Sto. Alexio Case Not Included) Fic. 3. DistrIBuTION, BY EpipeMIC WEEKS, OF OFFICIALLY RECOGNIZED CASES OF YELLow FEVER In Macé aNnD SANTO ALEIXO ginning with October, 1928, were: 185; 193; 234 (December); 224 (January—yellow fever probably present in December and January); 295; 544; 303 (yellow fever present in these months of February, March, April); 162 (May—yellow fever probably absent); 131 (June —yellow fever surely absent, present study begun). While the great increase in the number of calls noted in February, March, and April covers the period during which yellow fever is known to have been present in Magé, it also corresponds to the season of greatest prevalence of malaria. It is fortunate, from the standpoint of the present study, that Magé is one of the four places which had POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 353 been chosen in 1922 by Boyd™ for comparative studies on the epidemi- ology of malaria in the State of Rio de Janeiro. Boyd’s conclusions after three years’ study were (p. 111): “We have the villages of Sant’ Anna, Porto das Caixas, and Itamby with a very high incidence of endemic malaria, the fluctuations of which among the fixed population do not vary widely from year to year, while, on the other hand, we have in Magé an avea of low endemicity, subject from time to time to epidemic exacerbations.” A moderate epidemic of malaria had occurred in Magé in 1922 before Boyd’s first blood index had been taken and control measures begun. Boyd reports blood indices for the town in the months of July and August 1922-24 as follows: YEAR EXAMINED POSITIVE INDEX 1922 1,009 375 37.2 1923 1,352 139 "10.3 1924 1,208 59 4.9 Had Magé suffered a widespread epidemic of malaria in February, March and April, 1929, the blood index based on slides examined in June of the same year should have been high. Such was not the case ; only 32 of 932 blood films were positive. In spite of the fact that the 1929 index was taken one month nearer to the epidemic season than were Boyd’s indices, this index, 3.4 -+ 0.4, is the lowest recorded for Magé. The distribution of residence of the persons whose blood was found positive for malaria in June, 1929, shows no similarity to the distribution of cases of suspected yellow fever or of persons whose blood sera gave @ positive complement-fixation reaction. Furthermore, none of the persons whose blood was positive for malaria are among those listed as having had cases of suspected yellow fever. The authors believe that malaria may be eliminated as a cause of the large number of medical calls registered by the factory physician. It is unfortunate, so far as the present study is concerned, that deaths from a considerable district outside the town of Magé are registered with those of the town itself. The most recent official census (1920) gives the total population of the registry district as 6216, “ Boyd, M. F., et al.: Amer. Jour. Hyg., Monograph Series 5, 1926 (May). 354 SOPER, FROBISHER, KERR AND DAVIS which is more than double that of the town itself. However, even though rates cannot be ascertained, the raw data from the registry district are sufficiently interesting to be presented here, although no definite conclusions from them are justified. As will be seen from table 3 and figure 4, more deaths were registered in March, 1929, than in any other month during the ten-year period, 1920-29. The number of deaths registered during each month of 1929 was in excess of thenine-year average for that month; and in each TABLE 3 Registered deaths in Magé registration district by months, 1920 to 1929 tlie [Han 3 gs |HE8| | Eas eV latil glelsielg leis leis |i| © | Ree |g [ance January.......... 12} 6 sl 7/14) 31 8} 9] 8] 72] 8.0] 2.2) 9) 0.46 February......... ail 14] st of 71 4} 4) 8| 12] 77; 8.6] 2.2) 12) 1.6 March... 15} 11] 9| 12] 10] 9] 7} 10] 11) 94) 10.4} 1.4) 22) 8.1 April. ......... 00. ta] s| 4) of 10] S| 6] 10) 8] 71] 7.9} 2.0) 12| 2.08 May....c..ceeees | gl ataal gt 7} 7] 10] 5] 70] 7.8] 1.8] 10) 1.20 June.......... eee s| of st si 6} of 7 4| 13] 63] 7.0] 41.7) 8) 0.57 July... a 19] io} 6 4) 8} 7|/ 6} 9 80} 8.9] 43.9] 14) 1.30 August........... 3} of gs} 7 1) S| 3] 7} 14) 51] 5.7] 21.9) 12) 3.27 September........ 101 3} of a sl al 4) 6 7] Si] 5.7] 2.0) 11) 2.65 October........... al 6 tol 6 6 4] S| 8} 8] 55) 6.1) 4.5] 8] 1.25 November........ s| 4} 12} s| 31 5) 8 8} 7| 60] 6.7] 1.7] 14] 4.21 December......... s| 7| al gi 5] 7] 3) 10] 11) 60) 6.7) 41.7] 15) 4.88 Annual total....| 91] 98| 97] 99| 79| 68| 66| 96/110|804| 89.3/:10.7/147) 5.40 of five months (March, August, September, November and December) the number of deaths registered exceeded the previous maximum for that month. In March, August, November and December, the 1929 registered deaths exceeded the upper limits of the nine-year averages plus three times the probable errors of the distributions from 1920 to 1928 for these months. Deaths for March showed an in- crease over the nine-year average of more than eight times the probable error of the March distribution, and total deaths for the year 1929 showed an increase over the average of almost six times the probable POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 355 error of the annual distribution. The drop to normal in June was followed by an increase during the rest of the year, with the result that the mortality in the district continued much above normal even 60 40 30 2 AY A / ‘ 21° Sond, ‘ Lo TSS nn ALS ~N ~ mh TS, pone” 2 1 a MN NY AT m 69 N ey, a H a \ N/ LN M . 29 NZ a 6 4 8 2 —~— Mean wonthly deaths, 1920.28 ~— Maximum monthly deaths, 1920-28 come Monthly deaths, 1929 ---r- Mean monthly deaths plus 5 tines Probable error of the distribution by months, 1920.28 IZ 2 1 1 r “ 1 1 1 . Jan. Feb. Mar. Apr. May June July Auge Sept Oct. Nov, Dace Fic. 4, Dzatas mv Mack RecrsrraTion AREA BY Montus (Seumocanrmtc Scare) after the inauguration of mosquito control and the disappearance of yellow fever from the town itself. That this persistent high death rate in the district may have been associated with the persistence of yellow fever in the rural areas is suggested by the occurrence in Febru- 356 SOPER, FROBISHER, KERR AND DAVIS ary, March, April, May and June, 1930, of cases of yellow fever in Santo Aleixo, Andorinhas and nearby rural areas within the county of Magé”? without evidence of reintroduction of the disease. It is interesting to note that the peak of the known cases of yellow fever and the peak of sick calls, discussed above, fall in the same month as the maximum general mortality. Registered deaths by age groups for the period 1920 to 1929, in- clusive, are shown in table 4 and figure 5. Increases over the average for the previous nine-year period are noted for all age groups in 1929, such increases being more than three times the probable errors of the TABLE 4 Registered deaths in Magé registration district by age groups, 1920 to 1929 - Sak Saees8 & wy ao Be RO z |g, |#28| | fesse AGE S an gem Bo fame “ om | BEG ON < gm I Sa |] #2 often ea elaislSslalislisis{/sl] é aS |828/a/8,288e8 SlSiSlSiSpS;SlSls] 2 | Bs | gaa) S| SREP aS Under 17) 24) 24) 35, 24) 23) 30) 41] 35) 253 | 28.1 | 44.8) 30, 0.39 1 to 4 40) 21) 16) 22) 19) 15) 16; 17) 29) 195 | 21.7 | 45.2) 39) 3.36 5 to9 8} 7) 6 3) 6 1) 3) 3) 8) 645 5.0} +1.6) 7) 1.23 10 to 14 2) 5} 2) 1) OF 1) OF 2) 4) 17 1.9} 41.1) 7 4.75 15 to 19 3} 5S} 6 1) ty OF 3} 1) 4) 24 2.7 | +1.3) 6) 2.54 20 to 49 9| 17) 27/ 23] 17] 19] 9] 17] 16) 154 | 17.1 | 43.7] 28 2.95 50 and over | 12) 19} 16) 14) 12) 9) 5) 15) 14; 116 | 12.9 | +2.6) 30) 6.60 Total....... 