Secondary Logo

Journal Logo

Epidemiology and Social

Unconditional cash transfers for clinical and economic outcomes among HIV-affected Ugandan households

Mills, Edward J.a; Adhvaryu, Achyutab; Jakiela, Pamelac; Birungi, Josephined; Okoboi, Stephend; Chimulwa, Teddy Nabwire W.d; Wanganisi, Jonathane; Achilla, Tinaf; Popoff, Evang; Golchi, Shirina; Karlan, Deanh,i,j

Author Information
doi: 10.1097/QAD.0000000000001899

Abstract

Introduction

Recent scale-up of antiretroviral therapies (ARTs) has led to extensive delivery of ART to people living with HIV/AIDS. However, most public health efforts are aimed at improving the clinical care of people with HIV, and the economic burden associated with HIV infection is often unaddressed. HIV infection, especially in resource-limited settings, has profound economic costs [1,2]. Household welfare may be affected by the increased economic burden, resulting from both lost productivity due to poor health and from the loss of working family members [3]. In resource-limited settings, healthcare costs and the opportunity costs of lost productivity can amount to more than half of annual income per capita [4,5]. It has been estimated that HIV infection in Africa can lead to a 30–60% decrease in household income, a quadrupling of healthcare spending, and as much as a 50% decrease in the amount spent on food and school fees [6]. To cope with the financial demands associated with HIV, households may liquidate savings, business income, and household assets [7]. Moreover, a previous study in Uganda has also demonstrated a link between HIV infection and increased food insecurity at the household level that has important ramifications for children's human capital accumulation via school attendance and performance [8].

Various strategies, such as orphan adoption, community support groups, and microcredit, have been utilized to reduce the financial burden on HIV-affected families [9,10]. Microcredit has received considerable support from policy-makers seeking ways to improve economic outcomes among HIV-effected households [11–14]. However, people living with HIV/AIDS can often have difficulty accessing microcredit due to stigma or because borrowers or potential joint liability partners fear that future health expenses will limit their ability to repay loans [13]. A potential alternative to microcredit has been conditional cash grants, either as one-off transfers or on a recurring basis as part of a social protection program [15,16]. Conditions associated with receiving cash grants may include health behaviors, such as attending clinics or maintaining adherence to treatment, or social behaviors, such as school attendance. Conditional cash grants have been shown to improve psychological wellbeing and to increase schooling attainment and reduce HIV infection among adolescent girls [15]. Another approach to condition-based cash grants is unconditional receipt of cash; unconditional cash grants can improve targeting if some vulnerable individuals are unable to meet the requirements for conditional transfers [16,17].

To evaluate the effects of unconditional cash grants on both the clinical outcomes for people living with HIV/AIDS infection and the economic outcomes of HIV-affected households in rural Uganda, we conducted a 2 × 2 factorial randomized trial evaluating the effect of unconditional cash grants with and without financial assistance counseling. The goal of the study was to identify workable strategies for delivering cash grants to a marginalized population that may have lasting effects on income-generating capabilities. In this article, we report on health-related outcomes as well as outcomes related to household food security.

Methods

Trial design

In this 2 × 2 factorial trial, we recruited participants from two rural districts of Uganda, Masindi, and Soroti. This study was conducted in partnership with The AIDS Support Organization (TASO). TASO is a nationally representative HIV/AIDS clinical delivery and support organization that provides care to more than 100 000 HIV-infected patients and their families through 11 regional centers and 54 public health facilities across the country [18–21]. The TASO Masindi center provides HIV treatment to over 3800 HIV-positive patients within a catchment area that covers the districts of Masindi, Buliisa, Hoima, Nakasongola, and Kibale in the Albertine region of Uganda. TASO Soroti initiated the ART program in August 2006, and currently provides ART to over 5900 patients. Its catchment area is the Teso region of eastern Uganda, comprising the districts of Soroti, Kumi, Katakwi, Amuria, and Kaberamaido. TASO Soroti operates in an area where people were formally living in internally displaced camps.

