Association of the US President’s Emergency Plan for AIDS Relief’s Funding With Prevention of Mother-to-Child Transmission of HIV in Kenya

Key Points español 中文 (chinese)

Question  Is funding by the US President’s Emergency Plan for AIDS Relief for prevention of mother-to-child transmission of HIV associated with improvements in key prevention of mother-to-child transmission of HIV–related health outcomes in Kenya?

Findings  Using publicly available data sources, this population-based survey study found evidence that a $0.33 increase in the annual levels of President’s Emergency Plan for AIDS Relief funding for prevention of mother-to-child transmission of HIV per capita was associated with a significant 14% to 16% reduction in infant mortality in Kenya between 2004 and 2014.

Meaning  Identifying specific HIV prevention approaches, such as prevention of mother-to-child transmission of HIV, that are associated with improvements in health outcomes may help policy makers allocate funds to evidence-based approaches that are most likely to be effective.

Importance  From 2004 to 2014, the US President’s Emergency Plan for AIDS Relief (PEPFAR) invested more than $248 000 000 in the prevention of mother-to-child transmission (PMTCT) of HIV in Kenya. Concurrently, child mortality in Kenya decreased by half.

Objective  To identify the extent to which the decrease in child mortality in Kenya is associated with PEPFAR funding for PMTCT of HIV.

Design, Setting, and Participants  This population-based survey study conducted in Kenya estimated the association between annual per capita PEPFAR funding for PMTCT (annual PCF) and cumulative per capita PEPFAR funding for PMTCT (cumulative PCF), extracted using 2004-2014 country operational reports as well as individual-level health outcomes, extracted from the 2003, 2008-2009, and 2014 Kenya Demographic and Health Surveys and the 2007 and 2012 Kenya AIDS Indicator Surveys. The study included children of female respondents to the 2003, 2008-2009, and 2014 Kenya Demographic and Health Surveys who were born 1 to 60 months (for neonatal mortality) or 12 to 60 months (for infant mortality) before the survey, as well as female respondents who had recently given birth and reported on HIV testing during antenatal care (ANC) during the 2007-2014 surveys. Results were adjusted for year, province, and survey respondent characteristics. Statistical analysis was performed from July 8, 2016, to December 10, 2018.

Main Outcomes and Measures  Neonatal mortality was defined as death within the first month of life and infant mortality was defined as death within the first year of life. HIV testing during ANC was defined as receiving counseling on PMTCT, undergoing an HIV test, and receiving test results during ANC.

Results  The analysis included 33 181 neonates (16 870 boys), 26 876 infants (13 679 boys), and 20 775 mothers (mean [SD] age, 28.0 [6.7] years). PEPFAR funding was not associated with neonatal mortality. A $0.33 increase in annual PCF, corresponding to the difference between the 75th and 25th (interquartile range) percentiles of funding, was significantly associated with a 16% (95% CI, 4%-27%) reduction in infant mortality after a 1-year lag. A 14% to 16% reduction persisted after 2- and 3-year lags, and comparable reductions were observed for unlagged and 1-year lagged cumulative PCF. An increase of 1 interquartile range in cumulative PCF was associated with a 7% (95% CI, 3%-11%) increase in HIV testing during ANC, which intensified with subsequent lags. Between 2004 and 2014, sustained funding levels of $0.33 annual PCF could have averted 118 039 to 273 924 infant deaths.

Conclusions and Relevance  Evidence from publicly available data suggests that PEPFAR’s PMTCT funding was associated with a reduction in infant mortality and an increase in HIV testing during ANC in Kenya. The full outcome of funding may not be realized until several years after allocation.

Between 1988 and 2003, Kenya experienced a 32% increase in mortality among children younger than 5 years, which has been partially attributed to the HIV epidemic.^1,2 In response, Kenya established prevention of mother-to-child transmission (PMTCT) of HIV programs in more than 10 000 facilities.³ This achievement was supported by the US President’s Emergency Fund for AIDS Relief (PEPFAR), which contributed more than $248 million to PMTCT programs in Kenya between 2004 and 2014.⁴

Although PEPFAR’s investments in PMTCT coincided with a halving of the mortality rate among children younger than 5 years in Kenya,⁵ it is unknown whether this improvement can be attributed to PEPFAR funding for PMTCT. Programs for PMTCT provide a series of interventions, including provision of HIV testing during antenatal care (ANC), prescription of antiretroviral medication to HIV-positive mothers and exposed infants, and counseling on safe breastfeeding practices, which are critical to the survival of children born to HIV-positive mothers.⁶ Without PMTCT, 25% to 48% of children born to HIV-positive mothers become HIV positive, and, in low-resource settings, 50% of HIV-positive children who do not receive treatment die before 2 years of age.^7,8 However, child mortality decreased in most sub-Saharan African countries during the 2000s,⁹ and regional trends, rather than PEPFAR funding, could explain all or part of Kenya’s reductions in child mortality. In addition, PEPFAR-funded activities targeting adults could have displaced essential newborn and infant health services,^10,11 resulting in worsened child health outcomes.

