PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Health Aff (Millwood). Author manuscript; available in PMC 2013 November 6.
Published in final edited form as:
PMCID: PMC3819460
NIHMSID: NIHMS521266

Nearly Full Employment Recovery Among South African HIV Patients On Antiretroviral Therapy: Evidence From A Large Population Cohort

Abstract

Antiretroviral therapy for HIV may have important economic benefits for patients and their households. We quantified the impact of HIV treatment on employment status among HIV patients in rural South Africa who were enrolled in a public-sector HIV treatment program supported by the U.S. President’s Emergency Plan for AIDS Relief. We linked clinical data from more than 2000 patients in the treatment program with ten years of longitudinal socioeconomic data from a complete community-based population cohort of over 30,000 adults residing in the clinical catchment area. We estimated the employment effects of HIV treatment in fixed effects regressions. Four years after the initiation of antiretroviral therapy, employment among HIV patients had recovered to about 90 percent of baseline rates observed in the same patients three to five years before they started treatment. Many patients initiated treatment early enough that they were able to avoid any loss of employment due to HIV. These results represent the first estimates of employment recovery among HIV patients in a general population, relative to the employment levels that these patients had prior to job-threatening illness and the decision to seek care. We find large economic benefits to HIV treatment. For some patients, further gains could be obtained from initiating antiretroviral therapy earlier, prior to HIV-related job loss.

INTRODUCTION

Antiretroviral drug therapies have changed the prognosis for people with HIV/AIDS, enabling many patients to avoid morbidity and premature mortality due to HIV infection. Although clinical recovery on antiretroviral therapy is well documented,1 far less is understood about the economic benefits of treatment for HIV.2

We investigated the extent to which HIV patients receiving antiretroviral therapy were able to maintain employment or return to work after losing a job. We assessed employment recovery in a community in rural South Africa with very high rates of HIV infection—28 percent among adults3—and high rates of unemployment and temporary labor migration. We linked ten years of longitudinal socioeconomic data from a large, community-based population surveillance system with clinical records from a government HIV treatment program that serves the community. We provide the first estimates of the extent to which HIV patients on antiretroviral therapy in a general population were able to remain employed or return to work, comparing employment levels among the same patients several years before and after they initiated treatment.

This study provides some of the first rigorous evidence on the economic effects of one of the world’s largest health policy interventions, the expansion of public-sector HIV treatment in southern Africa, the region with the world’s highest HIV rates. Past studies on HIV treatment and employment have been of two types: clinical cohorts and firm-based studies. However, these study designs suffer from various limitations. Studies that recruit participants in clinical settings fail to observe employment of HIV patients prior to their decision to seek care.4-8 If care-seeking is correlated with employment status (for instance, because people have more time to seek care when they are unemployed), then estimates of employment recovery using a clinical baseline will be biased. Additionally, only one clinical study to date has adjusted for secular economic trends in the larger community.6 Some firm-based studies have overcome both of these limitations in estimating the effects of antiretroviral therapy on labor supply. However, these studies used highly selected samples of formally employed workers at firms in specific industries and with functioning workplace HIV treatment programs.9,10 Thus the results of these studies may not be generalizable to other workers or work environments, and nor to the experiences of many HIV patients in settings of high unemployment. Furthermore, existing clinical and firm-based studies may suffer from substantial attrition bias, because they did not observe patients who changed their job,9,10 migrated, or did not return for clinical follow-up for other reasons.4-8

In contrast, by observing HIV patients within a complete population cohort, we were able to avoid these limitations. First, we were able to assess employment recovery relative to levels observed three to five years before initiation of antiretroviral therapy, prior to any job-threatening HIV illness and the decision to seek care. This baseline enables us to avoid bias associated with the timing of clinical enrollment and represents a natural yardstick against which to measure employment recovery.

Second, we controlled for changes in local employment opportunities over time, using the employment experiences of community members who had not accessed treatment or other clinical care for HIV.

Third, our analysis included people representing the full range of labor market experiences in this community.

Finally, we observed employment outcomes regardless of whether HIV patients were retained in clinical care, took different jobs in the community, or migrated temporarily to other communities.

