Tuberculosis in a rural population of South India: a five-year epidemiological study*

TUBERCULOSIS Tuberculosis in a rural population of South India: a five-year epidemiological study* NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE 1 Tuberculosis is a serious health problem in India, but because ofinadequate facilitiesfor diagnosis in manyparts of the country clinical data are incomplete. However, data on preva- lence are available from a few sample surveys conducted in the general population during the last two decades. The study reported here provides information on the incidence of infection and disease as well as on the source and fate of cases. The information was obtained by means of surveys, repeated at intervals during 1961-68 in a randomly selected rural population of South India. In each survey, a tuberculin test was given and X-ray and sputum examinations were made. There were virtually no tuberculosis controlfacilities in the study area, since this was not covered by the national tuberculosis programme at the time. The annual rate of tuberculous infection in previously uninfected children and in adults was found to be about 1 %. The incidence of tuberculosis (confirmed by culture) was about 1 per thousand (excluding children below the age of 5 years). About 30 % of newly detected cases came from the population uninfected at an earlier survey. The prevalence of disease was four times as high as the incidence. Among cases found in the initial survey one-half died within 5 years and one-fifth continued to excrete bacilli after 5 years. Both infection and disease tended to decline during the observation period, though in an uneven manner that does not permit extrapolation. There was no evidence of an increase in drug-resistance among newly diagnosed cases. The findings of the study could be used in developing epidemetric models to predict the tuberculosis problem and to assess tuber- culosis control measures applied in the community. Considerable light was thrown on the epidemi- ology of tuberculosis in India by the studies reported by Ukil (1), Benjamin (2), Sikand & Raj Narain (3), the Indian Council of Medical Research (4) and Raj Narain et al. (5, 6). Data on the prevalence and incidence of infection and disease in a population benefiting from antituberculosis measures were pre- sented by J. Frimodt-M0ller (unpublished obser- vations, 1973) and Pamra et al. (7, 8). The National Tuberculosis Institute (NTI), Bangalore, started a * A more detailed report on this study is to appear in the WHO/TB/Techn. Information series of documents and in the Indian journal of tuberculosis. 1 The National Tuberculosis Institute, Bangalore, func- tions under the Ministry of Health and Family Planning, Government of India. The members of the scientific staff of the institute who had major responsibilities for the work reported are: Dr Raj Narain, Epidemiologist (up to 1966); Dr G. D. Gothi, Epidemiologist (from 1966 onwards); Mr S. S. Nair, Senior Statistical Officer (from 1962); Dr K. P. Rao, Bacteriologist; Dr P. Chandrasekhar, Medical Officer; Dr Pyarelal, Medical Officer; Dr A. Geser, WHO Epi- longitudinal study in 1961 to observe the natural history of pulmonary tuberculosis under the socio- economic conditions existing in an area that was not yet benefiting from any of the active control measures-such as BCG vaccination and chemo- therapy-forming part of the national tuberculosis programme. It was thought that the information obtained might help in the construction of epide- metric models and in the development of an assess- ment methodology for tuberculosis control pro- demiologist (up to 1961); Dr M. V. Jambunathan, Senior Statistical Officer (up to 1962); Mr N. Naganathan, Junior Bacteriologist; and Mr V. A. Menon, X-ray Engineer. The field work was carried out by the Epidemiological Teams, whose leaders were: Mr K. R. Rangaswamy; Mr M. S. Krishnamurthy; Mr A. N. Sashidhara; Mr K. Parthasarathy; Mr P. Srinivasan; and Mr G. Dwarakanath. The report was prepared by Dr G. D. Gothi; Mr S. S. Nair; Dr T. Ola- kowski, WHO Medical Officer; Dr D. R. Nagpaul, Director, National Tuberculosis Institute; Dr N. K. Menon, Adviser in Tuberculosis, Government of India; and Mr K. T. Ganapathy, Junior Statistical Officer. 3292 - 473 BULL. WORLD HEALTH ORGAN., 1974, 51 NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE grammes. The present paper is a brief report on the study for the 5-year period. OBJECTIVES The objectives of the study were: (1) to establish prevalence rates of tuberculous infec- tion and disease at different points of time; (2) to establish the annual incidence rates of tuber- culous infection and disease; (3) to study the fate of tuberculosis patients found in the surveys; and (4) to identify the principal epidemiological groups and to study the incidence of the disease in them. STUDY AREA AND POPULATION The study was conducted in 119 randomly selected villages in the taluks (subdivisions) of Magadi, Chennapatna, and Nelamangala, Bangalore District, South India. The population of the three taluks in 1961 was about 0.4 million, about 88% ofwhom lived in villages. The population of the villages ranged from 100 to 5 000 with an average of 500. From the socioeconomic point of view, the study population could be considered as similar to the rural popu- lation in most parts of the country. Normally, the rural population is comparatively stable, emigration (2.2%) being somewhat in excess of immigration (1.7%), according to Andersen & Banerjee (9). METHODS The population was surveyed four times from 1961 to 1968, with an interval of 11/2 years between the first and second surveys and second and third surveys. The interval between the third and fourth surveys was 2 years. At the first survey, a complete census of the population in each village was taken by house-to-house visits and recorded on individual cards and forms. At each subsequent survey persons registered earlier were identified and classified accord- ing to their present residential status. New individual cards were prepared for the newborn and for immi- grants. The field work for each survey lasted 2 years. All registered persons were examined at a place centrally located in the village. The shoulders of each person were examined for the presence of BCG scars. A tuberculin test with 1 TU of PPD RT 23 containing 0.05% Tween 80 in 0.1 ml was given on the