Secondary Logo

Journal Logo

Epidemiology and Social

Adult life expectancy trends in the era of antiretroviral treatment in rural Uganda (1991–2012)

Asiki, Gershim; Reniers, Georges; Newton, Robert; Baisley, Kathy; Nakiyingi-Miiro, Jessica; Slaymaker, Emma; Kasamba, Ivan; Seeley, Janet; Todd, Jim; Kaleebu, Pontiano; Kamali, Anatoli

Author Information
doi: 10.1097/QAD.0000000000000930



Antiretroviral therapy (ART) is now widely available in sub-Saharan Africa and there is strong evidence from clinical cohort studies in both developed and developing countries that ART has increased the life expectancy of HIV-infected people in treatment programmes [1,2]. Extrapolating these estimates to all people living with HIV (PLHIV) is, however, problematic because facility data lack information on those who never make contact with the health system, or, are lost to follow-up.

Here, we use data from a demographic surveillance site in Southwestern Uganda covering a period ranging from the early 1990s to 2012. Repeated community-based testing for HIV ensures that the HIV status is known for most adults, which allows us to estimate adult life expectancy by HIV status and show how these indices changed over time. Previous studies have used comparable data sources to describe mortality trends in terms of mortality rates or the probabilities of surviving adulthood [3–7], but detailed studies of population-level trends in (adult) life expectancy are less common [8]. The appeal of life expectancy as an index of mortality is that it is more sensitive to the shift in the age distribution of deaths, and therefore to the fact that ART prolongs the lives of PLHIV rather than eliminates AIDS deaths.

We describe the gains in adult life years in terms of the adult life expectancy of the population as a whole, and the adult life expectancy by HIV status. In addition, we quantify the remaining or residual burden of HIV on adult mortality, and label this the adult life expectancy deficit associated with HIV. A study from South Africa previously used verbal autopsy data and cause-deleted life tables for estimating the life expectancy deficit [8]; we measure it as the difference between the life expectancy of HIV-negatives and the life expectancy of the population as a whole [9].


Description of the Kyamulibwa General Population Cohort

The General Population Cohort (GPC) is a demographic surveillance site with around 23 000 current residents in 25 contiguous villages located in the Kalungu (formerly part of Masaka) district in rural southwestern Uganda. The GPC covered 15 villages between 1989 and 1999 and has since been expanded with an additional 10 villages, including four small trading centres [10]. HIV prevalence in the study site was just under 7% prior to the roll-out of ART in 2004, and increased to 9% in 2010 [10].

Because mortality may be under-reported during the start-up phase of the GPC, we report results for the period spanning 1991–2012. Data collection revolves around an annual census where information is gathered on changes to household membership (births, deaths, and in and out-migrations). The surveillance is complemented by village informants who record births and deaths in their registers and report on a monthly basis to the project office.

HIV testing and uptake of results

Medical surveys with HIV testing have been conducted on an annual basis since the inception of the GPC. Field workers contact all individuals aged 13 years and above who are resident for more than 3 months in the study villages, and obtain informed consent/assent before administering a study questionnaire and HIV testing. Until 2012, blood samples were drawn at home and transported for testing [two parallel enzyme-linked immunosorbent assay (ELISA) tests] at the Medical Research Council central laboratories in Entebbe. Results were made available within approximately 1 month of testing and participants could access their results at central counselling posts established in the study villages. Approximately 70% (range 61–89%) of the eligible residents that were contacted participated in the medical surveys between 1991 and 2012 [10]. Between 1989 and 1999, the proportion of those tested who accepted to receive their HIV results remained around 10%, despite an increase in the number of post-test counselling posts in 1993. The uptake of results increased to 36% in 2000/2001 when home delivery of results was introduced [11]. Following the introduction of ART in 2004, the uptake of test results increased to 46%, and to 56% in 2008 [12]. From 2012 onwards, rapid HIV testing with immediate return of test results on the same day was introduced per national guidelines [13], and the survey approach was also changed to invite participants for testing in a temporary hub in a central location in each village. The proportion of those contacted and accepting to test has not changed much, but the proportion of tested individuals who receive their result is now close to 99% (unpublished data). Participants who are diagnosed with HIV infection are immediately referred for care.

