Effectiveness of BCG vaccination against tuberculous meningitis: a case-control study in São Paulo, Brazil.

Effectiveness of BCG vaccination against tuberculous meningitis: a case-control study in Sao Paulo, Brazil V. Wunsch Filho,1 E.A. de Castilho,2 L.C. Rodrigues,3 & S.R.A. Huttly4 A case-control study was carried out in the Metropolitan Region of Sio Paulo, Brazil, to determine the protection against tuberculous meningitis conferred by BCG vaccination to children aged less than 5 years. The BCG vaccination coverage in the study area was about 88%. A total of 72 tuberculous meningitis patients were studied as well as 505 neighbourhood and 81 hospital controls. Analysis of the data using a conditional logistic regression for matched case-control studies indicated that the efficacy of BCG was similar for both groups of controls, that for neighbourhood controls (84.5%) being slightly greater than that for hospital controls (80.2%). No significant interactions were found between vaccination status and sex, age, or socioeconomic status. Introduction At the beginning of the last decade of the 20th century, tuberculosis still presents a public health challenge, particularly in developing countries. In Brazil the death rate from tuberculosis has dropped systematically since the introduction of specific chemotherapy. However, the current death rate of 5.9 per 100 000 per year from tuberculosis indicates that, apart from intestinal infections and pneumonia, it is still the infectious disease that causes most deaths in the country (14). According to official figures, the incidence of tuberculosis has shown a tendency to rise in Brazil (11). The annual risk of infection is, however, not precisely known because the policy of mass and indiscriminate vaccination with BCG has not facilitated such estimates. To reduce the global problem of tuberculosis, international health bodies have designated the identification and treatment of cases and vaccination with BCG as the principal components of control pro- grammes (25). Intradermal injection of BCG is con- sidered to be the best method of immunizing against the disease. Nevertheless, the effectiveness of BCG has been placed under doubt since several controlled trials reported contradictory results, with efficacies Department of Epidemiology, Faculdade de Saude Publica, Universidade de Sao Paulo, Av. Dr. Arnaldo, 715, Sao Paulo, 01255, Brazil. Requests for reprints should be sent to this address. 2 Fundac&o Oswaldo Cruz, Rio de Janeiro, Brazil. 3London School of Hygiene and Tropical medicine, London, England. 4 Department of Social Medicine, Faculdade de Medicina, Univer- sidade Federal de Pelotas, Pelotas, Brazil. Reprint No. 5047 that ranged from zero to 76% (2-5, 7-9, 12, 17, 18, 24). More recently, the case-control approach has been advocated for the evaluation of the effectiveness of BCG vaccination (21, 22); however, such studies have also reported a wide range of efficacies (13, 19, 23). Notwithstanding doubts about the effectiveness of BCG on the chain of transmission of tuberculosis, and therefore on the incidence of and death rate from the disease, the vaccine may still be useful if it gives protection against severe infantile forms of tuber- culosis, although such forms are not contagious (1, 15, 16). The results of a study in the United Kingdom in 1950 already indicated that BCG vaccine offered such protection (4); however, the results of a survey conducted in Chingleput, India, were unclear on this matter (1, 12). In Brazil, tuberculous meningitis is a notifiable disease. For all cases of the disease a surveillance form is completed, covering data from clinical and laboratory examinations and an epidemiological his- tory (6). Despite the difficulties in diagnosing this type of tuberculosis and the lack of confidence in the data collection in countries with socioeconomic characteristics such as Brazil, the seriousness of the disease probably results in all cases being notified. The hypothesis that BCG offers greater protec- tion against tuberculous meningitis than against pul- monary tuberculosis (4) is consistent with evidence from two Brazilian states with different schedules for BCG vaccination. For example, in Rio Grande do Sul, where children aged 7 years or older are vaccinated with BCG, the incidence of pulmonary tuberculosis in 1982 was about 12.5 times greater than that of tuberculous meningitis. In contrast, in Sao Paulo, where children receive BCG vaccine dur- ing the first year of life, the incidence of pulmonary Bulletin of the World Health Organization, 66 (1): 69-74 (1990) © World Health Organization 1990 69 V. Wunsch Fliho et al. tuberculosis to that of tuberculous