To investigate the timing of health recovery, we assessed trends in immunological status, physical functioning, and unemployment due to illness. We also estimated the effect of HIV treatment on migration because of the close link between migration and labor supply in this community. Trends were assessed up to four years after initiation of antiretroviral therapy.

DATA AND METHODS

Data

The Africa Centre for Health and Population Studies is a Wellcome Trust-funded research center at the University of KwaZulu-Natal, South Africa. Since 2000 the Africa Centre has collected longitudinal sociodemographic data on a cohort of more than 100,000 people living in a 167-square-mile area in northern KwaZulu-Natal, through routine household survey visits. In this community-based population surveillance system, demographic data on births, deaths, and migration are collected every six months, with response rates of greater than 99 percent.11 To capture the complexities of household living arrangements in South Africa, the Africa Centre collects data on people who reside outside the surveillance area but who continue to be members of a household within the area, such as those who migrate temporarily for work.11 As part of the population surveillance, the Africa Centre conducted seven waves of socioeconomic surveys over the period 2001-10 and four waves of health surveys over the period 2005-08.

Since 2004 the Africa Centre has collaborated with the South African Department of Health to implement the Hlabsia HIV Treatment and Care Programme at government facilities in the Hlabisa subdistrict of Umkhanyakude District, KwaZulu-Natal. The entire surveillance area is located within the program catchment area (see Appendix Exhibit 1).12 In the Hlabisa HIV Treatment and Care Programme, free antiretroviral therapy and other clinical services for HIV are provided by nurses at seventeen community-based health clinics and one subdistrict hospital. Since 2005 the program has received US financial support via the President’s Emergency Plan for AIDS Relief (PEPFAR).13-15

We linked clinical records of patients in the Hlabisa HIV Treatment and Care Programme with longitudinal socioeconomic and health data on these same patients collected in the Africa Centre’s population surveillance system. Patients were linked using their unique South African identification number, or by first name, surname, age, and sex. Ethics approval for data collection, linkage, and use was obtained from the Biomedical Research and Ethics Committee of the University of KwaZulu-Natal.

Study Population And Exclusion Criteria

The study population included all working-age people who were members of a household in the Africa Centre’s population surveillance area at some point during the ten-year follow-up period, 2001-10, and who themselves resided in the surveillance area from January 1 to June 30, 2004, the six-month period just prior to the establishment of the Hlabisa HIV Treatment and Care Programme.12 Observations of a given person from each wave of the population-based socioeconomic and health surveys were eligible for inclusion if the person was 18-59 years old on the date of the survey visit. If a person was observed multiple times in a single survey wave, we retained only the first response.

Coverage of the population in the socioeconomic survey was nearly complete. Of all working age adults in the surveillance area, 95 percent were observed in at least one survey wave. Response rates in the survey waves for which these individuals were age-eligible were also high. A small proportion of survey wave observations were missing due to the following four mutually-exclusive reasons: attrition (3.6 percent), mortality (3.6 percent), late entry into the cohort (1.9 percent), and lack of information on employment status for individuals under surveillance (11.3 percent) (Appendix Exhibit 2).12

People who began antiretroviral therapy in the Hlabisa HIV Treatment and Care Programme by December 31, 2010, were considered to be in what we called the “HIV treatment sample.” All working-age adults in the population cohort who had not accessed public sector HIV care or treatment services in Hlabisa were also included in the analysis, to adjust for secular changes in employment opportunities facing members of the community. People who accessed clinical services for HIV but had not yet initiated antiretroviral therapy by December 31, 2010 were excluded because they were likely to experience declines in health and employment over time—which would lead to upward bias in our estimates of employment recovery for HIV patients who initiated treatment.

Definition Of Exposure

The exposure of interest was the length of time from the date a person initiated antiretroviral therapy to the dates of the household survey visits when information on employment and other outcomes was collected. For analytical purposes, time since treatment initiation was divided into six-month intervals. We constructed indicator variables that had the value of one if an observation fell into a particular six-month interval, and had the value of zero otherwise. Observations of people not enrolled in the treatment program had the value of zero.

