The epidemic of HIV-associated tuberculosis in sub-Saharan Africa: does this also impact non-HIV-infected individuals?

Abstract

Annual tuberculosis notification rates have increased two to threefold in many countries in sub-Saharan Africa as a result of the HIV epidemic [1]. Despite huge increases in the incidence of HIV-associated tuberculosis, existing studies have differing conclusions as to whether this is causing increased transmission to non-HIV-infected individuals. Community-based studies have reported no or limited impact on tuberculosis notifications among the non-HIV-infected population [2,3], whereas data from Kenya suggest that the annual risk of tuberculosis infection has increased [4]. Two studies from South African gold mines differ; one reported substantial increases in tuberculosis among both HIV-infected and non-infected workers [5], whereas the other found no increase among non-infected workers [6]. Understanding the transmission of tuberculosis in African populations with high HIV prevalence is important for the further development of disease control strategies. We therefore read with interest the study recently reported by Egwaga and colleagues [7]. Their aim was to assess the impact of the HIV epidemic on tuberculosis transmission in Tanzania, and analysed the trend in the annual risk of tuberculosis infection using tuberculin skin testing among a randomized sample of children aged 6–14 years. This large well-conducted study reported trends in the annual risk of tuberculosis infection from 1983 to 2003, and found that despite substantial increases in HIV-associated tuberculosis notifications, the annual risk of tuberculosis infection among children decreased. The authors concluded that the overall population level effect of HIV on tuberculosis transmission is limited [7]. These data are helpful and provide some welcome news for tuberculosis control programmes. However, an important question is whether annual risk of tuberculosis infection data among children adequately reflect the overall transmission of tuberculosis within the community. Studies in high tuberculosis prevalence communities in South Africa, for example, have shown that much tuberculosis transmission occurs outside the home and is related to patterns of social interaction among adults such as that occurring in drinking hostels [8,9]. The annual risk of tuberculosis infection data among children, although important, may provide a limited insight into transmission risk among the adult population. We have recently reported on the devastating impact of HIV on the epidemiology of tuberculosis in a peri-urban community in South Africa, which has a well-functioning tuberculosis control programme [10]. Here the antenatal HIV seroprevalence has increased over 10 years to approximately 30%, and annual tuberculosis notification rates for the community exceeded 1400/100 000 in 2004. However, despite these huge increases in the overall community burden of tuberculosis, no significant change in tuberculosis rates has occurred among children aged 0–9 years, a finding that is consistent with those of Egwaga and colleagues [7]. However, we also reported an evolving tuberculosis epidemic among adolescents in the same community [10]. Adolescence has traditionally been described as the ‘golden period’ during which the risk of developing active tuberculosis is very low. However, in this community, rates have steadily risen from zero in 1996/1997 to 436/100 000 in 2003/2004; only a proportion of these cases were HIV associated. It thus seems that in this community the tuberculosis epidemic is spilling over into those traditionally at lowest risk. Moreover, more recently acquired data suggest that overall rates among non-HIV-infected young adults in this community are also rising, despite the maintenance of good treatment outcomes in the local tuberculosis service. Therefore, we suspect that the trend in the annual risk of tuberculosis infection data among children may provide a limited insight into the impact of HIV on the tuberculosis epidemic in this population as this is a highly age-specific effect and has not significantly affected children.