Based on estimated HIV-stratified TB case notification rates in 23 sub-Saharan African countries, TB notification rates among PLHIV have declined more than among people without HIV, concurrently with the expansion of ART. Additionally, TB notification rates among PLHIV declined more in countries with higher ART coverage. While this analysis was descriptive and could not explore causality, these results are consistent with a population-level effect of ART on decreasing TB incidence among PLHIV.
To our knowledge, this analysis is the first attempt to broadly examine the relationship between ART coverage and TB notification rates among PLHIV across sub-Saharan Africa. Reports from a few countries in eastern and southern Africa [15,16,17] had previously shown greater declines in TB notification rates among PLHIV than among people without HIV in the era of ART scale-up. Our results suggest this to be the case in other eastern and southern African countries as well. However, in most of the western African countries in our analysis, TB notification rates among PLHIV decreased less than among people without HIV, or actually increased. One possible contributor to this worrisome trend is the low ART coverage in several western African countries; there may be a threshold of ART coverage that is required before population-level declines in TB are observed. Another possible contributor is the fact that western African countries tend to have concentrated rather than generalized HIV epidemics, which leads to greater uncertainty in their adult HIV prevalence estimates. Furthermore, the marginalization of key populations with HIV in these countries may exacerbate their susceptibility to TB despite the availablity of ART [25]. Thus, without more information about the overlap of TB and HIV epidemiology in these settings, estimates made based only on population-level data are more difficult to interpret. Nonetheless, improving the accessibility and acceptability of HIV testing and ART to key populations in these countries may be critical to decreasing TB among PLHIV.
As countries move toward implementing the “Test and Start” policy [26] to treat all PLHIV with ART regardless of CD4 cell count, monitoring the impact of ART scale-up on TB trends will become increasingly important. The ability to do so depends on being able to stratify TB trends by HIV status. As our study demonstrates, it is possible to make crude estimates based on the types of data that are currently available. However, only half the countries in the region had sufficient data of adequate quality, with insufficient HIV testing coverage among TB patients being the most common reason for excluding countries from our analysis. Prioritizing HIV testing for TB patients is thus not only important to ensure appropriate clinical management, but also for monitoring trends in TB incidence among PLHIV.
Although ART is expected to play a critical role in reducing TB incidence among PLHIV, it is not the only important factor. For example, isoniazid preventive therapy (IPT) has been shown to reduce the development of TB, independent of ART, and is increasingly considered an integral component of routine care for PLHIV [27,28,29]. As scale-up of IPT also occurs across sub-Saharan Africa, monitoring the population-level impact of both ART and IPT on reducing TB incidence will be needed. Furthermore, focusing interventions only on PLHIV will not be sufficient to halt the incidence of TB among PLHIV. People without HIV tend to be a reservoir for TB transmission to PLHIV [1, 6, 30,31,32], so investment in general TB elimination strategies such as active case-finding to find and treat all cases are crucial to reducing TB incidence among PLHIV. Finally, to better understand the contribution of different factors on reducing TB incidence among PLHIV, research is needed that goes beyond analyzing population-level indicators. For instance, long-term cohort studies of PLHIV can help quantify the impact of IPT and ART delivered in programmatic settings, while molecular epidemiology studies can provide insight into transmission of TB from people without HIV to PLHIV.
Our study was subject to limitations that affect the conclusions that can be drawn from our results. Inherent in a descriptive analysis is the limitation that causal inferences cannot be made. Therefore, although TB incidence declined more among PLHIV than people without HIV, and although the declines in TB incidence were greater among PLHIV in countries with the highest ART coverage, we cannot claim that ART caused the declines. Other factors such as improved nutrition or housing conditions, as well as improved TB programs or health system improvements that occurred in the process of building stronger programs to deliver ART, may have affected TB incidence. It is also possible that increasing IPT coverage could have contributed to the greater declines in TB notification rates observed among PLHIV compared to people without HIV. However, we were unable to assess the potential impact of IPT, or even describe its scale-up, as notification data for the number of people receiving IPT were completely missing for a third of the countries in this analysis, and missing for two thirds of countries for the early years of the analytic period [19].
Our study was also subject to limitations related to the data that were available for analysis. Because we were limited to publicly available country-level data, adequate data were unavailable for over half of the countries in the WHO African region. As a result, our analysis was relatively complete in its coverage of eastern and southern African countries, but highly incomplete for the countries of western and central Africa. This limitation highlights the need to strengthen data in major public domains if we are to assess the impact of recent policy changes moving forward.
Finally, as with all analyses based on TB case notifications, we were unable to account for potential changes in case detection rates over time, or differences in case detection between PLHIV and people without HIV. While we do not know for sure how case detection rates have changed over time, it is likely that improvements in national reporting systems over time have led to increases in the likelihood of people with TB being notified to WHO; therefore, our analysis could underestimate the declines in case notification rates that have occurred. In addition, TB case detection among PLHIV and people without HIV may differ, but the direction of this difference is unknown. For example, given that HIV is a stigmatized disease and the majority of new HIV cases tend to present with advanced immunosuppression [33], delays in TB diagnosis (or missed TB diagnoses altogether) occur more frequently than for people without HIV. By contrast, PLHIV who are in care are routinely screened for TB, while people without HIV are generally not, so case detection among PLHIV may be higher than among people without HIV in some settings. Thus, the quantitative comparison between the case notification rates we estimate for PLHIV and people without HIV must be interpreted with these limitations in mind.