Large age shifts in HIV-1 incidence patterns in KwaZulu-Natal,South Africa

Recent declines in adult HIV-1 incidence have followed the large-scale expansion of antiretroviral therapy and primary HIV prevention across high-burden communities of sub-Saharan Africa. Mathematical modeling suggests that HIV risk will decline disproportionately in younger adult age-groups as interventions scale, concentrating new HIV infections in those >age 25 over time. Yet, no empirical data exist to support these projections. We conducted a population-based cohort study over a 16-y period (2004 to 2019), spanning the early scale-up of antiretroviral therapy and voluntary medical male circumcision, to estimate changes in the age distribution of HIV incidence in a hyperepidemic region of KwaZulu-Natal, South Africa, where adult HIV incidence has recently declined. Median age of HIV seroconversion increased by 5.5 y in men and 3.0 y in women, and the age of peak HIV incidence increased by 5.0 y in men and 2.0 y in women. Incidence declined disproportionately among young men (64% in men 15 to 19, 68% in men 20 to 24, and 46% in men 25 to 29) and young women (44% in women 15 to 19, 24% in women 20 to 24) comparing periods pre- versus post-universal test and treat. Incidence was stable (<20% change) in women aged 30 to 39 and men aged 30 to 34. Age shifts in incidence occurred after 2012 and were observed earlier in men than in women. These results provide direct epidemiological evidence of the changing demographics of HIV risk in sub-Saharan Africa in the era of large-scale treatment and prevention. More attention is needed to address lagging incidence decline among older individuals.

Despite remarkable progress in the expansion and scale-up of HIV treatment and prevention, 1.7 million new HIV infections and 770,000 deaths from AIDS-related illnesses occur every year globally (1). Almost half of new HIV infections and 40% of AIDS-related deaths are in East and southern Africa. South Africa, with an adult HIV prevalence of 20%, has the largest HIV epidemic in the world, with 7.7 million people currently living with HIV (1).Recent declines in adult HIV-1 incidence across high-burden communities of sub-Saharan Africa (SSA) are testament to the success of large-scale combination HIV prevention, including the expansion of antiretroviral therapy (ART) under universal test and treat (UTT) and the scale-up of voluntary medical male circumcision (VMMC) (2–5). When implemented at scale, large increases in viral load suppression from ART expansion are associated with substantial declines in population-level HIV incidence (2, 3, 5). The results from four large cluster randomized trials of UTT effectiveness demonstrated the feasibility of achieving high population ART coverage and viral load suppression through community mobilization (6–9). Yet, the results also showed the challenge of reducing population-level incidence with treatment alone (10–12).At the same time that HIV incidence has fallen, the burden of HIV infection (HIV prevalence) has more than doubled in those >age 25 (13–17), a result of extended life expectancy among people living with HIV (PLHIV) on suppressive ART (18, 19). Mathematical models suggest that new HIV infections will become increasingly concentrated in older adults over time, a result of prioritized HIV prevention in youth (20), increasing HIV prevalence in older adults, and delayed age at infection with declining risk (12), yet no empirical evidence exists to support these projections.Young age is one of the strongest predictors of HIV risk in SSA, a result of a sexual network structure that places younger individuals at greater risk of infection from older individuals (21–23), heterogeneous behavior and biology that increases exposure to and acquisition of HIV in younger women (24–28), and lower treatment coverage in younger PLHIV (29, 30). Disparities in HIV risk have supported efforts to target HIV prevention to younger individuals, especially to women 23). As the epidemic ages, however, older cohorts of HIV-negative individuals now face a greater probability of exposure to sexual partners living with HIV than they did a decade ago (31), albeit with a reduced viral load because of ART scale-up (13–17, 32, 33). The expansion of treatment and primary prevention, by lowering the overall force of infection (transmission rate), may also increase the average time to infection, a dynamic observed in other infectious diseases (34–37). The fewer individuals who become infected earlier in life means those individuals stay in the population at risk for longer, thereby shifting the relative share of infections to older individuals. Age-structured sexual contact patterns specific to sexually transmitted infections can further magnify age shifts in risk, as individuals tend to mix preferentially with partners of a similar age-group (21). Older cohorts may thus account for a growing proportion of HIV incidence as the overall transmission rate declines in the era of large-scale treatment and prevention.Despite enormous progress in efforts to scale-up treatment and prevention, HIV incidence remains well above epidemic control thresholds in high-burden communities (2–5, 38). More will be needed to identify subgroups with elevated risk to guide regional targets for HIV prevention over the next decade. The remarkable demographic and geographic heterogeneity characteristic of HIV epidemics across high-burden regions (39, 40) has ushered in an era of prioritizing the highest risk groups and geographies to achieve maximum population-level effect (41, 42). Understanding how the demographic landscape of HIV risk has shifted with the scale-up of combination HIV prevention is needed to realign prevention targets with current and future distributions of risk.Here, we measure nonlinear age- and sex-specific trends in HIV-1 incidence between 2004 and 2019 from one of the world’s largest ongoing population-based cohorts in rural KwaZulu-Natal, South Africa, a region with among the highest HIV incidence rates in the world. Following recent work showing significant declines in adult HIV incidence in the region (2), we use statistical approaches that can flexibly reveal a wide range of HIV incidence patters to test the hypothesis that incidence has declined differentially by age and gender over a period spanning the expansion and scale-up of ART and VMMC. We provide plausible epidemiological explanations for the observed changes in the age distribution of risk. Our results have major implications for expanding demographic targets for HIV prevention in the era of UTT.