Age and gender profiles of HIV infection burden and viraemia: novel metrics for HIV epidemic control in African populations with high antiretroviral therapy coverage

1 Introduction

The human immunodeficiency virus (HIV) continues to be a deadly infection. In 2022, there were an estimated 630,000 (480,000-880,000) deaths worldwide and 1.3 million new infections globally, of which more than half occurred in Eastern and Southern Africa (ESA) [1]. Combination antiretroviral treatment (ART) effectively suppresses HIV viraemia and is highly effective in reducing HIV-associated mortality[2–4] and preventing onward HIV transmission[5,6]. Consequently, treatment has been adopted as a public health strategy to prevent new HIV infections (treatment as prevention, TasP)[1,7,8]. Progress in the implementation of TasP programs is primarily measured through the Joint United Nations Programme on HIV/AIDS (UNAIDS) “95-95-95” targets, with the aim that 95% of people living with HIV (PLHIV) are aware of their status, 95% of those aware are on antiretroviral therapy, and 95% of those on ART achieve viral suppression. All countries in ESA have committed to achieving these targets by 2025[9], as part of the UN Sustainable Development Goals[10,11].

With the expansion of HIV testing and treatment services, many countries in ESA have either attained or are close to attaining the “95-95-95” targets at a population level[12,13]. Thus, determining novel effective methods to measure HIV epidemic control and direct future efforts to accelerate epidemic decline in such contexts is required. While stratification of “95-95-95” targets by age and gender represents one approach, current evidence suggests that these metrics may not be predictive of incidence declines[14–17]. Although populations stratifications of existing targets are useful in highlighting important inequalities in treatment coverage[18], they do not account for underlying demographic structures or distributions of HIV burden in the population. Consequently, they may not provide a complete understanding of the people most at risk of HIV-related morality and onward transmission, which is necessary to tailor and implement effective prevention interventions.

In this study, we consider an alternative metric for assessing HIV control, one that integrates demographic information, specifically the age and gender profiles of the population burdens of HIV infection and HIV viraemia. Using longitudinal data from the Rakai Community Cohort Study (RCCS), a large population-based HIV surveillance cohort in southern Uganda, we compare the proposed profiles against the HIV seroprevalence and the prevalence of viraemia among PLHIV implied by the “95-95-95” targets. Our analyses were conducted across two distinct geographic settings with unique demographic compositions, including agrarian villages and semi-urban trading centres with population pyramids typical of most ESA communities and hyper endemic Lake Victoria fishing communities with atypical population structures. The study timeframe, spanning 2013 to 2019, coincided with the national expansion of ART programs.

2 Methods

3 Results

3.1 Age and gender burden profiles of HIV infection

From July 2013 to May 2019, across the four survey rounds, a total of 37,743 individuals participated in the RCCS at least once (see Supplementary Table S1 for participation rates and Supplementary Figure S1 for study flowchart). Of these, 22,215 (58.86%) participated for the first time (Supplementary Table S2). Inland communities were characterized by expansive population pyramids with a predominantly young population, consistent with rapid population growth across Uganda and ESA[27]. In contrast, fishing communities were characterized by constrictive population pyramids comprising primarily working-age adults (20-40 years) and substantially fewer children or families than in inland communities (Figure 1b). Gender distributions also differed, with women accounting for 52.2% of the population in inland communities and 44.3% in fishing communities (Table 1).

View this table:

• View inline
• View popup

Table 1. Characteristics of populations with and without HIV in the Rakai Community Cohort Study, June 2018-May 2019 (round 19)

• Download figure
• Open in new tab

Figure 1. Age and gender profiles of HIV burden in the Rakai Community Cohort Study, June 2018-May 2019 (round 19).