volar surface of the left or right forearm, alter- nating the arms at each survey but choosing a dif- ferent site for the test. The longitudinal diameter of the induration was measured 3 or 4 days after the test. All persons aged 5 years and above were offered 70-mm chest photofluorograms. Two sputum speci- mens (one spot and one overnight specimen) were obtained from each person with an abnormal shadow on the photofluorogram, with a technically inade- quate photofluorogram, or with such physical dis- ability that an X-ray could not be taken. Sputa were collected at subsequent surveys from those who had had abnormal shadows in any of the pre- vious surveys even if their photofluorograms during the current survey were judged as normal. Photofluorograms were interpreted by two X-ray readers independently according to the following code: N = normal (no abnormality seen in lung fields) A = considered to be of nontuberculous etiology B = suggestive of tuberculous etiology; judged to be inactive C = suggestive of tuberculous etiology; judged to be pos- sibly active D = suggestive of tuberculous etiology; judged to be probably or definitely active Sputum smears were examined first by fluor- escence microscopy and then by the Ziehl-Neelsen technique. The sputum was then homogenized with 4% sodium hydroxide and inoculated on slopes of Lowenstein-Jensen medium. The slopes were exam- ined at intervals for 8 weeks. Persons with positive cultures that satisfied all the criteria for Myco- bacterium tuberculosis after a set of identification tests were considered as cases of tuberculosis if they had X-ray abnormality in the current survey. Tests of sensitivity to streptomycin, isoniazid, and para- aminosalicylic acid (PAS) were carried out. MATERIAL The study population at the first survey constituted about 15% of the population in the three taluks. Population registered during the four surveys is shown in Table 1. Permanent residents-i.e., those residing in a village for more than one year or since birth constituted about 96% of the registered popu- lation in each survey. The size of the permanently resi- dent population remained the same from the first to the fourth survey. There was a decrease in the propor- tion of children below the age of 15 years. Average annual death and migration rates per 1 000 during 474 TUBERCULOSIS Table 1. Registered population according to sex and residential status during four surveys (I-IV) 111 11 lv Residential status Sex No. % No. % No. % No. % permanent both sexes 61 663 95.8 62065 96.9 62789 95.6 61 581 95.7 residents males 31 465 97.2 31 653 97.9 32 051 96.9 31 138 96.8 females 30 198 94.3 30 412 95.8 30 738 94.4 30 443 94.7 temporarily both sexes 2 722 4.2 2 007 3.1 2 858 4.4 2 750 4.3 present males 913 2.8 671 2.1 1 027 3.1 1 032 3.2 females 1 809 5.7 1 336 4.2 1 831 5.6 1 718 5.3 both sexes 64385 100.0 64072 100.0 65647 100.0 64331 100.0 Total males 32 378 100.0 32 324 100.0 33 078 100.0 32 170 100.0 females 32 007 100.0 31 748 100.0 32 569 100.0 32 161 100.0 the 5-year study period, according to age group, are shown in Table 2. Differences in the age composition of the population between surveys were statistically significant for both sexes, but the magnitude of the changes was quite small and was not likely to affect the findings. There was a drought in the study area from 1965 to 1967, with a scarcity of food and localized epi- demics of cholera and smallpox that might have encouraged emigration. This period coincided with the end of the third survey and a major part of the fourth. No organized antituberculosis treatment was avail- able to the people in the area during the entire study period. During the second survey and the early part of the third, with a view to obtaining better cooperation, one month's supply of isoniazid tablets was issued to persons in whom pulmonary tuberculosis was diagnosed. For subsequent drug collections, they were advised to attend rural health institutions in the area, to which limited quantities of isoniazid had been provided. For a variety of reasons, particularly distance, very few patients took advantage of this advice and the practice was given up. The coverages for examination of BCG scars, reading of tuberculin tests, photofluorography of the chest, and sputum examination were fairly high at Table 2. Average annual death and migration rates per 1 000 during the 5-year study period, according to age group Males Females Age group (years) deaths emigration immigration deaths emigration immigration 0-4 19 44 38 20 45 39 5-14 3 61 37 3 67 33 15-34 4 73 41 6 76 64 35-54 9 37 23 10 30 18 55+ 49 23 17 44 33 26 All ages 11 54 34 11 57 40 475 NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE BOTH SEXES _ ltf -% - Ies __-s e==~~~~~~~~~.0-0' MALES FEMALES .- *ogeogoee geee so g 099eV t* 00100000. _-t _f. soe .om _____enin _ ....................................................... ,_ams ee-es-e-s Tuberculin test read (persons of all ages were eligible) ---- X-rayed (persons aged,>5 years were eligible) ee-999999 Sputum examined (persons with abnormal X-rays were eligible) 476 100 90 80 70 60 50 40 30 C: r_ a) 20 _ X-ray and tuberculin test read10 0 I 11 D] 117 I 11 IH N I 'n lu IV Surveys WHO 75371 Fig. 1. Coverage of the study population for sputum examinations and for the reading of tuberculin tests and X-rays, according to sex (surveys I-IV). all the four surveys (Fig. 1). Table 3 shows that the response group contained a higher proportion (range: 42.9-45.7Y.) of children aged 0-14 years than the non-response group (range: 29.2-30.7y.). Furthermore, the crude death rates at the subsequent survey in the non-response group were nearly double those found in the response group. Emigration rates were also considerably higher in the non-response group. Such differences in the death rate and emi- gration rate were observed in each age group. Persons without BCG scars in the response group among permanent residents have been included in the main analysis. Those with BCG scars, the pro- portion of whom ranged from 2.5 Y. at the first survey to 3.7 Y. at the fourth, were excluded. FINDINGS Prevalence of infection The distributions of indurations resulting from tuberculin testing at the first and subsequent surveys do not show a clear-cut demarcation between the infected and noninfected population. Therefore, the distribution of tuberculin indurations in 574 tuber- culosis