91} 98) 97| 99] 79] 68) 66) 961110} 804 | 89.3 |+10.7/147; 5.40 distributions for ages 1 to 4, 10 to 14, 20 to 49, and over 50. For all ages above 10 years, the total deaths in 1929 were equal to, or greater than, the maxima for these groups during the previous nine-year period. Santo Aleixo, in spite of its proximity to and dependence on Magé, was not known to harbor the virus of yellow fever until the first week in April, 1930, one year after the occurrence of the last registered case in Magé. Santo Aleixo, having no resident physician, depends on lay observers for certification of cause of death in those persons not seen by the cotton-mill company’s physician on his scheduled visits to the community. The Yellow Fever Service for the State of Rio de POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 357 Janeiro, alarmed at receiving from Santo Aleixo certificates in which the cause of death was given by laymen as “yellow-typhoid” and “hemorthagic-typhus,” ordered autopsy on all persons dying in the 50, 40 Se IN 2 py fs q oS j PON i | » 10 i m8 if | H 8 a) l < a 7 = i } e 6 1 cl | 5 \ a | . | 3 | 2 Lf Vv ~=—= ean deaths, 1920~28 —— Maximum deaths, 1920-28 -—— Deaths, 1929 woonee Joan deaths, 1920-28, plus 3 times probable error of the distribution by age groups, 1920.28 1 | O-1 i-f bod 10-14 Is - ie 2-0 0 Age Groups iio. 5. BzavHs iv Mact REGISTRATION AREA BY AGE GRours community. On April 5 it received notification of a death occurring on March 29, which autopsy had shown to be due to yellow fever." Retrospective investigation made at this time caused a diagnosis of yellow fever to be made for seven previous fatalities occurring without 358 SOPER, FROBISHER, KERR AND DAVIS medical attention between June 1, 1929, and March 25, 1930, with the following causes of death certified by laymen: malaria (2 cases), hemorrhagic-typhus (2 cases), pulmonary congestion, gastro-enteritis, and yellow-typhoid. After the discovery of the disease in Santo Aleixo in April, eleven additional cases were observed in this center-— three in Santo Aleixo proper and eight in the contiguous Andorinhas. The distribution of the eleven officially recognized cases in Santo Aleixo and the eight in Andorinhas is shown in figures 2 and 3. If Santo Aleixo and Andorinhas are considered as a unit, the official epidemic period is eighteen weeks, but if considered separately the period for Santo Aleixo is ten weeks and that for Andorinhas eight weeks, COLLECTION OF MATERIAL Magé. In June, 1929, a house-to-house canvass was made in the central part of Magé, and epidemiological data were collected regard- ing 1521 people living in 348 houses. During this survey 36 suspected cases of yellow fever were added to the official list of 22. Some of these were relatively mild cases and would not have suggested yellow fever under other circumstances. The distribution of the officially recognized cases and the additional suspected cases is roughly the same. At this same time (June, 1929) thick blood smears were taken from 932 residents of the study area to obtain the malaria index,!* and in July and August, 1929, blood samples of 10 to 30 cc. for complement- fixation tests were secured from 676 persons, 2 to 81 years of age, living in the area. In July, 1930, second blood samples were secured from 88 persons who had furnished samples in 1929; and in January, 1931, additional samples were taken from 34 persons who had furnished specimens on one or both of the previous occasions (tables 6 and 7). Santo Aleixo. In January, 1931, blood samples of 10 to 30 cc. were secured from 70 persons living in Santo Aleixo and Andorinhas, most of whom were less than 20 years of age. Piracicaba, State of Séo Paulo. As a control on the results of com- plement-fixation tests in Magé, 120 blood samples were secured in September, 1930, from persons of all ages living in Piracicaba, a busy town of some 25,000 residents in the interior of the State of Sio Paulo. 8 Results discussed above (page 353). POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 359 Piracicaba is situated about 140 miles from Santos at an altitude of more than 500 meters above sea level, and yellow fever is not known to have been present there during the past thirty-five years. METHODS All samples for tests were taken in vacuum venules, which are indis- pensable where many specimens are to be collected in homes and shops. The sera, after separation, were drawn off aseptically, placed in sterile tubes or ampules, and sent to the laboratory for as early examination as possible. The complement-fixation test was made, using an antigen prepared from the liver of monkeys infected with yellow fever. The technique - of this test has been adequately described elsewhere." The protection test consisted in injecting 3 or 4 cc. of serum intra- peritoneally into M. rhesus monkeys, one serum to each monkey. Within five minutes each animal received an infective dose of citrated blood of an animal in the early febrile stage of yellow fever, usually infected with the Asibi strain of virus. A control animal which re- ceived no serum was given a similar dose of infective blood at the same time as each group of test animals. Rectal temperatures of all ani- mals were taken twice daily. Autopsies were performed on animals that died, and the gross and microscopic pathological findings were recorded. Interpretation of tests on monkeys surviving febrile attacks was difficult, but after weighing evidence for and against protection, partial protection, and nonspecific reactions, most of these tests could be recorded as positive or negative. ; COMPLEMENT-FIXATION TESTS Comparative results in Magé and Santo Aleixo and in Piracicaba. The results of complement-fixation tests in Magé in July, 1929, in Santo Aleixo in January, 1931, and in Piracicaba in September, 1930, are presented in table 5. The percentages positive in Magé and Santo Aleixo are equal, wheregs the difference in percentages positive in Magé County and in Piracicaba is sufficiently large in proportion to its probable error to make it highly significant. Results four, sixteen and twenty-two months after the Magé epidemic. 