Global tuberculosis incidence and mortality during 1990-2000.

Research I Recherche Global tuberculosis incidence and mortality during 1990-2000 P.J. Dolin,1 M.C. Raviglione,2 & A. Kochi2 Forecasts of tuberculosis morbidity and mortality are presented for the decade 1990-99. An estimated 88 million new cases of tuberculosis, of which 8 million will be attributable to HIV infection, will occur in the world during the decade; 30 million people are predicted to die of tuberculosis in the same period, including 2.9 million attributable to HIV infection. The number of new tuberculosis cases occurring each year is predicted to increase from 7.5 million (143 cases per 100 000) in 1990 to 8.8 million (152 per 100 000) in 1995 and 10.2 million (163 per 100 000) in the year 2000. In 1990, 2.5 million persons were estimated to have died of tuberculosis; at the same level of availability of treatment, it is predicted that 3.0 million tuberculosis deaths will occur in 1995 and 3.5 million in 2000. Demographic factors, such as population growth and changes in the age structure of populations, will account for 79.5% of the predicted increases in new cases. Age-specific incidence rates in sub- Saharan Africa are increasing due to the HIV epidemic and will account for the remaining 20.5% of the forecast increase in new cases. In WHO's South-East Asian Region and in Central and South America the age-specific incidence rates are expected to fall during 1990-2000, but at a slower rate than in previous years because of the expected increase in HIV seroprevalence. In the Western Pacific and Eastern Mediterranean Regions the age-specific incidence rates are expected to fall during 1990- 2000 because of the effects of intervention strategies, but the total number of new cases will continue to increase until the end of the decade because of population growth. Estimates of tuberculosis incidence in 1990 for the developing countries by Murray (1) were extended by Sudre (2) to include the industrialized countries. Both authors used estimates of the annual risk of infection (ARI) in developing countriesa to calculate the expected incidence. The present study forecasts the future global and regional burden of tuberculosis morbidity and mortality using tuberculosis notifica- tion data (i.e., cases reported to the Ministries of Health) to complement the ARI-based calculations. This dual approach was used to provide a separate set of estimates from those based on ARI data alone. I Imperial Cancer Research Fund, Cancer Epidemiology Unit, Radcliffe Infirmary, University of Oxford, Oxford, England. 2 Tuberculosis Programme, World Health Organization, 1211 Geneva 27, Switzerland. Requests for reprints should be sent to Dr Raviglione at this address. a Cauthen GM, Plo A, ten Dam HG. Annual risk of tuberculosis infection (unpublished document WHO/TB/88.154, 1988). This document is available upon request to the Tuberculosis Pro- gramme, WHO, Geneva, Switzerland. Reprint No. 5470 The impact of demographic factors (such as global population growth and aging of the world's popula- tions) and epidemiological factors (such as adverse effects of the human immunodeficiency virus (HIV) epidemic and the beneficial effects of intervention programmes) is discussed. Methods Tuberculosis incidence In 1990. WHO routinely collects data on the number of tuberculosis cases in Member States each year.b Within each WHO region, except the African region, an overall regional crude incidence rate was calcula- ted by estimating the incidence in the most populated countries. Notification data were considered reliable when provided by programmes with an established b World Health Organization. Tuberculosis notification update, July 1992 (unpublished document WHO/TUB/92.169, 1992). This document is available upon request to the Tuberculosis Pro- gramme, WHO, Geneva, Switzerland. Bulletin of the World Health Organization, 1994, 72 (2): 213-220 © World Health Organization 1994 213 P.J. Dolin et al. surveillance system. Reliable notification data were preferentially used for these estimates. For countries with unreliable notification data, the