cases diagnosed during the study was exam- ined. This suggested that an induration of I0 mm or more is evidence of infection. Raji Narain et al. (10) had observed that persons retested in subsequent surveys generally showed a larger tuberculin induration. It was observed that between the first and second surveys an average increase of 0.6-3.0 mm had occurred in various five-year age groups. Between the second and third surveys an additional increase of 0.3-2.3 mm took place and a further increase of 0.02-2.0 mm occurred at the fourth survey. A new demarcation line be- tween the infected and noninfected was adopted for the second survey by adding the mean increase in tuberculin sensitivity seen in previously tested per- sons to the 10-mm level decided on for the first survey, the increases being cumulated for the third and the fourth surveys and added to the 10-mm level. The prevalence of infection in the population was about 30yo: 25Y. for females and 35% for males. This rate increased with age up to the age of 45-54 years. The increase for females slowed down after the age of 15 years in contrast to that for males (Table 4). The overall prevalence of infection was fairly constant throughout the surveys (29.3-30.4y.), for TUBERCULOSIS Table 3. Age-sex composition of the response and non-response groups in four surveys Males Females Survey (years) response non-response response non-response No. % No. % No. % No. % 0-4 4207 15.2 484 12.8 4115 15.2 502 15.8 5-14 8 284 29.9 600 15.8 8 078 29.9 557 17.5 1 15-34 8244 29.8 1 441 38.0 7970 29.5 1 243 39.1 35-54 4 640 16.8 881 23.2 4 601 17.0 547 17.2 55+ 2 297 8.3 387 10.2 2 254 8.3 331 10.4 Total 27 672 100.0 3 793 100.0 27 018 100.0 3180 100.0 0-4 3764 14.3 708 13.1 3678 14.5 740 14.7 5-14 8120 30.9 874 16.2 8 043 31.7 844 16.7 11 15-34 7 973 30.4 2003 37.1 7561 29.8 1 851 36.7 35-54 4374 16.7 1 188 22.0 4213 16.6 951 18.9 55+ 2024 7.7 625 11.6 1 878 7.4 653 13.0 Total 26 255 100.0 5398 100.0 25 373 100.0 5039 100.0 0-4 3844 14.9 711 11.5 3824 15.1 658 12.2 5-14 7 609 29.4 1 037 16.8 7 686 30.3 993 18.4 III 15-34 8159 31.5 2267 36.7 7675 30.3 1 986 36.7 35-54 4266 16.5 1 522 24.6 4257 16.8 1110 20.5 55+ 1 993 7.7 643 10.4 1 889 7.5 660 12.2 Total 25 871 100.0 6 180 100.0 25 331 100.0 5 407 100.0 0-4 3 664 14.0 568 11.6 3 738 14.2 498 1 1.9 5-14 7 470 28.5 776 15.8 7 635 29.1 811 19.4 IV 15-34 8119 30.9 1 851 37.8 7971 30.4 1 513 36.2 35-54 4691 17.9 1 161 23.7 4673 17.8 793 19.0 55+ 2 294 8.7 544 11.1 2 244 8.5 567 13.6 Total 26 238 100.0 4 900 100.0 26 261 100.0 4 182 100.0 males as well as for females. The standardized rates, worked out for surveys I-IV, taking the age-sex structure of the permanent population of the first survey as standard, were 30.4%, 31.0%, 30.6%, and 31.2%, respectively-almost similar to the observed rates shown in Table 4. There was a statistically significant decrease in the prevalence of infection in the age groups 0-14 and 15-24 years. Incidence of infection The incidence of infection has been defined as the number of newly infected persons found in a subsequent survey among the noninfected of the preceding survey per 100 persons per year. Raj Narain et al. (11) suggested a method for separating out the newly infected. This method was slightly modified by fitting a normal curve to the major 477 478 NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE Table 4. Prevalence of infection according to age and sex in the four surveys 11 III IV Age group (years) test read infected test read infected test read infected test read infect Both sexes 0-4 7 981 2.1 7 207 1.8 7 453 1.3 7 216 1.0 5-9 7 470 7.9 7 624 7.6 7 470 7.0 7 562 6.4 10-14 7 259 16.5 6 697 16.9 6301 16.1 6 021 15.4 15-24 7 161 33.2 6 523 32.1 7 077 29.6 7366 31.9 25-34 7 569 46.4 7 235 48.3 6 974 48.7 6 640 47.3 35-44 4 831 52.3 4 526 54.5 4 571 56.1 4 961 54.8 45-54 3 652 56.1 3 207 58.2 3 131 60.1 3 367 60.7 55+ 4 223 56.6 3 606 60.3 3 568 59.3 4 074 62.1 Total 50 146 29.5 46 625 30.4 46 545 29.3 47 207 30.4 Males 0-4 4031 2.2 3635 1.6 3729 1.0 3567 0.5 5-9 3 580 7.5 3 741 7.1 3 779 7.0 3 817 6.2 10-14 3 718 17.8 3 270 16.9 2 920 17.0 2 788 15.1 15-24 3 639 39.8 3 288 36.8 3 656 37.5 3 684 37.1 25-34 3 530 59.6 3294 60.4 3 126 63.1 2 933 62.3 35-44 2 344 64.5 2 162 64.0 2 178 67.9 2 354 68.6 45-54 1 759 65.5 1 551 67.6 1 482 71.8 1 597 72.3 55+ 2083 63.4 1 817 65.8 1 778 66.1 1 992 68.3 Total 24 684 34.7 22 758 34.5 22 648 34.1 22 732 34.9 Females 0-4 3 950 2.0 3 572 2.0 3 724 1.6 3 649 1.6 5-9 3890 8.3 3883 8.4 3 691 6.9 3 745 6.6 10-14 3 541 15.2 3427 16.9 3381 14.7 3233 15.2 15-24 3 522 26.3 3 235 27.3 3 421 25.1 3 682 25.2 25-34 4039 34.8 3 941 36.5 3 848 37.0 3707 38.3 35-44 2 487 40.8 2 364 43.3 2 393 43.2 2 607 43.3 45-54 1 893 47.2 1 656 49.3 1 649 50.7 1 770 50.3 55+ 2140 50.0 1 789 54.7 1 790 52.8 2082 57.5 Total 25 462 24.6 23 867 25.3 23 897 24.7 24 475 26.2 part of the observed distribution (12). The number The incidence of infection generally increased of newly infected persons could thus be estimated with age (Table 5). In the three intervals between in a more objective manner than was done by Raj surveys it generally decreased for all ages combined Narain et al. (11). Newly infected persons generally as well as for each individual age group except for show a substantial increase in induration. the 10-14-year age group (both sexes) and the age TUBERCULOSIS Table 5. Incidence of infection, according to age and sex, between surveys Both sexes Males Females Age group non- non- nn(years) infected at No. newly annual rate infected at No. newly annual rate inon No newly annual rate earlier round infected (% earlier round infected (% inflecteduat infected (% Between surveys & II (correction factor for annual incidence: 0.648) 0-4 6 295 78 0.80 3 198 34 0.69 3 097 44 0.92 5-9 5699 94 1.07 2720 34 0.81 2979 60 1.30 10-14 4 495 88 1.27 2 261 40 1.15 2 234 48 1.39 15-34 6 093 204 2.17 2 367 88 2.41 3 726 116 2.02 35-54 2 971 145 3.16 993 51 3.33 1 978 94 3.08 55+ 1 234 65 3.41 513 24 3.03 721 41 3.68 0-34 22 582 464 1.33 10 546 196 1.20 12 036 268 1.44 35+ 4205 210 3.24 1 506 75 3.23 2699 135 3.24 Total 26 787 674 1.63 12 052 271 1.46 14 735 403 1.77 Between surveys 11 & III (correction factor for annual incidence: 0.636) 0-4 7 714 79 0.65 3 905 44 0.72 3 806 35 0.58 5-9 5 146 73 0.90 2 435 32 0.84 2 696 41 0.97 10-14 3544 76 1.36 1 842 51 1.76 1 702 25 0.93 15-34 5 287 139 1.67 1 963 67 