Both HIV infection and silicosis, one of the most common industrial lung diseases world-wide, are strong risk factors for tuberculosis (TB) [1–4]. The risk of silicosis increases with increasing years of employment and cumulative exposure to silica dust, but can be minimized by dust control measures . Although the prevalence of silicosis in more affluent countries is now very low following improvements in occupational safety standards during the second half of the 20th century, the prevalence in developing countries is poorly quantified and it is likely that many workers remain at high risk [5,6]. Large industries in developing countries tend to attract migrant workers, who have been shown to be at higher risk of acquiring HIV infection than non-migrant colleagues . As such, migrant workers in industrializing countries such as South Africa are at disproportionately high risk of acquiring two of the strongest known risk factors for TB. The nature of the combined effects of these two risk factors has not previously been investigated, but will be one of the determining factors of TB incidence in any workforce where both are highly prevalent.
The South African gold mining industry employs about 300 000 miners, mainly migrant workers from rural South Africa and neighbouring countries. Most miners live in densely populated hostel accommodation based around their workplace, returning to their home areas few times a year . Gold is found in silica rich seams, and the prevalence of silicosis in South African gold miners is known to be high [8–11]. TB incidence among miners has been correspondingly high throughout this century, in the region of 600 per 100 000 miners per year , and both silicosis and TB are recognized as occupational hazards by compensation legislation. During the 1990s South Africa experienced a severe HIV epidemic . Over the same period, both the incidence of TB and prevalence of HIV infection among miners diagnosed with TB have risen progressively to the extent that annual TB case rates now exceed 2000 per 100 000 employees in some workforces, with about half being HIV-associated [14,15].
This study is a retrospective cohort analysis of TB and Mycobacterium kansasii disease incidence in South African miners, the main aim of which was to investigate the combined effects of HIV and silicosis on TB incidence. The study was designed to include as many older HIV-positive men as possible, in order to maximize the number of HIV-positive silicotics, and thus the power to investigate interactions.
Miners working for a single company in Welkom, Free State, South Africa were provided with medical care by one hospital and several peripheral clinics. The hospital provided a centralized TB diagnosis and control programme and the clinics provided primary care, including syndromic sexually transmitted disease (STD) management. HIV testing was offered to all STD patients at the peripheral clinics and all TB patients at the hospital, with informed consent and pre- and post-test counselling. The workforce size reduced from 70 000 to 50 000 employees during the period covered by the study, mainly as a consequence of the fall in the price of gold.
Miners who had their first HIV test at a primary health clinic between January 1991 and December 1996 were identified from HIV test records. A sample of 4438 men was provisionally entered into a retrospective cohort on the basis of their first HIV test result and their age. The selection procedure was based on random selection into three age groups, as summarized in Figure 1, so that a total of 3000 HIV-negative miners were randomly selected from men aged under 30 years, 30–39 years, and 40 years or older. Men who had a subsequent positive HIV test result during cohort follow-up were then excluded. Similarly, 500 HIV-positive men were randomly selected from each of the two younger age groups, together with all of the 438 known HIV-positive men aged 40 years or older. Attempts were made to identify the test indication for all HIV-positive and HIV-negative men. Men documented to have been HIV tested because of suspicion of TB, and those already on TB treatment at the start of follow-up were excluded. Cohort follow-up was started 60 days after the HIV test date, in order to further minimize the risk of inadvertently including men who were HIV tested because of symptomatic TB. Follow-up continued until a diagnosis of TB or non-tuberculous mycobacterial (NTM) disease, termination of employment, or to 1 August 1997.
Dates of birth, first employment in the South African mining industry and job details were obtained from company records and the miners’ national employment agency. The study was conducted with prior approval of the mining company concerned. Ethical approval was obtained from the Ethics Committees of the Ernest Oppenheimer Hospital, South Africa and the London School of Hygiene and Tropical Medicine.
Mini-radiographs, taken annually for routine screening purposes, were scored for rounded upper zone opacities suggestive of silicosis using a five-point scale modified from the International Labour Office (ILO) system . ‘Possible', ‘probable’ and ‘early’ silicosis grades were intended to correspond to ILO grades of 0/1, 1/0 and 1/1 respectively, with ‘advanced’ covering all higher ILO grades. Radiographs were graded independently for silicosis by two readers, without reference to personal details. Final grades were reached by consensus in cases of initial disagreement. The ILO system was designed for use with standard-sized films, but use of mini-radiographs has been validated in a separate study involving this same workforce and readers, in which the ability to distinguish silicotic from tuberculous scarring was also demonstrated .