annual risk of infection was used to estimate incidence. Notifica- tion data are relatively poor for the African region (sub-Saharan Africa), so a slightly different approach was used; the region was divided into four geogra- phical areas and within each area, a crude incidence rate was estimated, based on the most reliable notifi- cation data (e.g., United Republic of Tanzania for East Africa, South Africa for southern Africa, C6te d'Ivoire for West Africa). This approach differs from previous estimates of the 1990 incidence (1, 2), which were based solely on annual risk-of-infection dataa and gave equal consid- eration to data from countries with different popula- tion sizes. While notification data are of poor quality for many countries, and any estimates based on such data will risk underestimating the incidence, reliable data are available from other countries, particularly those where good tuberculosis control programmes are established. The estimates presented in this paper must be considered conservative owing to the fact that tuberculosis cases are generally underreported. In 1995 and 2000. In estimating future incidence, allowances were made for demographic factors (changes in the size and age structure of regional populations) and epidemiological factors (changes in underlying incidence rates). To accurately allow for both demographic and epidemiological factors, regional age-specific incidence rates for the years 1995 and 2000 were estimated and then applied to regional age-specific population projections for 1995 and 2000. This was undertaken in two steps. First, data available at WHO on the age distribution of notified cases in each region during 1990 were applied to the 1990 regional crude incidence rates to derive the 1990 regional age-specific incidence rates. Second, trends in regional notification rates during 1985-90 were applied to the 1990 regional age-specific inci- dence rates to derive estimates of regional age- specific incidence rates for the years 1995 and 2000. This assumes that future age-specific trends will remain unchanged. These rates were then applied to regional age-specific population projections (3) to calculate the number of incident cases expected in 1995 and 2000. A more detailed description of these methods is reviewed elsewhere.c c Dolin PJ, Raviglione MC, Kochi A. A review of current epi- demiological data and estimation of future incidence and mortal- ity from tuberculosis. This document (WHO/TB/93.173) is avail- able upon request to the Tuberculosis Programme, WHO, Geneva, Switzerland. Mortality in 1990, 1995 and 2000 Published case-fatality rates of 7% for industrialized countries (4) and reported rates of 15% for Eastern Europe were used. These deaths probably resulted from late presentation for treatment and failure to diagnose tuberculosis as the underlying disease. For Central and South America around 20% of tubercu- losis cases were assumed to die of the disease. For other regions, case-fatality rates of 35-40% were estimated, based on a case-fatality rate of 15% for cases receiving treatment and 55% for those not treated (1, 5-7). It was further assumed that all cases notified to WHO were treated, that 5% of treated patients are not reported, and that the proportion of cases receiving treatment remains at the 1990 level. In the South-East Asian and Western Pacific regions around half of all cases are estimated to have received treatment in 1990. HIV-attributable tuberculosis Current data suggest that between 5% and 10% of persons co-infected with HIV and Mycobacterium tuberculosis will develop tuberculosis each year, compared with less than 0.2% of persons infected with M. tuberculosis but not HIV (8, 9). From these data, we estimate that around 95% of HIV-infected tuberculosis cases are attributable to HIV infection (i.e., attributable risk = 95%), and the remaining 5% of co-infected cases would have developed tubercu- losis regardless of their HIV status. This attributable risk was applied to estimates of HIV seroprevalence among patients with tuberculosis (9) to derive the number of HIV-attributable tuberculosis cases (i.e., population attributable risk) in 1990, 1995 and 2000. The numbers of tuberculosis deaths attributed to HIV infection were estimated by applying the regional case-fatality rates to the estimated number of HIV- attributable cases. Total incidence and mortality during 1990-99 The total number of new cases and deaths in the 10- year period 1990-99 was calculated using the 1990 estimates for the period 1990-mid-92, the 1995 esti- mates for mid-1992-mid-97, and the 2000 estimates for the period mid-1997-1999. Results Tuberculosis incidence In 1990. It is estimated that there were 7 537 000 incident cases of tuberculosis in 1990 (Table 1). Over 4.9 million cases (65%) occurred in the South- East Asian and Western Pacific regions, including 214 WHO Bulletin OMS. Vol 72 1994 Global tuberculosis Incidence and mortality Table 1: Estimated tuberculosis (TB) Incidence and HIV-attributable tuberculosis cases In 1990, 1995 and 2000, by region 1990 1995 2000 Total HIV-attributed Total HIV-attributed Total HIV-attributed Region TB cases Ratea TB cases TB cases Ratea TB cases TB cases Ratea TB cases South-East Asia 3 106 000 237 66 000 3 499 000 241 251 000 3 952 000 247 571 000 Western Pacificb 1 839 000 136 19 000 2 045 000 140 31 000 2 255 000 144 68 000 Africa 992 000 191 194 000 1 467 000 242 380 000 2 079 000 293 604 000 Eastern Mediterranean 641 000 165 9 000 745 000 168 16 000 870 000 168 38 000 Americasc 569 000 127 20 000 606 000 123 45 000 645 000 120 97 000 Eastern Europed 194 000 47 1 000 202 000 47 2 000 210 000 48 6 000 Industrialized countries" 196 000 23 6 000 204 000 23 13 000 211 000 24 26 000 Total 7 537 000 143 315 000 8 768 000 152 738 000 10 222 000 163 1 410 000 (4.2%) (8.4%) (13.8%) Increase since 1990 16.3% 35.6% a Crude incidence rate per 100 000 population. b Includes all countries of the Western Pacific Region of WHO, except Japan, Australia and New Zealand. c Includes all countries of the American Region of WHO, except USA and Canada. d Eastern European countries, and independent states of the former USSR. e Western European countries, USA, Canada, Japan, Australia and New Zealand. 2.1 million in India, 1.3 million in China, and 0.4 million in Indonesia. One million cases are estimated to have occurred in Sub-Saharan Africa, 0.2 million cases in Eastern Europe and independent states of the former USSR, and 0.2 million cases in Western Europe and other industrialized countries. In 1995 and 2000. The predicted number of incident cases of tuberculosis in 1995 and 2000 is shown in Table 1. Global incidence is predicted to increase from 7.5 million new cases annually in 1990 to 10.2 million new cases by the year 2000, an increase of 36%. Over 3.9 million new cases annually are expected in South- East Asia, 2.3 million in the Western Pacific, and 2.1 million in sub-Saharan Africa by 2000. Table 2 shows the age-specific incidence in each region for the years 1990 and 2000; 70% of new cases are aged between 15 and 59 years (when people are economically most productive), 20% at least 60 years, and 10% under 15 years. In the Western Pacific and Eastern Mediterra- nean regions, small decreases in age-specific inciden- ce rates are expected. However, the actual number of new cases in each age group is forecast to increase due to population growth. In addition, the crude inci- dence rates (all ages) for these regions are also fore- cast to increase, even with decreasing age-specific rates, because of the demographic aging of regional populations. In sub-Saharan Africa, age-specific rates are expected to increase during the decade. In spite of the effects of some national tuberculosis programmes, the incidence of tuberculosis in sub-Saharan Africa is rapidly increasing because of the HIV epidemic. The number of new cases per year in sub-Saharan Africa is predicted to double by the year 2000. For South-East Asia and Central and South America, age-specific and crude incidence rates are expected to fall during 1990-2000. However, the rate of decline is slower than in previous years because of the expected spread of the HIV epidemic. The actual number of new cases will continue to