8 Frobisher, M., Jr.: Jour. Prev. Med., 1931, 5: 65-78 (Jan.). 360 SOPER, FROBISHER, KERR AND DAVIS The results of complement-fixation tests on the sera of 54 persons in Magé four and sixteen months, and four and twenty-two months, after an epidemic of yellow fever are given in tables 6 and 7. A large number of sera were tested, but only those giving entirely satisfactory readings in both the tests are included. A statistical analysis of these tables shows that, although there is considerable tendency for early positive complement-fixation tests to become negative, there is, nevertheless, a high degree of correlation between the results of suc- cessive tests. Resulis of parallel complement-fixation and protection: tests. | The results of protection tests with sera from Magé (1929) giving clearly TABLE 5 . Summary of complement-fixation tests, Magé, Santo Aleixo, and Piracicaba speci- | SATIS LOCALITY BLOOD DRAWN ee ORT | ros awe TESTED | rons* Magé........0.0cccee eens July, 1929 "370 | 250t | 105 | 42.0 42.1 Santo Aleixo............... January, 1931 60 50 21 | 42:024.7 Total, Magé County.............0.00.0000- | 430 | 300 | 126 | 42.041.9 Piracicaba............2005: September, 1930 120 101 3 3.0 41.1 Difference in per cent positive, Magé County and Piracicaba........... 39.0 42.2 * That is, not anticomplementary. t These sera had been stored in ampules for some time, which may account for the fact that only 68 per cent of the reactions were satisfactory. positive or clearly negative complement-fixation readings are shown in table 8. It will be seen that positive complement-fixation results were generally confirmed by positive protection tests, whereas an appreciable number of sera giving negative complement-fixation reactions protected M. rhesus against yellow fever virus. In spite of the tendency toward an excess of positive protection tests over positive complement-fixation tests, statistical analysis of table 8 shows a high degree of correlation between complement-fixation and protection test results. In table 9 are presented comparative results of complement-fzation and protection tests with the sera of 18 persons in Magé giving a POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 361 history of an illness in 1929 suspected to have been yellow fever. Of these, 3 were anticomplementary and could not be read, 12 were positive, one was negative, and two were doubtful. Protection tests on M. rhesus with the sera of 9 of these 18 persons confirmed 7 positive TABLE 6 ‘ Resulis of successive complement-fization tests with sera of the same persons taken at M agé Sour and sixteen months* after an epidemic of yellow fever 1930 PER CENT TOTAL CHANGE From Postrive | NEGATIVE 1929 RESULTS 1929: Positive.......... 0c ccc e cence cence 11 15 26 57.7 +6.5 Negative...........-.- cece ecceceees 1 27 28 3.6 42.4 Total. ...... ccc cece cece c nce ner eees 12 42 54 25.9 47.8 * No cases of yellow fever are known to have occurred in the town of Magé after April 8, 1929, although cases were confirmed for the county in June, 1929, and from February to June, 1930. Data from Lintz and Parreiras: “Notas e estudos epidemiologicos sobre a febre amarella (1928-1930). Nictheroy, 1930. TABLE 7 Comparative results of complement-fixation tests with sera of same persons taken at Mi ‘agé four and twenty-two months after an epidemic of yellow fever (1929 and 1931) 1931 PER CENT TOTAL CHANGE FROM P Necativs 1929 exsuLrs 1929: Positive... 0... cc ccc ccc ccecceeenees 10* 3t 13 23.1 +8.6 Nogative....0.0. 22... cc ccc ceceeeaees 2t 9|| 11 18.2 +8.7 Total... cece eee ee eee cease 12 12 24 4.2 49.8 * Two of six of these tested in 1930 were negative, t One of these tested in 1930 was negative, $ Two of these tested in 1930 were both negative. I Three of these tested in 1930 were all negative. and one negative complement-fixation test and indicated that one doubtful complement-fixation test should have been negative. In other words, nine persons with recent history of suspected yellow fever gave closely corresponding results by the two test methods. Of 362 SOPER, FROBISHER, KERR AND DAVIS TABLE 8 . . Results of protection tests with sera giving readable complement-fixation reactions, Magé COMPLEMENT-FIXATION TEST PER CENT CONFIRMED BY COMPLEMENT- Positive Negative Total FIXATION TEST Protection tests: . Positive........0:eeeee cere ree 28 14 42 *| 66.7 44.9 Negative. .......-eeeeeeeeeree 3 31 34 91.2 +3.3 Total. ....cceceee cere eeeeeeee 31 45 76: . Per cent confirmed by protection . Te 00.3 £3.6 | 68.9 44.7 TABLE 9 Results of complement-fization and protection tests in 18 persons affected in Magé in 1929 with an illness suspected to have been yellow fever case son eaTNIrY Date oF COMPLEMENT-FIXATION' raorzcrion 1929 | 1930 15 52 Portugal March AC not done not done 17 26 Italy March + not done + 194 24 Syria March + not done + 195 21 Brazil March + not done + 266 32 Portugal March + + not done 267 29 Portugal March + + not done 268 6 Portugal March + + , not done 655 48 Portugal March + not done + 657 18 Brazil March + not done - + ” 658 16 Brazil March - not done - 1181 35° Syria April AC not done not done 1187 12 Brazil February AC not done. | not done 1188 15 Brazil April ? not done - 1235 39 Brazil March + not done | + 1238t 13 Brazil April + not donet | not done 1239 8 Brazil March + not done . + 1241 8 Brazil March -? not done -{ not done 1248 38 Brazil March + not done j not done * AC = Anticomplementary. + Complement-fixation test January, 1931, positive. the persons whose sera gave positive complement-fixation reactions in 1929, three were reéxamined in 1930 and one in 1931, and all again gave positive reactions. POSTEPIDEMIC IMMUNITY TO YELLOW FEVER | 363 YELLOW FEVER IMMUNITY DISTRIBUTION IN MAGE IN 1929 AS INDICATED, BY COMPLEMENT-FIXATION AND M. RHESUS-PROTECTION TESTS _ The distribution of positive and negative complement-fixation reac- tions, according to the part of Magé in which the donor of the blood sample resided, does not, to casual inspection, show a marked differ- ence. However, if the results of tests are grouped with respect to occurrence of officially recognized cases of yellow fever, there appears to be a definite relationship between known infection of a group of con- tiguous houses and a high percentage of positive complement-fixation reactions, especially in the more closely built-up central part of the town. . TABLE 10 Results of complement fixation tests in Magé, according to place of residence COMPLEMENT-FIXATION PER CENT RESIDENCE EXAMINED POSETIVE Positive Negative In central blocks with cases 6, 7, 8, 9, 10, 12, 13, 14, 15, 16, 17, 19, 20, 21, and 22*.