Bacteriology and case definitions
Men with suspected mycobacterial disease were investigated to a standard protocol with three sputum specimens taken over 2 days. Slides were made from concentrated sputum and stained with auramine for fluorescent microscopy. Positive slides were confirmed with Ziehl–Neelsen (ZN) staining. Sputum was inoculated onto Lowenstein–Jensen (LJ) slopes and incubated for up to 8 weeks. From 1996 an initial identification step for Mycobacterium tuberculosis was carried out on LJ slopes with more than five colonies, using a colorimetric ribosomal RNA hybridization test (Accuprobe 1 M. tuberculosis complex probe kit, Gen-Probe, San Diego, California, USA). Otherwise, LJ slopes were sent directly to the South African Institute of Medical Research (SAIMR) mycobacteriology laboratory for biochemical species identification.
Mycobacterial disease episodes were identified from a database containing records of all cases from 1984. Case definitions for culture-negative TB were: (i) pulmonary – compatible radiographic changes plus two of the following: smear positive; no response to amoxycillin; radiological response to anti-tuberculous drugs; (ii) pleural effusion – exudate plus one of: tissue granulomas; radiological response to anti-tuberculous drugs; (iii) meningitis – cerebrospinal fluid lymphocytic pleocytosis with raised protein, low glucose and a negative cryptococcal antigen test; (iv) lymphadenitis – localized lymphadenopathy plus tissue acid-fast bacilli and/or granulomas; (v) pericarditis – fibrinous effusion plus response to anti-tuberculous drugs.
Mycobacterium kansasii disease was defined as one or more sputum isolate(s) in association with new, otherwise unexplained, chest radiograph changes plus two of the following: smear positive; no response to amoxycillin; radiological response to regimens containing rifampicin and ethambutol.
The TB database was also used to identify previous TB treatment, therefore men treated before 1984 or at another location would not have been identified.
Data were analysed with STATA 5.0 software (STATA Corporation, College Station, Texas, USA). Individual records were expanded into person–year follow-up periods according to age, employment duration, calendar period, interval since previous TB treatment and interval since HIV test. Poisson regression was used to calculate univariate and multivariate-adjusted TB and M. kansasii disease incidence rate ratios (IRR) for different variables from the expanded data set. Tests for trend were calculated for variables with more than two categories using likelihood ratio tests.
Poisson regression assumes a multiplicative combination of effects in individuals with multiple risk factors. This assumption is made for mathematical convenience, and does not necessarily reflect the true way in which risks combine. In a separate analysis, this assumption was investigated with regard to the joint effects of HIV and silicosis by fitting TB incidence data to both additive and multiplicative models. In order to evaluate consistency with the data, likelihood ratio tests were used to compare models with and without an interaction term between HIV and silicosis.
A total of 1374 HIV-positive and 2648 HIV-negative men were included in the main analysis, as shown in Figure 1. Reasons for exclusion in the other 64 HIV-positive and 352 HIV-negative men were: seroconversion in 265; HIV test taken for suspected TB in 19; already on TB treatment in 97; left employment in 35. The indication for HIV testing was not identified in all men due to difficulties locating outpatient case records. Indications were identified for 937 (58%) HIV-positive and 1761 (67%) HIV-negative men. The most common indications were: STDs (85% of HIV-positive and 92% of HIV-negative men); herpes zoster (5% of HIV-positive and 0.5% of HIV-negative men); other rashes (2% of both groups).
Baseline characteristics by HIV group are shown in Table 1. HIV-negative men were significantly more likely to have entered in the earlier years (when HIV prevalence in STD patients was still low), but were otherwise similar to HIV-positive men.
Radiographs could not be found for 74 men (2%). Silicosis was strongly correlated with age and employment duration. The combined prevalence of early/advanced silicosis increased from 0.37% in men under 30 years to 4.6% in 30- to 39-year-olds and 22.8% in men aged 40 years or older.
TB cases and incidence
TB was diagnosed in 148 HIV-positive and 85 HIV-negative men, and met study case definitions in 135 and 78 respectively. The site and proportion of smear and/or culture positive disease are shown in Table 2. There were 15 men with smear positive pulmonary TB (nine HIV-positive and six HIV-negative, respectively) for whom cultures were either negative, contaminated or had not been requested.
Overall TB incidence was 4.9 and 1.1 per 100 person–years in HIV-positive and HIV-negative men respectively [IRR, 4.5; 95% confidence interval (CI), 3.4–6.0]. Incidence rates and rate ratios for risk factors and time period are shown by HIV group in Table 3. Incidence increased significantly in both HIV-positive and HIV-negative men with increasing silicosis grade, age, employment duration, and was significantly higher in underground compared with surface workers and in men previously treated for TB. Silicosis and employment duration were both significantly correlated with age, and there was confounding apparent between these variables (data not shown). However, TB incidence in miners aged 40 years or more was high even among those with normal radiographs (6.4 and 1.1 per 100 person–years in HIV-positive and HIV-negative men respectively). TB incidence was significantly higher in HIV-positive, but not HIV-negative, men for the period 1995–1997 compared to 1991–1994. Incidence in HIV-positive silicotics (early/advanced) was 16.1 (95% CI, 10.7– 24.2) per 100 person–years during the later period.
Multivariate-adjusted IRRs for effects of risk factors on TB incidence are shown in Table 4. There was a significant interaction between HIV effect and time period (P = 0.028) and so the effect of HIV was estimated separately for each time period. The IRR for HIV infection increased from 2.8 in 1991–1994 to 5.9 in 1995–1997. Other significant risk factors were: silicosis (IRR, 1.8, 2.2 and 2.5 for probable, early and advanced grades respectively), increasing age, underground job (IRR, 0.5 for surface workers) and previous TB (IRR, 1.3 for the first 5 years after treatment and 2.6 thereafter). There was no significant interaction between the effects of HIV infection and silicosis (P = 0.4), indicating that the data were consistent with the associated risks combining multiplicatively.
The joint effects of HIV infection and silicosis were further evaluated by comparing the predicted rates from two simple regression models that included HIV and silicosis as the only exposure variables. The models differed in that one assumed that the effects of HIV infection and silicosis on TB incidence combined in a multiplicative way in men with both risk factors, whereas the other assumed that they combined in an additive way. Observed and predicted TB incidence rates are shown in Figure 2. Rates predicted by the multiplicative model were similar to those observed. Introducing an interaction term between HIV and silicosis (so allowing the data to be fitted with no assumption about the nature of the relationship) did not significantly improve the multiplicative model fit (P = 0.5 for interaction term), indicating that the data were consistent with the model. By contrast, the additive model underestimated TB incidence in HIV-positive silicotics, and model fit was significantly improved when an interaction term was included (P = 0.011), indicating that the data were not consistent with a simple additive combination of risks.
Pulmonary M. kansasii disease was diagnosed in nine HIV-positive and seven HIV-negative men, 14 of whom were smear positive. There were no other cases of NTM disease, apart from one HIV-negative man diagnosed with pulmonary Mycobacterium scrofulaceum disease. Mycobacterium kansasii disease incidence was 0.32 (95% CI, 0.17–0.62) and 0.10 (95% CI, 0.05–0.20) per 100 person–years for HIV-positive and HIV-negative men, respectively. Univariate and multivariate-adjusted IRRs for effects of risk factors on M. kansasii incidence are shown in Table 5. Significant risk factors after adjustment were: HIV infection (IRR, 4.1), silicosis (IRR, 3.3), age over 40 (IRR, 6.0) and previous TB treatment (IRR, 3.9).
Since the earliest years of the industry South African gold miners have suffered high rates of TB, with silicosis and crowded living conditions being historically the main identified risk factors [3,8]. The impact of the HIV epidemic on TB incidence in miners will, in part, depend on the nature of the interaction between silicosis and HIV as risk factors for TB, as well as on the prevalence rates of both risk factors. The results from this study demonstrate a high prevalence of silicosis together with TB incidence rates in HIV-positive miners that are consistent with a multiplicative, but not with an additive, effect of HIV infection and silicosis. As such, HIV-associated TB in miners remains as strongly dependent on silicosis as TB among HIV-negative miners, and, at the individual level, TB incidence rates among HIV-positive silicotics are considerably higher than rates reported from HIV-positive Africans with no occupational risk factors [17–21].