increase because of population growth. For Central and South America, it is estimated that the effects of falling age-specific incidence rates will counter pop- ulation growth by around the year 2005, at which point the number of new cases in Central and South America should fall. In Eastern Europe and states of the former USSR, notifications fell during 1985-90. However, it is not clear that this decline in incidence rates can be maintained under current social conditions in these countries. It is predicted that tuberculosis inci- dence rates in Eastern Europe and states of the for- mer USSR will remain at their 1990 level in the near future. Notification rates for some Western European and other industrialized countries (e.g., Netherlands, Norway and USA) have recently increased after years of decline (4). Rates in other Western Europe- an and industrialized countries may also increase in the near future, partly because of increasing migra- tion of people from regions of the world with a high- er incidence of tuberculosis, and partly because of the increasing number of persons with dual HIV and M. tuberculosis infection. It is estimated that tuber- culosis incidence rates in Western Europe and the WHO Bulletin OMS. Vol 72 1994 215 P.J. Dolin et al. Table 2: Estimated total tuberculosis Incidence In 1990 and 2000, by region and age group 0-14 years 15-34 years 35-59 years .60 years All ages Region and year Cases Ratea Cases Ratea Cases Rate" Cases Ratea Cases Rateb South-East Asia: 1990 155 000 31 932 000 207 1 398 000 515 621 000 740 3 106 000 237 2000 175 000 30 1 126 000 206 1 835 000 514 816 000 739 3 952 000 247 Western Pacific (excluding Japan, Australia and New Zealand): 1990 248 000 70 423 000 80 579 000 166 589 000 487 1 839 000 136 2000 274 000 68 420 000 79 776 000 165 785 000 486 2 255 000 144 Africa: 1990 99 000 42 447 000 266 347 000 380 99 000 398 992 000 191 2000 447 000 142 857 000 366 606 000 480 169 000 498 2 079 000 293 Eastem Mediterranean: 1990 64 000 41 224 000 170 257 000 346 96 000 435 641 000 165 2000 67 000 40 294 000 164 372 000 340 137 000 429 870 000 167 Americas: 1990 57 000 35 199 000 125 228 000 240 85 000 272 569 000 127 2000 36 000 20 206 000 110 295 000 225 108 000 257 645 000 120 Eastern Europe and former USSR: 1990 19 000 20 58 000 45 68 000 56 49 000 75 194 000 47 2000 19 000 20 55 000 45 81 000 56 55 000 75 210 000 48 Industrialized countries:c 1990 20 000 10 59 000 22 68 000 27 49 000 37 196 000 23 2000 19 000 10 55 000 22 82 000 27 55 000 37 211 000 24 a Age-specific incidence rate per 100 000 population. b Crude incidence rate per 100 000 population. c Western Europe, USA, Canada, Japan, Australia and New Zealand. other industrialized countries will either remain at their current levels or increase slightly during the decade. For both Eastem and Western Europe, the number of new cases per year is expected to increase because of population growth. HIV-attributable tuberculosis It is estimated that 0.3 million (4%) of the 7.5 mil- lion new tuberculosis cases in 1990 were attributable to HIV infection (Table 1). Around 0.2 million of the HIV-attributable cases occurred in sub-Saharan Afri- ca. By the year 2000, 1.4 million (14%) of the 10.2 million new cases occurring each year will be attri- butable to HIV infection. Around 40% of these HIV attributable cases will occur in sub-Saharan Africa and a further 40% in South-East Asia. A more detai- led analysis of the impact of the HIV epidemic on tuberculosis incidence and mortality is presented elsewhere.d Impact of demographic factors on incidence Table 3 shows the predicted number of additional cases of tuberculosis in 1995 and 2000, compared with the 1990 incidence, due to demographic and epidemiological factors. Over three-quarters of the dSee footnote c on p. 214. predicted increase in incidence will result from demographic factors, such as population growth and changing age structure of populations. Less than 25% of the predicted increase will result from changes in underlying incidence rates. In sub-Saharan Africa, epidemiological factors (rising incidence rates due to HIV epidemic) are