Antiretroviral therapy services availability

ART became available in the study site in 2004. Most PLHIV choose to go to the Medical Research Council study clinic for treatment, but ART is now also available in nearby public facilities. As per national guidelines, ART eligible individuals have three adherence counselling visits and a medical examination prior to treatment initiation. PLHIV are also required to identify a treatment adherence supporter. Following ART initiation, patients are seen at the clinic after 2 weeks, 4 weeks, and then at 3-monthly intervals, or more frequently if there is a medical need. As for the other aspects of ART services provision, the study clinic closely follows the guidelines of the Ministry of Health for determining treatment eligibility (based on CD4+ cell count thresholds), and in terms of treatment regimens. The study site may, however, be atypical in the sense that the research clinic is better resourced than most public health facilities in the area. This, for example, also facilitates the proactive follow-up of patients who miss their scheduled appointments. This is not limited to HIV, but also includes patients with other chronic conditions, including diabetes and hypertension. ART coverage in our study population has always been relatively high, and probably somewhat higher than in the rest of Uganda. For example, based on ART initiation at CD4 cell counts ≤200 cells/μl, the estimated coverage in the GPC in 2008 was 69 versus 53% for Uganda as a whole [12].

Life expectancy estimates

We estimate trends in three adult life expectancy measures: the adult life expectancy of the general population, the adult life expectancy by HIV status (positive, negative, and unknown), and the adult life expectancy deficit associated with HIV. All estimates are disaggregated by sex.

The adult life expectancy is defined as the number of additional years that a 15-year old can expect to live under the mortality rates that prevail in a particular year or period. It is a synthetic and age-standardized measure of mortality, and accounts for the fact that ART prolongs the lives of PLHIV rather than avoids AIDS-associated deaths altogether. Using continuous-time survival analysis methods, the adult life expectancy is computed as the area under the Kaplan–Meier product limit survivor curve between ages 15 and 100. As done elsewhere, we impose an upper age limit because the contribution of the years lived by centenarians to the overall life expectancy is often trivial but could be biased by age overreporting [8]. Percentile-based confidence bounds around life expectancy estimates are obtained via bootstrapping with 1000 replications. Adult life expectancy estimates are presented for groups of three calendar years because the population is relatively small for single year estimates by sex and HIV status.

The adult life expectancy deficit quantifies the effect of HIV-associated mortality on the population-wide life expectancy. It treats the life expectancy of HIV-negatives as a benchmark of achievable life expectancy and subtracts the observed life expectancy for the population as a whole. In other words, it is the average or per person number of adult life years lost to HIV. The adult life expectancy deficit depends on the epidemic magnitude and maturity as well as the mitigating effects of ART scale-up on HIV-related mortality. Non-HIV mortality will moderate the life expectancy deficit as fewer life years are lost to HIV in a population where the death rates from other causes are high. Younger ages at HIV infection and death will widen the life expectancy gap.

The age at infection and the presence of competing mortality risks are important considerations for understanding sex differences in the adult life expectancy trends and deficit because women have lower non-HIV-related mortality in adulthood and they are usually infected at a younger age than men [14–16]. Elsewhere, we provide greater detail on factors that invalidate the interpretation of the difference between the population-level life expectancy and the life expectancy of the HIV-negatives as the life expectancy deficit attributable to HIV [9]. This includes correlated mortality risks (e.g. PLHIV have higher mortality rates from causes other than HIV), and spillover effects that may result from increased investments in HIV services, or, the utilization of these services (e.g. PLHIV experience larger mortality reductions from other causes because of their greater engagement with health services) [8,17].

We complement our analyses of adult life expectancy trends with an age decomposition of life expectancy differences to better understand the contribution of mortality differences in each age group to the overall life expectancy difference using a method described by Arriaga [18]. We conduct the age decomposition for two quantities: the overall life expectancy gains that have been made compared with the pre-ART period and the current adult life expectancy deficit.