Outcome Measures

We estimated trends in the following economic outcomes: employment status; unemployment due to illness; and whether a person resided in the surveillance area, an indicator of migration. These variables were observed in up to six socioeconomic survey waves, from 2003 to 2010, with employment additionally observed in 2001. Data were reported for all resident and nonresident household members by a single proxy respondent, typically the household head or another senior member of the household.

To assess the timing of health recovery, we estimated trends in physical functioning and immunological status. Physical functioning was assessed through the four health surveys mentioned above, which were conducted over the period 2005-08 as part of the population surveillance. Respondents were asked to report whether they could “walk 5km (around 45 min) without stopping,” “carry heavy objects, like a 20 litre water container, for 20 metres without stopping,” and “participate in heavy or vigorous activities such as planting, harvesting, or construction work”. We constructed a summary indicator for physical functioning that had the value of one if the respondent could walk 5km, carry heavy objects, and do hard work; otherwise, it had the value of zero. Only resident household members and a 10 percent sample of nonresidents were eligible for the health surveys, which were conducted with each respondent in face-to-face interviews.

Immunological status was assessed among HIV patients using data on CD4+ lymphocyte counts collected via routine, semi-annual clinical monitoring, following treatment initiation. CD4+ lymphocytes are white blood cells involved in the immune system’s defense against infections and tumors.14 Because these cells are targeted and destroyed by HIV, counts of the cells are a commonly used indicator of HIV disease progression and immunological recovery among patients on treatment.

All outcomes other than CD4+ counts were coded as dichotomous indicators—that is, with a value of zero or one. Details on the assessment and construction of the outcome measures are provided in the Appendix.12

Statistical Methods

To assess the effect of HIV treatment on physical functioning, employment, unemployment due to illness, and migration, we estimated linear probability regression models with individual fixed effects. The predictors of interest were the six-month intervals of time relative to the date when a person began antiretroviral therapy.

The individual fixed effects controlled for all time-invariant characteristics of study participants. We additionally controlled for the following time-varying factors: month of survey visit (to adjust for seasonality), day of survey visit (to adjust for pay-day effects); sex- and education-specific age profiles (to account for the evolution of labor supply over a person’s working life, for different demographic groups); and survey wave fixed effects, interacted with sex and education (to account for changes in the employment opportunities for specific demographic groups, such as due to the opening of a local coal mine in 2009). We accounted for serial correlation in the error terms by clustering standard errors at the individual level.16

In all fixed effects regression models, the period three to five years prior to the initiation of antiretroviral therapy was the reference group, with which we compared all other periods. An indicator for five to eight years before the start of treatment was included to test the hypothesis that the reference group was observed prior to any employment loss due to HIV illness. As an extension of our main analysis, we assessed employment recovery for men and women separately. Employment opportunities in the community are highly segregated by sex, and these differences may affect the ability of HIV patients to maintain or return to work.

For HIV patients who lost their jobs prior to initiating treatment, a critical question is how they fared in the labor market, relative to other people suffering job loss. HIV patients may face barriers to reemployment, including incomplete health recovery, side effects of treatment, discrimination from employers, time spent seeking care, and difficulties relocating for work. We conducted survival analysis to assess the length of jobless spells for HIV patients who lost work within the three years before they initiated treatment, relative to jobless spells of controls matched according to birth year, sex, education, place of residence prior to job loss, and year of job loss. We estimated hazard regression models and constructed Kaplan-Meier failure curves to assess rates of reemployment. Further details are provided in the online Appendix.12

Crude trends in CD4+ counts were estimated for a panel of treatment initiators who were observed continuously for a full four years following initiation of antiretroviral therapy. We used the statistical software Stata, version 11, for all statistical tests.