stronger than demographic fac- tors. For the South East Asian, Western Pacific and Eastern Mediterranean regions, and Central and South America, the decreases in age-specific rates will yield a reduction in new cases, but this will be outweighed by additional new cases due to popula- tion growth. Total incidence during 1990-99 For the 10-year period 1990-99, it is estimated that 88.2 million people will develop tuberculosis, 8.0 million of which will be attributable to HIV infec- tion. In South-East Asia, 35.1 million new cases of tuberculosis are expected during the decade, includ- ing 2.8 million (8%) HIV-attributable cases (Fig. 1). Around 20.5 million new cases are predicted for the Western Pacific region, including 0.4 million (2%) HIV-attributable cases, while 15.0 million cases are expected in sub-Saharan Africa during the decade, including 3.9 million (25%) cases attributable to HIV infection. 216 WHO Bulletin OMS. Vol 72 1994 Global tuberculosis Incidence and mortality Table 3: Estimated additional cases of tuberculosis In 1995 and 2000, compared with 1990, by region, attributable to changes In demographic factors (population growth and changing age distribution) and epidemiological factors (changing Incidence rates) 1995 2000 Region Demographic Epidemiologic Demographic Epidemiologic South-East Asia 399 000 -6 000 860 000 -14 000 Western Pacifica 218 000 -12 000 442 000 -26 000 Afrca 171 000 304 000 378 000 709 000 Eastern Mediterranean 119 000 -15 000 266 000 -37 000 Americasb 74 000 -37 000 157 000 -81 000 Eastern Europec 8 000 0 16 000 0 Industrialized countriesd 8 000 0 15 000 0 All regions 997 000 234 000 2 134 000 551 000 (81.0%) (19.0%) (79.5%) (20.5%) a Excluding Japan, Australia and New Zealand. b Excluding USA and Canada. c Eastern Europe and independent states of former USSR. d Western Europe, USA, Canada, Japan, Australia and New Zealand. Mortality from tuberculosis The estimated number of tuberculosis deaths in 1990, 1995 and 2000 is shown in Table 4. Some 2 530 000 tuberculosis deaths occurred in 1990, including 1.1 million in the South-East Asian region and 0.6 million in the Western Pacific region. Globally, 116000 tuberculosis deaths (4.6%) were attributable to HIV infection in 1990, most of which occurred in sub-Saharan Africa. By the year 2000 it is predicted, assuming the proportion of cases receiving treatment remains at the 1990 level, that 3 509 000 tuberculosis deaths will occur annually, 39% more than in 1990. In South- East Asia, 1.4 million deaths annually are anticipated by the year 2000. Globally, 0.5 million tuberculosis deaths per year will be attributable to HIV infection by then. The HIV-related deaths will occur mainly in sub-Saharan Africa and South-East Asia. Total deaths during 1990-99. During the 10-year period 1990-99 it is estimated that 30.0 million people will die of tuberculosis; 2.9 million of these deaths (9.7%) will be attributed to HIV infection (Fig. 2). In South-East Asia, 12.3 million tuberculo- sis deaths will occur during the decade, including 1.0 million HIV-attributable deaths. Around 6.0 million tuberculosis deaths are expected in sub-Saharan Afri- ca, 1.5 million (25%) of which will be attributed to HIV infection. Discussion Nearly 90 million new tuberculosis cases and 30 mil- lion tuberculosis deaths are expected to occur during the present decade at the present level of interven- tions. For a disease where intervention is known to be cost-effective (10), this is truly staggering. These estimates are based on notification data and, because of underreporting of tuberculosis cases, must be considered conservative. This is reflected in our estimated 1990 incidence of 7.5 million new cases being slightly lower than previous estimates, which were based on annual risk-of-infection data (2). Similarly, the estimates of tuberculosis mortality should be considered conservative, e.g., the estimated 2.5 million tuberculosis deaths in 1990, compared with previous estimates of 2.9 million deaths (2). While the exact number of new cases and deaths is not known, the current and previous estimates are consistent in suggesting