meningitis was 52:1; regional differences may, however, confound this apparent relationship. BCG vaccine coverage varies markedly in dif- ferent regions of Brazil. Vaccine coverage in Sao Paulo city is reported to be very high, exceeding 100% according to official estimates; a survey con- ducted in 1982-3, however, estimated that the coverage was 88% for children less than 18 months of age.a In view of these facts and of the necessity to better define the role of BCG in tuberculosis control programmes, we carried out a case-control study in the Metropolitan Region of Sao Paulo to evaluate the effectiveness of BCG against tuberculous menin- gitis. Our findings are reported here. Materials and methods Study area The study was conducted in the Metropolitan Region of Sao Paulo (MRSP), which comprises 37 muni- cipalities within an area of 8053 km2. The population of MRSP recorded in the 1980 census was 12.5 million, which corresponded to 50.3% of that of the State of Sao Paulo and 10.6% of the entire Brazilian population. A wide range of living standards prevail in the study region, with some areas having only very basic environmental and social amenities. MRSP is very urbanized with only a few rural areas in the outskirts of the city (20).b Tuberculous meningitis Between 1979 and 1983, about 150 cases of tuber- culous meningitis per annum were notified in the State of Sao Paulo, and of these a third were among children under 5 years of age (incidence, approx- imately 1.7 per 100 000 per annum). The majority of cases of the disease notified in the State of Sao Paulo are admitted to two hospitals: Emilio Ribas or Man- daqui. ' [Investigation of vaccination coverage in the municipality of Sio Paulo]. Centre for Health Information. Government of the State of Sao Paulo. Unpublished document, 1982 (in Portuguese). b [Health care in the Metropolitan Region of Sio Paulo]. Paper presented at the WHO/PAHO Regional Meeting of the Technical Consultation on Primary Health Care and Development Services in Urban Areas and Large Cities, Washington, DC, 20 November 1981. PAHO unpublished document (in Portuguese). BCG vaccination Routine BCG vaccination (Moreau-Rio de Janeiro strain)c is compulsory in Brazil as part of a programme established by the Ministry of Health. It is recommended that the vaccine be given to children, without a previous tuberculin test, between birth and the end of their first year of life. Although BCG is widely used the protection afforded by it has never been assessed in Brazil. Selection of cases All notified cases of tuberculous meningitis which were admitted to the Emilio Ribas or Mandaqui Hospital from 1 January 1981 to 31 December 1983 that involved patients who were born after 1978 (coverage with intradermal BCG became high in Brazil only from 1979) were ascertained. Because of the difficulties in diagnosing tuberculous meningitis unequivocally, criteria were defined based on clinical and epidemiological findings, as well as on the results of bacilloscopy, culture of cerebrospinal fluid (CSF), and necropsy. All cases selected were residents of MRSP. Data on cases were collected in two question- naires: one based on the hospital records and the other on household interviews to clarify and com- plement the hospital record data, particularly in relation to the BCG vaccination status of the chil- dren. The vaccination status of children was deter- mined from their vaccination cards and, if possible, by the presence of vaccination scars. Cases were classified as BCG-positive only if their date of vaccin- ation preceded that of diagnosis of tuberculous men- ingitis. In situations where a case died, the child concerned was still included in the study and a household visit was made. The mother was then questioned about the dead child's BCG vaccination status and was asked to produce the vaccination card. If the mother had lost the child's card, health centre archives were searched. The mother's word was accepted only if she affirmed that her child had never been vaccinated, and such children were clas- sified as unvaccinated. Cases were excluded from the analysis if they could not be located or information could not be obtained about their vaccination status. Selection of controls Neighbourhood controls. Cases and controls were matched by home area and socioeconomic stratum. In order to obtain a minimum of four suitable controls per case, eight potential controls were c Fundag&o Ataulpho de Paiva, Rio de Janeiro, Brazil. 