Mortality data and cause of death profiles enable rational planning of public health services and interventions, and contribute to clinical awareness. In low and middle income countries, however, these statistics are usually sparse and often of poor quality: coverage of death registration in Africa has been reported to be less than 10% . HIV/AIDS has had a major impact on mortality in Africa [2,3]; the number of adult deaths in South Africa almost doubled from 1989 to 2003 . Even though South Africa has made a concerted effort to improve death registration, the underreporting of deaths and the misclassification of causes of death continue to undermine the validity of reported mortality statistics [5,6].
Tuberculosis is widely recognised as the leading cause of death in Africans with HIV/AIDS , but very few reports provide data on specific causes of death in populations affected by HIV/AIDS in Africa. A few record reviews have been published with cause of death information [8–10], but most other studies have been small hospital-based autopsy series [11–14]. One HIV prevalence cohort study reported on autopsy data combined with record review , but no study has been conducted on individuals with known dates of seroconversion to HIV.
In this paper we explore cause-specific mortality by duration of HIV infection in a large cohort of HIV-positive and negative men employed in four South African gold mines, and followed after leaving the mines. We have previously shown that the all-cause mortality rates for the men in this cohort were similar to those in the west before HAART was available . We now present data on causes of death, as well as on the accuracy of clinical diagnoses in a subset of men for whom autopsy data were available. We present mortality rates for natural, unnatural, and tuberculosis-related deaths, and analyses of causes of death in those who have died.
The establishment of the cohort in this study has been described in detail in previous papers [16,17]. In brief, the cohort of HIV-positive and negative men was established using routinely collected data on miners employed by four gold mines in South Africa, linked using unique industry numbers. HIV tests were conducted in surveys of the workforce in the early 1990s, and when clinically indicated in sexually transmitted disease and other clinics and in the hospital. All HIV test results are confidential and testing is performed with informed consent and counselling. The mining company treated miners with AIDS-related opportunistic infections during the period of the study (and continues to do so); antiretroviral therapy (ART) has been offered since 2004.
The HIV-positive cohort consisted of known seroconverters: individuals for whom at least one negative HIV test was recorded, followed by an HIV-positive test. To define the HIV-seroconverter cohort, tests from medical and tuberculosis clinics were excluded, to avoid the biased inclusion of fast progressors. All men with an interval of 3 years or less between the last negative and the first positive test were included, and the seroconversion date was estimated as the midpoint of the seroconversion interval. Men who tested HIV negative in a survey, with no subsequent evidence of HIV from tests, or clinical or autopsy diagnoses, were used as a comparison group.
Information on vital status and cause of death was obtained from several sources: mine personnel records, mine accident records kept by the assurance company, mine hospital records for those who died while employed or soon after, records kept by the Employment Bureau of Africa, the national South African death register, and the autopsy database at the National Institute for Occupational Health.
According to South African law and provided that the next of kin agrees, the cardiorespiratory organs of miners are examined at autopsy for compensation purposes, regardless of the clinical cause of death. The organs are submitted, together with the clinicians' assessment of the cause of death, to the National Institute for Occupational Health in Johannesburg, where the autopsies are conducted. Tissue sections of the lungs are stained with alcian blue (for cryptococcus), Ziehl–Neelsen (for mycobacteria) and Grocott (for Pneumocystis and other fungi), in addition to haematoxylin and eosin. All the autopsy reports and histological slides were reviewed for this study.
The study was approved by the Ethics Committees of the University of the Witwatersrand, South Africa, and the London School of Hygiene and Tropical Medicine, UK.
Cause-specific death rates by HIV status and time since seroconversion were compared. Survival methods and Poisson regression were used, and age and time since seroconversion were treated as time-varying covariates using Lexis expansion. HIV-positive men were included in the analysis from the date of the first positive test (left truncation). Deaths were coded as resulting from natural or unnatural (external) causes. The data on cause of death were more complete and more reliable, often being available from several sources, for those who died while still employed by the mine. Analysis by cause of death was therefore restricted to deaths occurring while working on the mine, or within 6 months of leaving. Some miners leave employment because they are terminally ill, so censoring the analysis at the actual date of leaving employment would underestimate the death rate. Observations were ‘censored’ at death, 6 months after leaving the mine, or at the end of the study (October 2002). For estimating cause-specific mortality rates, individuals who died of other causes were censored at the date of death.