Sensitivity Analyses

To assess the robustness of our fixed effects regression estimates to modeling assumptions, we estimated conditional logistic (fixed effects) models. To assess the robustness of our main employment results to attrition, mortality, late entry into the cohort, and missingness, we imputed data for all survey waves when a cohort member was age-eligible but the person’s employment status was not observed. Details are provided in the Appendix.12

RESULTS

Descriptive Statistics

The socioeconomic surveys collected 138,666 observations of 32,321 people ages 18-59 between 2001 and 2010, who resided in the Africa Centre surveillance area during the first six months of 2004. Of these 32,321 cohort members, 2027 initiated antiretroviral therapy in the Hlabisa HIV Treatment and Care Programme by December 31, 2010.

Appendix Exhibit 212 presents descriptive statistics for the sample. Relative to the general working-age population, those who initiated HIV treatment were older, more likely to be female, less likely to have completed secondary school, and less likely to reside outside the surveillance area. Unemployment was very high among both the HIV treatment group and other members of the larger population, with 66 percent of working-age adults not employed.

Immune Recovery For HIV Patients

Patients initiated antiretroviral therapy with an average CD4+ count of 122 cells/mm3, well below the treatment eligibility threshold of 200 cells/mm3. Immune functioning improved rapidly among most patients (Exhibit 1). Within the first six months, the mean CD4+ count among these patients recovered to 235. Thereafter, CD4+ counts continued to rise, but at a decreasing rate. In clinical studies, patients with CD4+ counts above 200 have dramatically lower risk of adverse health events17 compared to those with CD4+ counts below that level, and those with CD4+ counts above 500 have survival rates similar to the general population.18

EXHIBIT 1
Immunological Recovery Among HIV Patients Receiving Antiretroviral Therapy

Physical Functioning

Of 32,321 people in the socioeconomic sample, 22,401 were invited to participate in the 2005-08 health surveys (Appendix Exhibit 3).12 In contrast to the socioeconomic surveys, which had very high participation levels, participation in the health surveys was low, with 48 percent of eligible respondents refusing to be interviewed for the survey. However, among people who participated in the survey, the response rate for the physical functioning questions was over 99 percent. The analytical sample for the health surveys consisted of 29,551 observations of 16,538 individuals.

Appendix Exhibit 412 shows regression results for physical functioning and its component indicators. From a baseline of 79 percent 3-5 years before treatment initiation, the proportion of respondents who could walk five kilometers, carry a heavy object, and do hard work declined by 16 percentage points in the year prior to HIV treatment initiation. That change represented a 76 percent relative increase in the proportion of respondents who had at least one limitation. On average, physical functioning for these respondents recovered to baseline levels within one year after they began treatment, although the effects are imprecisely estimated due to the fact that these data were only collected through 2008, so the number of observations on physical functioning after treatment initiation is relatively small.

Employment

Both crude and regression-adjusted trends in employment for HIV patients receiving antiretroviral therapy (Exhibits 2 and and3)3) display a consistent pattern: There was a sharp decline in employment in the year before the patient started treatment, followed by a slow but nearly complete recovery. From a baseline of 37.0 percent, 3-5 years before treatment initiation, employment fell by 14.1 percentage points, a 38 percent relative decline. Four years after treatment began, employment among HIV patients was only 3.8 percentage points lower than in the baseline reference period, a 90 percent relative recovery.

EXHIBIT 2
Crude Employment Trend Among HIV Patients Receiving Antiretroviral Therapy
EXHIBIT 3
Regression-Adjusted Employment Trend Among HIV Patients Receiving Antiretroviral Therapy

Employment levels five to eight years before treatment initiation were no different from the reference period of three to five years before initiation, suggesting that any loss of employment associated with HIV illness occurred within the three years before initiation (Exhibit 4). Trends in employment were not substantively different in conditional logistic models (Appendix Exhibit 5)12 nor in linear probability models using imputation to account for attrition, mortality, late entry into the cohort, and data missingness (Appendix Exhibit 6).12

EXHIBIT 4
Changes In Employment and Place of Residence

Employment trends were qualitatively similar for men and women (Appendix Exhibits 7 and 8),12 recovering nearly completely by four years after initiation for both sexs. Compared to women, men had substantially higher baseline employment rates (49 percent versus 32 percent) and experienced both larger absolute declines prior to treatment initiation and larger recoveries.