that between 7.5 and 8.0 million new cases and 2.5-3.0 million tuberculosis deaths occurred in 1990. Current intervention strategies are expected to result in substantial reductions in age-specific inci- dence rates in the Eastern Mediterranean region and Central and South America, and to a lesser degree in the Western Pacific and South-East Asian regions. However, the total number of new cases in these regions is predicted to increase in the near future because of population growth. The impact of the HIV epidemic is most evident in sub-Saharan Africa where the number of new cases per year is forecast to double by the end of the decade. In South-East Asia and other regions there has been little impact, to date, of the HIV epidemic on tuberculosis. However, by the year 2000 over 500 000 new cases and 200 000 deaths in South-East Asia will be attributable to HIV infection. A number of assumptions were made in these analyses. It was estimated that 5% of all treated WHO Bulletin OMS. Vol 72 1994 217 P.J. Dolin et al. Fig. 1. World map showing estimated cumulative tuberculosis cases, 1990-99. Total cases 88 MILLION Table 4: Estimated total tuberculosis deaths and HIV-attributable tuberculosis deaths in region, assuming that regional treatment coverage rates remain at their 1990 level 1990, 1995 and 2000, by Deaths in 1990 Deaths in 1995 Deaths in 2000 Attributed Attributed Attributed Region Total to HIV Total to HIV Total to HIV South-East Asia 1 087 000 23 000 1 225 000 88 000 1 383 000 200 000 Western Pacifica 644 000 7 000 716 000 11 000 789 000 24 000 Africa 393 000 77 000 581 000 150 000 823 000 239 000 Eastern Mediterranean 249 000 4 000 290 000 6 000 338 000 15 000 Americasb 114 000 4 000 121 000 9 000 129 000 19 000 Eastern Europec 29 000 <200 30 000 <600 32 000 <900 Industrialized countriesd 14 000 <500 14 000 1 000 15 000 2 000 All regions 2 530 000 116 000 2 977 000 266 000 3 509 000 500 000 (4.6%) (8.9%) (14.2%) Increase since 1990 17.7% 38.7% a Excluding Japan, Australia and New Zealand. b Excluding USA and Canada. c Eastern Europe and independent states of former USSR. d Western Europe, USA, Canada, Japan, Australia and New Zealand. WHO Bulletin OMS. Vol 72 1994218 Global tuberculosis incidence and mortality Fig. 2. World map showing estimated cumulative tuberculosis deaths, 1990-99. Eastern Europe 302,000 South & South-East Asia ,Japan 42,000 Total deaths 30 MILLION PO- cases are not reported to WHO and that 100% of reported cases were treated. Limited global data are available on the completeness and quality of notifi- cations. These levels were chosen as conservative estimates of the global situation. Earlier mortality estimates (2) used a case-fatality rate of 50% for HIV-positive tuberculosis cases, whereas the current estimates did not assume that mortality was different between HIV-positive and HIV-negative cases. Forecasting future incidence and mortality is difficult and can only be based on data available at the time of the modelling. Substantial changes in epi- demiological factors, such as greater than expected increases in the seroprevalence of HIV among per- sons infected with M. tuberculosis, would increase the future burden of disease. Conversely, increased availability of treatment would reduce the forecast number of future cases and deaths. It has been demonstrated that effective applica- tion of short-course chemotherapy in well-managed national tuberculosis programmes produces excellent results, even under the most adverse conditions (11). Short-term chemotherapy of smear-positive tubercu- losis cases is one of the most cost-effective health interventions available (10). A higher priority must be given to this disease, both by the countries most severely affected and by donor countries which invest in health care programmes in those countries. Acknowledgements We thank Mr R. Bumgarner, Dr M. Felten, Dr P. Graf, Dr P. Nunn, Dr R. O'Brien, Dr S. Spinaci, Dr B. Vareldzis, and Mrs D. Weil from the WHO Tuberculosis Programme, Geneva, Switzerland, Dr Peter Smith from the London School of Hygiene and Tropical Medicine, London, England, and Dr K. Styblo