The study was submitted to and approved by the TASO Uganda Research Ethics Committee, the University of Ottawa Ethics Review Board in Canada, and the Innovations for Poverty Advisory (IPA) Review Board. The protocol is registered at the Ugandan National Council of Science and Technology. All study participants provided informed consent. This trial was registered with the American Economic Association Registry for Randomized Controlled Trials (AEARCTR-0000049).

Participants

Participants were recruited through TASO centers, including clinics and Community Drug Distribution Point meetings [22] and through community outreach home visits to participants who had not visited a center recently. The trial was described as a lottery, and participants were aware they could be allocated to one of four possible arms.

To be recruited for the study, participants had to be HIV-positive, aged 18–60 years old (between the legal age of maturity and the retirement age), living in Masindi or Soroti District, and receiving antiretroviral care from TASO. Only one person per household was permitted to participate.

Interventions

Participants were randomized to one of four interventions arms: Unstructured Grant (T1), Mental Planning + Grant (T2), Pure Control (T3), and Expectations/Control (T4). Details about each arm are provided in Table 1. In brief, T1 and T2 are the intervention groups; participants assigned to those arms received unconditional cash grants with or without financial counseling. T3 and T4 arms are the control groups. Participants in the T3 control arm were told that they would not receive grants, while those in T4 arm were told that they would receive cash grants 12 months after the start of the program (after the follow-up survey). The cash grants for T1 and T2 were 350 000 Ugandan Shillings (equivalent to about 140 USD), approximately equivalent to 2 months of average household income; grants were distributed in 2013–2014 in a single payment. Figure 1 displays the allocation of the intervention and describes the intervention arms.

Table 1
Table 1:
Details to intervention arms.
Fig. 1
Fig. 1:
Participant flow diagram.

Outcomes

Baseline data was collected prior to intervention. Three surveys were done during the study to measure outcomes before and after intervention, and a follow-up survey was conducted approximately one year after the intervention.

Our primary outcomes were changes in CD4+ cell count, sexual behavior, and adherence to ART. Secondary outcomes were changes in health expenditures (seven questions addressing household spending on health), household food security and adult mental health. Sexual behavior was measured using a sexual behavior questionnaire designed by the Joint Clinical Research Centre, Uganda (comprising six questions on sexual activity, partners, and condom use). The questionnaire was derived from the Marriage and Sexual Activity module of the Uganda Demographic and Health Survey. Adherence was measured using 3-day adherence recall to ART. Household food security measured using the Household Hunger Scale, a 6-question scale that we dichotomized as food insecure or not. For regression analysis, we used health expenditure on a continuous scale as health expenditure over the previous 3 weeks, while the other scales were considered binary (due to noncombinability) with sexual behavior assessed as condom used or not during last sexual activity, adherence as meeting a threshold of more than 95% possession ratio or not, and food insecurity. Adult mental health was measured using the Kessler-10 Mental Health Scale. Mental health outcomes will be reported separately.

Recruitment and randomization

Recruitment and randomization took place between October 2013 and May 2014. Participants were followed until July 2016. 2170 eligible patients were randomized in this study (Fig. 1).

Using computer generated random numbers, we stratified randomization by district (Masindi and Soroti), by gender (male and female), and by age (18–35, 36–50, and 51–65) resulting in 12 blocks (2 × 2 × 3).

Blinding

Allocation was concealed from recruiters, participants, and study staff, but randomization posttreatment was not blinded due to the nature of this study.

Statistical methods

We analyzed this trial as a 2 × 2 factorial design using an intent-to-treat approach. Factorial designs permit assessment of an individual intervention effect and allow for the analysis to determine if additive effects exist for interventions with greater intensity [23,24]. In our primary analysis, we collapsed both treatment and control arms and tested for differences within the collapsed groups [25]. For our secondary analysis, we performed a between-group analysis to determine whether adding financial counseling and planning to the unconditional cash grant treatment changed the way cash gets used and then downstream effects (i.e. T1 compared to T2), and to determine whether expectations of a future lump-sum cash transfer changed pretransfer investment behavior (i.e. T3 compared to T4). Finally, we tested whether our factorial assumption of combinability was correct by combining T1 and T3 compared to T2 and T4.