Other Economic Outcomes

Unemployment due to illness nearly doubled, from 3.5 percent at baseline to 6.6 percent in the six months before treatment initiation, i.e. a 3.1 percentage point increase (Exhibit 4). Ten percent of men and 5 percent of women were unemployed due to illness in the last six months prior to treatment initiation (Appendix Exhibit 8). After one year of antiretroviral therapy, however, HIV patients (or their household proxies) were no more likely to report that illness or disability was a barrier to employment than at baseline. These trends are consistent with the rapid health improvements indicated by data on CD4+ counts and physical functioning.

In the years following treatment initiation, patients were more likely to reside in the population surveillance area, indicating lower levels of temporary migration, a possible obstacle to re-employment (Exhibit 4).

Survival Analysis Of Jobless Spells

How did HIV patients fare in the event of job loss, relative to the general population? Following job loss, the median time out of work was 3.3 years for matched controls and 3.7 years for patients receiving HIV treatment (Exhibit 5). HIV patients found work at a slightly lower rate than controls (Hazard Ratio = 0.88), but this difference was not significant (Appendix Exhibit 9).12

EXHIBIT 5
Probability Of Finding Work After Job Loss For HIV Patients Receiving Antiretroviral Therapy And Matched Controls

DISCUSSION

This study is the first to evaluate the economic impact of antiretroviral therapy for HIV in a complete population cohort in sub-Saharan Africa. We assessed employment recovery among patients who were receiving HIV treatment relative to employment levels observed in the same patients, years before any job-threatening HIV illness and the decision to seek care. Four years after initiating treatment, patients were 90 percent as likely to be employed as they were at baseline, 3-5 years before initiating antiretroviral therapy. Both men and women undergoing HIV treatment experienced nearly complete recovery of employment.

Many patients initiated treatment early enough that they were able to avoid any loss of employment due to HIV illness. Given that these patients would have experienced rapid declines in health and economic productivity in the absence of treatment,9,10,19 this is an important finding. A majority of patients who lost work prior to initiating treatment had regained employment four years later.

Although employment recovered nearly to baseline levels, the recovery lagged substantially behind observed health gains. This finding contrasts with results of previous studies, which found rapid gains in labor supply following the initiation of antiretroviral therapy.6,7,9,10 However, those studies were conducted among employees of firms with workplace HIV treatment programs9,10 or in clinical cohorts in settings with lower unemployment rates than the community where our study took place.6,7

To understand to what extent the lag in employment recovery for HIV patients on antiretroviral therapy was due to general labor market conditions in this community, we compared the length of jobless spells among HIV patients with those of matched controls. HIV patients who lost their job prior to treatment initiation faced long periods without work. However, for the average patient, almost 90 percent of the time out of work could be explained by factors faced by the general population.

Novel Aspects Of The Study

We advance the literature on the economic impacts of HIV treatment and avoid several critical limitations of previous studies. First, because the decision to seek care may be correlated with changes in employment status (unrelated to HIV illness), studies that recruited patients in clinical settings4-7 cannot disentangle the effects of treatment from the normal churning of the work force. For example, it is plausible that HIV-infected individuals are more likely to access antiretroviral therapy when they have lost employment, even for reasons unrelated to their health, because they no longer face employment-related time constraints in accessing health care. The employment gains observed in clinical studies could then be completely unrelated to the effect of antiretroviral therapy and simply reflect normal patterns of re-employment after job loss.

Second, even if the gains in employment in clinical studies were only due to antiretroviral therapy, it would be unclear how to interpret these observed employment gains. This is because baseline employment status measured in clinical studies depends on when, in the course of HIV disease, patients seek care. As access to HIV treatment improves, people with HIV will on average seek care earlier in their disease course and will thus be less likely to have lost work due to HIV-related disease. In clinical studies, the “treatment effect” during times of good access would thus be attenuated in comparison to the effect observed during times of bad access, even if the true effect of antiretroviral therapy on employment does not change.