from the Tuberculosis Sur- veillance Research Unit, the Hague, Netherlands, for their useful suggestions and comments on this paper. WHO Bulletin OMS. Vol 72 1994 219 P.J. Dolin et al. Rdsum6 Incidence mondiale de la tuberculose et mortalit de 1990 A 2000 Des pr6visions de la morbidite et de la mortalit6 dues a la tuberculose sont presentees pour les annees 1990-99. On estime que, pendant cette decennie, 88,2 millions de nouveaux cas de tuber- culose seront enregistres dans le monde, dont 8,0 millions imputables a l'infection a VIH; il est prevu que 30 millions de personnes mourront de la tuberculose pendant la meme periode, dont 2,9 millions par suite de l'infection a VIH. On prevoit que le nombre des nouveaux cas de tuberculose se produisant chaque annee pas- sera de 7,5 millions (143 cas pour 100000 per- sonnes) en 1990 a 8,8 millions (152 pour 100 000) en 1995 et 10,2 millions (163 pour 100 000) en 2000. D'apres les estimations, 2,5 millions de per- sonnes sont mortes de la tuberculose, en 1990; si les traitements disponibles restent au meme niveau, il est prevu que 3,0 millions de d6ces par tuberculo- se auront lieu en 1995 et 3,5 millions en 2000. L'accroissement prevu du nombre de cas nouveaux sera dO pour 77% a des facteurs demo- graphiques, par exemple I'augmentation de la population et l'evolution de la pyramide des ages. Les taux d'incidence specifiques de l'age sont en voie d'accroissement en Afrique subsaharienne par suite de l'epid6mie d'infection a VIH et ils repr6senteront les 23% restants en ce qui concer- ne I'augmentation prevue des cas nouveaux. Dans la R6gion OMS de l'Asie du Sud-Est, ainsi qu'en Am6rique centrale et en Amerique du Sud, on pense que les taux d'incidence specifiques de l'age baisseront pendant la periode 1990-2000, mais plus lentement que les annees pr6cedentes a cause de l'accroissement pr6vu de la s6ropre- valence du VIH. Dans les R6gions du Pacifique occidental et de la Mediterran6e orientale, les taux d'incidence sp6cifiques de l'age devraient baisser durant la periode 1990-2000 grace aux strategies d'intervention, mais le nombre total des cas nouveaux continuera a augmenter jusqu'a la fin de la d6cennie a cause de 1'expansion demo- graphique. References 1. Murray CJ. Health sector priorities review: tubercu- losis. In: Jamison DT, Mosley WH, eds. Disease control priorities in developing countries. New York, Oxford University Press, 1993. 2. Sudre P, ten Dam HG, Kochi A. Tuberculosis: a global overview of the situation today. Bulletin of the World Health Organization, 1992, 70: 149-159. 3. United Nations. Global estimates and projections of population by sex and age, the 1988 revision (ST/ESA/SER.R/93). New York, United Nations, 1989. 4. Raviglione MC et al. Secular trends of tuberculosis in Western Europe. Bulletin of the World Health Organization, 1993, 71: 297-306. 5. Lindhart M. The statistics of pulmonary tuberculosis in Denmark, 1925-1934. A statistical investigation on the occurrence of pulmonary tuberculosis in the period 1925-1934, worked out on the basis of the Danish National Health Service file of notified cases and deaths. Copenhagen, E. Munksgaard, 1939. 6. Humphries MJ et al. Deaths occurring in newly notified patients with pulmonary tuberculosis in Eng- land and Wales. British journal of diseases of the chest, 1984, 78: 149-158. 7. Springett VH. Ten-year results during the introduc- tion of chemotherapy for tuberculosis. Tubercle, 1971, 52: 73-87. 8. Selwyn PA et al. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. New Eng- land journal of medicine, 1989, 320: 545-550. 9. Narain JP, Raviglione MC, Kochi A. HIV-associat- ed tuberculosis in developing countries: epidemiolo- gy and strategies for prevention. Tubercle and lung disease, 1992, 73: 311-321. 10. Murray CJ et al. Cost effectiveness of chemothera- py for pulmonary tuberculosis in three sub-Saharan African countries. Lancet, 1991, 338: 1305-1308. 11. Styblo K. The impact of HIV infection on the global epidemiology of tuberculosis. Bulletin of the Interna- tional Union against Tuberculosis and Lung Dis- ease, 1991, 66: 27-32. 220 WHO Bulletin OMS. Vol 72 1994