Bayesian models were used to analyze the data. The endline CD4+ cell counts were modeled as Poisson counts with latent Gaussian intensities (constrained to be positive) linked to the treatment effects with age and gender as covariates through a linear model. Two-way interactions between treatment and age, as well as two-way interactions between treatment and gender were also included in the model. Health expenditure was modeled as a Gaussian random variable with the linear model used as its mean. The binary outcomes were linked to the linear model using a logistic function. The baseline CD4+ cell counts were also used in the model when available. The results of a previous Kenya study on microfinance loans were used as an informative prior for CD4+ cell counts [26]. For the other outcomes and other model parameters, noninformative priors were used. Further details of the Bayesian model and prior specification with a prior sensitivity analysis are included in the supplementary material, http://links.lww.com/QAD/B296. We used the prior information on the effect of cash grants on endline CD4+ cell counts to specify the sample size [26]. To be conservative, however, we assume a smaller mean difference (100) than estimated in a previous study (165) and use a standard deviation of 500. Under these assumptions to achieve 90% power and 5% type I error rate, the sample size for each arm needs to be about 525 patients. The available resources allowed for a larger sample size than this minimum requirement. Therefore, a total sample size of 2000 in the four-arm design is a conservative figure that assures retaining power and controlling error rates in a more complex design.

Role of funding source

The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Results

We randomized 2170 patients to participate, with 1081 receiving a cash grant. Baseline characteristics were similar between intervention groups (Table 2). Over the 24-month study period, a total of 77 participants were lost to follow-up (19, 23, 22, 13 in T1, T2, T3, and T4, respectively), and 24 participants died during the trial (6, 7, 5, 6 in T1, T2, T3, and T4, respectively).

Table 2
Table 2:
Baseline characteristics.

The effect sizes of unconditional cash grants for primary analysis (i.e., T1 + T2 vs. T3 + T4) are presented in Table 3, and the distribution of effects are shown in Figure 2. We found no significant intervention effects on CD4+ cell count changes; the mean change in CD4+ cell count was 35.48 [95% Credible Intervals (CrIs) −59.91 to 131.62] cells/μl and the probability of a positive intervention effect for CD4+ cell count was 0.77 (Fig. 2a). The mean change in health expenditure was $2.65 (95% CrIs −9.30 to 15.69) dollars spent in the previous 3 weeks with 0.66 probability of a positive intervention effect (Fig. 2b). We found no improvements on sexual behavior, (Fig. 2c, 0.10 probability of a positive effect) and food security (Fig. 2c, 0.67 probability of positive effect). Moreover, we found 0.91 probability of a positive intervention effect for ART adherence (Fig. 2c) but the OR for ART adherence overlapped the null effect of 1 (OR: 3.15, 95% CrIs: 0.58 to 18.15). In our secondary analyses (frequentist analysis and comparison of individual arms), we found no important intervention effects for all study outcomes (Tables A1–A4 in supplementary materials, http://links.lww.com/QAD/B296). Our assumption of factorial combinability was demonstrated. The effect of mental planning on CD4+ cell count, where we identified a change of 104.20 (CrI: 5.99–202.16) cells between groups, with a probability of 0.98, favoring mental planning (Table 4) (Table A2, supplementary material, http://links.lww.com/QAD/B296).

Table 3
Table 3:
Primary analysis results (T1 + T2 vs T3 + T4).
Fig. 2
Fig. 2:
Estimated effect of distribution of cash grants on (a) CD4+ cell count (b) health expenditure, and (c) sexual behavior, adherence, and food security.Adhere, adherence; food, food security; sexual, sexual behaviors.
Table 4
Table 4:
Secondary analysis results (individual arms vs. pure control arm).

Discussion

We aimed to assess whether immediately improving household income with unconditional cash grants would lead to improvements on outcomes related to health and household security. We examined whether cash transfers of a sizeable amount (approximating 2 months adult income in these settings), coupled with financial counseling, can lead to improved health outcomes and household food security. In our study, we found no important intervention benefits of the cash grants on health outcomes. We found that the mental planning component of the trial may have increased CD4+ cell count, although we did not elucidate why. These results should be of interest to decision-makers, especially to those interested in providing cash transfers or microcredit as a public health intervention to improve clinical and household economic outcomes among HIV patients.