(a) Map of the fishing (triangles) and inland (circles) study communities within the Rakai region in south-central Uganda, produced using the ‘ggmap’ package version 3.0.2[28] in R (©Stadia Maps ©Stamen Design ©OpenMapTiles © OpenStreetMap contributor) (b) Population pyramids in fishing and inland communities. For each 1-year age band, the number of census-eligible individuals (colored bars) are shown for men(blue) and women (pink). Colored lines represent smoothed population size estimates. (c) The profile for the burden of HIV in the population is shown in black (line: posterior median estimate, linerange: 95% credible interval) and values shown on the right y-axis, for both community types (top, bottom) and both genders (left, right). For comparison, HIV seroprevalence within each 1-year age band is shown in orange (bars: posterior median estimates, error bar: 95% credible interval), and values shown on the left y-axis.

To describe the burden of HIV infection, we initially followed the typical approach to reporting HIV seroprevalence and estimated the proportion of individuals with HIV infection (i.e., positive HIV serostatus) in each 1-year age band (Figure 1c, orange). Despite close geographical proximity, HIV seroprevalence was higher in fishing communities along Lake Victoria than in the inland communities[19]. In 2019, HIV seroprevalence was 32.8% (95% posterior credible interval: 31.5%-34.2%) in fishing communities, versus 11.0% (10.5%-11.5%) in inland communities (Table 1), and it was larger among first-time participants (Supplementary Figure S3).

The distinct population pyramids of these communities also prompted us to characterize the HIV infection burden profile (Figure 1c). We found that this profile centered more strongly on younger populations than statistics of HIV seroprevalence within each 1-year age band might suggest, such that some younger populations with relatively lower HIV seroprevalence accounted for a higher fraction of the HIV infection burden. For example, by 2019, in inland communities half (51.4% (49.3%-53.5%)) of all women with HIV were 27-39 years old and half (53.1% (49.5%-56.3%)) of all men with HIV were 34-45 years old.

Differences in the gender burden of HIV infection across community types depended both on differences in seroprevalence and population structures. In 2019, HIV seroprevalence among women was 1.53 (1.40-1.66) fold higher than among men in fishing communities and 1.78 (1.61-1.98) fold higher among women than among men in inland communities. Further, as women accounted for a smaller proportion of the population in fishing communities (44.3%) as opposed to in inland communities (52.2%), women made up a smaller fraction of the PLHIV in fishing communities (55.2% (53.1%-57.3%)) rather than in inland communities (66.2% (63.8%-68.4%)).

3.3 Age and gender profiles of the burden of HIV viraemia

To characterize the age and gender-specific populations in which viraemia concentrates, we first considered the proportion of PLHIV exhibiting viraemia stratified by age and gender. By 2019, in inland communities, the proportion of PLHIV exhibiting viraemia was highest in men aged 23-27 years (49.8% (39.9%-60.3%)) (Figure 2), showing also that conventional 5-year age banding (15-19 years, 20-24 years …) can mask the actual age groups in which prevalence of viraemia peaks. In fishing communities, the proportion of PLHIV exhibiting viraemia in 2019 was highest in men aged 28-32 years (28.2% (20.0%-41.2%)) (Figure 2). Across all age groups, the proportion of PLHIV exhibiting viraemia was larger in first-time participants (Supplementary Figures S4-5), and lower, or equal in women than men. By 2019, among women, this proportion was highest in women aged 15-17 years in inland communities (30.3% (18.7%-44.0%)) and in women aged 18-22 years in fishing communities (23.9% (16.2%-34.0%)).

• Download figure
• Open in new tab

Figure 2: Trends in the proportions of PLHIV who have unsuppressed viraemia in the Rakai Community Cohort Study.

(a) Estimated proportions of people with HIV in 1-year age bands who have unsuppressed viraemia, in men and women of inland communities. On the left, time trends in the proportion of PLHIV with unsuppressed viraemia from round 16 (July 2013-January 2015) to round 19 (June 2018 to May 2019) are shown for men and women by 1-year age bands (points: posterior median estimates, linerange: 95% CrIs). On the right, posterior median estimates in the proportion of PLHIV with unsuppressed viraemia in round 19 are shown for men and women by 5-year age bands (bars: posterior median estimates, linerange: 95% CrIs). The horizontal dotted line represents the 14% suppression threshold implied by 95-95-95 targets. (b) Same for PLHIV in fishing communities.