At the population level, the combined effects of an HIV epidemic in a workforce that is already highly predisposed to TB from inadequately controlled silica-dust exposure has been loss of TB control, with incidence increasing rapidly to the current extreme rates of over 2000 per 100 000 employees per year . There are major regional public health implications, in addition to those of individual morbidity and mortality, because of the large size and economic importance of the South African mining industry. There is some evidence that migration between South African mines contributed towards the rapid dissemination of TB throughout Southern Africa at the beginning of this century . The effects of mineworker migration on present day TB and HIV epidemiology are unknown, but could be of regional significance. For example, living in the same household as a migrant employee of the South African mines has been identified as a risk factor for tuberculin reactivity in Botswanan children .
TB incidence in HIV-positive men increased significantly during the study period, from 2.2 to 5.8 per 100 person–years, whereas incidence in HIV-negative men showed no time trend. The HIV epidemic has only recently reached South Africa , and one possibility is that incidence increased due to an increasing proportion of more advanced HIV infection in the later period. A further possibility, however, is increasing TB transmission within the workforce with rapid progression to disease in HIV-positive men, so that a rising proportion of HIV-associated TB was from recently acquired infection. Both factors may be contributing. Most miners live in hostels, sharing a room with several others. Such close living conditions result in a high potential for epidemic TB transmission, as has occurred in HIV wards and housing facilities in the United States and Europe .
The incidence of pulmonary M. kansasii disease among both HIV-positive and HIV-negative miners was also considerably higher than estimates from other populations . A prospective cohort study has confirmed that most isolates of M. kansasii in the study workforce are associated with disease, and that M. kansasii is the most common NTM species isolated from HIV-positive as well as HIV-negative miners . With the current incidence data, this indicates that the main effect of the HIV epidemic has so far been to increase the incidence of mining-associated NTM disease, rather than leading to the emergence of previously uncommon NTM species. Associations between mining and M. kansasii have been noted before [26,27]. HIV infection, although most strongly associated with disseminated Mycobacterium avium complex disease, has previously been associated with M. kansasii disease in the United States and Europe [24,28]. The emergence of HIV-associated M. kansasii disease at high incidence among South African miners, however, is unique for sub-Saharan Africa.
One possible source of bias in this study comes from sub-optimal information concerning seroconversion to HIV-positive among initially HIV-negative men. Ideally HIV-negative men would have been routinely re-tested and follow-up stopped at the time of a positive test. However, routine re-testing was not carried out and instead all identified seroconverters were excluded from the study, regardless of interval to or indication for the first positive HIV test. As such, some genuinely HIV-negative person–years were discarded, but some unidentified HIV-positive person–years will have been left in the HIV-negative group. This may have resulted in either over- or under-estimation of TB incidence in the HIV-negative group, depending on the difference between the numbers of genuine HIV-negative person–years removed and unidentified HIV-positive person–years left behind. However, since this applies to a minority, the potential bias is small and is unlikely to account for the high TB incidence in HIV-negative miners, particularly as a cohort study predating the HIV epidemic found similar rates of TB among older silicotic and non-silicotic miners .
The severity of the HIV epidemic in South Africa during the last few years and increasing incidence of TB in HIV-positive miners indicate that the South African mining industry is now in the early stages of a major TB epidemic. The two immediate priorities are to reduce both HIV transmission and ongoing dust exposure through improved STD management, educational programmes, and environmental control. The role of migration needs to be investigated with respect to the risk of HIV infection in miners. The high prevalence of silicosis found in this and other studies implies that dust control has been far from adequate in South African mines [8,9,11]. Improvements on the current system for monitoring dust control are required, and should include routine surveillance of silicosis prevalence as well as measurements of dust exposure.