The most likely cause of death was determined from the information available. Autopsy information was used when possible, and autopsy findings were used in preference to other information for cardiorespiratory causes of death. Hospital records were assumed to be the next most reliable source, followed, in order, by fatal mine accident data , death certificates, and personnel records.
The proportion of deaths from different causes was calculated to investigate the occurrence of HIV-related and other diseases by the duration of HIV infection. In addition, the presence at death of tuberculosis, Pneumocystis jirovecii pneumonia (PCJ) and cryptococcal disease was investigated, whether or not they were the most likely causes of death. For comparison of proportions, chi square or Fisher's exact tests were used, as appropriate. The chi square test for trend was used to test for trends by duration of HIV infection. All confidence levels were calculated at the 95% level.
Clinicopathological correlations were assessed for deaths from respiratory disease, among those for whom autopsy records were available. Taking autopsy as the gold standard, the sensitivity and specificity of clinical diagnoses were estimated.
Overall 1950 HIV-positive and 6164 HIV-negative miners were included in the study; 219 HIV-positive and 219 HIV-negative miners died while employed, and a further 89 HIV-positive and 35 HIV-negative miners died within 6 months of leaving the mines. Whether the death was from natural or unnatural causes was known for 91% (279/308) of HIV-positive miners and 92% (234/254) of HIV-negative miners.
The information on cause of death was often available from several sources. Autopsy data were available for 89 HIV-positive (29%) and 104 HIV-negative (41%) men, of whom 66 HIV-positive and 38 HIV-negative men died from natural causes. The main reason for autopsies not being performed was that deaths occurred away from the mines (at home or in another hospital): 70% of 244 deaths occurring on the mine were followed by autopsy, compared with 7% of 318 deaths occurring elsewhere. The lower autopsy rate in HIV-positive miners was explained by the lower autopsy rate for deaths from natural than from unnatural causes (29 versus 56%) and the lower rate for those dying after leaving the mines. Among HIV-positive miners, autopsy rates were similar for those with different durations of HIV infection.
For HIV-positive miners the source of information on cause of death was autopsy records for 27%, hospital records for 26%, death certificate for 12% and personnel and other records for 35%. Equivalent figures for HIV-negative miners were 38, 44, 5, and 13%.
Excluding those with unknown causes of death, 87% of HIV-positive miners (242/279) and 49% of HIV-negative miners (115/234) died from natural causes (Table 1). The mortality rates from unnatural causes in HIV-positive men were constant over time, and were slightly higher than in HIV-negative men, whereas the rates from natural causes increased with time since seroconversion, and were much higher in the HIV-positive than in the HIV-negative men. The mortality rate ratio for deaths from natural causes, comparing HIV-positive and HIV-negative men, was very high: eight within the first five years from seroconversion and 19 thereafter. This was almost twice as high as the rate ratio for deaths from all causes in each period.
Among those dying from natural causes, further information on the likely cause was available for 202 of the HIV-positive men (83%) and 87 of the HIV-negative men (76%; Table 2). Among these deaths, HIV-positive men were twice as likely as HIV-negative men to have died from infections and were much less likely to have died from cancers (other than Kaposi's sarcoma) or cardiovascular conditions. Similar patterns were seen in different age groups, except that the lower proportion of deaths from cancer in the HIV-positive group was only seen in the older age groups (above 35 years).
The mortality rates from tuberculosis were 7.8/1000 person-years at risk for HIV-positive men and 0.6/1000 person-years at risk for HIV-negative men (Table 1). Rates also increased with age. After adjusting for age, rates were 17 times higher in HIV-positive than in HIV-negative men.
Among those who died from natural causes, there was little evidence of trends in causes of death by duration of HIV infection: the proportion of deaths from any infection, or from tuberculosis, PCJ, or cryptococcal disease were all similar in those infected with HIV for less than 3 years, 3–5 years, or more than 5 years (Table 2), although numbers were small in some of these groups. The proportion of deaths from Kaposi's sarcoma increased with duration of HIV infection but remained small, and other cancers were rare. Among the HIV-positive men who died from natural causes the proportion dying with any infection increased with age at seroconversion, from 75% in those aged less than 25 years, to 94% in the over 45 year age group; this was largely caused by increases in bacterial and other pneumonias. There was no trend with age at seroconversion in the proportion of deaths from tuberculosis.