Then, we accounted for the distinct population structure in inland and fishing communities to characterize the population subgroups in which viraemia is concentrating. Figure 3a compares the age and gender specific prevalence of viraemia among PLHIV to the viraemia burden profile in the population. We found that viraemia did not center in the age groups currently furthest from achieving “95-95-95” targets. Specifically, in inland communities, women aged 15-19 years were furthest from achieving the “95-95-95” targets in 2019, but we estimate they contributed only 4.9% (3.0%-7.3%) to people exhibiting viraemia. Instead, women aged 20-24 years and 25-29 years each contributed approximately 10% to people exhibiting viraemia in 2019, although these age groups were either close to or had already exceeded the “95-95-95” targets. Similarly, in inland communities, men aged 20-24 years were furthest from achieving “95-95-95” targets in 2019, but we estimate they contributed only 5.4% (3.9%-7.4%) to the burden of viraemia. Rather, men aged 25-29 years and 30-34 years each contributed approximately 10% to the burden of viraemia in 2019. Similar patterns were observed in fishing communities. In 2019, women aged 15-19 were one of the few groups that had not yet met “95-95-95” targets, but this group only contributed 5.1% (2.5%-7.8%) to the burden of viraemia. Instead, women aged 30-34 had met the targets but contributed 8.9% (6.7%-12.6%) to the burden of viraemia. Among men, age groups from 20 to 34 years exhibited similar levels of viral suppression among PLHIV. However, there were more men aged 30-34 exhibiting viraemia as compared to men aged 20-24. The former amounted to 19.6% (14.8%-25.9%) of the burden of viraemia as compared to 3.0% (1.6%-5.2%) in the latter group.

• Download figure
• Open in new tab

Figure 3. Age and gender metrics for people exhibiting viraemia.

(a) Estimates for the prevalence of viraemia among PLHIV (median: purple bars, lines: 95% CrI, right y-axis) in each gender and age band are compared to the burden profile of viraemia (median: black curve, lines: 95% CrI, left y-axis) for inland (top row) and fishing communities (bottom row). (b) Time trends in the burden profiles of viraemia for both men and women (facets), colored by community type (inland or fishing) and survey round (16 or 19) colors (line: posterior median estimate, ribbon: 95% credible interval). The profiles were multiplied by the population prevalence of viral suppression to visualize shifts in the size of the population exhibiting viraemia.

Figure 3b depicts temporal changes in absolute suppression levels, as well as shifts in the profile of the burden of viraemia. Even though we found a steady increase in the median age of PLHIV by gender and community strata, the age profile of people exhibiting viraemia remained relatively stable over time (Supplementary Table S3).

4. Discussion

This study estimated the age and gender burden profiles of HIV infection and HIV viraemia in south-central Uganda using data from communities with HIV risk and demographic profiles typical of rural and semi-urban ESA and data from Lake Victoria fishing communities with high HIV seroprevalence and atypical demographic profiles. Overall, the burden profiles revealed important differences by geographic location, gender and age group. Following the earlier expansion of TasP programs, fishing communities witnessed faster increases in suppression levels among PLHIV than in inland communities, with overall levels surpassing the “95-95-95” targets by 2019. However, the burden of HIV viraemia remained larger in fishing communities, where the proportion of individuals exhibiting viraemia was approximately 2.5 times larger than in inland communities, reflecting higher HIV prevalence in fishing communities. When analyzing the burden of viraemia by gender, we found that in both community types, most individuals exhibiting viraemia were men, although fewer men have HIV than women. When also considering age, we found that the age groups that contributed most to the burden of viraemia were not necessarily the ones that were furthest from achieving the “95-95-95” targets. For example, in inland communities, individuals aged 25-34 contributed to approximately 40% of the burden of viraemia but had higher suppression levels than younger age groups.