This study contributes to a disparate understanding of the role of credit market interventions for HIV/AIDS targeting programs. A recent pilot cluster-RCT from Kenya that enrolled 140 patients receiving a water pump, locally coordinated and condition-based loan (150 USD), and education on sustainable farming practices and financial management found positive treatment effects on virological suppression and CD4+ T-cell rebound as well as improved household food security [26]. Previous conditional cash transfer studies that have assessed the role of microcredit on the risk of HIV acquisition among HIV negative individuals have shown disparate effects [15,27–30]. Given the vast improvements in the standard practices for HIV support counseling over the three plus decades of the HIV epidemic, it is possible that standard of care, which entails services like intensive community support provided by TASO, may already have led to improvements on health outcomes, and additional interventions, such as microcredit, may only lead to marginal improvements.

A potential concern for policy makers is that the recent randomized trials on microfinance for non-HIV specific populations have demonstrated no effects or modest effects, at best, on household income [31,32]. It is important to recognize that in one study, microcredit to participants with prior business experience led to strong business growth [31]. However, this still poses a problem for HIV/AIDS targeting programs because even trials that showed modest benefits associated with microfinance may be an overestimate of the likely impacts on HIV-positive populations despite their preexisting economic vulnerability.

Our trial has strengths and limitations. Strengths include the novel interventional approach that reflected real-life use of cash, and consideration of outcomes important to both households and policy makers. However, our study was limited in the absence of viral load monitoring. Viral monitoring, which we recognize as an important clinical measure, is not part of routine clinical care within TASO, and instead CD4+ cell count and adherence recall were used. Viral load monitoring is a more clinically useful measure of treatment response than CD4+ measurement or adherence recall. We asked specific questions about sexual behaviors and it is possible that participants did not provide truthful answers due to stigma or other reasons. We found, for example, that questions on condom use were frequently skipped. We also recognize that it is possible that placing conditions on the use of cash transfers may have led to different results; however, the use of unconditional cash grants allowed us to evaluate whether giving choice to the individuals in need could improve their health and household outcomes. We permitted only one family member per household to participate. However, given the complex inter-household networks in Uganda, it is possible that in some circumstances more than one household member received an intervention. Another limitation may be that we did not limit the study eligibility to the person who manages the overall finances of the household, as it is possible that giving grants to the person in charge of the household may have led to different outcomes in terms of household security. There is a well recognized sense of community in many settings in Africa, whereby individuals assist each other with financial necessities. Our trial examined the expenditure of recipients and also participants on healthcare for others and the receipt of healthcare financial assistance. It is possible that finances were spent on assisting others on nonhealthcare issues and that this sharing of resources contributed to muted effects on individual health. Our study examined a large unconditional cash transfer in a one-time disbursement. It is possible that behavior change may have occurred had we provided smaller cash supplements over time, as more traditional cash transfer programs typically do, such as the Brazilian Bolsa Familia Programme [33], perhaps the best known long-term cash transfer program. The long terms cash transfer programs tend to be condition based and therefore test a different hypothesis than ours. Our trial identified a treatment effect on CD4+ cell count change from mental planning. However, this was identified during a variety of sensitivity analyses and, therefore, should be explored in future trials.

In conclusion, our study does not find an effect of unconditional cash transfers on a number of measured health outcomes among HIV-infected patients in this Ugandan setting. Substantial increases in income did not lead to increased spending on health or improved health outcomes in our sample. Results could differ in other settings, particularly since the individuals in our study were already being served by TASO, which provides them with reliable healthcare and reliable access to antiretroviral drugs. Nonetheless, our results suggest that a range of interventions that directly or indirectly increase income in the short-term should not be expected to importantly improve health outcomes for HIV-positive populations.

Acknowledgements

We would like to thank the staff at TASO that have helped to oversee and conduct this trial. Jonna Bartfelt, Mamta Pimoli, and Gean Spektor provided excellent research assistance.