We were able to avoid both of these problems by using, as a baseline, employment status observed several years before treatment initiation. This reference period is observed prior to job-threatening HIV illness and the decision to seek care, so neither factor can bias our treatment effect estimates. Our early baseline is thus a natural yardstick against which to measure the effect of antiretroviral therapy on employment, enabling us to express employment recovery as a percent of the employment levels that patients had many years earlier, and likely would have had in the absence of HIV.

A third strength of our study is that it assesses, for the first time, the effect of HIV treatment on employment in southern Africa, while controlling for secular employment trends in the general population. Because local employment rates fluctuate with agricultural cycles and the opening (or closing) of large firms, estimates that do not control for secular trends may be biased.

A fourth strength is that, in contrast to clinical and firm-based studies, we observed people even after clinical attrition, job changes, or migration. A fifth strength is that we assessed long-term employment recovery, up to four years after the initiation of antiretroviral therapy. As our results show, long-term follow-up is necessary to observe the effects of treatment on employment in settings where jobs are scarce and recovery of employment may be slow.

Sixth, to assess whether patients receiving HIV treatment are disadvantaged in finding re-employment opportunities, we provide the first survival analysis of jobless spells for HIV patients undergoing treatment, relative to matched controls. We find that HIV patients who lose work face jobless spells that are similar to the general working-age population. However, we caution that, in contrast to the fixed effects models of employment recovery, in this survival analysis we were unable to control for all unobserved time-invariant factors.

Finally, our findings were observed in a community with high levels of unemployment and temporary labor migration, and a generalized HIV epidemic with very high infection rates. Such conditions characterize many areas in southern Africa. However, the generalizability of our results may be limited in settings where migration and unemployment are rare, or where HIV prevalence is low and concentrated among particular high-risk groups (see the online Appendix for a discussion of study limitations).12

Policy Implications

Our findings have a number of important policy implications. First, in the context of uncertain future funding for HIV treatment,20 the results support the notion that the economic benefits of treatment are large. These benefits should be considered along with the direct health benefits in cost-benefit calculations and political decisions regarding funding.21,22

Furthermore, in a region where nearly half the population lives with someone who has accessed HIV care or treatment, the benefits of employment recovery for HIV patients are likely to be widely enjoyed.23 In ongoing analyses of this population cohort, we are investigating the economic spillover effects of HIV treatment on the labor supply of other household members and school enrollment of children in affected households. It is important to note that although the private economic benefits of treatment are large, there appears to be little scope for private financing, as a result of the lag in employment recovery and the large number of people who initiate treatment but remain unemployed. Public subsidies and, in some situations, donor assistance, will be required to sustain and expand access to HIV treatment.

Second, information on the economic benefits of treatment could be used to recruit patients into HIV care and treatment before they get sick. Early treatment uptake has positive spillover effects through reduction of HIV transmission.24 Providing information on the economic benefits of treatment could help increase HIV testing and achieve high levels of treatment coverage. The appeal to economic benefits may be particularly salient for working-age men who have little contact with health systems and lower HIV treatment uptake than women.25

Third, people who lose their job before initiating HIV treatment experience lengthy jobless spells in this setting of high unemployment. Our results suggest that further gains in economic well-being for HIV patients could be achieved through recruitment into care and treatment prior to job loss. Many patients seek care only after they have become quite sick; interventions to improve take-up of HIV testing, care, and treatment are thus needed.26

Relaxing the clinical eligibility requirements for treatment initiation would also be likely to facilitate earlier recruitment into care and treatment.27 In 2011 South Africa raised the CD4+ count eligibility threshold to 350 cells/mm3 for all patients. Future analysis of this cohort will assess the impact of this policy change on employment loss and recovery.

Fourth, when it comes to improving the economic welfare of HIV patients, antiretroviral therapy is a complement, rather than a substitute, for sound macroeconomic and fiscal policies that emphasize job creation. In rural South Africa, employment of HIV patients receiving treatment recovered nearly to baseline levels. But with baseline employment at only 37 percent, it is clear that job creation - for all South Africans - must be emphasized in conjunction with the expansion of treatment access.