HIV status imputation procedures

To allocate exposure time to the respective HIV status categories we classify time prior to the first recorded HIV test as HIV status unknown. Failure to do so introduces downward bias in mortality estimates because survivors would be the only ones contributing person-time prior to the first test. Time following an HIV-positive test remains positive until censoring or death. Any time between two negative tests is always counted as negative, irrespective of the interval between tests. The time following the last negative test is considered negative for a duration corresponding to the probability that 95% of their age group remains uninfected given the sex-specific HIV incidence rates or death, whichever occurs first.


In total, 27 925 adults aged 15 years and above have ever been resident in the GPC between 1991 and 2012 and included in the analysis (Table 1). A little over half (53.8%) of all adults are women and they also contribute the majority of persons-years of exposure (52.4%). Because the HIV medical survey with HIV testing is repeated each year, the HIV status information in the dataset is fairly complete (83.2% of the person-years lived). The average person-year prevalence – defined as the HIV-positive person years divided by the HIV-positive plus HIV-negative person years – is 6.8%.

Table 1:
Study site characteristics and death rates by age, sex, HIV status, and period (Kyamulibwa General Population Cohort, 1991–2012).

A total of 2359 deaths occurred in the entire study period (1991–2012). Over half (57.3%) of the deaths occurred after the age of 45 years. The proportion of adults who died was highest among PLHIV (29.5%), followed by those with HIV-negative status (5.5%). Among people with unknown HIV status, 3.1% died (Table 1).

Table 1 also gives a first indication of the changes in adult mortality that have taken place following the introduction of ART services in the study site. Overall adult mortality rates have declined from 17.8/1000 person-years [95% confidence interval (CI): 16.9–18.8] in the pre-ART era to 11.9/1000 (95% CI: 11.2–12.7) for the years following the introduction of ART. The most pronounced declines in mortality have been registered among PLHIV where the death rates declined from 117.2/1,000 (95% CI: 107.2–128.2) to 39.1/1,000 (95% CI: 34.0–44.9). Detailed trends in mortality rates by HIV status have been published elsewhere [3].

The mortality rate reductions have also translated in important gains in adult life expectancy, as illustrated in Figure 1 (a–c). Detailed estimates along with their 95% CI are presented in Supp. Tables Overall female life expectancy increased from 39.3 years (95% CI: 35.9–42.8) in 1991–1993 to 56.1 (95% CI: 54.0–58.5) in 2009–2012, for a total gain of 16.8 years. The life expectancy for men rose from 38.6 (95% CI: 35.4–42.1) to 51.4 years (95% CI: 49.2–53.7) over the same period, for a total gain of 12.8 years. Among women, 82.1% of these gains were achieved after 2000, and 38.1% after 2005. The corresponding figures for men are 53.1% and 29.7%, respectively.

Fig. 1:
(a–d): Adult LE trends with 95% - confidence intervals by sex, period and HIV status (Kyumalibwa GPC, 1991–2012).

The adult life expectancy of HIV-negative women increased by 6.7 years between 1991–1993 and 2010–2012; from 55.4 (95% CI: 51.7–59.4) to 62.1 (95% CI: 59.8–64.4). In contrast, the life expectancy of HIV-negative men remained around 54 years: 54.6 years in 1991–1993 (95% CI: 50.2–59.1) and 54.2 years in 2009–2012 (95% CI: 51.9–56.8) (Fig. 1a).

The life expectancy gain among PLHIV is pronounced for both sexes (Fig. 1b). Before the availability of ART at the local health facility, the adult life expectancy of HIV-positive women and men hovered around or just above 10 years. Between 2000–2002 and 2009–2012 the adult life expectancy increased by 22.9 years for women [from 13.0 (95% CI: 10.9–15.4) to 35.9 years (95% CI: 30.8–46.0)] and by 20.0 years for men (from 14.3 (95% CI: 6.6–20.6) to 34.3 (95% CI: 18.3–47.3) years). As was the case for the overall life expectancy and the life expectancy for HIV-negatives, HIV-positive women have gained more life-years in the past decade than HIV-positive men, but the differences are smaller.

The trend in the life expectancy of adults with an unknown HIV status is similar to that of the population as a whole, but life expectancy estimates are on average a little lower (Fig. 1c). This suggests that PLHIV are overrepresented among those whose HIV status is unknown.