The impact on TB incidence rates from improved HIV prevention and dust control will not be apparent for a number of years, however, because of the chronic course of both diseases, and control measures likely to be effective in the short-term are also urgently needed. Active radiological case finding and directly observed therapy are already in place. Reports from other settings with high HIV prevalence, however, indicate that rapid diagnosis and treatment do not prevent TB incidence from rising . The additional burden of silicosis and unusually high potential for TB transmission will add considerably to the difficulties of controlling TB in miners. Isoniazid preventive therapy (IPT) is effective and should be offered to tuberculin positive silicotics  and HIV-positive men , but would need wide coverage to achieve an appreciable impact on overall incidence and may need to be targeted more widely, or combined with other measures such as increased active case-finding. If these fail, then synchronized mass case-finding plus IPT campaigns, similar to those successfully used to gain control of TB epidemics in the past , will need to be considered. One further option would be to make highly active antiretroviral therapy (HAART) available for miners known to be HIV-positive, and to link this with vigorous promotion of voluntary counselling and HIV testing. The incidence of HIV-associated TB has been shown to be significantly reduced by HAART, probably reflecting recovery of the ability to mount a protective immune response .
TB is an increasing occupational hazard for South African miners because of the combined effects of poor dust control and the recent HIV epidemic. Companies need to accept responsibility for the consequences of unsafe working and living conditions in South African mines, including the high risk of HIV infection in migrant workers. The emphasis of occupational health programmes needs to be changed to ensure adequate disease prevention as well as compensation of affected individuals. It is likely, however, that interactions between silica dust exposure and HIV infection are already affecting other workforces throughout the industrializing countries of Asia and Africa, generating hidden epidemics of occupational TB. Strategies capable of controlling epidemics of TB in such highly predisposed workforces need to be developed and implemented. Further research with the aims of reducing the risk of HIV infection in migrant workers and improving occupational health in developing countries is urgently needed.
This paper is dedicated to the memory of Daan Mulder, who died in October 1998 having made a major contribution towards the understanding of TB and HIV infection in developing countries during the course of his lifetime. We thank Carol Van Blommestein, Themba Moyake, Jake Mjandana, Tuso Ramolahloane, Louis van Rensburg and Hetta Steyn for help with microbiology data and locating radiographs and occupational records, and Philip Herselman for reading radiographs. Sponsorship: The study was funded by the Wellcome Trust through a Training Fellowship in Clinical Tropical Medicine awarded to E.L.C.
1. De Cock KM, Chaisson RE. Will DOTS do it? A reappraisal of tuberculosis control in countries with high rates of HIV infection.
Int J Tuberc Lung Dis 1999, 3: 457 –465.
2. Snider DE. The relationship between tuberculosis and silicosis.
Am Rev Respir Dis 1978, 118: 455 –460.
3. Cowie RL. The epidemiology of tuberculosis in gold miners with silicosis.
Am J Respir Crit Care Med 1994, 150: 1460 –1462.
4. Corbett EL, Churchyard GJ, Clayton T. et al. Risk factors for pulmonary mycobacterial disease in South African gold miners.
:A case-control study.
Am J Respir Crit Care Med 1999, 159: 94 –99.
5. Becklake MR. The mineral dust diseases.
Tuber Lung Dis 1992, 73: 13 –20.
6. Saiyed HN, Chatterjee BB. Rapid progression of silicosis in slate pencil workers: II.
:A follow-up study.
Am J Ind Med 1985, 8: 135 –142.
7. Mbizvo MT, Machekano R, McFarland W. et al. HIV seroincidence and correlates of seroconversion in a cohort of male factory workers in Harare, Zimbabwe.
AIDS 1996, 10: 895 –901.
8. Leger JP. Occupational diseases in South African mines–a neglected epidemic?
S Afr Med J 1992, 81: 197 –201.
9. Murray J, Kielkowski D, Reid P. Occupational disease trends in black South African gold miners.
:An autopsy-based study.
Am J Respir Crit Care Med 1996, 153: 706 –710.
10. Trapido AS, Mqoqi NP, Williams BG. et al. Prevalence of occupational lung disease in a random sample of former mineworkers, Libode District, Eastern Cape Province, South Africa.
Am J Ind Med 1998, 34: 305 –313.
11. Corbett EL, Murray J, Churchyard GJ. et al. Use of miniradiographs to detect silicosis: comparison of radiological and autopsy findings.
Am J Respir Crit Care Med 1999, 160: 2012 –2017.
12. Cowie RL, Langton ME, Becklake MR. Pulmonary tuberculosis in South African gold miners.
Am Rev Respir Dis 1989, 139: 1086 –1089.