70 WHO Bulletin OMS. Vol 68 1990. Effectiveness of BCG vaccination against tuberculous meningitis sought from children in the neighbourhood of each case. The mother of the index child was asked to nominate two children of neighbours, each of whose mothers nominated two more. This process was continued until eight controls were identified. The only requisite was that the children nominated should have been born after 1978. Information on controls was collected during home visits, using a very similar questionnaire to that used for cases. To determine the BCG vaccination status of the neigh- bourhood controls, the same procedure was used as for cases, and only those children whose vaccination status was known with certainty were accepted. Chil- dren who had a past or present history of tuberculosis were rejected as controls. Hospital controls. In order to detect any biases that might have been introduced by the neighbourhood controls, a second series of controls was selected from patients in Emilio Ribas Hospital. In order to ensure that there was one suitable hospital control for each case, attempts were made to identify three potential controls per case. Hospital controls who were suspected to have had tuberculosis were excluded. Also, children with diseases which could have been prevented by vaccin- ation, e.g., measles or diphtheria, were also excluded, because it was considered likely that those not vac- cinated against these diseases would have had less chance of having received BCG vaccine. The hospital controls were selected from patients admitted with meningitis caused by Streptococcus pneumoniae or Haemophilus influenzae and who also satisfied the matching requirements of sex, home area, date of birth, and date of admission to hospital +6 months with respect to that of the index case. Two questionnaires were used for hospital con- trols, one to obtain hospital records data and the other, which was similar to that for the neighbour- hood controls, for the home visit. The BCG vaccina- tion status of the hospital controls was determined in the same way as that of the cases and neighbourhood controls. Sample size and estimation of vaccine efficacy With the number of cases that were expected to be found during the study period and the level of BCG coverage, it was concluded that vaccine efficacy above 50% could be detected at the 5% level of significance at a power of 80%. Vaccine efficacy was estimated from the relation- ship 1-RR (21), where RR is the relative risk of tuberculous meningitis among the vaccinated com- pared to the unvaccinated children, estimated from the odds ratio. The odds ratio was calculated by conditional logistic regression analysis for matched case-control studies, using the EGRET software package (10). Results Characteristics of the study population During the study period, a total of 474 cases of tuberculous meningitis were notified in Sao Paulo State, 196 (41.4%) of which involved 0-4-year-olds. In the two hospitals where the investigation was carried out, 271 diagnosed cases of tuberculous meningitis were admitted, 115 (42.4%) of which were children aged less than 5 years. Any case that trans- ferred between the two hospitals or which was re- admitted was counted only once, using the initial admission to the first hospital in the respective year. A total of 94 eligible cases remained, 19 of which were omitted from the study (11 were born outside the study area, while eight could not be located). Of these 75 cases that were located and visited, there was uncertainty about the true vaccination status of two children, who were therefore also eliminated from the study. Also a control could not be obtained for one case, which was therefore removed. Altogether, 72 cases fulfilled the study criteria (25 cases from 1981, 24 from 1982, and 23 from 1983). For these 72 cases, which formed the basis of the study, the history of contact with tuberculosis patients was traced for 46 of them (63.9%). Chest X- rays were available for only 50 of the 72 cases, and 46 (92%) were positive for tuberculosis. The results of CSF cultures were available for all cases, and Myco- bacterium tuberculosis was isolated from 10 children (13.9%). The results of CSF smears for acid-fast bacilli were available for all cases and were positive for four children (5.5%). The overall case fatality rate was 50%, but was higher among those aged less than 1 year (60.5%). Of children who survived, many suffered neurological sequelae from which they recuperated with difficulty. Only 11 of the children investigated recovered without exhibiting apparent neurological abnormalities (Table 1). A total of 520 neighbourhood controls and 83 hospital controls were visited. Of the neighbourhood controls, 15 were omitted (12 were born outside the defined study area, one was being treated for pulmo- nary tuberculosis during the period in which he was visited, and for two others doubts remained as to their true vaccination status). Neighbourhood con- trols could not be located for four cases. Of the 83 hospital controls, two were omitted (one was born outside MRSP and there were doubts about the vaccination status of the other), leaving 81. Hospital controls could not be obtained for 12 cases. WHO Bulletin OMS. Vol 68 1990. 