Author contributions: E.M., A.A., P.J., J.B., W.J., T.A., E.P., S.G., D.K. contributed to study concept and design. E.M., A.A., P.J., J.B., W.J., T.A., S.O., T.C., E.P., S.G., D.K. contributed to acquisition, analysis, or interpretation of data. All authors drafted the report. All authors contributed to critical revision of the report for important intellectual content. E.M., A.A., E.P., and S.G. did the statistical analysis. E.M. and D.K. obtained funding. All authors provided administrative, technical, or material support. E.M., J.B., S.O., and D.K. supervised the study.

Conflicts of interest

Declaration of interest: D.K. is a founder of Innovations for Poverty Action (IPA), who administered the study. J.B., S.O., T.C., W.J., and T.A. are/were employees of TASO, and oversaw the study and permitted access to patients.

Statement of data availability: De-identified data is available to qualified investigators upon request.

Registration: American Economic Association Registry for Randomized Controlled Trials (AEARCTR-0000049)

Funding Source: The Canadian Institutes of Health Research (CIHR).

References

1. Gillespie S, Greener R, Whiteside A, Whitworth J. Investigating the empirical evidence for understanding vulnerability and the associations between poverty, HIV infection and AIDS impact. AIDS (London, England) 2007; 21 (suppl 7):S1–S4.
2. Taleski SJ, Ahmed K, Whiteside A. The relationship between economic evaluations and HIV and AIDS treatment policies. Curr Opin HIV AIDS 2010; 5:204–209.
3. Veenstra N, Whiteside A. Economic impact of HIV. Best Pract Res Clin Obstet Gynaecol 2005; 19:197–210.
4. Mahal A, Canning D, Odumosu K, Okonkwo P. Assessing the economic impact of HIV/AIDS on Nigerian households: a propensity score matching approach. AIDS (London, England) 2008; 22 (suppl 1):S95–101.
5. Tekola F, Reniers G, Haile Mariam D, Araya T, Davey G. The economic impact of HIV/AIDS morbidity and mortality on households in Addis Ababa, Ethiopia. AIDS Care 2008; 20:995–1001.
6. McDonagh A. Microfinance strategies for HIV/AIDS mitigation and prevention in sub-Saharan Africa. No. 993514113402676. International Labour Organization, 2001.
7. Donahue J, Kubbucho K, Osinde S. Responding to a silent economic crisis among microfinance clients in Kenya and Uganda. Nairobi, Kenya: Microsave; 2001.
8. Bukusuba J, Kikafunda JK, Whitehead RG. Food security status in households of people living with HIV/AIDS (PLWHA) in a Ugandan urban setting. Br J Nutr 2007; 98:211–217.
9. Bennell P. The impact of the AIDS epidemic on the schooling of orphans and other directly affected children in Sub-Saharan Africa. J Development Studies 2005; 41:467–488.
10. Decroo T, Van Damme W, Kegels G, Remartinez D, Rasschaert F. Are expert patients an untapped resource for ART provision in sub-Saharan Africa?. AIDS Res Treat 2012; 2012:749718.
11. de Walque D, Dow WH, Nathan R, Abdul R, Abilahi F, Gong E, et al. Incentivising safe sex: a randomised trial of conditional cash transfers for HIV and sexually transmitted infection prevention in rural Tanzania. BMJ Open 2012; 2:e000747.
12. Fernald LC, Hamad R, Karlan D, Ozer EJ, Zinman J. Small individual loans and mental health: a randomized controlled trial among South African adults. BMC Public Health 2008; 8:409.
13. Mohindra KS, Haddad S. Women's interlaced freedoms: a framework linking microcredit participation and health. J Hum Dev 2005; 6:353–374.
14. Mohindra KS, Haddad S, Narayana D. Can microcredit help improve the health of poor women? Some findings from a cross-sectional study in Kerala, India. Int J Equity Health 2008; 7:2.
15. Baird SJ, Garfein RS, McIntosh CT, Ozler B. Effect of a cash transfer programme for schooling on prevalence of HIV and herpes simplex type 2 in Malawi: a cluster randomised trial. Lancet (London, England) 2012; 379:1320–1329.