} 118 66 52 56.0 +3.4 Outside above listed blocks..........--- 132 39 93 29.6 242.7 Total. occ. cece cece eee e eee eee renee 250 105 145 42.0 +2.1 In house with case......-.--e--eseeeeee 18 10 8 56 +:7.9 In neighboring house.........--++++++-- 43 20 23 47 +5.1 Others. .....0. cece eee e rere c cee eerees 189 75 114 40 +:2.4 Total. ..... cece cece e cece eee eee eeee 250 105 145 40 +2.1 * Cases numbered in order of date of onset. In those central blocks in which occurred fifteen known cases (table 10) a significantly higher percentage of complement-fixation tests was posi- tive than in other parts of Magé, although the earliest known cases and possibly the longest period of infection occurred in these other areas. The differences in percentages positive in the. groups “in house with case,” “in neighboring house,” and “others” (table 10) are very sug- gestive, but an analysis of the relationship of these differences to their probable errors indicates that such differences cannot be considered statistically significant. The classification of results according to the color of the donor of the blood specimen (table 11) reveals that there is a significant difference 364 SOPER, FROBISHER, KERR AND DAVIS in percentages positive in whites and blacks. There is no racial segre- gation in Magé, although there may be some tendency for blacks to live away from the center of population. To minimize this possible effect of residence, all persons tested were classified as to color and to proximity to known cases of yellow fever, and whites still seem to give a higher percentage of positive reactions than do either mulattoes or negroes, although the numbers in each group are too small to give sig- nificance to the differences noted. Protection tests on M. rhesus with sera from Magé, although few in number, seemingly confirm the TABLE 11 Results of complement-fixation tests in Magé, according to color and residence RESIDENCE PER CENT POSITIVE® In . tone peigh- Others | Total COLOR suspect | ‘house Inhouse | In neigh- 7 > > 2 with : boring Others Total é A2la| a é ZzlalZ . White......... 7| 6 | 10] 8) 44) 52] 61) 66/54 +9.3/56 +7.9146 +3.4/48 42.9 Mulatto....... 3/1] 9} 11] 21) 38) 33] 50/75 414.645 + 7.5136 +4.2/40 3.6 Black......... Qj 1 1) 4] 10] 24) 11] 29 0 20 +12.0/29 +5.2/28 +5.0 Total seeeeee 10) 8 | 20) 23) 75/114/105/145/56 +7.9147 +5.1140 42.4142 42.1 * No statistically significant differences are shown above when probable errors of differ- ences are taken into consideration, except between total blacks and total whites, with difference of 20 +:5.8 per cent. results of the complement-fixation test, protection being afforded by sera from twenty (60.6 +:5.7 per cent) of thirty-three whites, four- teen (56 +6.7 per cent) of twenty-five mulattoes, and only eight (44 +7.9 per cent) of eighteen blacks tested. This apparent difference in racial susceptibility is in accord with clinical observation, but an inspection of probable errors shows that these results are not statis- tically significant. Although the official list of yellow fever cases occurring in “Magé i in 1929 shows only two cases in foreigners, at least five others occurred in foreign families whose children born in Brazil are listed as Brazilian. Likewise; among the 36 additional suspected cases discovered in the POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 365 present study, 14 occurred in foreigners or in children of foreign parents. This number is out of proportion to the number of foreigners in Magé and emphasizes the greater susceptibility of those of foreign blood to clinical yellow fever. TABLE 12 Results of complement-fixation and protection tests in males and females, Magé COMPLEMENT-FIXATION PROTECTION Positive! Nee | Percent [positine| Newer | Pesstee Males........c cece cence cece rene 49 80 |3842.9) 21 17 «| 55.3 5.4 Females. .......-.2:ceeeeeeeeeee 56 65 |4643.3| 21 17 | $5.3 45.4 Total. 2... eee eee eee 105 | 145 (4242.1) 42 34 155.3 43.8 TABLE 13 Results of complement-fixation and protection tests by age groups, Magé COMPLEMENT-FIXATION PROTECTION TEST AGE (YEARS) . Positive |Negative | Total Penitive Positive Nega- Total Peritive Under 4 2 3 § 140 419.1 0 0 0 §to9 11 12 23 «| 48 410.4 3 2 5 |60 +19.1 10 to 14 13 13 26 |50 +6.5 5 2 7 \71 s11.5 15 to 19 13 30 43 130 +4.7 5 14 19 |26 =+6.8 20 to 29 27 32 59 |46 +4.5; 11 5 16 | 68.7 +7.8 30 to 39 22 22 44 |50 +6.1 11 4 15 |73.5 +7.7 40 to 49 11 14 25 |44 +6.7 4 3 7 | 57.1 414.9 50 and over 6 19 28 |24 +5.7 3 4 7 142.92414.9 | Total...... 105 145 | 250 |42 42.1] 42 34 | 76 155.3 43.8 Under 19 | 39 58 | 97 |40 43.4] 13 | 18 | 31 [41.9 +6.0 20 and over | 66 87 153 |/43 42.7] 29 16 45 |64.5 +4.8 Total...... 105 145 | 250 |42 42.1] 42 34 76 155.3 43.8 Males and females show the same percentage of positive protection tests (table 12). The difference between percentages in positive com- plement-fixation tests is less than two times its probable error and hence not highly significant. Classification in accordance with whether the donor was above or 366 SOPER, FROBISHER, KERR AND DAVIS | below 20 years of age (table 13), a grouping based on the probable date of the last previous outbreak of yellow fever in Magé, shows that the results of complement-fixation tests are practically uniform in the two groups, whereas the results of protection tests jump from 41.9 +6.0 per cent positive in the younger group, to 64.5 +4.8 per cent positive in the groupsaver 20 years of age. This difference (22.6 +:7.7 per cent) is almost three times its probable error. The difference be- tween the percentage positive by complement-fixation tests and that positive by protection tests in the group over 20 years of age (21.5 +5.5 per cent) is significant. When only complement-fixation tests are considered, a significant difference (20 +6.1 per cent) is found - between the percentage positive in the age group under 49 years and that positive in the group 50 years and over. The results of complemient-fixation tests by age groups for Santo Aleixo are as follows: AGE (YEARS) EXAMINED POSITIVE Under4 0 0 5 to9 11 3 10 to 14 22 8 15 to 19 13 7 20 to 29 2 1 30 to 39 1 1 40 to 49 1 1 50 and over Total. 00... eee ee 50 21 Per cent positive.........:esseeeeee seers deceeeeecsaeedeeeeeseaee 42.0 44.7 DISCUSSION In addition to possible technical errors in performing the comple- ment-fixation and protection tests, there appear to be other factors influencing the results of these tests. As the complement-fixing and protection bodies, although formed in response to the same stimulus, are apparently not identical, factors causing an increase or decrease in one element may not affect the other. Errors in either protection or complement-fixation tests, then, should affect the degree of correlation found between results of the two tests. POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 367 M. rhesus-protection tests may be falsely positive because of normally resistant animals. These are estimated at “not more than 6 per cent” by Beeuwkes, Bauer and Mahaffy," who have undoubtedly had the largest experience with the test. The same authors, who feel the necessity of using duplicate animals for each test, report that in 274 duplicate tests there were 38 instances (13.9 +1.4 per cent) in which only one of two test ariimals died. Evidence of the difficulty of interpreting protection tests was also encountered in our tests. Twenty-five of 76 protection tests (33 +:3.6 per cent) were not entirely clear cut and were interpreted on the basis of previous experience with the test. Falsely negative M. rhesus-protection tests do occur, and consistent results are not always obtained with serum from the same individual. Patient N. C. D., who had yellow fever in April, 1929 (Burke and Davis),"* has failed to react clinically to repeated feedings of numerous batches of infected mosquitoes over a period of eighteen months. Serum of N. C. D. has given the following results in monkey protection tests during 1929: May and July, prevented death but not fever; ‘September, no protection; October, full protection; November, severe fever and recovery; December, full protection. Beeuwkes, Bauer and Mahaffy"’ also cite a case in which the serum from a person known to have had yellow fever failed to protect one of a pair of M. rhesus monkeys, after having repeatedly protected M. rhesus in previous tests. It is not clear whether these inconsistencies are due (1) to uncertainties of the test itself, because of variation in suscepti- bility of individual monkeys and of variation in activity of virus, either of which might cause variable results in working with serum of low ti- ter of protection bodies, or (2) to variation in the titer of protection bodies in the serum of the same individual, possibly dependent on opportunity for reinfection. By analogy with results obtained in n monkeys, complement-fixing bodies may be expected to appear in the sera of human yellow fever patients several days after the onset of the disease, or later, and to increase gradually to a maximum, to be followed in many ¢ases by a 1” Beeuwkes, H.; Bauer, J. H.; and Mahaffy, A. F.: Amer. Jour. Trop. Med., 1930, 10: 305-333 (Sept). 1* Burke, A. W., and Davis, N.C.: Amer. Jour. Trop. Med., 1930, 10: 419-426 (Nov.). 368 SOPER, FROBISHER, KERR AND DAVIS decline.!*. ®. 21 It seems probable, however, that a considerable num- ber of persons will maintain a supply of complement-fixing bodies over a long period of time. Hudson has reported complement-fixation tests on the sera of five persons who had contracted yellow fever in the laboratory. One of these persons failed to develop complement-fixing bodies during eight months of observation, while another showed persistence of complement-fixing bodies twenty-one months after infection. No estimate can be made at the present time of the per- centage of persons who will fail to develop complement-fixing bodies. _ Macacus rhesus monkeys in which the complement-fixing titer of the blood has fallen very low some months after infection have, when reinoculated with the virus of yellow fever, shown an increase of com- plement-fixing titer to the previous maximum titer or beyond, without exhibiting other signs of reinfection (G. E. Davis,® Frobisher). Similar reactions may be expected to occur in human beings. Whether such repeated exposures to the virus are capable of producing a per- manently higher titer of complement-fixing bodies in the blood is _ unknown. , It should be noted that the first three of the above-mentioned factors leading to a disparity between the results of complement-fixation and protection tests (namely, resistant monkeys, delayed appearance or absence of complement-fixing properties, and nonpermanence of complement-fixing properties) all tend to give a greater number of positive protection tests than of positive complement-fixation tests. The effect of possible variations in titer is more difficult to foretell. In spite of the several factors of variation, the results of the Magé complement-fixation tests show a high degree of correlation with the Magé protection tests (tables 8 and 9), and the results of the Magé post- epidemic complement-fixation tests are widely different from those obtained with sera from Piracicaba, where yellow fever is not endemic. The results of postepidemic complement-fixation tests in Magé also correspond closely, as to percentage positive, with the results of similar tests at Santo Aleixo. In Magé, 90 +3.6 per cent of positive comple- ment-fixation tests and only 69 -£4.6 per cent of negative complement- 29 Frobisher, M., Jr.: Amer. Jour. Hyg., 1931, 13: 586-613 (March). 20 Davis, G. E.: Amer. Jour. Hyg., 1931, 13: 79-128 (Jan.). 21 Monteiro, J. L., and Travassos, J.: Memorias do Instituto Butantan (Sao Paulo), 1930, 5: 173-191. % Hudson, N. P.: Proc. Soc. Exper. Biol. & Med., 1931, 28: 937-939. POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 369 fixation tests were confirmed by protection tests. Analyzing the figures of table 8 from the opposite point of view, it is noted that 67 +4.9 per cent of positive protection tests and 91 +3.3 per cent of negative protection tests were confirmed by the results of complement- fixation tests. In other words, 90 +-3.6 per cent of positive comple- ment-fixation tests were confirmed by protection tests and 91 +3.3 per cent of negative protection tests were confirmed by complement- fixation tests, but only 69 +4.6 per cent of negative complement-fixa- tion tests were confirmed by protection tests and only 67 +4.9 per cent of positive protection tests were confirmed by complement- fixation tests. These results are important and indicate that the complement-fixing properties in the blood of convalescents are less constantly produced and less permanent than are the elements respon- sible for protection. This suggestion is in part borne out by the results of repeated complement-fixation tests on the same individuals, which indicate that there may be a rapid decline of positive reactions in the first years after an epidemic (tables 6 and 7). In Magé the per- centages of positive complement-fixation tests in the age groups below and above 20 years were equal, whereas a much higher percent- age of protection tests was positive in the age group over 20 years. If the presence of complement-fixing bodies always represents a recent infection, one would be forced to conclude that the older age group, in spite of residence during childhood in an endemic area, had produced as high a percentage of infections as had the younger age group. However, if complement-fixing bodies are in most cases only present for a short period of time and are produced anew in the presence of the virus without production of disease—which, as has been shown above, is what occurs in M. rhesus—an equal percentage of positive comple- ment-fixation results should be obtained in all age groups following reintroduction of the virus after a long absence. The spread between positive complement-fixation tests and positive protection tests in the older age group, then, may represent roughly the number of immunes in Magé who were not subjected to the virus anew in 1929. An at- tempt was made to calculate the probable percentage of immunity in the age group above 20 years, previous to the most recent introduc- tion of the disease, using the formula: 2+r(100—2) =y 370 SOPER, FROBISHER, KERR AND DAVIS in which r is the apparent attack rate of the present epidemic as indicated by protection tests in the age group ugder 20 years, and by complement-fixation tests in all groups, and y is Me final postepidemic percentage of immunity indicated by the protection test for the age group over 20 years. The solution of the equation: z+ .4(100 — 2) = 64.5 indicates that approximately 41 per cent of the age group over 20 years were immune before the last outbreak. . G. E. Davis,?° although reporting excellent results with the com- plement-fixation test in monkeys, concludes on the basis of a small series of largely negative results that the complement-fixation test can be of no assistance in determining endemic areas of yellow fever. Although there are at present difficulties in the interpretation of indi- vidual complement-faation results in endemic areas, comparison of the results in Magé, Santo Aleixo and Piracicaba (table 5) indicates that the complement-fixation test may have a definite place in out- lining endemic, or at least recently epidemic, areas. Results obtained by Monteiro and Travassos" ‘support this belief. These authors report thirty-three (49 +-4.2 per cent) positive complement-fixation reactions—twenty (29.8 +-3.8 per cent) strongly positive and an addi- tional thirteen weekly positive—in tests on sera of 67 persons in Bahia, an old endemic center of yellow fever, but no positive reactions in tests on sera of 20 Lithuanian and Japanese immigrants from whom blood samples were secured within 24 hours of arrival in Brazil. The Piracicaba data, taken in conjunction with the high percentage of positive complement-fixation reactions in Magé, which were con- firmed by positive protection tests, indicate that very few falsely positive complement-fixation tests will be read. Results are sufh- ciently striking to warrant making further comparative studies of pro- tection and complement-fixation tests in areas where the history of yellow fever is known, before discarding the complement-fixation test or assuming that its results are not important in the study of endemic areas. m The percentage of positive protection tests in Magé (55.3 +3.8) is not statistically different from that (68.0 +6.3) reported by Beeuwkes, Bauer and Mahaffy” from Ife in Nigeria after an epidemic POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 371 of yellow fever. Endemic areas may be expected to show a greater divergence between positive complement-fixation and positive protec- tion tests than do areas recently epidemic. From a study of material herewith presented Magé is believed to have been subject, during the first four months of 1929, to a rather extensive outbreak of yellow fever, characterized by a large number of mild immunizing infections and by very few severe and fatal cases. The results of complement-fixation and protection tests, taken in con- junction with the low malaria index found, are interpreted as indicat- ing that the number of medical calls in January, February, March and April, when yellow fever is known to have been present, was probably due to so-called abortive attacks of yellow fever. The analysis of immunity distribution in Magé, as shown by the complement-fixation- test in the present study, failed to reveal statistically significant differ- ences between groups variously classified by age, sex, and race. Even if the excess of total deaths for the epidemic period in the Magé district be attributed to yellow fever, the evidence still indicates that the fatality rate in Magé was actually very low, although the official fatality rate was 59 +7.0 (13 deaths in 22 cases). As the strain of virus responsible for the Magé epidemic was most probably imported from Rio de Janeiro, it is interesting to note that the official fatality rate in Rio during the same year was 59 +1.3. Since virus, climate and previous history of yellow fever are similar for Magé and Rio de Janeiro, may it not be possible that a large number of abortive infections were occurring in the latter city also? The official fatality rate in Santo Aleixo is believed, in the light of known facts and the results of the complement-fixation test, to be greatly exaggerated because of incomplete registry of cases. Many clinical observers have in the past expressed a belief in the occurrence of mild cases of yellow fever in endemic areas, but the con- ditions under which these take place have never been established. The tendency of epidemiologists has been to ignore the possibility of their occurrence except in infants and in negroes, and these two groups have largely been held responsible for all that is difficult to explain in the epidemiology of yellow fever in endemic areas. Practically all observers agree that the foreigner, newly arrived from an area where yellow fever is not endemic, runs the greatest risk of 372 SOPER, FROBISHER, KERR AND DAVIS infection and suffers most severely from the disease. On no other . factor believed to influence susceptibility, however, is there unanimity of observation and opinion. Many authors assert that infants and small children are relatively immune to the disease, whereas Hansen* reports a high fatality rate in this group. The negro race is generally believed to suffer less from yellow fever than do other races, and yet Boyce,* who