Overall, our results show that the metrics considered by the “95-95-95” targets are insufficient to characterize the population exhibiting viraemia, and hence at risk of transmitting HIV. For example, by 2019, fishing communities had an overall prevalence of suppression among PLHIV that surpassed “95-95-95” targets, while inland communities had not yet achieved them. However, individuals exhibiting viraemia represented a larger proportion of the population in fishing communities as opposed to inland communities, consistent with HIV incidence remaining higher in the former communities despite additional interventions such as PrEP[29]. This is largely explained by differences in HIV prevalence and implies that reaching “95-95-95” by 2025 may be insufficient to lower incidence below elimination threshold[30], especially in countries with elevated burden of HIV[31].

Moreover, since 2020 the updated UNAIDS targets recommend the “95-95-95” goals to be met within subpopulations stratified by age and gender[32]. In practice, this supports the prioritization of interventions to those subgroups that are furthest from achieving the targets and improving barriers and inequalities in treatment coverage[18]. However, these subgroups may not disproportionately contribute to the burden of HIV viraemia. For example, in inland communities, we found that young individuals aged 15-24 constituted approximately two fifths of the census-eligible population, but only one fifth of the population exhibiting viremia. In fact, the lower suppression levels commonly reported among young PLHIV[12,33–40] are counterbalanced by the small absolute number of young PLHIV. This observation is typically missing from studies supporting prioritization of younger age-groups in ESA based on the UNAIDS targets⁴⁵. However, there are instances in which higher suppression levels and HIV seroprevalence lead to balanced profiles. For instance, although most PLHIV are women, there are similar numbers of men and women exhibiting viraemia in inland communities. Taken together, these findings support the need to reach and treat a wide subset of the population: men aged 25-34 and women aged 20-39, who disproportionately contribute to unsuppressed viraemia.

The longitudinal nature of the study further allowed us to analyze temporal dynamics in the age profile for the burden of HIV. From 2013 to 2019, we found an increase in the median age of PLHIV of at least 2 years across community types and genders. Although incidence and HIV mortality are decreasing, a shift in this direction is not guaranteed. In inland communities, younger individuals constitute an increasing proportion of the population, and therefore small reductions in incidence may still be consistent with increasing numbers of PLHIV in these age groups. More generally, the population is expected to grow at an exponential rate in Sub-Saharan Africa[41], implying that stagnating incidence rates may potentially shift the age profile to younger age groups. Continued efforts to track the changing demographics of countries in the region, for example through population censuses, would provide helpful in planning long-term HIV interventions, as well as generalizing to other Sustainable Development Goals[42].

While we assessed both age and gender burden profiles of HIV infection and HIV viraemia in this study, the latter metric provides a cross-sectional description of individuals at most at risk of HIV mortality and onward transmission. Hence, we propose the viraemia burden profile as a new metric for HIV control. Critically we observed that the burden profiles of viraemia revealed similar relative contributions of men and women, and larger relative contributions of middle-aged individuals (25-34) than younger individuals (15-24). We further show that these important observations are obscured by age-gender stratified “95-95-95” targets. Furthermore, unlike the “95-95-95” targets[14,15], population burden of HIV viraemia has been shown to be strongly predictive of population HIV incidence[17,43].

This study has several limitations. First, due to the RCCS definition of census-eligible individuals in the study period, our study population is limited to individuals aged 15-49 years. Accordingly, we are not able to quantify the growing seroprevalence of HIV among individuals aged 50 or more, nor viral suppression outcomes in children (<15 years) living with HIV. Importantly, the reported age and gender specific percentages in the burden profiles in the census-eligible population aged 15-49 years should not be interpreted as fractions of the entire population, as they would be overestimates. Second, our study only considered data from Uganda, and our results may not directly generalize to other settings. However epidemiological patterns reported here are typical of ESA, such as: expansive population pyramids[27], patterns in HIV prevalence by age and gender[44], and gender gaps in suppression levels[12,33–40]. These observations suggests that the major contributions of individuals aged 25-34 to the burden of viraemia may generalize across the subcontinent. Third, to generalize our results to the entire census-eligible population, we assumed that first-time study participants were representative of non-participants in terms of age and gender-specific prevalence and viral suppression levels. Despite similar HIV seroprevalence estimates, our estimates of viral suppression among first-time participants were lower than for all participants. Generalizing the results to the census-eligible population avoided overestimation of viral suppression levels and allowed us to match our analyses’ denominators to the ones defined in the UNAIDS goals. Fourth, the age profile of the population exhibiting viraemia may not match the age profile of the sources of HIV infections, but rather the potential sources of HIV transmission. The difference between the two populations can be explained by heterogeneities in transmission risk among individuals exhibiting viraemia. For example, phylogenetic studies in Uganda[21], Zambia[45], and Botswana[46] have estimated highest transmission risks in men aged 30-34 and women aged 20-24, and much lower for example among individuals aged more than 50. These results hence reinforce the importance of the subpopulations contributing most strongly to the burden of viraemia described in this paper.