71 V. Wunsch Fliho et al. Table 1: Clinical outcome of the 72 cases of tuberculous meningitis that formed the basis of the study, Sao Paulo, Brazil Outcome No. of cases Died in hospital 36 (50.0)' Died after discharge from hospital" 2 (2.8) Neurological sequelae 23 (31.9) No apparent neurological sequelae 11 (15.3) Total 72 (100.0) ' Figures in parentheses are percentages. b Death was verified at a home visit. Table 2 shows the distribution of cases and controls by age, sex and socioeconomic status (defined in terms of family income, area of residence, degree of domestic crowding, and mother's education level), and vaccination status. For ease of presenta- tion, unmatched data are shown; however, matching Table 2: Characteristics of the study cases and controls, Sio Paulo, Brazil No. of No. of No. of neighbourhood hospital Characteristic cases controls controls Age <6 months 14 (19.4)' 35 (6.9) 20 (24.7) 6-11 months 30 (41.7) 36 (7.1) 34 (42.0) 12-23 months 21 (29.2) 104 (20.6) 18 (22.2) A24 months 7 (9.7) 330 (65.3) 9 (11.1) Sex Male 49 (68.1) 250 (49.5) 51 (63.0) Female 23 (31.9) 255 (50.5) 30 (37.0) Socioeconomic statuse 1 58 (80.6) 431 (85.3) 55 (67.9) 2 13 (18.1) 74 (14.7) 26 (32.1) Unknown 1 (1.4) Vaccination status Vaccinated 42 (58.3) 463 (91.7) 72 (88.9) Unvaccinated 30 (41.7) 42 (8.3) 9 (11.1) Total 72 505 81 ' Figures in parentheses are percentages. b Children were classified as socioeconomic status 1 if they satisfied at least three of the following conditions: per capita household income less than one minimum wage (about USS 58.26 in May 1986); greater than four persons per bedroom; resident in the peripheral area of Sao Paulo municipality or other mun- icipalities of the Metropolitan Region of Sao Paulo; and mother illiterate or only partly literate. was preserved to estimate vaccine efficacy. Two- thirds of subjects came from the municipality of Sao Paulo, most from the peripheral poorer areas, while the remainder were from the other 36 municipalities of MRSP. Hospital controls were matched by sex and by age (within 6 months) but differed from cases with respect to their socioeconomic status, more frequently coming from higher status groups. Neigh- bourhood controls, as expected, had the same socioeconomic status as cases, but were older and had a higher proportion of females. Efficacy of BCG vaccination Vaccine efficacy was calculated separately for each group of controls using a conditional logistic re- gression analysis, and the results are shown in Table 3. The efficacies obtained were similar and high, the efficacy for neighbourhood controls (84.5%) being slightly greater than that for hospital controls (80.2%). No significant interactions were found between vaccination status and sex, age, or socio- economic status. Discussion To the best of our knowledge, this is the first large study to quantify the effect of BCG vaccination against tuberculous meningitis. Previously, Miceli et al. in a case-control study in Argentina reported that BCG vaccination had an efficacy of 100% against this form of tuberculosis, although the sample size was small (13). Our results indicate that in the study community in Sao Paulo, BCG vaccination was highly effective against tuberculous meningitis in chil- dren below 5 years of age. This finding is very encouraging for the prevention of a disease that has a high fatality rate and serious neurological sequelae among many of those who survive. Tuberculous meningitis predominantly affects children from Table 3: Efficacy of BCG vaccination against tuberculous meningitis for matched pairs of cases and controls, Sio Paulo, Brazil Vaccine efficacy (%) Cases and neighbourhood controls" 84.5 (66.7-92.8%)b Cases and hospital controlsc 80.2 (40.6-93.4%) ' Adjusted for age, sex, and socioeconomic status. Figures in parentheses are the 95% confidence intervals. c Adjusted for socioeconomic status. WHO Bulletin OMS. Vol 68 1990.72 Effectiveness of BCG vaccination against tuberculous meningitis poorer socioeconomic and environmental back- grounds, who may also be less likely to have been vaccinated with BCG. Since the majority of cases occur among children aged 3-11 months, we recom- mend that BCG vaccine be administered within the first 3 months of life and that efforts are made to achieve a high coverage across all socioeconomic strata. Use