Although PEPFAR-focus countries have experienced greater reductions in adult mortality than nonfocus countries,^12,13 most previous studies have not found corresponding reductions in child mortality.^11,14,15 However, prior research has not specifically assessed the association of PEPFAR funding with PMTCT, one of PEPFAR’s activities most directly linked to children, and child health outcomes. In this study, we evaluated whether the amount of PEPFAR funding for PMTCT is associated with decreased probability of neonatal and infant mortality and increased probability of the receipt of HIV counseling, testing, and test results as part of ANC in Kenya.

Health outcome data came from the publicly available Kenya Demographic and Health Surveys and Kenya AIDS Indicator Surveys (KAIS), which use stratified 2-stage cluster random sampling to select nationally representative samples. Our analysis included data from the 5 most recent surveys: the 2003, 2008-2009, and 2014 Kenya Demographic and Health Surveys and the 2007 and 2012 KAIS.^3,5,16-18 Neonatal mortality, defined as death within the first month of life, and infant mortality, defined as death within the first year of life, were assessed using female respondents’ birth histories from the 2003, 2008-2009, and 2014 surveys. For each live birth, mothers reported their child’s birth date, vital status, and death date, if applicable. Because maternal HIV is a common cause of maternal and infant death, birth histories can underestimate infant mortality in generalized HIV epidemics¹⁹; however, bias can be reduced by considering recent births.²⁰ Therefore, neonatal mortality was assessed among children born 1 to 60 months prior to the interview date and infant mortality was assessed among children born 12 to 60 months prior to the interview date. Although the 2012 KAIS gathered data on abbreviated birth histories, we excluded it in our primary analysis because neonatal and infant mortality was not included in the final report of the survey (eAppendix and eTable 1 in the Supplement).³ Rates of HIV testing during ANC, which is one mechanism through which PMTCT programs can help prevent vertical transmission of HIV and subsequent child mortality, were measured among female respondents who had given birth within 5 years of the 2007, 2008-2009, and 2012 surveys and within 2 years of the 2014 survey and is defined as receiving counseling on PMTCT, undergoing an HIV test, and receiving test results during ANC. For women with multiple births, data on HIV testing during ANC were gathered for the most recent birth. Data on the independent variable, PEPFAR funding for PMTCT, were extracted from publicly available Country Operational Plans (COPs), which describe annual planned expenditures²¹ (eAppendix, eFigure 1, and eTable 2 in the Supplement).

This research involved the collection and analysis of preexisting, publicly available, deidentified data and received an exemption for review from the Harvard T.H. Chan School of Public Health Institutional Review Board and a “not research” designation from the US Army Medical Research and Materiel Command, Human Research Protection Office. The protocol was additionally reviewed and approved by the Kenya Medical Research Institute’s Scientific and Ethics Review Unit. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. Additional details on the sampling frame, response rates, and other ethical and methodological considerations for the Kenya Demographic and Health Surveys and KAIS can be found elsewhere.^3,5,16-18

Statistical analysis was performed from July 8, 2016, to December 10, 2018. We assessed the association between the amount of per capita PEPFAR funding for PMTCT and health outcomes using regression models.²² For annual per capita funding (annual PCF), each individual’s exposure equaled the amount of PEPFAR funding allocated to their province of residence in their year of birth (or, for HIV testing during ANC, in the year they gave birth) divided by the province’s population.²³ For cumulative per capita funding (cumulative PCF), each individual’s exposure to PEPFAR funding was calculated using the cumulative total of PEPFAR funding for PMTCT allocated to their province from the beginning of PEPFAR until their year of birth. We investigated 0- to 3-year lags between funding allocation and the time that funding was hypothesized to have effects (eTable 2 in the Supplement). Individuals residing in province-years without funding data were excluded from the primary analyses but were included in sensitivity analyses.