As multiple infections occur, Table 3 examines the trends in tuberculosis, PCJ, and cryptococcal disease in more detail in those who died from natural causes, including all those with autopsy or clinical evidence of these infections even if they were not the main cause of death. Among HIV-negative miners this added one additional individual with tuberculosis, and among HIV-positive miners this added 25 individuals with evidence of tuberculosis, two with PCJ and three with cryptococcal disease. There was still no evidence of trends in the occurrence of these infections with the duration of HIV infection among those who died from natural causes.
Silicosis was noted at autopsy in 15 HIV-positive men (23%) and in eight HIV-negative men (21%) who died from natural causes, and in an additional seven HIV-negative men (no HIV-positive men) who died from unnatural causes. In two HIV-negative men silicosis was the major pathological diagnosis and was taken to be the cause of death.
Of those who died from natural causes and had autopsies performed, 61 of 66 HIV-positive men (92%) had respiratory disease at autopsy compared with 26 of 38 HIV-negative men (68%; P = 0.001). Most of these diseases were infections: 55 of 66 HIV-positive (83%) and 14 of 38 HIV-negative (37%) men had evidence of respiratory infection (P < 0.001; Table 4). In addition, four of 22 HIV-positive men (18%) and eight of 66 HIV-negative men (12%) who died from unnatural causes had evidence of tuberculosis at autopsy (P = 0.5).
Table 4 shows the correlation between clinical diagnosis and autopsy findings for those dying from natural causes. Many of the infections diagnosed at autopsy had not been noted clinically (35/69; 51% of cases). Pulmonary tuberculosis was missed clinically in 11 of 20 men (55%) in whom it was diagnosed at autopsy, and PCJ and cryptococcal pneumonia in eight of nine (89%) and 12 of 12 (100%) cases, respectively. The sensitivity of clinical diagnosis was similar for HIV-positive and HIV-negative men, but the specificity was lower for HIV-positive men. In HIV-positive men only six of 23 clinical diagnoses of pulmonary tuberculosis were confirmed at autopsy, whereas three of three cases in HIV-negative men were confirmed.
This retrospective study of a seroconverter cohort examined cause-specific mortality in an African population before the roll-out of ART: both mortality rates and the proportion of deaths from different causes were calculated. The difficulties in obtaining reliable death data in HIV-positive populations in Africa have been well described; even when there is good coverage for death registration, AIDS deaths continue to be underreported and diseases are often misclassified [5,6]. This adds to the difficulty in quantifying the contribution of HIV to mortality in Africa [4,18].
The completeness of the cause of death information in this study was maximized by having many sources of information. Causes of death were categorized into two major groups, natural and unnatural causes, and natural causes were then analysed in more detail.
The mortality rate from unnatural causes was similar in HIV-positive and HIV-negative men. We previously reported that deaths from work-related injuries in this cohort did not differ by HIV status . Work-related injuries contributed approximately one death per 1000  of the three to four per 1000 person-years at risk, accounting for the slightly higher mortality rate from unnatural causes than in the South African adult male population of 2.8 per 1000 in 2000 . Other deaths were mainly road traffic accidents and homicides. The proportions of deaths from natural causes were also comparable to those found in a previous study in different South African gold mines . The mortality rate from unnatural causes remained constant with the duration of HIV infection.
The age-adjusted mortality rate ratio for natural causes, comparing HIV-positive and HIV-negative men was 11 and, as expected, increased with the duration of HIV infection. Infections were the leading cause of death among HIV-positive men but, among those who died from natural causes, the proportion who died from infectious diseases did not increase significantly with the duration of HIV infection. The infections differed markedly from the leading causes of death in HIV-infected individuals in high income countries in the pre-ART era, where PCJ, Mycobacterium avium intracellularae, cytomegalovirus and bacterial pneumonias were the most frequent .
Tuberculosis was the leading cause of death in HIV-positive men as in other studies in Africa that have analysed cause-specific mortality [11–14]. The mortality rate from tuberculosis was low among HIV-negative men, despite the increase in the incidence of tuberculosis since the early 1990s . The death rate from tuberculosis in HIV-positive men was 16 times that in HIV-negative men within the first five years since HIV infection. This is consistent with the early increase in the incidence of tuberculosis soon after HIV infection that we have previously shown in this cohort , but the rate ratio for tuberculosis mortality was much higher than that for disease (previously estimated as 2.9).