13. Williams B, Campbell C. Understanding the epidemic of HIV in South Africa: analysis of the antenatal clinic survey data.
S Afr Med J 1998, 88: 247 –251.
14. Churchyard GJ, Kleinschmidt I, Corbett EL, Mulder D, De Cock KM. Mycobacterial disease in South African gold miners in the era of HIV infection.
Int J Tuberc Lung Dis 1999, 3: 791 –798.
15. Murray J, Sonnenberg P, Shearer SC, Godfrey-Faussett P. Human immunodeficiency virus and the outcome of treatment for new and recurrent pulmonary tuberculosis in African patients.
Am J Respir Crit Care Med 1999, 159: 733 –740.
16. International Labour Office. Guidelines for the Use of ILO International Classification of Radiographs of Pneumoconiosis.
Series 22: Occupational
Safety and Health. Geneva: ILO; 1981.
17. Bucher HC, Griffith LE, Guyatt GH. et al. Isoniazid prophylaxis for tuberculosis in HIV infection: a meta-analysis of randomized controlled trials.
AIDS 1999, 13: 501 –507.
18. Braun MM, Badi N, Ryder RW. et al. A retrospective cohort study of the risk of tuberculosis among women of childbearing age with HIV infection in Zaire.
Am Rev Respir Dis 1991, 143: 501 –504.
19. Allen S, Batungwanayo J, Kerlikowske K. et al. Two-year incidence of tuberculosis in cohorts of HIV-infected and uninfected urban Rwandan women.
Am Rev Respir Dis 1992, 146: 1439 –1444.
20. Leroy V, Msellati P, Lepage P. et al. Four years of natural history of HIV-1 infection in African women: a prospective cohort study in Kigali (Rwanda), 1988–1993.
J Acquir Immune Defic Syndr Hum Retrovirol 1995, 9: 415 –421.
21. Gilks CF, Godfrey-Faussett P, Batchelor BI. et al. Recent transmission of tuberculosis in a cohort of HIV-1-infected female sex workers in Nairobi, Kenya.
AIDS 1997, 11: 911 –918.
22. Lockman S, Tappero JW, Kenyon TA, Rumisha D, Huebner RE, Binkin NJ. Tuberculin reactivity in a pediatric population with high BCG vaccination coverage.
Int J Tuberc Lung Dis 1999, 3: 23 –30.
23. Daley CL, Small PM, Schecter GF. et al. An outbreak of tuberculosis with accelerated progression among persons infected with the human immunodeficiency virus.
:An analysis using restriction-fragment-length polymorphisms.
N Engl J Med 1992, 326: 231 –235.
24. Bloch KC, Zwerling L, Pletcher MJ. et al. Incidence and clinical implications of isolation of Mycobacterium kansasii: results of a 5-year, population-based study.
Ann Intern Med 1998, 129: 698 –704.
25. Corbett EL, Blumberg L, Churchyard GJ. et al. Nontuberculous mycobacteria: defining disease in a prospective cohort of South African miners.
Am J Respir Crit Care Med 1999, 160: 15 –21.
26. Falkinham JO. Epidemiology of infection by nontuberculous mycobacteria.
Clin Microbiol Rev 1996, 9: 177 –212.
27. Cowie RL. The mycobacteriology of pulmonary tuberculosis in South African gold miners.
Tubercle 1990, 71: 39 –42.
28. Klein JL, Corbett EL, Slade PM, Miller RF, Coker RJ. Mycobacterium kansasiiand human immunodeficiency virus co-infection in London.
J Infect 1998, 37: 252 –259.
29. Hong Kong Chest Service/Tuberculosis
Research Centre MBMRC. A double-blind placebo-controlled clinical trial of three antituberculosis chemoprophylaxis regimens in patients with silicosis in Hong Kong.
Am Rev Respir Dis 1992, 145: 36 –41.
30. Comstock GW, Baum C, Snider DE. Isoniazid prophylaxis among Alaskan Eskimos: a final report of the bethel isoniazid studies.
Am Rev Respir Dis 1979, 119: 827 –830.
31. Jones JL, Hanson DL, Dworkin MS, De Cock KM. HIV associated TB in the era of HAART.1999 National HIV Prevention Conference.
Atlanta, Georgia, 1999 [abstract 164].