Antiretroviral therapy does not only postpone mortality. It enables HIV patients to lead economically productive lives, benefiting patients’ households and communities, as well as the patients themselves. For some HIV patients, further economic benefits could be attained by initiating treatment earlier, prior to job-threatening illness and disability. Most importantly, however, the full economic benefits of HIV treatment will be realized only with a long-term commitment to universal treatment access.

Supplementary Material

Supplementary Material

ACKNOWLEDGMENTS

Versions of this research were presented at the International HIV Treatment as Prevention (TasP) Workshop, Vancouver, British Columbia, April 25, 2012 and the Health Care Policy Health Economics Seminar at Harvard University, Boston, Massachusetts, May 7, 2012. Jacob Bor was supported by the Global Infectious Diseases Program at the Harvard Global Health Institute. Till Bärnighausen and Frank Tanser were supported through Grant Nos. R01 HD058482-01 from the National Institute of Child Health and Development, National Institutes of Health (NIH) and 1R01MH083539-01 from the National Institute of Mental Health, NIH, and by the Wellcome Trust. Funding for the Africa Centre’s longitudinal population surveillance was received from the Wellcome Trust. The Hlabisa HIV Treatment and Care Programme has received support from the US Agency for International Development and the President’s Emergency Plan for AIDS Relief (PEPFAR). Till Bärnighausen led the PEPFAR-supported contribution to the Hlabisa HIV Treatment and Care Programme in 2006-07. The authors thank David Canning, David Cutler, Nicoli Nattrass, Brook Baker, Jeremy Barofsky, and Sicelo Sengwayo for helpful feedback on this work. The authors also thank Colin Newell, Kevi Naidu, Nokuthula Skhosana, and the entire staff of the Africa Centre for Health and Population Studies and the Hlabisa HIV Treatment and Care Programme. Most of all, the authors thank the respondents who gave their time to this research. All errors are our own.

Notes

1. Egger M, May M, Chêne G, Phillips AN, Lederberger B, Dabis F, et al. Prognosis of HIV-1-infected patients starting highly active antiretroviral therapy: a collaborative analysis of prospective studies. Lancet. 2002;360:119–29. [PubMed]
2. Beard J, Feeley F, Rosen S. Economic and quality of life outcomes of antiretroviral therapy for HIV/AIDS in developing countries: a systematic literature review. AIDS Care. 2009;21:1343–56. [PubMed]
3. Zaidi J, Grapsa E, Tanser F, Newell ML, Bärnighausen T. HIV prevalence trends after scale-up of antiretroviral treatment. 2012. Unpublished paper.
4. Rosen S, Ketlhapile M, Sanne I, DeSilva MB. Differences in normal activities, job performance and symptom prevalence between patients not yet on antiretroviral therapy and patients initiating therapy in South Africa. AIDS. 2008;22:S131–9. [PubMed]
5. Rosen S, Larson B, Brennan A, Long L, Fox M, Mongwenyana C, et al. Economic outcomes of patients receiving antiretroviral therapy for HIV/AIDS in South Africa are sustained through three years on treatment. PloS One. 2010;5:e12731. [PMC free article] [PubMed]
6. Thirumurthy H, Zivin JG, Goldstein M. The economic impact of AIDS treatment. J Hum Resour. 2008;43:511–52. [PMC free article] [PubMed]
7. Thirumurthy H, Jafri A, Srinivas G, Arumugam V, Saravanan RM, Angappan SK, et al. Two-year impacts on employment and income among adults receiving antiretroviral therapy in Tamil Nadu, India: a cohort study. AIDS. 2011;25:239–46. [PubMed]
8. Coetzee C. The impact of highly active antiretroviral treatment (HAART) on employment in Khayelitsha. S Afr J Econ. 2008;76:S75–85.
9. Larson BA, Fox MP, Rosen S, Bii M, Sigei C, Shaffer D, et al. Do the socioeconomic impacts of antiretroviral therapy vary by sex? A longitudinal study of Kenyan agricultural worker employment outcomes. BMC Public Health. 2009;9:240. [PMC free article] [PubMed]
10. Habyarimana J, Mbakile B, Pop-Eleches C. The impact of HIV/AIDS and ARV treatment on worker absenteeism. J Hum Resour. 2010;45:809–39.
11. Tanser F, Hosegood V, Bärnighausen T, Herbst K, Nyirenda M, Muhwava W, et al. Cohort profile: Africa Centre Demographic Information System (ACDIS) and population-based HIV survey. Int J Epidemiol. 2008;37:956–62. [PubMed]
12. To access the Appendix, click on the Appendix link in the box to the right of the article online.
13. Houlihan CF, Bland RM, Mutevedzi PC, Lessells RJ, Ndirangu J, Thulare H, et al. Cohort profile: Hlabisa HIV Treatment and Care Programme. Int J Epidemiol. 2010;40:318–26. [PMC free article] [PubMed]
14. During the observation period, all people infected with HIV who had fewer than 200 cells/mm3 of CD4+ lymphocytes—white blood cells involved in the immune system’s defense against infections and tumors—or who had stage-IV HIV disease according to the World Health Organization’s classification (see Note 15), were eligible to start antiretroviral therapy. Beginning in 2010 pregnant women and patients with active tuberculosis were eligible to do so if their CD4+ counts were below 350 cells/mm3. Department of Health, Republic of South Africa. The South African antiretroviral treatment guidelines. Pretoria: The Department; 2010.
15. World Health Organization Interim proposal for a WHO Staging System for HIV Infection and Disease. Wkly Epidemiol Rec. 1990;65:221–24. [PubMed]
16. Bertrand M, Duflo E, Mullainathan S. How much should we trust differences-in-differences estimates? Q J Econ. 2004;119(1):249–75.
17. Baker JV, Peng G, Rapkin J, Abrams DI, Silverberg MJ, MacArthur RD, et al. CD4+ count and risk of non-AIDS diseases following initial treatment for HIV infection. AIDS. 2008;22:841–8. [PMC free article] [PubMed]
18. Lewden C, Chêne G, Morlat P, Raffi F, Dupon M, Dellamonica P, et al. HIV-infected adults with a CD4 cell count greater than 500 cells/mm3 on long-term combination antiretroviral therapy reach same mortality rates as the general population. J Acquir Immune Defic Syndr. 2007;46:72–7. [PubMed]
19. Barnett T, Whiteside A. AIDS in the twenty-first century: disease and globalization. Palgrave Macmillan; New York (NY): 2002.
20. Leach-Kemon K, Chou DP, Schneider MT, Tardif A, Dieleman JL, Brooks BPC, et al. The global financial crisis has led to a slowdown in growth of funding to improve health in many developing countries. Health Aff (Millwood) 2012;31:228–35. [PubMed]
21. Resch S, Korenromp E, Stover J, Blakley M, Krubiner C, Thorien K, et al. Economic returns to investment in AIDS treatment in low and middle income countries. PloS One. 2011;6:e25310. [PMC free article] [PubMed]
22. Over M. Achieving an AIDS Transition: Preventing Infections to Sustain Treatment. Center for Global Development; Washington, DC: 2011.
23. Bor J, Bärnighausen T, Newell C, Tanser F, Newell ML. Social exposure to an antiretroviral treatment programme in rural KwaZulu-Natal. Trop Med Int Health. 2011;16:988–94. [PubMed]
24. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour M, Kumarasamy N, et al. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365:493–505. [PMC free article] [PubMed]
25. Muula AS, Ngulube TJ, Siziya S, Makupe CM, Umar E, Prozesky HW, et al. Sex distribution of adult patients on highly active antiretroviral therapy (HAART) in southern Africa: a systematic review. BMC Public Health. 2007;7:63. [PMC free article] [PubMed]
26. Bärnighausen T, Tanser F, Dabis F, Newell ML. Interventions to improve the performance of HIV health systems for treatment-as-prevention in sub-Saharan Africa: the experimental evidence. Curr Opin HIV AIDS. 2012;7:140–50. [PubMed]
27. Schwartländer B, Stover J, Hallett T, Atun R, Avila C, Gouws E, et al. Towards an improved investment approach for an effective response to HIV/AIDS. Lancet. 2011;377:2031–41. [PubMed]