Other studies in Africa have reported similar patterns of infectious causes of death in HIV-positive individuals: tuberculosis, cryptococcal disease, PCJ, bacterial pneumonia, and unspecified meningitis and gastroenteritis [11,12,23]. Cryptococcal disease was the second most common infectious cause of death in HIV-positive men in this study, but occurred at a much lower frequency than tuberculosis, and at a similar frequency to cryptococcal disease found in other African autopsy series [10–13]. In an earlier cohort study in gold miners from different South African mines , cryptococcal disease was the leading cause of death. Other infections, such as bacterial pneumonia and PCJ, each accounted for less than 10% of deaths from natural causes in HIV-positive men.
Although the proportion of deaths from natural causes that were caused by infections was significantly higher in HIV-positive than in HIV-negative men, the proportions of deaths from infections overall, or specific infections, did not change significantly with the duration of HIV infection. In particular, although tuberculosis is known to occur earlier in the course of HIV infection than opportunistic infections , among those who die, there is no evidence that tuberculosis is any more likely to contribute to the deaths of faster progressors than those of slower progressors. This suggests that the patterns of disease are similar among those dying of natural causes at different times after HIV infection.
One of the strengths of this study was the availability of autopsy data, which are known to play an important role in the completeness and validity of cause of death information . Although autopsies were not available for all, and the autopsy rate was lower in HIV-positive men than in HIV-negative men, the autopsy rate was high for those who died while in employment (70%). Men who die after leaving the mine seldom have autopsy examinations because facilities for removing the cardiorespiratory organs are scarce in rural areas of South Africa. Despite this, there was no selection bias with regard to clinical cause of death, in contrast to other settings in which autopsy is often requested only when the cause of death is unclear.
In this study clinicopathological discrepancies were high, both for pulmonary diseases that are widely recognized as being associated with HIV in African adults, such as pulmonary tuberculosis, and for those that have been less frequently reported, e.g. PCJ and cryptococcal disease. High rates of discrepancy have previously been reported in both HIV-infected and uninfected populations, and in both resource-rich [25–27] and resource-poor settings. Several autopsy studies have reported that tuberculosis had not been diagnosed before death in up to 50% of cases in whom it is found at autopsy [13,14,28], with obvious implications for tuberculosis control programmes, as well as for interpreting the results of this and other studies of causes of death.
The strengths of this study were the inclusion of an HIV-negative cohort for comparison of causes of death, the large size of the cohort, the many sources available for ascertainment of the cause of death, and the availability of autopsy data. One of the limitations is that specific causes of death (beyond the classification of natural and unnatural) were known only for those dying while still employed. The patterns of causes of death may thus not be generalizable to men who die after leaving the mine. Rates were therefore not calculated for individual causes of death other than tuberculosis, as they are likely to be underestimated. Autopsies were performed only on the cardiorespiratory organs so death from other causes would not have been noted at autopsy.
Causes of death among miners may not be the same as those in the general population. As we have reported previously, however, the age-specific death rates in HIV-negative miners were similar to those in the male South African population before HIV was widespread . The high mortality rate from unnatural causes among HIV-negative men, seen in South Africa generally as well as in this cohort, masks the full effect of HIV on all-cause mortality. Relative mortality rates for deaths from natural causes are much higher than those for all causes, and may be more readily compared with populations with lower rates of mortality from unnatural causes.
Among those with natural causes of death, the particular causes might also differ. Tuberculosis rates, in particular, are high among miners, but tuberculosis was the cause of death in a similar proportion (32%) of HIV-infected adults in Cote d'Ivoire , and tuberculosis accounted for 20% of all adult deaths from natural causes in South Africa in 2003 . The comparison of the relative importance of specific causes of death at different times since HIV infection should be generalizable to other populations.
This cause of death profile provides a baseline against which the impact of ART on mortality in terms of specific causes can be assessed. Many studies from high income countries report a reduction in AIDS-defining events, with an increase in non-AIDS-defining causes such as trauma, and cardiovascular and liver disease [29–31]. Nevertheless, opportunistic infections remain important in the era of ART [32–34]. There are only a few studies on causes of death in ART cohorts in Africa, but it has been reported that tuberculosis is likely to remain the leading cause of death in HIV-infected patients on ART .
In conclusion, HIV has had a dramatic impact on overall mortality rates in this population as a result of its effect on deaths from natural causes. Tuberculosis is the leading cause of death in both HIV-positive and HIV-negative men who die from natural causes. Among HIV-positive men mortality rates increased with the duration of HIV infection; however, the specific natural causes of death were similar in those infected for different lengths of time. This suggests that the rate of development of immunosuppression affects the timing of death but not its cause. It remains to be seen how these findings will be influenced by ART.
Phyllis Back assisted with the review of the autopsy data.
Sponsorship: This study was funded by the Colt Foundation, UK. J.G. was supported by the UK Department of Health (public health career scientist award).
Conflicts of interest: None.
1. Mathers CD, Fat DM, Inoue M, Rao C, Lopez AD. Counting the dead and what they died from: an assessment of the global status of cause of death data. Bull WHO 2005; 83:171–177.
2. Porter K, Zaba B. The empirical evidence for the impact of HIV on adult mortality in the developing world: data from serological studies. AIDS 2004; 18(Suppl. 2):S9–S17.
3. UNAIDS/WHO. AIDS epidemic update: 2005
. Geneva: UN–AIDS/WHO; 2005.
4. Bradshaw D, Laubscher R, Dorrington R, Bourne DE, Timaeus IM. Unabated rise in number of adult deaths in South Africa. S Afr Med J 2004; 94:278–279.
5. Groenewald P, Bradshaw D, Dorrington R, Bourne D, Laubscher R, Nannan N. Identifying deaths from AIDS in South Africa: an update. AIDS 2005; 19:744–745.
6. Groenewald P, Nannan N, Bourne D, Laubscher R, Bradshaw D. Identifying deaths from AIDS in South Africa. AIDS 2005; 19:193–201.
7. Corbett EL, Marston B, Churchyard GJ, De Cock KM. Tuberculosis in sub-Saharan Africa: opportunities, challenges, and change in the era of antiretroviral treatment. Lancet 2006; 367:926–937.
8. Etard JF, Ndiaye I, Thierry-Mieg M, Gueye NF, Gueye PM, Laniece I, et al
. Mortality and causes of death in adults receiving highly active antiretroviral therapy in Senegal: a 7-year cohort study. AIDS 2006; 20:1181–1189.
9. Lindan CP, Allen S, Serufilira A, Lifson AR, Van de Perre P, Chen-Rundle A, et al
. Predictors of mortality among HIV-infected women in Kigali, Rwanda. Ann Intern Med 1992; 116:320–328.
10. Okongo M, Morgan D, Mayanja B, Ross A, Whitworth J. Causes of death in a rural, population-based human immunodeficiency virus type 1 (HIV-1) natural history cohort in Uganda. Int J Epidemiol 1998; 27:698–702.
11. Ansari NA, Kombe AH, Kenyon TA, Hone NM, Tappero JW, Nyirenda ST, et al
. Pathology and causes of death in a group of 128 predominantly HIV-positive patients in Botswana, 1997–1998. Int J Tuberc Lung Dis 2002; 6:55–63.
12. Lucas SB, Hounnou A, Peacock C, Beaumel A, Djomand G, N'Gbichi JM, et al
. The mortality and pathology of HIV infection in a west African city. AIDS 1993; 7:1569–1579.
13. Nelson AM, Perriens JH, Kapita B, Okonda L, Lusamuno N, Kalengayi MR, et al
. A clinical and pathological comparison of the WHO and CDC case definitions for AIDS in Kinshasa, Zaire: is passive surveillance valid? AIDS 1993; 7:1241–1245.
14. Rana FS, Hawken MP, Mwachari C, Bhatt SM, Abdullah F, Ng'ang'a LW, et al
. Autopsy study of HIV-1-positive and HIV-1-negative adult medical patients in Nairobi, Kenya. J Acquir Immune Defic Syndr 2000; 24:23–29.
15. Corbett EL, Churchyard GJ, Charalambos S, Samb B, Moloi V, Clayton TC, et al
. Morbidity and mortality in South African gold miners: impact of untreated disease due to human immunodeficiency virus. Clin Infect Dis 2002; 34:1251–1258.
16. Glynn JR, Sonnenberg P, Nelson G, Bester A, Shearer S, Murray J. Survival from HIV-1 seroconversion
in Southern Africa: a retrospective cohort study in nearly 2000 gold-miners over 10 years of follow-up. AIDS 2007; 21:625–632.
17. Murray J, Sonnenberg P, Nelson G, Shearer S, Bester A, Begley A, Glynn JR. Effect of HIV on work-related injury rates in South African gold miners. AIDS 2005; 19:2019–2024.
18. STATSSA. Mortality and causes of death in South Africa, 2003 and 2004: findings from death notification
. Pretoria: Statistics South Africa; 2006.
19. Anderson BA, Phillips HE. Adult mortality (age 15–64) based on death notification data in South Africa: 1997–2004.
Report no. 03-09-05. Pretoria: Statistics South Africa.
20. Chan IS, Neaton JD, Saravolatz LD, Crane LR, Osterberger J. Frequencies of opportunistic diseases prior to death among HIV-infected persons. Community Programs for Clinical Research on AIDS. AIDS 1995; 9:1145–1151.
21. Sonnenberg P, Glynn JR, Fielding K, Murray J, Godfrey-Faussett P, Shearer S. HIV and pulmonary tuberculosis: the impact goes beyond those infected with HIV. AIDS 2004; 18:657–662.
22. Sonnenberg P, Glynn JR, Fielding K, Murray J, Godfrey-Faussett P, Shearer S. How soon after infection with HIV does the risk of tuberculosis start to increase? A retrospective cohort study in South African gold miners. J Infect Dis 2005; 191:150–158.
23. Wong ML, Back P, Candy G, Nelson G, Murray J. Pneumocystis jirovecii
pneumonia in African miners at autopsy. Int J Tuberc Lung Dis 2006; 10:756–760.
24. Burton JL, Underwood J. Clinical, educational, and epidemiological value of autopsy. Lancet 2007; 369:1471–1480.
25. d'Arminio Monforte A, Vago L, Lazzarin A, Boldorini R, Bini T, Guzzetti S, et al
. AIDS-defining diseases in 250 HIV-infected patients; a comparative study of clinical and autopsy diagnoses. AIDS 1992; 6:1159–1164.
26. Klatt EC. Diagnostic findings in patients with acquired immune deficiency syndrome (AIDS). J Acquir Immune Defic Syndr 1988; 1:459–465.
27. Wilkes MS, Fortin AH, Felix JC, Godwin TA, Thompson WG. Value of necropsy in acquired immunodeficiency syndrome. Lancet 1988; 2:85–88.
28. d'Arminio Monforte A, Vago L, Gori A, Antinori S, Franzetti F, Antonacci CM, et al
. Clinical diagnosis of mycobacterial diseases versus autopsy findings in 350 patients with AIDS. Eur J Clin Microbiol Infect Dis 1996; 15:453–458.
29. Crum NF, Riffenburgh RH, Wegner S, Agan BK, Tasker SA, Spooner KM, et al
. Comparisons of causes of death and mortality rates among HIV-infected persons: analysis of the pre-, early, and late HAART (highly active antiretroviral therapy) eras. J Acquir Immune Defic Syndr 2006; 41:194–200.
30. Mocroft A, Brettle R, Kirk O, Blaxhult A, Parkin JM, Antunes F, et al
. Changes in the cause of death among HIV positive subjects across Europe: results from the EuroSIDA study. AIDS 2002; 16:1663–1671.
31. Sackoff JE, Hanna DB, Pfeiffer MR, Torian LV. Causes of death among persons with AIDS in the era of highly active antiretroviral therapy: New York City. Ann Intern Med 2006; 145:397–406.
32. Jain MK, Skiest DJ, Cloud JW, Jain CL, Burns D, Berggren RE. Changes in mortality related to human immunodeficiency virus infection: comparative analysis of inpatient deaths in 1995 and in 1999–2000. Clin Infect Dis 2003; 36:1030–1038.
33. Krentz HB, Kliewer G, Gill MJ. Changing mortality rates and causes of death for HIV-infected individuals living in Southern Alberta, Canada from 1984 to 2003. HIV Med 2005; 6:99–106.
34. Smit C, Geskus R, Walker S, Sabin C, Coutinho R, Porter K, Prins M. Effective therapy has altered the spectrum of cause-specific mortality following HIV seroconversion
. AIDS 2006; 20:741–749.