5. Conclusions

In summary, prioritizing HIV prevention interventions across age and gender subgroups solely based on the “95-95-95” targets may not be the most effective way to reduce the number of people exhibiting viraemia, and hence at risk of HIV mortality and transmitting HIV. Importantly, these metrics ignore underlying population structures and differences in HIV seroprevalence by subgroups, which are integral to the calculation of the age and gender profile for the burden of HIV viraemia. Future surveillance efforts should prioritize population censuses and population-level measures of HIV viraemia to better direct HIV control and elimination efforts, particularly in settings at or achieving the “95-95-95” targets.

Competing interests

MKG, LCW report grants from the National Institutes of Health during the conduct of this study. ABr, MM report an EPSRC PhD studentship during the conduct of this study. OR acknowledges grants from the Bill and Melinda Gates Foundation (OPP1175094 to C.Fraser., OPP1084362 to D. Pillay), the Engineering and Physical Sciences Research Council (EP/X038440/1), and the Moderna Charitable Foundation during the conduct of this study. This arrangement has been reviewed and approved by the Johns Hopkins University in accordance with its conflict-of-interest policies.

Authors’ contributions

OR, MKG, SJR, RS designed the study. OR, MKG, SJR, JK oversaw the study. RG, RS, GK, OL, TQ, FN, LWC, SJR, JK, GN, MKG oversaw and performed data collection. ABr, IK, ABl, MM, MKG, OR contributed to the analysis. ABr, MKG, OR wrote the first draft. All authors discussed the results and contributed to the revision of the final manuscript.

Author information

[Optional]

Funding

This study was supported by the National Institute of Allergy and Infectious Diseases [U01AI075115 to RHG, R01AI087409 to RHG, U01AI100031 to RHG, ZIAAI001040 to TCQ, R01AI43333 to LWC]; the National Institute of Mental Health [F31MH095649 to Dr Jennifer Wagman, R01MH099733 to Ned Sacktor and MJW, R01MH107275 to LWC]; the Division of Intramural Research of the National Institute for Allergy and Infectious Diseases [TCQ, OL, SJR], NIAID [K01AA024068 to Dr Jennifer Wagman]; the Johns Hopkins University Center for AIDS Research [2P30AI094189 to Dr Richard Chaisson]; the U.S. President’s Emergency Plan for AIDS Relief (PEPFAR) through the Centers for Disease Control and Prevention [NU2GGH000817 to RHSP]; the Engineering and Physical Sciences Research Council through the EPSRC Centre for Doctoral Training in Modern Statistics and Statistical Machine Learning at Imperial and Oxford [EP/S023151/1 to Prof Axel Gandy]. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Disclaimer

For the purpose of open access, the author has applied a ‘Creative Commons Attribution’ (CCBY) license to any Author Accepted Manuscript version arising.

Data Availability Statement

Data necessary to reproduce the analyses are available on Zenodo (https://doi.org/10.5281/zenodo.10955672) as open access dataset under the CC-BY-4.0 license. Code to reproduce all analyses is freely available on GitHub version 1.1.2 under the CC-BY-4.0 license at the repository (https://github.com/MLGlobalHealth/longi_viral_loads).

Supporting Information

[Optional]

Supporting Information file 1: Title of Supporting Information file

Information on file format. Brief description of file content.