of the case-control approach has been recommended for studies of the effectiveness of BCG vaccination (15, 21, 22). Compared with controlled trials, the case-control approach is both quicker to carry out and cheaper. In accord with the reports of other workers (13, 19, 23), our findings indicate that the case-control method is useful for evaluating the effectiveness of BCG vaccination. It is particularly encouraging that similar results were obtained with the two groups of controls. Hospital controls, although easy to locate, differed from cases with respect to socioeconomic status. The neighbourhood controls, although they had the same socioeconomic status as the cases, differed from the latter in their age and sex distributions; they were, however, a more plentiful source of controls. The analysis with each group of controls therefore still required some con- trol of confounding variables, which led to similar estimates of BCG efficacy. The vaccine efficacy was slightly higher for neighbourhood controls, but this may have arisen because of further confounding. No evidence was found for interactions between vaccina- tion status and age, sex, or socioeconomic status, but the sample size was rather small to study these effects. In most countries where tuberculosis is endemic, its incidence has remained fairly constant. For such countries, BCG vaccination is an attractive policy to protect children against tuberculosis, although it does not significantly decrease transmission of the disease. Our results are encouraging for the preven- tion of tuberculous meningitis and should hopefully stimulate further case-control studies of BCG vaccin- ation and childhood tuberculosis in other countries. Acknowledgements We are grateful to Dr A. Pio and Dr H.G. ten Dam, both of WHO, Dr P.G. Smith, London School of Hygiene and Tropical Medicine, and Dr C.G. Victora, Universidade Federal de Pelotas, for their helpful comments and tech- nical assistance. This work would not have been possible without the full cooperation of the medical staff of Emilio Ribas and Mandaqui Hospitals. We are grateful also to Mr L. Walsh for his contribution to the preparation of this manuscript. This investigation was partly supported by a grant from WHO. R4sum6 Efficacit6 de la vaccination par le BCG contre la m6ningite tuberculeuse: etude cas-temoins A S8o Paulo, Br6sil A l'approche de la fin du XXe siecle, la tuberculose pose encore un probleme de sante publique, notamment dans les pays en developpement. Pour y remedier, les organismes sanitaires internatio- naux ont decide d'axer les programmes de lutte sur l'identification et le traitement des cas de maladie et sur la vaccination par le BCG. Bien que le vaccin BCG soit largement utilise dans le monde a titre de mesure preventive contre la tuberculose, sa valeur a ete remise en question. L'article presente une discussion du role du BCG en tant que mesure de lutte contre la tuberculose et etudie la protection qu'il confere contre la menin- gite tuberculeuse. Les politiques de vaccination par le BCG dans les Etats bresiliens de Rio Grande do Sul et de Sao Paulo, qui ont differents calendriers vaccinaux, ont ete evaluees; on a pour cela examine l'incidence de la meningite tuberculeuse dans chacun de ces Etats. Dans l'Etat de Rio Grande do Sul, o'u les enfants sont vaccines a l'age de sept ans, l'in- cidence de la meningite tuberculeuse est environ quatre fois plus elevee qu'a Sao Paulo o'u les enfants sont vaccines avant l'age d'un an. La couverture vaccinale du BCG varie sensiblement d'une region a l'autre du Bresil. Par exemple, en ville de Sao Paulo, elle est tres elevee, depassant meme 100% selon les chiffres officiels; toutefois, une enquete realisee en 1982-1983 conduit a estimer a 88% la couverture vaccinale chez les enfants de moins de 18 mois. L'article rapporte les resultats d'une etude cas-temoins menee dans la zone urbaine de SAo Paulo (Bresil) afin de determiner la protection que confere le BCG contre la meningite tuberculeuse chez les enfants de moins de cinq ans. L'etude a porte sur 72 cas de meningite tuber- culeuse, 505 temoins de voisinage et 81 temoins hospitaliers. On a calcule l'efficacite du vaccin selon la formule 1-RR, dans laquelle RR est le risque relatif de meningite tuberculeuse chez les sujets vaccines par rapport aux sujets non vac- cines, exprime par le odds ratio. Avec une analyse de regression logistique conditionnelle, nous avons calcule que l'efficacite du vaccin etait analogue dans los deux groupes de temoins, legerement plus grande toutefois chez les temoins de voisinage (84,5%) que chez les t6moins hos- pitaliers (80,2%). On n'a observe aucune relation WHO Bulletin OMS. Vol 68 1990. V. WOnsch Fliho et al. significative entre l'etat vaccinal et le sexe, I'Age ou le niveau socio-economique. Bien que des r6serves aient ete exprim6es quant a l'efficacit6 du BCG sur la chaine de trans- mission de la tuberculose et par consequent sur l'incidence et la mortalit6 g6n6rale dues i cette maladie, nos observations montrent qu'il peut 6tre utile s'il prot6ge contre les formes infantiles graves de tuberculose. Comme la plupart des cas de m6ningite tuberculeuse chez l'enfant survien- nent chez les nourrissons de 3 c 11 mois, nous recommandons d'administrer le BCG au cours des trois premiers mois de la vie. References 1. Anela, K.S. Immunoprophylaxis in tuberculosis. Jour- nal of communicable diseases, 16(1): 49-53 (1983). 2. Aronson, J.D. et al. A twenty-year appraisal of BCG vaccination in the control of tuberculosis. Archives of internal medicine, 101: 881-893 (1958). 3. Medical Res4arch Council Tuberculosls Vaccines Clinical Trials Committ". BCG and vole bacillus vaccines in the prevention of tuberculosis in adoles- cents. First Report. British medical journal, 1: 413- 427 (1956). 4. MedIcal Research Council Tuberculosis Vaccines Clinical Trlals Committee. BCG and vole bacillus vaccines in the prevention of tuberculosis in adoles- cence and early adult life. Third Report. British medical journal, 1: 973-978 (1963). 5. Medical Research Council Tuberculosis Vaccines Clinical Trials Commitee. BCG and vole bacillus vaccines in the prevention of tuberculosis in adoles- cence and early adult life. Fourth Report. Bulletin of the World Health Organization, 46: 371-385 (1972). 6. Centre for Health Information. [Manual for epidemiological surveys. Rules and instructions.] Sao Paulo, Secretary of State for Health, Government of the State of Sao Paulo, 1978 (in Portuguese). 7. Clemens, J.D. et al. The BCG controversy. Journal of the American Medical Association, 17: 2362-2369 (1983). 8. Comstock, G.W. & Palmer, C.E. Longterm results of BCG vaccination in the southern United States. American review of respiratory disease, 93: 171-183 (1966). 9. Comstocic, G.W. & Webster, R.G. Tuberculosis studies in Muscogee County, Georgia. American review of respiratory disease, 100: 839-884 (1969). 10. EGRET epidemiological graphics, estimation and testing package. Statistics and Epidemiology Research Corporation, Seattle, WA, 1988. 11. 1Brazillan Institute- for Geography and Statstice. [Annual statistics for Brazil]. Brasilia, Secretariat for Planning of the Office of the President of the Republic, 1983 (in Portuguese). 12. Indian MedIcal Re"arch Council. Tuberculosis Prevention Trial, Madras. Trial of BCG vaccine in South India. Indian journal of medical research, 72 (suppl.): 1-74 (1980). 13. MIcell, I. et al. Evaluation of the effectiveness of BCG vaccination using the case-control method in Buenos Aires, Argentina. International journal of epi- demiology, 17: 629-634 (1988). 14. [Mortality statistics. Brazil 1980.] Brasilia, National Secretariat for Basic Health Care, National Division of Epidemiology, 1983 (in Portuguese). 15. WHO Technical Report Series No. 651, 1980. (Vacci- nation against tuberculosis: report of an ICMR/WHO Scientific Group). 16. WHO Technical Report Series No. 652, 1980 (BCG vaccination policies: report of a WHO Study Group). 17. Palmer, C.E. et al. Community trials on BCG vaccina- tion. American review of tuberculous and pulmonary diseases, 77: 877-907 (1958). 18. Rosenthal, S.R. et al. BCG vaccination against tuber- culosis in Chicago. A twenty-year study statistically analyzed. Pediatrics, 28: 622-641 (1961). 19. Shapiro, C. et al. A case-control study of BCG and childhood tuberculosis in Cali, Colombia. Inter- national journal of epidemiology, 14: 441-446 (1985). 20. [Studies and investigations-Part 13.] Sao Paulo, Coordinator for Analysis of Data, Secretary for Econ- omics and Planning, Government of the State of Sao Paulo, 1977 (in Portuguese). 21. Smith, P.G. Retrospective assessment of the effec- tiveness of BCG vaccination against tuberculosis using the case-control method. Tubercle, 63: 23-35 (1982). 22. Smith, P.G. et al. Assessment of the protective efficacy of vaccines against common diseases using case-control and cohort studies. International jour- nal of epidemiology, 13: 87-93 (1984). 23. Smith, P.G. Case-control studies of the efficacy of BCG vaccine in children. Bulletin of the International Union against Tuberculosis and Lung Disease, 62(3): 70-76 (1987). 24. ton Dam, H.G. Research on BCG vaccination. Advan- ces in tuberculosis research, 21: 79-106 (1984). 25. WHO Technical Report Series No. 671, 1982 (Tuber- culosis control: report of a joint IUAT/WHO Study Group). 74 WHO Bulletin OMS. Vol 681990.