The adult life expectancy deficit associated with HIV is presented in Fig. 1d. In the early 1990s, the average number of life-years lost still amounted to 16.1 years (95% CI: 12.7–19.8) for women and 16.0 years (95% CI: 12.1–19.9) for men. By 2009–2012, the life expectancy deficit had declined to 6.0 years (95% CI: 4.1–7.8) for women and to 2.8 years (95% CI: 1.2–4.6) for men. Worth noting is that the life expectancy deficit started declining before introduction of treatment: by 2000–2002, the life expectancy deficit had already declined to 13.1 years (95% CI: 10.3–15.9) among women and to 11.0 years (95% CI: 8.2–14.2) among men.

The upper panel of Figure 2 illustrates the contributions of each age group to the adult life expectancy gains that have been made between 1991–2002 (the pre-ART period) and 2009–2012 (the most recent period with data). The lower panel depicts the age-group contributions to the life expectancy deficit for 2009–2012. The total life expectancy gain that is decomposed in the upper panel amounts to 13.2 years for women and 8.8 year for men, and illustrates that mortality reductions in women and men of reproductive age have contributed most to the life expectancy gains that have been made compared with the pre-ART years. The age profile of these contributions to the overall adult life expectancy is younger for women than for men, and that is indeed what one would expect given their earlier age at infection [15]. The age-group contributions to the current adult life expectancy deficit associated with HIV (lower panel) are more evenly spread across the adult age range, and that is the case for both women and men.

Fig. 2:
Age group contributions to the adult life-years gained between 1991–2001 and 2009–2012 (upper panel), and the adult LE deficit associated with HIV in 2009–2012 (lower panel) (Kyamulibwa GPC).


The population in rural Uganda has experienced large mortality reductions between 1991 and 2012, amounting to an adult life expectancy gain of 17 years for women and 13 years for men. The gains in adult life expectancy are largely driven by reductions in HIV-related mortality; a finding that is supported by the life expectancy trends among HIV-positive and HIV-negative individuals: among PLHIV, the life expectancy rose by 20 and 23 years among men and women, respectively. Over the same period, the adult life expectancy among HIV-negatives improved by 6.7 years for women and did not change for men.

The large declines in HIV-associated mortality are also captured by the difference between the adult life expectancy of the HIV-negative population and the life expectancy of the population as a whole. In the early 1990s, this life expectancy deficit was still 16 years or more for both sexes, but has now shrunk to 6 years for women and 3 years for men. However, not all of the decline in the life expectancy deficit is driven by the expansion of ART programmes and could also result from historical declines in HIV incidence [9]. Improvements in the screening for, and management of opportunistic infections may have contributed to the declining mortality among PLHIV prior to treatment availability, but the effect of earlier declines in HIV incidence are probably more important. Mbulaiteye and colleagues [19] reported a decline in HIV incidence in the study population in the early 1990s, which is expected to have caused a decline in mortality around 10 years later [16]. This explanation is corroborated by our results because the adult life expectancy deficit started declining before the introduction of ART in 2004. Incidentally, differences in the timing of the epidemic peak are also likely to explain the apparent discrepancy with a prior study on adult life expectancy trends following the expansion of ART programmes in KwaZulu-Natal [8]. HIV prevalence is three times as high in the South African study site, and one might therefore expect that the expansion of treatment programmes made a larger difference for adult mortality. However, Bor and colleagues estimated that adult life expectancy increased by 11.3 years between 2004 and 2011, which is comparable with the life expectancy gains recorded in this study population, even though its HIV prevalence is much lower. In the Ugandan case, declining mortality because of historical declines in HIV incidence augments the life expectancy gains achieved by the expansion of treatment. In South Africa, the HIV epidemic peaked later and adult life expectancy would have continued to decline in the absence of treatment [8,9].