observed the Barbados epidemic of 1909, calls attention (p. 238) to the large percentage of cases occurring in blacks, with a high death rate. The same author comments on the outbreak of yellow fever in Freetown, West Africa, in 1884, in which cases of the disease, but no deaths, were reported among the native-born blacks. Boyce attributes the mildness of the disease in blacks in Freetown to previous attacks of the disease, a position untenable in the light of recent laboratory experience with the protection test. Ramsey” cites the Asibi case (Stokes, Bauer and Hudson)! as a proof that yellow fever in the black may be a very mild disease, and inter- prets data from Senegal on the basis of race. However, in Brazil the yellow fever virus has several times been isolated from whites whose attacks of the disease were so mild that a positive diagnosis could not have been made on clinical grounds. The first two reported infections of M. rhesus with South American virus (Aragao)?” were produced with blood from immigrants who had had mild cases of the disease; and the S.R. strain of virus reported upon by N. C. Davis and Burke’* was obtained from a Spanish woman whose infection was very light. Dr. P. J. Crawford succeeded (October, 1930) in establishing the yellow fever virus from a native white child (Morlin), aged 12 years, who had a three-day fever and headache, without vomiting, jaundice, albumin- uria or other symptoms suggestive of yellow fever. 23 Hansen, H.: Amer. Jour. Trop. Med., 1929, 9: 233-239 (July). Tt must be remembered that the so-called “negro race” is composed of many widely variant racial strains which probably are not uniform in resistance to yellow fever. Saw- yer and Lloyd (Jour. Exper. Med., 1931, 54: 533-555—Oct.) have demonstrated wide variations in resistance to yellow fever encephalitis in various strains of white mice. oe Sir R. W.: “Yellow Fever and Its Prevention,” 1911, London (J. Murray), % Ramsey, G. H.: Amer. Jour. Hyg., 1931, 13: 129-163 (Jan.). 2? Aragio, H. de B.: Mem. Inst. Oswaldo Cruz (Rio de Janeiro), Suppl. No. 2, 1928, pp. 23-46 (Oct. 15). %8 Davis, N. C., and Burke, A. W.: Jour. Exper. Med., 1929, 49: 975-984 (June). POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 373 The present study indicates that, during an epidemic in a previously endemic region, a large part of the native population may develop a specific immunity to the virus of yellow fever without presenting the classical symptoms of the disease. This unsuspected immunization occurs in various age groups and is not limited to the black race. No definite indication is given, however, as to the factors which deter- mine the mild or negative symptoms which accompany this immun- ization. . : The epidemiology of yellow fever has had no more devoted observer and student than the late Dr. H. R. Carter’ who, after presenting widely varying fatality rates for different outbreaks (p. 1244), asks: “Ts the fever milder among the natives of countries in which it is endemic or of frequent occurrence—t.e., is there an hereditary immunity? Ihave thought so from what I have seen in purely white communities of the United States on the Gulf of Mexico, but it is nowise proven.” The influence of individual heredity in determining resistance to disease is again being emphasized by the scientific world after several decades during which the causative organism alone had been con- sidered. Jennings? goes so far as to state: “It is probable that there is no disease whatever, acute or chronic, infectious or noninfectious, whose occurrence is not influenced by the nature of the individual’s genetic constitution.” An hereditary immunity, or rather resistance, to the effects of yellow fever infection, whether due to the direct transmission of protection from parents to children, as suggested by Carter’s observations, or to the selective action of the virus with elimination of a large part of the ‘more susceptible genetic combinations from the race, would explain a surprising number of observed facts and would harmonize many of the divergent opinions held by students of the disease. First, it would constitute a most logical explanation for the one point on which all observers are agreed, namely, that yellow fever is most severe among foreigners. Negroes have not in the past moved in appreciable num- bers from areas where yellow fever does not occur to zones where it is endemic. The disease in negroes has usually been observed, then, in natives of endemic, or repeatedly epidemic areas and might be * Jennings, H. S.: “The Biological Basis of Human Nature,” 1930, p. 149. New York (W. W. Norton & Co.). THE JOURNAL OF PREVENTIVE MEDICINE, VOL. 6, NO. 6 374 SOPER, FROBISHER, KERR AND DAVIS expected, on the hypothesis of hereditary resistance, to show a low fatality rate. Assuming an inherited resistance to yellow fever, the Barbados (1909) and the Freetown (1884) experiences cited above can be reconciled on the basis of limited previous outbreaks with little inherited resistance or incomplete elimination of susceptible strains in Barbados, and of extensive previous outbreaks in Freetown. An inherited resistance might be expected to be most effective in early childhood, and the number of cases observed in infants of newly- arrived foreigners must be very limited; both these factors would help to explain the impression that yellow fever is mild in children. On the other hand, a population highly resistant because of the elimination of the more susceptible gene combinations might be expected, since such human population, to show: occasional typically severe and fatal cases even in the early age groups, elimination can never be complete as _inasmuch as, on this hypothesis, resis ce is based on genetic constitu: tion and not on gradually diminishing protection bodies inherited from the immediate parents. Such seyere cases in infants are known to occur in highly endemic areas where both parents are of native. stock and apparently possess long-standing immunity to yellow fever. That the apparent mildness of the di: se in children is not entirely due to age is evident from the reports on fatality of yellow fever in children in nonendemic regions. Hansen® reports 19 deaths in 23. cases in children under two years of ag (a fatality rate of 82.6 +5.3), and 14 deaths in 27 cases in children between two and five years of age (a fatality rate of 52.0 +6.5) durin occurring in Peru after the disease years. an epidemic of yellow fever d been absent for twenty-five The observation that the mortality in old people is very high may be due in part to the fact that very few natives of endemic regions live to an old age without having had the disease. Hence the classification of “old age” here will often carry th hidden classification of “for eigner.” Either of the above suggested hereditary mechanisms would admir- ably explain the fact that when yellow fever was reintroduced into Rio de Janeiro in 1928 and 1929, after lapse of two decades, there did not occur among the natives of the cases or of deaths. On a larger scale, ity a large number of reported the hypothesis of inherited resistance to yellow fever virus and the consequent occurrence of many POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 375 subclinical? cases in the natives of endemic areas, irrespective of race ° dage, explains observed facts much better than does the assumption “mild attacks limited to negroes and infants. However, an affirma- tive answer to Carter’s query need not carry the implication that his corollary of an inherited resistance necessarily holds, or that, even if present, such hereditary resistance is the most important factor in the production of mildinfections. The hypothesis of hereditary resistance fails to explain the low incidence of clinical yellow fever in foreigners more than five and less than twenty years in Brazil in the recent Rio outbreak. Neither does it explain the belief in the loss of immunity by natives and old-time residents living for some years in nonendemic areas. . More and more epidemiologists are coming to realize the possible importance of nonspecific factors—such as chemical composition of diet, vitamin content of diet, variation in basal metabolic rate, expo- sure to sunlight, variations in temperature and humidity—in the production of seasonal and group variations in attack and fatality rates of infectious disease, although the mechanism whereby such factors intervene in the infectious process has not been determined. For example, Aycock*! calls attention to a number of seasonal variations in physiological processes and shows that such seasonal variations con- form to the same law as that which determines the occurrence of polio- myelitis. Without attempting to link resistance to the virus of polio- myclitis with any one of these processes, he suggests that there may be some variation in the physiological activity of the body which influ- ences resistance to the disease. This author has coined the word autarcests to distinguish this nonspecific “protective power against disease which exists in the body by reason of a normal or balanced physiological activity,’ from specific immunity, the result of actual invasion of the body by the disease-producing agent. Variations in temperature and humidity influence the prevalence of yellow fever by their direct action on the insect host of the virus, but it has also been suggested that climatic factors and exposure to the uninfected insect host over long periods of time may result in a sum total of acclimatization capable of giving the population of certain ; * Not necessarily without symptoms, but without symptoms recognizable as yellow ever, - "Aycock, W. L.: Jour. Prev. Med., 1929, 3: 245-278 (May). 376 SOPER, FROBISHER, KERR AND DAVIS regions an increased relative resistance to the dangerous results of infection with the yellow fever virus. The possible influence of variations in metabolism and of antecedent dietary habits on the course of an infection which may produce such pronounced destruction of liver tissue as does yellow fever should not be overlooked in view of the known protective action of certain food substances against the liver destruction caused by such agents as chloroform (N. C. Davis)* and carbon tetrachloride (N. C. Davis,* Minot and Cutler). In this connection it is interesting to note that some of the highest fatality rates reported-for yellow fever have been observed in ‘‘closed groups,” such as companies of soldiers and crews of ships, which, in addition to fulfilling the requirement of “newly- arrived foreigners,” had also been subjected to uniform and probably well-below-optimum diets. SUMMARY AND CONCLUSIONS 1, Complement-fixation tests on postepidemic blood samples taken from 300 persons in Magé and Santo Aleixo and on 101 samples from Piracicaba, lying outside the yellow fever zone, show a highly signifi-_ cant difference between the percentages of positive reactions in the postepidemic samples and in the samples from the nonendemic area. The conclusion is drawn that the complement-fixation test may be use- ful in field studies of yellow fever, especially in outlining areas recently epidemic. 2. A small number of comparative results of complement-fixation . tests on the same individuals in Magé at intervals of four, sixteen and twenty-two months after an epidemic of yellow fever indicate that there is a tendency, especially in the first year after infection, for the titer of the complement-fixing bodies in the blood to decrease. 3. Comparative results of complement-fixation and M. rhesus- protection tests with 76 sera indicate that positive complement- fixation tests will generally be confirmed by positive protection tests and that negative protection tests will usually be confirmed by nega- tive complement-fixation tests. However, an appreciable number of # Davis, N. C.: Arch. Int. Med., 1919, 23: 612-635 (May). *3 Davis, N. C.: Jour. Med. Res., 1924, 44: 601-614 (Sept.). . “ Minot, A. S., and Cutler, J. T.: Jour. Clin. Investigation, 1928, 6: 369-402 (Dec.).- POSTEPIDEMIC IMMUNITY TO YELLOW FEVER 377 sera giving protection showed negative results with the complement- fixation technique. These findings indicate that complement-fixing bodies demonstrable by present methods are less constantly produced. than are protection bodies, and are less permanent. 4, The combined results of complement-fixation and protection tests given in this report are interpreted as indicating that, at the time of known outbreaks of yellow fever in Magé and Santo Aleixo, rela- tively large percentages of the local populations were acquiring a specific immunity to the yellow fever virus without apparent attacks of the disease. A statistical analysis of the available data failed to show that such invisible acquisition of active immunity was related in any way to age, sex or race. No data are presented to indicate what factors are responsible for this ability of the population of these two towns, lying in previously endemic areas, to acquire active immunity with low fatality rates. 5. The possibility of low fatality rates with many unnoted infections in natives of endemic or previously endemic yellow fever areas, without regard to age, sex, or color, must be taken into consideration in any future attempt to study the distribution of the yellow fever virus.