We estimated risk ratios and 95% CIs associated with PEPFAR funding for PMTCT using weighted generalized estimating equations²⁴ that created a representative sample across surveys and stepwise selection of restricted cubic splines that assessed for potential nonlinear associations between funding and health outcomes.²⁵ All models treated individual birth as the unit of analysis and adjusted for province and controlled for calendar year using restricted cubic splines. To address confounding, we present both minimally adjusted models, which adjusted only for province and calendar year, and fully adjusted models, which included additional factors that were believed to be potentially important confounders that were known or potential risk factors for the outcomes. Fully adjusted models further controlled for household wealth quintile, water and sanitation access, urban vs rural status, mosquito net ownership, respondent educational level, ethnicity, religion, marital status, age, parity, and exposure to mass media, and, for neonatal and infant mortality, child’s sex, short interval preceding birth, and birth order rather than parity (eAppendix in the Supplement). To estimate the number of lives saved by PEPFAR funding for PMTCT, we used the 1-year lagged model to determine the number of infants who would have died between 2004 and 2014 under different levels of annual PCF (eAppendix in the Supplement).

We evaluated the sensitivity of our results to missing exposure data by conducting analyses assuming that province-years with missing funding data received $0.04 in annual PCF (reflecting the minimum observed funding level), $0.23 in annual PCF (reflecting the 25th percentile of observed funding levels), $0.32 in annual PCF (reflecting the median of observed funding levels), $0.56 in annual PCF (reflecting the 75th percentile of observed funding levels), and $0.93 in annual PCF (reflecting the maximum observed funding level) and, for infant mortality, using inverse probability weighting (eAppendix in the Supplement).²⁶ We also assessed whether the association between PEPFAR funding and infant mortality differed by maternal HIV status among the subset of infants with known maternal HIV status based on HIV testing conducted in the 2003 and 2008 surveys (eAppendix in the Supplement).

The 2003 survey included data on 8195 female respondents, the 2007 survey included data on 10 244 female respondents, the 2008-2009 survey included data on 8444 female respondents, the 2012 survey included data on 7954 female respondents, and the 2014 survey included data on 31 079 female respondents.

Between 2004 and 2006, all financial data were redacted from the COPs. Overall, we assigned 53% of total planned PEPFAR expenditures for PMTCT to specific provinces; the remainder could not be assigned to specific provinces either because they had been allocated to nationwide programs (16%) or implementing partners working across province borders (10%) or because there was insufficient information to assign funding to a specific implementing partner (20%) (percentages total 99% owing to rounding; see eFigure 1 in the Supplement).

Our analysis included 33 181 neonates (16 311 girls and 16 870 boys) born 1 to 60 months before the survey, 822 of whom died during the first month of life. Neonatal mortality was not associated with annual PCF or cumulative PCF (Table 1; eFigure 2 in the Supplement).

The 2003, 2008-2009, and 2014 surveys included birth histories for 128 199 children (13 197 girls and 13 679 boys), 26 876 of whom were born 12 to 60 months before the survey. Of these, 1222 died within the first year of life. After a 1-year lag, a $0.33 increase in annual PCF, corresponding to the difference between the 75th and 25th percentiles (interquartile range) of observed annual PCF data, was associated with a significant 16% (95% CI, 4%-27%) reduction in infant mortality in the fully adjusted model (Table 2). This reduction was sustained after 2- and 3-year lags (Figure 1), but was not significant in the unlagged model. For cumulative PCF, a $0.83 increase (reflecting the interquartile range of cumulative PCF) was similarly associated with a 14% (95% CI, 1%-25%) decrease in infant mortality in the unlagged model. This reduction was sustained after a 1-year lag, with associations attenuating and becoming nonsignificant after subsequent lags.

The 2007, 2008-2009, 2012, and 2014 surveys included data on 57 721 female respondents (mean [SD] age, 28.0 [6.7] years), 21 488 of whom had recently given birth. Of the 20 775 women with data on HIV testing, 11 984 (57.7%) received HIV testing during ANC. Annual PCF was not associated with HIV testing during ANC in any model, although there were associations with cumulative PCF (Table 3). After a 1-year lag, a $0.83 increase in cumulative PCF was associated with a 7% (95% CI, 3%-11%) increase in HIV testing during ANC. This association intensified with subsequent lags: a $0.83 increase in cumulative PCF was associated with a 9% (95% CI, 5%-14%) increase after a 2-year lag and a 19% (95% CI, 11%-28%) increase after a 3-year lag (Figure 2).