Sex or gender differences in the life-years gained and lost offer a novel perspective on an ongoing debate about equitable access to services [20]. The adult life-years gained by PLHIV is 3 years higher for women than it is for men. As suggested elsewhere, differences in the engagement with health services could explain some of the sex differences in the life expectancy trends of PLHIV [21]. However, this phenomenon is also related to variability in the age of infection and to differences in the background mortality of men and women [9]. Women, for example, are usually infected at younger ages than men [14–16], and have lower non-HIV mortality in adulthood. In the absence of treatment, a female HIV infection and death will, therefore, incur a larger loss in life expectancy. Conversely, preventing a female HIV death will entail a larger gain in life-years.

Despite the larger gains in adult life expectancy, women still have a larger life expectancy deficit than men, which suggests that the disproportionate burden of HIV on women persists. The matter is probably more complex than is presented here, however, as the expansion of HIV services may produce spillover effects with possibly different repercussions for male and female mortality from causes that are unrelated to HIV (e.g. via the strengthening of maternal health services provision) [17].

The study pertains to a relatively small population that has been under active surveillance for the last 25 years, and for these reasons we cannot claim that it is representative for the country or even rural areas within the country. However, with the exception of the community outreach with HIV testing, the medical services provided to this population closely adhere to the national guidelines and are indicative of their potential impact. The results from this study also provide an empirical counterpoint for estimates that rely more heavily on modelling. The WHO, for example, reports changes in Uganda's national-level life expectancy at age 15 between 1990–2012 of 3 years (from 46 to 49) and 6 years (from 42 to 48) for women and men, respectively [22]. Our results from southwestern Uganda are indicative of much larger mortality reductions and corresponding increases in adult life expectancy that are at least twice as high. Bor and colleagues [8] noted equally large differences between a demographic surveillance site in KwaZulu-Natal and national-level trends for South Africa, which suggests that the WHO may have underestimated the life expectancy gains following the expansion of treatment programmes in populations with generalized epidemics.


This research is jointly funded by the UK Medical Research Council (MRC) and the UK Department for International Development (DFID) under the MRC/DFID Concordat agreement (G0700837) and benefited from two grants to the ALPHA Network ( from Wellcome Trust (085477/Z/08/Z), and the Bill and Melinda Gates Foundation (BMGF- OPP1082114).

K.B. receives support from the MRC UK and DFID (MRC grant number G0700837). We also like to thank the residents of the study area and the entire study team.

Authors contributions: Conceived and designed the study: G.A., G.R., and A.K. Analysed the data: G.R., E.S., K.B., J.N., and I.K. Contributed to the writing of the manuscript: G.A., G.R., R.N., K.B., J.N., E.S., I.K., J.S., J.T., P.K., and A.K.

Conflicts of interest

There are no conflicts of interest.


1. Antiretroviral Therapy Cohort Collaboration. Life expectancy of individuals on combination antiretroviral therapy in high-income countries: a collaborative analysis of 14 cohort studies. Lancet 2008; 372:293–299.
2. Mills EJ, Bakanda C, Birungi J, Chan K, Ford N, Cooper CL, et al. Life expectancy of persons receiving combination antiretroviral therapy in low-income countries: a cohort analysis from Uganda. Ann Intern Med 2011; 155:209–216.
3. Kasamba I, Baisley K, Mayanja BN, Maher D, Grosskurth H. The impact of antiretroviral treatment on mortality trends of HIV-positive adults in rural Uganda: a longitudinal population-based study, 1999-2009. Trop Med Int Health 2012; 17:e66–73.
4. Reniers G, Slaymaker E, Nakiyingi-Miiro J, Nyamukapa C, Crampin AC, Herbst K, et al. Mortality trends in the era of antiretroviral therapy: evidence from the Network for Analysing Longitudinal Population based HIV/AIDS data on Africa (ALPHA). AIDS 2014; 28 (Suppl 4):S533–542.
5. Slaymaker E, Todd J, Marston M, Calvert C, Michael D, Nakiyingi-Miiro J, et al. How have ART treatment programmes changed the patterns of excess mortality in people living with HIV? Estimates from four countries in East and Southern Africa. Global Health Action 2014; 7:22789.
6. Floyd S, Marston M, Baisley K, Wringe A, Herbst K, Chihana M, et al. The effect of antiretroviral therapy provision on all-cause, AIDS and non-AIDS mortality at the population level: a comparative analysis of data from four settings in Southern and East Africa. Trop Med Int Health 2012; 17:e84–93.
7. Jahn A, Floyd S, Crampin AC, Mwaungulu F, Mvula H, Munthali F, et al. Population-level effect of HIV on adult mortality and early evidence of reversal after introduction of antiretroviral therapy in Malawi. Lancet 2008; 371:1603–1611.
8. Bor J, Herbst AJ, Newell ML, Barnighausen T. Increases in adult life expectancy in rural South Africa: valuing the scale-up of HIV treatment. Science 2013; 339:961–965.
9. Reniers G, Eaton J, Nakiyingi-Miiro J, Kabudula C, Crampin A, Herbst K, et al. The impact of antiretroviral therapy on adult life expectancy in sub-Saharan Africa. I Annual Conference on Retroviruses and Opportunistic Infections (CROI). Seattle; 2015.
10. Asiki G, Murphy G, Nakiyingi-Miiro J, Seeley J, Nsubuga RN, Karabarinde A, et al. The general population cohort in rural south-western Uganda: a platform for communicable and noncommunicable disease studies. Int J Epidemiol 2013; 42:129–141.
11. Wolff B, Nyanzi B, Katongole G, Ssesanga D, Ruberantwari A, Whitworth J. Evaluation of a home-based voluntary counselling and testing intervention in rural Uganda. Health Policy Plan 2005; 20:109–116.
12. Kazooba P, Kasamba I, Baisley K, Mayanja BN, Maher D. Access to, and uptake of, antiretroviral therapy in a developing country with high HIV prevalence: a population-based cohort study in rural Uganda, 2004-2008. Trop Med Int Health 2012; 17:e49–57.
13. Downing RG, Otten RA, Marum E, Biryahwaho B, Alwano-Edyegu MG, Sempala SD, et al. Optimizing the delivery of HIV counseling and testing services: the Uganda experience using rapid HIV antibody test algorithms. J Acquir Immune Defic Syndr Hum Retrovirol 1998; 18:384–388.
14. Glynn JR, Carael M, Auvert B, Kahindo M, Chege J, Musonda R, et al. Why do young women have a much higher prevalence of HIV than young men? A study in Kisumu, Kenya and Ndola, Zambia. AIDS 2001; 15 (Suppl 4):S51–60.
15. Gregson S, Nyamukapa CA, Garnett GP, Mason PR, Zhuwau T, Caraël M, et al. Sexual mixing patterns and sex-differentials in teenage exposure to HIV infection in rural Zimbabwe. Lancet 2002; 359:1896–1903.
16. Todd J, Glynn JR, Marston M, Lutalo T, Biraro S, Mwita W, et al. Time from HIV seroconversion to death: a collaborative analysis of eight studies in six low and middle-income countries before highly active antiretroviral therapy. AIDS 2007; 21:S55–S63.
17. Grepin KA. HIV donor funding has both boosted and curbed the delivery of different non-HIV health services in sub-Saharan Africa. Health Aff (Millwood) 2012; 31:1406–1414.
18. Arriaga EE. Measuring and explaining the change in life expectancies. Demography 1984; 21:83–96.
19. Mbulaiteye S, Mahe C, Whitworth J, Ruberantwari A, Nakiyingi J, Ojwiya A, et al. Declining HIV-1 incidence and associated prevalence over 10 years in a rural population in south-west Uganda: a cohort study. Lancet 2002; 360:41–46.
20. Johannessen A. Are men the losers of the antiretroviral treatment scale-up?. AIDS 2011; 25:1225–1226.
21. Braitstein P, Boulle A, Nash D, Brinkhof MW, Dabis F, Laurent C, et al. Gender and the use of antiretroviral treatment in resource-constrained settings: findings from a multicenter collaboration. J Womens Health (Larchmt) 2008; 17:47–55.
22. World Health Organisation. Trends of life expectancy at age 15 years in Uganda (1990–2012). In Global Health Observatory Data Repository. Geneva, Switzerland: World Health Organisation; 2012.

adult mortality; antiretroviral therapy; HIV; life expectancy; sub-Saharan Africa; Uganda

Supplemental Digital Content

Copyright © 2016 Wolters Kluwer Health, Inc.