2.17 3376 72 1.36 35-54 2 477 63 1.62 881 34 2.45 2 568 29 0.72 55+ 940 21 1.42 404 14 2.20 521 7 0.85 0-34 21 691 367 1.08 10145 194 1.22 11 580 173 0.95 35+ 3417 84 1.56 1 285 48 2.38 3089 36 0.74 Total 25108 451 1.14 11 430 242 1.35 14669 209 0.91 Between surveys III & IV (correction factor for annual incidence: 0.503) 0-4 5 796 67 0.58 2 905 35 0.61 2 891 32 0.56 5-9 5 536 73 0.66 2 761 36 0.66 2 775 37 0.67 10-14 3817 80 1.05 1 659 42 1.27 2 181 38 0.88 15-34 5 829 103 0.89 2 316 36 0.78 3 629 67 0.93 35-54 2520 65 1.30 777 21 1.36 1 775 44 1.25 55+ 966 25 1.30 373 10 1.35 612 15 1.23 0-34 20978 323 0.77 9641 149 0.78 11 476 174 0.76 35+ 3486 90 1.30 1 150 31 1.36 2387 59 1.24 Total 24 664 413 0.84 10 936 180 0.83 13 852 233 0.85 479 NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE Table 6. Prevalence of cases, according to age and sex, at surveys l-IV Age IV group(years) No. No. of Rate per No. No. of Rate per No. No. of Rate per No. No. of Rate per examined cases 100000 examined cases 100000 examined cases 100 000 examined cases 100 000 Both sexes 5-14 14 866 14 94 13 843 10 72 13 749 5 36 13 333 5 37 15-34 15 665 59 377 14 765 53 359 14 818 45 304 14 707 38 258 35-64 9084 56 616 8382 44 525 8275 50 604 8989 68 756 55+ 4274 49 1 146 3643 44 1 208 3563 36 1 010 4184 51 1 219 Total 43889 178 406 40633 151 372 40405 136 337 41 213 162 393 Males 5-14 7 453 1 0 134 6 828 5 73 6 780 1 1 5 6 555 3 46 15-34 8 000 40 500 7 595 32 422 7 635 30 393 7 365 23 312 35-54 4582 40 873 4293 34 792 4180 37 885 4530 54 1 192 55+ 2202 41 1 862 1 947 39 2003 1 893 33 1 743 2193 42 1 915 Total 22 237 131 589 20 643 110 533 20 488 101 493 20 643 122 590 Females 5-14 7 413 4 54 7 015 5 71 6 969 4 57 6 778 2 30 15-34 7 665 1 9 248 7190 21 292 7183 1 5 209 7 342 1 5 204 35-54 4 502 16 355 4 089 10 245 4 095 13 317 4 459 14 314 55+ 2072 8 386 1696 5 295 1670 3 180 1 991 9 452 Total 21 652 47 217 19990 41 205 19917 35 176 20570 40 194 groups 0-4 and 5-9 years among males. The overall decline from 1.63% to 0.84% and the decrease in the 0-34-year age group from 1.33% to 0.77% were stat- istically significant. However, for individual age groups the changes failed to attain significance. The standardized rates, on the basis of the age-sex distri- bution of noninfected persons in the first survey, showed a similar decline (1.6% to 1.23% and 0.87 %). Prevalence of disease The overall prevalence of bacteriologically con- firmed disease ranged from 337 at the third survey to 406 at the first survey per 100 000 population aged 5 years or more (Table 6). At each survey, the number of cases in females was about 1/3 of that in males. Of the total number of " prevalence cases I, about 50% were in males aged 35 years or more, whereas only 10% were in females of the same age. The prevalence rates increased with age, especially for males. These rates in both sexes were similar in the 5-14-year age group but were 4 to 9 times as high among males in the age group 55 years or more. The prevalence of disease decreased from the first to the third survey and increased in the fourth. Standardized rates, calculated on the basis of the age-sex distribution of the permanent population of the first survey, were almost similar to the observed rates at all the surveys. The younger age groups, 5-14 and 15-34 years, showed a con- tinuous decrease. The decline in the former was statistically significant but failed to attain signifi- cance in the latter (0.1< P < 0.2). For the combined age group 5-34 years the decline was significant. Among the older age groups, for either sex, fluctu- ations were observed from one survey to the next. The proportion of cases positive on culture only was generally higher than that of cases positive on culture and smear at the first three surveys (53.0- 480 TUBERCULOSIS Table 7. Prevalence, according to age and sex, of cases positive on culture but not on direct smear microscopy at surveys llVa I1 III lv Age group No. only Rate per % No. only Rate per % No. only Rt e % No. only Rt e(years) culture- 1R00000 of all culture- 1 00 000 of all culture- Rateopoe of all culture- 1Rte0per of allpositive cases positive cases positive cases positive cases Both sexes 5-14 10 67 71.4 8 58 80.0 5 36 100.0 3 23 60.0 15-34 28 179 47.5 28 190 52.8 23 155 51.1 1 7 116 44.7 35-54 30 330 53.6 22 262 50.0 27 326 54.0 24 267 35.3 55+ 27 632 55.1 22 604 50.0 20 561 55.6 23 550 45.1 Total 95 216 53.4 80 197 53.0 75 186 55.1 67 163 41.4 Males 5-14 6 80 60.0 4 59 80.0 1 1 5 100.0 3 46 100.0 15-34 17 212 42.5 13 172 40.6 14 183 46.7 9 122 39.1 35-54 22 480 55.0 1 7 396 50.0 22 526 59.5 21 464 38.9 55+ 23 1 044 56.1 18 924 46.2 18 951 54.5 18 821 42.9 Total 68 306 51.9 52 252 47.3 55 268 54.5 51 247 41.8 Females 5-14 4 54 100.0 4 57 80.0 4 57 100.0 - - - 15-34 1 1 144 57.9 1 5 209 71.4 9 125 60.0 8 109 53.3 35-54 8 178 50.0 5 122 50.0 5 122 38.5 3 67 21.4 55+ 4 193 50.0 4 236 80.0 2 120 66.7 5 251 55.6 Total 27 125 57.4 28 140 68.3 20 100 57.1 16 78 40.0 a For the number examined and the prevalence of all cases, see Table 6. 55.1 % of all cases), but was lower at the fourth survey (41.4%). This observation applied to both sexes and most age groups (Table 7). Of the prevalence cases at three successive sur- veys, 53.8%, 37.0% and 49.2% were newly detected at the second, third, and fourth surveys, respectively. About 25% of the prevalence cases found at the fourth survey had existed from the first survey. Incidence of disease Persons who had an X-ray abnormality but were bacteriologically negative, or who had had normal X-rays in all the previous surveys and were found to have tuberculosis at a subsequent survey, were defined as " incidence cases ". The incidence of disease was defined as the number of incidence cases of tuberculosis found at a subsequent survey per 100 000 population per year among persons exam- ined at the preceding survey. The overall annual incidence of disease per 100000 population was 132 between the first and second sur- veys, 79 between the second and third surveys, and 99 between third and fourth surveys (Table 8). The cor- responding standardized rates, on the basis of the age-sex distribution of the permanent population at the first survey, were 136, 80, and 104, respecti- vely. The incidence rates increased with age gene- rally for both males and females, the increase in older age groups being much greater among males. The overall rate for males was nearly double that for females. More than half of the new male cases were 35 years of age or more, whereas more than 481 NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE Table 8. Incidence of tuberculosis between surveys, according to age and sex, and corresponding annual incidence rates Incidence between surveys Age group III-llllII-IV Average (years) annual rate No. No. of Annual No. No. of Annual No. No. of Annual per 100 000 examined cases rate per examined cases rate per examined cases r00000iooooo cases 100000 a exmnd s rt per00 Both sexes 5-14 12456 11 57 11 427 4 22 11 008 3 14 31 15-34 11 612 24 134 11 273 13 73 11 088 15 68 92 35-54 7 320 15 133 6745 13 123 6 695 27 - 203 153 55+ 3 088 20 420 2 749 10 231 2 590 17 330 327 Total 34476 70 132 32194 40 79 31 381 62 99 103 Males 5-14 6 265 6 62 5 649 1 1 1 5 314 1 9 27 15-34 5 992 10 108 5 810 8 88 5 697 9 79 92 35-54 3 680 9 158 3 367 8 151 3 317 20 303 204 55+ 1 654 17 666 1 492 10 426 1 397 14 504 532 Total 17 591 42 155 16 318 27 105 15 725 44 141 134 Females 5-14 6 191 5 52 5778 3 33 5694 2 18 34 15-34 5 620 14 161 5 463 5 58 5 391 6 56 92 35-54 3 640 6 107 3 378 5 94 3378 7 104 102 55+ 1 434 3 136 1 257 - - 1 193 3 126 87 Total 16885 28 107 15876 13 52 15656 18 58 72 a The annual incidence rates were calculated by applying the following correction factors for the time intervals: surveys I-ll: 0.648; surveys 11-111: 0.636; surveys III-IV: 0.503. half of the female cases were below the age of 35 years. The incidence rates for the three periods decreased significantly in the age groups below 35 years in both sexes. The proportion of incidence cases positive on culture ranged from only 55.8% between the third and fourth surveys to 75.0%/ between the second and third surveys. Incidence in some epidemiological groups The incidence of tuberculosis is expected to vary with tuberculin sensitivity and/or the presence of pathology in the lungs, as revealed by X-ray. There- fore, the population was subdivided arbitrarily into 9 homogeneous epidemiological groups on the basis of these characteristics (Table 9). Within any radio- logical group the incidence rate was highest among those with indurations of 20 mm or more. In that category of reactors, the incidence of disease was 45 times as high in the CD group, and 5 times as high in the AB group, as it was in the N group. The CD group with an induration of 20 mm or more, constituting 0.4% of the total followed-up population, contributed 26.9% of the incidence cases (Table 9). The next largest contribution (22.4%) to the total incidence was from noninfected persons with a normal X-ray, who constituted 61 00 of the total population. The lowest contribution (1.5 %) came from the noninfected AB group, accounting for only 5% of the total population. 482 TUBERCULOSIS Table 9. Incidence of disease between surveys I and tuberculin status at survey I and 11 according to radiological Tuberculin Radiological status at No. % No. % Annual status at survey followed up of total of cases of total incidence survey a (induration edp(per 1 00 000) in mm) 0-9 19 419 61.0 15 22.4 50 10-19 4 767 15.0 2 3.0 27 N 20+ 4 201 13.2 12 17.9 185 Total 28 387 89.2 29 43.3 66 10+ 8 968 28.2 14 20.9 101 0-9 1 592 5.0 1 1.5 41 10-19 767 2.4 3 4.5 253 AB 20+ 744 2.3 11 16.4 958 Total 3103 9.8 15 22.4 313 10+ 1 511 4.7 14 21.1 600 0-9 128 0.4 3 4.5 1 519 10-19 60 0.2 2 3.0 2160 CD 20+ 141 0.4 18 26.9 8 272 Total 329 1.0 23 34.3 4 530 10+ 201 0.6 20 29.0 6 448 0-9 21 139 66.4 19 28.4 58 10-19 5 594 17.6 7 10.4 81 Total 20+ 5 086 16.0 41 61.2 522 Total 31 819 100.0 67 100.0 136 10+ 10 683 33.6 48 71.6 291 a For details of groups N, A, B, C, and D, see page 474. Isoniazid resistance The proportion of those resistant to isoniazid among the prevalence cases showed a significant increase, from 11.2% at the first survey to 27.2% at the third survey, but dropped to 21.6% at the fourth survey. However, the proportion of freshly discovered resistant cases decreased from 80.0% at the second survey to 34.4% at the fourth survey. Among the incidence cases, isoniazid resistance did not increase during the interval between successive surveys. The proportions of those with resistance were 11.4%, 10.0%, and 6.5% for the incidence cases found at the second, third, and fourth surveys, respectively. These findings suggest that the main reason for the steady increase in the resistance rate may be a tendency among patients with resistance to isoniazid to continue to live as bacillary cases. 483 NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE Fate of 126 cases Fate of 55 Fate of 35 cases Fate of 154 cases Fate of 136 cases fate of 119 cases of survey I oontinuing cases of survey Ill of I of s1urveycI of survey lil at survey II of survey 11 (continuing at survey I at survey lIl at survey IV at survey II from survey I ) (adjusted for at survey IV 1-5 ears) . Fig. 2. Fate of cases discovered at the first survey and of patients who were still excreting bacilli when examined at subsequent surveys. Fate of cases Out of the 178 cases found at the first survey, 126 could be followed up at the three subsequent surveys during the study period. The death rate and the cure rate were highest during the first 11/2 years: by the end of this period, 30.2% of the 126 patients were dead, 27.8% had been cured, and 42% were still excreting bacilli. By the end of the second 1½/2-year period, the corresponding propor- tions were 40.5%, 34.9%, and 24.6%. Two years later-i.e., at the end of the 5-year period-62 pa- tients (49.2%) were dead, 41 (32.5%) were cured, and 23 (18.3%) remained sputum-positive (9 sensi- tive to drugs and 14 drug-resistant). Fig. 2 shows Fig. 3. Fate of prevalence cases discovered at surveys 1, II, and III over a period of 1.5 years. the fate of the 126 cases found at the first survey, and more particularly that of the cases found at subsequent surveys to be excreting tubercle bacilli. The mortality and relapse rates for patients who had become abacillary at the second and third sur- veys were not appreciably different at subsequent surveys. Average mortality was about 7 %, the relapse rate was 10%, and 83% of patients remained abacillary. The fate of cases during the 1½/2-year period after detection could be studied in three different cohorts between successive surveys (Fig. 3). Statistically sig- nificant variations were observed in the cure rates, from 24.4% for the third cohort to 39.0% for the Table 10. Fate of incidence cases Incidence cases Total No. followed up Cured Dead Remainedbetween: bacillary Surveys I and 11 70 63a 33 9 21(52.4%) (14.3%) (33.3%) Surveys II and Iil 40 39 b 11 16 13(28.2%) (38.5%) (33.3%) a Followed up for 1% years. b Followed up for 2 years. 484 TUBERCULOSIS second cohort. The variations in cure rates could not be accounted for by differences in age, sex, or bacillary status. However, in the third cohort a higher proportion of cases showed extensive disease and cavitation. Between the first and the second surveys, 70 inci- dence (new) cases were discovered, 63 of which were followed up at the third survey. Over the 1 1/2-year period 52.4% were cured, 14.3 % died, and 33.3% continued to be bacillary (Table 10), com- pared with 39.0%, 16.9%, and 44.1 %, respectively, for prevalence cases of the second survey that were followed up at the same time. Between the second and third surveys, 40 incidence cases were diagnosed, 39 of which were followed up at the fourth survey- i.e., after 2 years. Their fate (Table 10) was similar to that of prevalence cases of the third survey, which had a cure rate of 24.4% and a death rate of 31.1 %, 44.5% of patients remaining bacillary, over the 2-year period. DISCUSSION At the time of selection, the study area was seen to have a population structure similar to that of the country as a whole. During the study it was observed that: (1) the expected population increase did not occur; (2) the age-sex composition of the population varied somewhat at different surveys; and (3) the drought that occurred towards the end of the period caused considerable privation to the population. The main reason for the non-occurrence of the expected population increase was the excess of emigration over immigration. Pamra et al. (8) also did not find the expected increase in study population between their third and fourth surveys. Furthermore, although the coverages for various examinations were quite high, the non-response group was somewhat dif- ferent from the response group in terms of the age-sex composition and of death and migration rates at subsequent surveys. Horwitz & Knudsen (13) had observed similar differences. The above-men- tioned changes in the age-sex composition owing to migration or non-response do not matter when age-sex specific rates are compared. Moreover, the magnitude of the changes was so small that the standardized rates did not differ appreciably from the overall rates. The overall prevalence of infection ranged from 29.3 % to 30.4% from the first to the fourth survey. These rates were significantly lower than the 38.3% reported by Raj Narain et al. (5) for a neighbouring district. The overall incidence of infection ranged from 1.61 % to 0.84%. This broadly corresponds to the estimate of 1-2% reported by Frimodt-M0ller (14). In the higher age groups, the incidence of infection continuously increased, unlike the prevalence of in- fection. The actual number of noninfected persons at risk being fewer in higher age groups, additions to the earlier prevalence are small. Furthermore, it is believed that allergy wanes more readily in higher age groups. These factors contribute to the flattening of the curve for the prevalence of infection in those age groups. The prevalence of bacillary tuberculosis at the first survey was 406 per 100 000 population aged 5 years or more. The rate was similar to that reported by Raj Narain et al. (5) and Frimodt-M0ller (14), and was within the range found in different parts of the country by the Indian Council of Medical Research (4). The average annual incidence of disease was 103 per 100 000. Between the first and second surveys, the rate for the infected of the first survey was 0.29%, compared with 2.1 % for the newly infected. Thus, the incidence among the newly infected was 7 times the rate for those already infected. This is higher than the comparative risk (5 times) reported by Palmer et al. (15) and by Groth Petersen (16), but considerably lower than that (27 times) reported by Trauger (17). Most of the cases among the newly infected were in the age group 5-34 years. The present study has shown that a substantial proportion (about 1/3) of cases may be cured with very little or no antituberculosis treatment. Baily et al. (18) have shown that some cases are cured even with a short course of treatment. These findings indicate that tuberculosis cases are not a uniform entity. There can be different gradations from the point of view of diagnosis and ability to benefit from treatment. The differences between male and female patients with regard to death and cure rates support this view. The prevalence and incidence of infection showed a significant decrease during the 5 years in the age groups 0-24 and 0-34 years, respectively. At each survey, the infection rate among the newborn, who were tested for the first time at that survey, also showed a decline. The rates were 0.56%, 0.37%, 0.21 %, and 0.18% for surveys I-IV, respectively. The overall prevalence of disease also decreased from the first to the third survey, but it showed a slight increase at the fourth. However, the preva- lence at the fourth survey was below that at the 485 NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE first survey. The incidence of disease was highest between the first and second surveys and lowest between the second and third surveys. The decline in both the prevalence and the incidence of disease was significant in the age group 5-34 years. Pamra et al. (8) reported a decline in the prevalence of dis- ease from 400 to 210 per 100 000 over a period of 8 years in an urban area with organized antituber- culosis measures. Similarly J. Frimodt-M0ller (un- published observations, 1973) also reported a decline in disease rates in rural areas with antituberculosis measures. During the 5-year period, the various rates were observed to decline in different age groups-i.e., the prevalence of infection in the age group 0-24 years, the incidence of infection in all age groups, and the prevalence and incidence of disease in the age group 5-34 years. Any comparison of the degree of decline in the four rates for all ages together may not be meaningful. Moreover, there are inherent difficulties in comparing the degrees of decline in the preva- lence and incidence rates. For example, the former are calculated for different points of time, whereas the latter refer to fairly long intervals of 11/2 or 2 years; the denominators for these rates differ; and the numerator of a prevalence rate includes the earlier prevalence and current incidence in vary- ing proportions depending on the age and on whether infection or disease is the criterion. Thus, it is not surprising that, over a brief period-5 years-the overall prevalence of infection did not differ much when the overall incidence of infection showed a sharp fall. In the age group 0-4 years, the above- mentioned limitations are considerably smaller (though not completely eliminated). Therefore, the rate of decline in the prevalence of infection (1000 per year) approaches that for the incidence of infec- tion (800 per year). However, these high rates of decline may not persist for long. It is also possible that, within a brief period, the different rates may not all change uniformly at the same time. This hypothesis is supported by the findings of Pamra et al. (8) that, whereas the prevalence of disease showed a sharp fall of 500 in 21/2 years between their third and fourth surveys, the incidence of dis- ease did not decrease. It is known that when tuberculosis declines in a community the reduction in the prevalence and inci- dence of infection and disease is first seen in the younger age groups (19, 20). This phenomenon of decline in younger age groups was also observed in the present study. It could be hypothesised, on the basis of observations, that tuberculosis in the study area is following a slow downward course with occasional spurts influenced by environmental fac- tors. Such spurts could be observed from the fig- ures presented by Stephen-Hall (20), Pagel et al. (21), and Waaler (22) on the trend of tuberculosis in coun- tries where the disease has been on the decline. But it is difficult to extrapolate the overall decline in the trend in tuberculosis observed in the study area, because the period of observation was too short and drought may have interfered with the natural trend. Such interference has been reported also by Garai & Sen (23). Another survey, carried out much later- for example, 12-15 years after the initial sur- vey-would be needed for definite evidence of the long-term trend. The ratio between the incidence and prevalence of infection for the age group 0-14 years was 1: 9 (1: 7 for females and 1: 10 for males). The ratio between the prevalence of disease and of infection was 1: 80, compared with 1: 93 observed from the data of Raj Narain et al. (5) and 1: 104 from the data of Frimodt-M0ller (24). That ratio among males (1: 59) was twice as high as among females (1: 114). It was 1: 190 for the age group 5-14 years and gradually increased to 1: 52 for persons aged 55 years or more. The ratio between the prevalence of disease and the incidence of infection was 1: 4. It can be assumed that 1 case results in the infection of 4 previously noninfected persons per year. The ratio between the incidence and the prevalence of disease was 1: 4 among males and 1: 3 among females. It had been observed earlier that the prevalence and incidence were greater among males than among females. Therefore, a higher inci- dence in relation to prevalence of disease in females implies a quicker turnover of cases compared with males. The ratio between incidence and prevalence rates of disease was 1: 2 for the age group 5-34 years and 1: 5 for the age group 35 years or more, which suggests that the turnover was more rapid in younger persons. The incidence cases showed a natural cure rate of 520% and a mortality of 14% over the immediate observation period of 11/2 years. Prevalence cases (a mixture of incidence and left-over cases) showed a cure rate of 390% and a mortality of 170% during the same period. This finding is in keeping with the general belief that incidence cases have a more favourable prognosis. In the initial phase of the disease, the cure and death rates were higher (Fig. 2). 486 TUBERCULOSIS Wide variations in incidence rates were noted among the 9 epidemiological groups formed on the basis of tuberculin test indurations and radiological status. This raises the possibility of adding some important epidemiological classes to epidemetric models (25, 26, 27). Furthermore, for tuberculosis case-finding, subsequent examination of persons in the AB and CD groups, constituting 11% of the population, can reveal only 57% of the total inci- dence of disease. These cases can be identified only by examining the entire population-an expensive and time-consuming procedure that may not be justifiable in many countries. On the other hand, examination of persons with a tuberculin induration of 20 mm or more, who constitute 16% of the popu- lation, can reveal 61% of the incidence cases. The identification of such persons is cheaper and more practicable. ACKNOWLEDGEMENTS The credit for the success of the study goes to the epidemiological survey teams, without whose untiring efforts it would not have been possible to carry out this work; to the X-ray technicians and laboratory technicians for their meticulous work; to the statistical personnel for their painstaking care in scrutinizing, processing, and analysing the data; to the transport and administration staff for their regular and timely assistance in the conduct of the study; to the stenographers for carrying out the voluminous typing work; and to the BCG assessment team for training tuberculin testers and readers. Among those mentioned above, the following deserve special mention: Mr G. Ramanatha Rao, Mr V. V. Krishna Murthy, Mrs M. J. Jayalakshmi, Mr N. Srikan- taramu, Mr R. Channabasaviah, Mr V. Jaigopal, Mr R. P. Bhatnagar, Mr D. E. Johnson, Mr Narasimha Rao, Mr Rajan Mathew, Mr Lakshminarayana Rao, Miss Padmalatha Krishnan, Miss K. R. Prameela, and Miss C. S. Indira Bai. Valuable comments and suggestions were received from the members of the Technical Co-ordination Committee of the Institute, particularly Dr G. V. J. Baily and Dr Kul Bhushan, and from Dr K. Styblo, Dr S. Grzy- bowski, Dr J. Guld, and Dr K. L. Hitze (WHO con- sultants and officers). The study could not have been undertaken without the assistance of WHO and UNICEF. RESUME LA TUBERCULOSE DANS UNE POPULATION RURALE DE L'INDE ME'RIDIONALE: ETUDE EPIDEMIOLOGIQUE DE CINQ ANS En vue d'etudier l'epidemiologie de la tuberculose dans une region ou aucune lutte antituberculeuse n'etait men6e, on a examin6 une population rurale de 65 000 personnes dans le sud de l'Inde a 4 reprises entre 1961 et 1968. Des 6preuves tuberculiniques, des examens radiologiques et des examens de crachats ont ete pratiques. L'intervalle entre la Ire et la 4e enqu8te a 6te de 5 ans, et plus de 80% de la population a et6 examine chaque fois. La composition par age et par sexe de l'echantillon a subi quelques modifications lors des differentes enquetes. La pr6valence de l'infection tuberculeuse etait d'environ 30% (sexe f6minin: 25%; sexe masculin: 35%Y.). Elle augmentait avec l'age jusqu'A 55 ans. Les taux globaux sont restes relativement inchanges lors des enquetes successives mais on a note un declin continu de la prevalence dans le groupe d'age 0-24 ans. L'incidence annuelle de l'infection etait en moyenne de 1 %, les taux augmentant avec I'age. Pendant la duree de l'etude, l'incidence globale a decru de 1,63% A 0,8 %. Le taux de prevalence de la maladie a varie de 337 a 406 pour 100000 pendant la duree de l'etude; il etait maximal lors de la lre enquete et le plus faible lors de la 3e. La proportion des cas dans le sexe feminin etait de 1 pour 3 dans le sexe masculin, et 500% de l'ensemble des malades etaient des hommes de 35 ans et plus. Les taux s'accroissaient avec l'age surtout chez les hommes. Dans les groupes d'age 5-34 ans, les taux de prevalence de la tuberculose ont diminue continuellement pendant la duree de l'etude. Le taux d'incidence annuelle de la maladie a vari6 de 79 A 132 pour 100000, avec un maximum entre la lre et la 2e enquete et un minimum entre la 2e et la 3e. Les taux augmentaient avec l'age et etaient pres de deux fois plus eleves dans le sexe masculin. L'incidence, chez les sujets de moins de 35 ans, a diminue pendant la duree de l'6tude. Le taux annuel d'incidence a et6 le plus eleve (522 pour 100000) chez les sujets presentant une reaction 487 488 NATIONAL TUBERCULOSIS INSTITUTE, BANGALORE de 20 mm ou plus a l'epreuve tuberculinique. Ce groupe, correspondant a 16% de l'ensemble de la population, a foumi 61 % des nouveaux cas. Dans les trois groupes radiologiques consid6res au sein de la population (normal; anomalies de type tuber- culeux, mais inactives, et anomalies non tuberculeuses; images de tuberculose active ou probablement active, mais abacillaire), les taux annuels d'incidence de la tuberculose bact6riologiquement confirm6e ont ete respectivement de 165, 958 et 4530 pour 100 000. Le 3e groupe, correspondant a 1% de l'ensemble de la population, a fourni 34% des nouveaux cas. Dans chacun des trois groupes radiologiques precites, l'incidence a ete la plus 6levee chez les sujets reagissant par une induration de 20 mm ou plus a la tuberculine. Dans le 3e groupe, les sujets de cette categorie, represen- tant 0,45% de l'ensemble de la population, ont fourni 27% de l'ensemble des nouveaux cas. Le taux de prevalence des cas resistant a l'isoniazide a vari6 de 11,2% (lre enquete) a 27,2% (3e enquete). Parmi les nouveaux cas, la proportion des r6sistances a ce m6dicament a pass6 de 11,4% a 6,5% pendant la duree de 1'etude. Sur 126 patients r6examin6s a 3 reprises, on a enre- gistr6 49,2% de deces, 32,5% de guerisons, tandis que 18,3% des malades continuaient a excr6ter des bacilles. Les d&ecs et les guerisons ont ete les plus nombreux pendant la premiere periode de 11/2 an. Parmi les 63 nouveaux cas enregistres entre la Ire et la 2e enquete, et revus lors de la 3e, on a not6 52,4% de guerisons, 14,3% de deces, tandis que 33,3% des malades continuaient a excreter des bacilles. Les diff6rences constatees dans le d6clin des divers taux, pendant les 5 ans de l'6tude, n'autorisent aucune extrapolation au-dela de cette periode qui est trop courte pour qu'on puisse juger des tendances de la tuberculose. On peut cependant emettre l'hypothese que la maladie est probablement en regression dans la region oiu ont eu lieu les enquetes. REFERENCES 1. UKIL, A. C. Indian journal of medical research, 17 (January) (1930). 2. BENJAMIN, P. V. Indian medical gazette, 73: 540 (1938). 3. SIKAND, B. K. & RAJ NARAIN. In: Proceedings of the 9th Tuberculosis and Chest Diseases Workers' Conference, Lucknow. New Delhi, Tuberculosis Association of India, 1952, p. 106. 4. INDIAN COUNCIL OF MEDICAL RESEARCH. A national sample survey, 1959 (ICMR Special report No. 34). 5. RAJ NARAIN ET AL. Indian journal oftuberculosis, 10: 85 (1963). 6. RAJ NARAIN ET AL. Bulletin of the World Health Organization, 29: 641 (1963). 7. PAMRA, S. P. ET AL. In: Proceedings ofthe 23rd Tuber- culosis and Chest Diseases Workers' Conference, Bombay. New Delhi, Tuberculosis Association of India, 1968, p. 61. 8. PAMRA, S. P. ET AL. Indian journal of tuberculosis, 20: 57 (1973). 9. ANDERSEN, S. & BANERJI, D. Population review, 6: 69 (1962). 10. RAJ NARAIN ET AL. Indian journal of tuberculosis, 13: 43 (1966). 11. RAJ NARAmIN ET AL. Bulletin of the World Health Organization, 34: 605 (1966). 12. HALD, A. Statistical theory with engineering appli- cations. London, Chapman & Hall, 1957, p. 144. 13. HORwITz, 0. & KNUDSEN, J. Bulletin of the World Health Organization, 23: 669 (1960). 14. FRIMODT-M0LLER, J. Bulletin of the World Health Organization, 22: 111 (1960). 15. PALMER, C. E. ET AL. American review of respiratory diseases, 76: 514 (1957). 16. GROTH-PETERSEN, E. Bulletin of the World Health Organization, 21: 5 (1959). 17. TRAUGER, D. A. American review of respiratory dis- eases, 87: 582 (1963). 18. BAILY, G. V. J. ET AL. Indian journal of tuberculosis, 21: 153 (1973). 19. GRIGG, E. R. N. ET AL. American review of respira- tory diseases, 78: 151, 426, & 583 (1958). 20. STEPHEN-HALL, A. In: Heaf, F. R. G., ed. Sym- posium on tuberculosis. London, Cassell, 1957, p. 67. 21. PAGEL, W. ET AL. Pulmonary tuberculosis, 4th ed. London, Oxford University Press, 1964, p. 469. 22. WAALER, H. T. The risk of tuberculous infection in Norway. Oslo, Statens Skjermbildefotografering, 1972. 23. GARAI, R. & SEN, P. K. Indian journal oftuberculosis, 20: 125 (1973). 24. FRIMODT-M0LLER, J. In: Proceedings of the 17th Tuberculosis and Chest Diseases Workers' Confer- ence, Cuttack. New Delhi, Tuberculosis Association of India, 1961, p. 19. 25. WAALER, H. T. ET AL. American journal of public health, 52:1001 (1962). 26. WAALER, H. T. & PIOT, M. A. Bulletin of the World Health Organization, 41: 75 (1969). 27. REVELLE, C. S. ET AL. American review of respiratory diseases, 96: 893 (1967).