Based on the 1-year lagged model, increasing PEPFAR funding for PMTCT from $0 to $0.33 in annual PCF may be associated with an estimated reduction of 118 039 infant deaths between 2004 and 2014, and increasing funding from $0 to $0.93 in annual PCF may be associated with an estimated reduction of 286 438 infant deaths during the same period. These estimates are conservative and do not include benefits accrued among mothers or among children older than 1 year. When data from KAIS 2012 were included, the estimated number of lives saved was appreciably larger: a $0.33 increase in annual PCF was associated with a 36% (95% CI, 12%-53%) reduction in infant mortality after a 1-year lag, which, during 2004-2014, may be associated with 273 924 fewer infant deaths at funding levels of $0.33 in annual PCF and 547 179 fewer infant deaths at funding levels of $0.93.

Sensitivity analyses for our unlagged, 1-year lagged, and 3-year lagged models generally corresponded with the main results: for infant mortality, annual PCF remained associated with a 13% to 19% reduction when province-years with missing data were assigned $0.04 to $0.32 in annual PCF or with inverse probability weighting (eTables 3-5 in the Supplement). When we investigated the association between PEPFAR funding and infant mortality by maternal HIV status, PEPFAR was significantly associated with lower rates of mortality among infants born to HIV-negative mothers in the unlagged and 1-year lagged model but not among infants born to HIV-positive mothers (eTable 6 in the Supplement). However, owing to the small number of deaths among infants with known maternal HIV status (214 deaths among the 3803 infants born to HIV-negative mothers and 45 deaths among the 369 infants born to HIV-positive mothers), there was no evidence that PEPFAR funding had significantly different associations in the 2 groups.

Our study joins a growing body of literature that suggests that PEPFAR has benefited population health.^12-14,27-29 Using publicly available data, we found evidence that PEPFAR funding for PMTCT is associated with reduced infant mortality and increased HIV testing during ANC in Kenya. Our analytic approach provides stronger evidence for the benefits of PEPFAR funding than do other designs, such as pre-post designs that fail to account for secular trends, which have been used previously to assess PEPFAR’s effectiveness.³⁰

Earlier studies of PEPFAR’s effectiveness focused on a shorter period (2004-2010) and had not found significantly greater reductions in infant mortality among PEPFAR-focus countries relative to nonfocus countries.^14,15 However, a recent multicountry ecologic study investigating the effect of PEPFAR funding during 2004-2014 reported significant associations between PEPFAR-focus country status and reduced child mortality.³¹ Our significant findings for infant mortality similarly may stem from the fact that our study period (2004-2014) was longer than for most previous studies and may have been better able to investigate the outcome of funding over a longer period in the past than was possible in previous studies. Our significant findings may also stem from 1 or more other factors: funding for PMTCT may be more directly associated with infant mortality than overall PEPFAR funding; our study was not confounded by country-level factors; and Kenya received more PEPFAR funding than most other PEFPAR countries and may have experienced correspondingly greater benefits, providing greater statistical power to detect associations than in previous studies. Despite its association with improvements in infant mortality and HIV testing during ANC, PEPFAR funding was not associated with neonatal mortality, perhaps because neonatal mortality may be more resistant than postneonatal mortality to public health interventions.⁹

Annual funding was not associated with reduced infant mortality in the year of allocation but became beneficial at later lags (Figure 1). Although not significant, similar trajectories were observed for HIV testing during ANC and neonatal mortality, suggesting that the full outcome of annual funding may not be observable for several years across many outcomes. This delay may reflect logistical delays as PEPFAR funds are absorbed by local PMTCT programs but may also reflect biological realities. There is a 9-month lag between conception and birth and another 12-month lag between birth and ascertaining infant survival. Both logistic and biological factors should be considered when defining a program evaluation’s time horizon. Our finding that annual funding is significantly associated with infant mortality, but not with HIV testing during ANC, may suggest that other PMTCT-related activities, such as provision of antiretroviral therapy to HIV-positive mothers and exposed infants, are more proximally associated with reduced child mortality. However, because cumulative funding was positively associated with increased HIV testing during ANC, it is also possible that sustained investments in PMTCT programs, rather than high levels of annual funding, are necessary to result in increased access to HIV testing during ANC.

When using cumulative funding rather than annual funding, we observed more significant associations after shorter lags. Compared with annual funding, cumulative funding may better reflect sustained investments in infrastructure and personnel. The 2- and 3-year lagged cumulative PCFs were not significantly associated with infant mortality; however, these estimates should be interpreted with caution, as many of the 95% CIs for these associations between PEPFAR funding levels and infant mortality were wide and the point estimates still suggest a strong association between funding and reduced mortality. Alternatively, the association of cumulative funding with infant mortality may fade over time. The significant nonlinear association between 3-year lagged cumulative funding and HIV testing during ANC may suggest a threshold effect for cumulative PCF. However, in our data extracted from the Country Operational Reports, as of 2014 only 1 of 8 provinces (Nyanza) had received more than $2.68 in cumulative PCF; this apparent threshold may be driven largely by a single province. The joint findings of the annual PCF and cumulative PCF models suggest that there is a delayed association of annual funding with PMTCT-related outcomes, that the association of cumulative funding is greater than the association of spending in any individual year, and that there is no evidence of diminishing returns to PEPFAR’s investment in PMTCT programs.

Our findings were relatively robust to sensitivity analyses assessing possible bias owing to missing funding data. Only under unlikely scenarios where missing province-years received very high annual PCF ($0.56-$0.93) did our results for infant mortality become attenuated across all lags. Discrepancies between the main analysis and sensitivity analyses either reflected the detection of new nonlinear associations or they occurred under unlikely scenarios in which missing province-years received very high annual PCF. These analyses suggest that, despite limitations of COPs as a source of regional funding data, our findings are unlikely to be explained by missing data patterns.

When we examined whether the association between PEPFAR funding and infant mortality differed by mothers’ HIV status, PEPFAR was not associated with significantly lower infant mortality among infants of HIV-positive mothers; however, this analysis was limited by the small sample of mothers with known HIV status and because HIV status was ascertained at the time of the interview, which would have been prior to the child’s birth, and could have induced misclassification of women’s HIV status during pregnancy. In contrast, infants of HIV-negative mothers experienced reduced mortality, possibly reflecting positive spillover effects including improved health literacy, health professional training, and health system resources.

This study illustrates how cost-effective, large-scale program evaluations can be conducted using preexisting data sources. We relied exclusively on COPs, which are produced annually by all PEPFAR focus countries, and the Demographic and Health Survey and the AIDS Indicator Survey. At least 1 Demographic and Health Survey or AIDS Indicator Survey has been conducted in all 31 countries producing COPs. Consequently, our approach may be applied in other geographical settings and adapted for other HIV-related health outcomes. PEPFAR and other international donors seeking to evaluate programmatic outcomes might consider routinely collecting annual data on financial expenditures disaggregated by geography. Linking these financial data to the Demographic and Health Survey, AIDS Indicator Survey, or ongoing PEPFAR-funded population-based HIV impact assessment surveys would enable future evaluations of outcomes. Recently released subnational-level data on PEPFAR’s programmatic activities for 2015-2016⁴ could also be used in future evaluations.

Investigating the association between the amount of funding and health outcomes addresses policy questions about the effectiveness of programs and can inform allocation of future funds. Our study follows 3 previous studies investigating the amount of PEPFAR funding as an exposure and resulting associations with related health and behavioral outcomes, with varying findings.^27,28,33 For example, Lo and colleagues³³ did not observe an association between funding for abstinence and being faithful activities, which are designed to delay age of first sexual activity and reduce individuals’ number of sexual partners, and reduced high-risk sexual behavior. Access to a range of literature on the associations between funding for specific HIV prevention approaches and population-level health improvements can help policy makers decide which evidence-informed activities to fund and increase allocation of funds to approaches likely to be effective.

This study has limitations. As with many observational studies, our analysis is vulnerable to bias owing to residual confounding and missing exposure data, making the establishment of causality difficult. We sought to minimize unmeasured confounding by adjusting for fixed effects of province, which controls for unobserved time-invariant province characteristics, and for year, which controls for nationwide secular trends. However, our analysis is vulnerable to confounding by province characteristics that both vary over time and are correlated with PEPFAR funding for PMTCT, such as investment in antimalarial campaigns or child health programing, both of which increased drastically during the early 2000s.^2,32 We strove to partially capture time-varying province characteristics by adjusting for individual variables, such as mosquito net ownership. However, bias from time-varying confounding is possible, particularly if PEPFAR preferentially allocated funds to implementing partners working in provinces with improving health outcomes. As evidenced by their relatively wide 95% CIs, there is still substantial uncertainty around many of our point estimates, and we may not have had sufficient power to detect all important associations. In addition, the data source for our exposure, COPs, describes annual planned expenditures, which may not reflect actual expenditures, and disaggregate funding by implementing partner rather than by geography, limiting our ability to assign funding to provinces. We expect these administrative processes to result in random measurement error and underestimate associations. However, systematic bias could occur if, for example, implementing partners working in provinces with strong health systems reported higher-quality information in COPs than those working in provinces with weak health systems. Furthermore, our analysis links province-level funding to respondent-level outcomes and does not account for variation in the distribution of PEPFAR-funded activities within a province or variation in individuals’ engagement with PEPFAR-funded PMTCT programs. Thus, these findings are best interpreted as the benefits associated with living in a province receiving a given level of PEPFAR funding and likely underestimate the benefits of interacting directly with a PEPFAR-funded PMTCT program. Similarly, because health outcome data were unavailable after 2014, our analysis does not capture the outcome of recent PEPFAR-funded PMTCT-related activities. Finally, although we investigated multiple definitions of the funding measures, we did not formally adjust for multiple comparisons. Although not all our findings reach statistical significance, the point estimates for the association of funding with both infant mortality and HIV testing in ANC generally provide a consistent picture from a mechanistic point of view, which helps to strengthen confidence in our findings. By presenting our results for these analyses in graphical and tabular form, we hope to allow readers to consider the totality of the evidence when interpreting our results.

Funding by PEPFAR for PMTCT was associated with reduced infant mortality and increased HIV testing during ANC in Kenya. This research illustrates how preexisting data sources can be used to conduct cost-effective, large-scale program evaluations in a robust and timely manner.

Accepted for Publication: June 30, 2019.

Published: September 13, 2019. doi:10.1001/jamanetworkopen.2019.11318

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Barnhart DA et al. JAMA Network Open.

Corresponding Author: Donna Spiegelman, ScD, Center for Methods in Implementation and Prevention Science, Department of Biostatistics, Yale School of Public Health, 60 College St, Ste 201, New Haven, CT 06510 (donna.spiegelman@yale.edu).

Author Contributions: Drs Barnhart and Spiegelman had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Barnhart, Tsikhutsu, Kirui, Sawe, Muli, Sugut, Birx, Hamm, Coakley, Hickey, Spiegelman.

Acquisition, analysis, or interpretation of data: Barnhart, Kirui, Abboud, Birx, Hickey, Wolfman, Lee, Spiegelman.

Drafting of the manuscript: Barnhart, Kirui, Sawe, Muli, Abboud, Coakley, Hickey.

Critical revision of the manuscript for important intellectual content: Barnhart, Tsikhutsu, Sugut, Birx, Hamm, Hickey, Wolfman, Lee, Spiegelman.

Statistical analysis: Barnhart, Lee, Spiegelman.

Obtained funding: Spiegelman.

Administrative, technical, or material support: Tsikhutsu, Kirui, Muli, Sugut, Abboud, Birx, Coakley, Hickey, Wolfman, Lee, Spiegelman.

Supervision: Sawe, Hickey, Spiegelman.

Conflict of Interest Disclosures: Drs Barnhart and Spiegelman reported receiving grants from the National Institutes of Health during the conduct of the study. Drs Hamm, Coakley, Wolfman, and Lee reported receiving grants from US Army Medical Research Acquisition Activity during the conduct of the study. Dr Hickey reported receiving organizational funding support from US Army Medical Research Acquisition Activity and the US President’s Emergency Plan for AIDS Relief (PEPFAR) during the conduct of the study and institutional and programmatic funding support related to the work described in this article from the US government, through PEPFAR. Dr Lee reported receiving grants from PEPFAR and the US Department of State during the conduct of the study. Dr Spiegelman reported receiving grants from the US National Institute of Environmental Health Sciences and grants from US National Institute of Allergy and Infectious Diseases during the conduct of the study. No other disclosures were reported.

Funding/Support: This study was funded by grant DP1ES025459 from the US National Institute of Environmental Health Sciences and grant R01AI112339 from the US National Institute of Allergy and Infectious Diseases, by PEPFAR, and by a cooperative agreement (W81XWH-07-2-0067) between the Henry M. Jackson Foundation for the Advancement of Military Medicine Inc and the US Department of Defense.

Role of the Funder/Sponsor: The funding sources had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation or approval of the manuscript. The Military HIV Research Program and Office of the Global AIDS Coordinator reviewed the manuscript prior to publication; the decision to submit the manuscript remained with the investigators.

Disclaimer: The views expressed are those of the authors and do not necessarily represent the official positions of the US government, the US Army, or the Department of Defense.