16. Haushofer J, Shapiro J. The short-term impact of unconditional cash transfers to the poor: Experimental Evidence from Kenya. Q J Econ 2016; 131:1973–2042.
17. Haushofer J, Shapiro J. Household response to income changes: evidence from an unconditional cash transfer program in Kenya. Cambridge, MA: M.I.T.; 2013.
18. Bakanda C, Birungi J, Mwesigwa R, Ford N, Cooper CL, Au-Yeung C, et al. Association of aging and survival in a large HIV-infected cohort on antiretroviral therapy. AIDS (London, England) 2011; 25:701–705.
19. Bakanda C, Birungi J, Mwesigwa R, Nachega JB, Chan K, Palmer A, et al. Survival of HIV-infected adolescents on antiretroviral therapy in Uganda: findings from a nationally representative cohort in Uganda. PloS One 2011; 6:e19261.
20. Chu R, Mills EJ, Beyene J, Pullenayegum E, Bakanda C, Nachega JB, et al. Impact of tuberculosis on mortality among HIV-infected patients receiving antiretroviral therapy in Uganda: a prospective cohort analysis. AIDS Res Ther 2013; 10:19.
21. Mills EJ, Bakanda C, Birungi J, Chan K, Hogg RS, Ford N, et al. Male gender predicts mortality in a large cohort of patients receiving antiretroviral therapy in Uganda. J Int AIDS Soc 2011; 14:52.
22. Okoboi S, Ding E, Persuad S, Wangisi J, Birungi J, Shurgold S, et al. Community-based ART distribution system can effectively facilitate long-term program retention and low-rates of death and virologic failure in rural Uganda. AIDS Res Ther 2015; 12:37.
23. McAlister FA, Straus SE, Sackett DL, Altman DG. Analysis and reporting of factorial trials: a systematic review. JAMA 2003; 289:2545–2553.
24. Montgomery AA, Peters TJ, Little P. Design, analysis and presentation of factorial randomised controlled trials. BMC Med Res Methodol 2003; 3:26.
25. Montgomery AA, Astin MP, Peters TJ. Reporting of factorial trials of complex interventions in community settings: a systematic review. Trials 2011; 12:179.
26. Weiser SD, Bukusi EA, Steinfeld RL, Frongillo EA, Weke E, Dworkin SL, et al. Shamba Maisha: randomized controlled trial of an agricultural and finance intervention to improve HIV health outcomes. AIDS (London, England) 2015; 29:1889–1894.
27. Pettifor A, MacPhail C, Hughes JP, Selin A, Wang J, Gomez-Olive FX, et al. The effect of a conditional cash transfer on HIV incidence in young women in rural South Africa (HPTN 068): a phase 3, randomised controlled trial. The Lancet Global health 2016; 4:e978–e988.
28. Yotebieng M, Thirumurthy H, Moracco KE, Kawende B, Chalachala JL, Wenzi LK, et al. Conditional cash transfers and uptake of and retention in prevention of mother-to-child HIV transmission care: a randomised controlled trial. The lancet HIV 2016; 3:e85–93.
29. Pronyk PM, Kim JC, Abramsky T, Phetla G, Hargreaves JR, Morison LA, et al. A combined microfinance and training intervention can reduce HIV risk behaviour in young female participants. AIDS (London, England) 2008; 22:1659–1665.
30. Pronyk PM, Hargreaves JR, Kim JC, Morison LA, Phetla G, Watts C, et al. Effect of a structural intervention for the prevention of intimate-partner violence and HIV in rural South Africa: a cluster randomised trial. Lancet (London, England) 2006; 368:1973–1983.
31. Banerjee A, Breza E, Duflo E, Kinnan C. Do credit constraints limit entrepreneurship? Heterogeneity in the returns to microfinance. Evanston, IL: Department of Economics Northwestern University; 2015.
32. Karlan D, Zinman J. Microcredit in theory and practice: using randomized credit scoring for impact evaluation. Science 2011; 332:1278–1284.
33. Rasella D, Aquino R, Santos CA, Paes-Sousa R, Barreto ML. Effect of a conditional cash transfer programme on childhood mortality: a nationwide analysis of Brazilian municipalities. Lancet (London, England) 2013; 382:57–64.
Keywords:

factorial randomized trial; HIV; microfinance; unconditional cash grant

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc.