Can money prevent the spread of HIV? A review of cash payments for HIV prevention

Abstract

Cash payments to improve health outcomes have been used for many years, however, their use for HIV prevention is new and the impact not yet well understood. We provide a brief background on the rationale behind using cash to improve health outcomes, review current studies completed or underway using cash for prevention of sexual transmission of HIV, and outline some key considerations on the use of cash payments to prevent HIV infections. We searched the literature for studies that implemented cash transfer programs and measured HIV or HIV-related outcomes. We identified 16 studies meeting our criteria; 10 are completed. The majority of studies have been conducted with adolescents in developing countries and payments are focused on addressing structural risk factors such as poverty. Most have seen reductions in sexual behavior and one large trial has documented a difference in HIV prevalence between young women getting cash transfers and those not. Cash transfer programs focused on changing risky sexual behaviors to reduce HIV risk suggest promise. The context in which programs are situated, the purpose of the cash transfer, and the population will all affect the impact of such programs; ongoing RCTs with HIV incidence endpoints will shed more light on the efficacy of cash payments as strategy for HIV prevention.

Background

Thirty years into the global HIV epidemic, effective methods to prevent new HIV infections remain limited to biomedical interventions. Recently, several effective, new interventions, including male circumcision, oral and vaginal pre-exposure prophylaxis, and treatment as prevention have emerged.^1–5 If taken to scale, these interventions have the potential to significantly reduce the number of new infections globally; however, there are numerous barriers to the widespread uptake and impact of these interventions which include economic, behavioral, structural and biological factors.^6–9 At the structural level, barriers to successful implementation of biomedical interventions may include financial barriers to individual uptake or to government ability to offer interventions, lack of skilled personnel to offer new interventions, and/or cultural norms that are not supportive of the behaviors required for uptake or use of interventions. Effective combination prevention approaches will require interventions that address structural and behavioral risk factors to have a maximum impact on the epidemic.

Recently, the use of cash payments to reduce HIV risk, either by addressing structural risk factors such as poverty or by incentivizing behavior change, has emerged as a novel prevention tool receiving significant attention.^10–12 While the use of cash to improve health outcomes has a long history in development, social protections, and psychology (e.g. contingency management), less is known about its potential impact on HIV prevention. We provide a brief background on the rationale behind using cash as a means to improve health outcomes, review current studies that have been completed or are underway using cash for HIV prevention, and outline some key considerations with regard to the use of cash payments as a means to prevent new HIV infections.

Historical Overview

Programs that use cash payments to improve individual well-being have been in existence for a number of decades. Some of the first programs using cash to improve health outcomes come from the development and social protection fields where cash transfers have traditionally been implemented to alleviate the impact of poverty.¹³ Cash transfer programs are currently estimated to reach over 1 billion people in the developing world.¹⁴ These payments have been both unconditional (payments normally go to households deemed ‘poor’ but individuals are not required to do anything to receive payments) and conditional (payments are tied to behaviors deemed beneficial to the individual). It is theorized that providing cash to poor households improves health outcomes by making health care, food, or education more affordable by increasing household income.¹⁴ Such programs have aimed to increase uptake of prevention health services (e.g. antenatal care services, immunizations), and to improve growth outcomes, primarily in children. Overall these programs have been effective in increasing the use of preventive health services although the evidence on improving health outcomes is mixed. ^15,16

Behavioral economists also theorize that improving recipients’ socio-economic status and access to better food and educational opportunities can affect recipients’ expectations of their life expectancy, outlook for the future, and preferences for ‘healthier behaviors.¹³ Further, they hypothesize that cash transfers conditioned on investments in social goods such as education and health may offset myopia whereby individuals may invest more in goods or services with immediate rewards rather than those where benefits may only vest in the future, such as education.^13,14 Building on behavioral economic theory, lessons learned from research on incentives and health suggest include findings that small, tangible incentives provided frequently and close to the observed outcome are more effective than larger payments made less often.¹⁷

Psychologists have also used cash to improve health outcomes through contingency management (CM). Such interventions are based on the theory that behaviors targeted for change should be monitored frequently and rewarded with tangible incentives when desired behavior change is demonstrated.¹⁸ Contingency management strategies have been used to address a number of unhealthy behaviors for the past 30–40 years, including smoking cessation¹⁹, obesity management²⁰, and most commonly substance abuse^21–23, and have shown varying levels of success. Lessons learned from CM interventions are similar to those from the behavioral economics literature and also emphasize the importance of measuring readily detectible target behaviors, and providing tangible incentives when desired behaviors are demonstrated and withholding when behaviors not demonstrated.²⁴ Some other strategies used in CM found to be effective for controlling substance abuse include escalating incentives that reset to zero when the desired behavior is not achieved and the use of lottery or “fish-bowl” drawing based rewards.²⁵ The efficacy of such methods has yet to be tested in the HIV prevention field.

Methods

To provide a comprehensive summary of the evidence regarding cash for HIV prevention, we aimed to review all studies that have been completed or are underway that use cash or financial incentives and measure HIV or HIV-related outcomes. We included studies which aim to reduce the risk of sexual transmission of HIV by either providing participants with cash transfers (both conditional and unconditional); providing incentives for particular risk reduction outcomes; or reducing financial barriers to schooling. Schooling is the focus of many cash transfer programs due to the numerous benefits found to be associated with increased years of schooling, in particular among women. Women with more education have been found to be at lower risk of HIV infection, have fewer children, and have greater earning potential.²⁶ We identified studies for inclusion in multiple ways: to identify completed studies, we searched the PubMed and EconLit databases using search terms (cash transfer, cash incentive, cash reward, monetary reward, economic assets, contingency management, or school fee) and (HIV, STD, STI, pregnancy, or sexual behavior); to identify unpublished studies and studies currently underway, we also searched the websites of The World Bank, the NIH Research Portfolio Online Reporting Tool, archived abstracts from AIDS and APHA conferences (2000–2011), and relied on personal communications. From each of the studies identified, we abstracted information about the target population (age, sex), the study (design, sample size, location), the intervention (transfer type, conditionality), and the major results, if available.

Results

Review of the evidence

As of April 2012, we identified 16 studies that aimed to reduce HIV risk with cash transfers, providing incentives for particular risk reduction outcomes, or reducing financial barriers to schooling (Table 1). Nine of the studies were completed in the last seven years, six are underway, and one was completed more than twenty years ago. We posit that these interventions fall into one of two categories with regard to their mechanism of action: 1) interventions that provide cash to address up-stream structural risk factors for HIV (e.g. payments to relieve poverty or increase education) and are thus more directly related to development/social protection programs or 2) interventions that provide cash incentives for immediate measureable outcomes related to HIV (e.g. cash for HIV tests or negative HIV or STD test results) and are thus most similar to contingency management. The majority of studies fall into category 1: that is, they aim to address larger structural barriers to HIV prevention by reducing poverty or alleviating economic barriers to schooling. Two studies, CAPRISA 007 and Yo Puedo, fall into both categories, as they aim to address structural risk factors through cash transfers for school-related activities, while also providing cash rewards for specific outcomes such as HIV testing or negative pregnancy tests.^27,28 The majority of the studies are conducted in developing countries (14/16) and focus on adolescents (15/16). Nine studies provide cash on a conditional basis, six provide cash on an unconditional basis, and one examines the effect of both. The purpose of the transfer varies by study (and conditionality or not), but include: school attendance, school completion, poverty alleviation, and completion of health promotion activities, such as STI/HIV testing. The majority of studies are randomized controlled trials (15/16) and thus controlled for confounding in the study design. The only observational study controlled for appropriate factors at a range of levels.²⁹ Studies measure a variety of outcomes, including sexual behaviors and intentions (12/16), STI status (2/16), and HIV status (5/16). Overall, the majority of studies that have measured a change in sexual behaviors found a positive impact (9/10); however, one pilot study found a negative impact in men (Malawi Incentives Project): immediately after receiving their cash incentive, men reported more sex acts (although they also reported more condom use).³⁰ Only one large RCT, the SIHR trial, has released results on HIV outcomes. At the 18-month follow up, the study found lower HIV prevalence among the intervention participants compared to the controls.³¹ Three large RCTs are currently underway among adolescents in South Africa (CAPRISA 007 and HPTN 068) and Tanzania (Iringa Combination HIV Prevention Trial) to determine the impact of cash transfers on HIV incidence ^27,32,33.

Table 1.

Studies providing cash payments/incentives to reduce HIV risk behaviors as of April 2012 (n=16)

Target Population		Primary author	Study Years	Location	Study Base	Study Design	Sample Size	Study Name	Transfer Type	Conditionality	Main Outcome	Measure of Effect
Orphans and Vulnerable Adolescents and Youth	Females and males Average age of 13.7 Grade 6–7	Ssewamala43	2005–2008	Uganda	School based	Cluster randomized controlled trial Randomized at school level	260 individuals 15 schools	SUUBI	Savings account for secondary schooling or a family business	Conditional Attending 12 asset building and financial planning workshops Contributing to matched savings	Intention to engage in sexual risk taking behavior	β: −1.64 (95% CI: −2.92, −0.35)**
	Females 10–16 years of age Grade 6	Hallfors44	2007–2009	Zimbabwe	School based	Cluster randomized controlled trial Randomized at school level	329 individuals 25 schools		Assistance with school costs	Unconditional	School dropout	Adjusted OR: 8.48 (95% CI: 3.6,198)***1
	Females and males 12–14 years of age	Cho45	2008–2009	Kenya	School based	Cluster randomized controlled trial Randomized at household level	105 individuals 79 households		Assistance with school costs (uniforms and fees and money to address other problems resulting in absenteeism)	Unconditional	School dropout	Absolute change: 8% point decrease**
											Begin sexual intercourse	Absolute change: 14% point decrease*
	Females and males 15–24 years of age	Handa46	2007–2011	Kenya	Population based	Randomized controlled trial	6,000 individuals (estimate)	Kenya Cash Transfer for Orphans and Vulnerable Children (CT-OVC)	Cash transfer for parent/guardian	Unconditional	Sexual behavior	Study ongoing
	Females and males 13–17 years of age Poor female-headed household or household with OVC	American Institutes for Research47	2011–2013	Zambia	Population based	Cluster randomized controlled trial Randomized at the Community Welfare Assistance Committees (CWACS) level	2000 individuals (estimate) 90 CWACS	Zambia Vulnerability grant	Cash transfer for guardian/caregiver/household head	Unconditional	Sexual behavior	Study ongoing
Adolescents	Females 13–22 years of age Never married Students in school at baseline ("school girl") Students dropped out of school at baseline ("drop out")	Baird31	2008–2009	Malawi	School based	Cluster randomized controlled trial Randomized at the enumeration area level	1,289 individuals 176 enumeration areas	Zomba Cash Transfer Program (also known as Schooling, Income, and HIV Risk (SIHR))	Assistance with school costs and cash transfer for parent/guardian and child	3 treatment arms Conditional on school attendance for drop out; conditional on school attendance for school girl Conditional on school attendance for drop out; unconditional for school girl Conditional on school attendance for drop out; no transfer for school girl	HIV prevalence	Adjusted OR: 0.36 (95% CI: 0.14,0.91)**2
											HSV-2 prevalence	Adjusted OR: 0.24 (95% CI: 0.09,0.65)**2
	Females 13–20 years of age Grade 8–11	Pettifor32	2011–2013	South Africa	School based	Randomized controlled trial	2,900 individuals (estimate)	Swa Koteka (HPTN 068)	Cash transfer for parent/guardian and child	Conditional School attendance	HIV incidence	study ongoing
											Sexual behavior
	Females and males 13 years of age and older Grade 9–10	Karim27	2010–2012	South Africa	School based	Cluster randomized controlled trial Randomized at the school level	4,000 individuals (estimate) 14 schools	CAPRISA 007 (Reducing HIV in Adolescents (RHIVA))	Cash transfer for child	Conditional Improved academic performance School attendance Passing examinations HIV test Participating in after school program	HIV incidence	Study ongoing
											Sexual behavior
	Females and males 16–21 years of age Same-aged members of their social network Self-identify as Latino	Minnis28	2010–2012	San Francisco, United States	Population based	Randomized controlled trial (pilot study) Randomized at the social network level	180 individuals (estimate) 60 social networks	Yo Puedo: Future Opportunities for Youth	Cash transfer for child	Conditional (participant defined) Educational and skill development Participation in reproductive health activities	Use of reproductive health care services	Study ongoing
											Pregnancy and childbearing attitudes and norms
											Behaviors associated with pregnancy
	Females and males Average age: 14 Grade 6	Duflo48	2003–2005	Kenya	School based	Cluster randomized controlled trial Randomized at the school level	70,000 individuals 328 schools		Assistance with school costs (school uniforms)	Unconditional	Childbearing (girls)	Absolute change: 1.5% point decrease* Relative change: 10% decrease*
	Females Primiparous Under age 18	Stevens-Simon49	1991–1993	United States	Clinic based	Randomized controlled trial	286 individuals	Dollar-a-Day Program	Cash transfer for individual	Conditional (2 groups) Remaining non-pregnant Remaining non-pregnant AND participation in weekly peer support groups	Repeat pregnancy	Absolute change: 4% point decrease† Relative change: 9% decrease†
Adolescents and Adults	Females and males 15 years of age and older	Thornton36	2004	Malawi	Population based	Randomized controlled trial	2,812 individuals	Malawi Diffusion and Ideational Change Project (MDICP)	Cash transfer for individual tested	Conditional Learning HIV status	Learning HIV test results	Absolute change: 43% point increase***
	Females and males 12–24 years of age	Galárraga29	2004	Mexico	Population based	Observational	3,743 individuals	Oportunidades	Cash transfer for parent/guardian	Conditional Participation in health promotion activities School attendance	Sexually active (girl)	Absolute change: 6.1% point increase
											Condom at last sexual intercourse (girls)	Absolute change: 17.3% point increase
											Sexually active (boys)	Absolute change: 32.6% point decrease
											Condom at last sexual intercourse (boys)	Absolute change: 64.8% point increase
	Females and males 16–75 years of age	Kohler30	2006	Malawi	Population based	Randomized controlled trial Randomized at the individual or couple level	1,307 individuals	Malawi Incentives Project (part of the Malawi Diffusion and Ideational Change Project (MDICP))	Cash transfer to individual tested	Conditional • Negative HIV test after 1 year	HIV incidence (women)	No effect (no estimate given)
											Condom use (women)	No effect (Absolute change: 0.0% points)
											Risky sexual behavior (women)	Absolute change: 6.7% point decrease*
											HIV incidence (men)	No effect (no estimate given)
											Condom use (men)	Absolute change: 5.2% point increase*
											Risky sexual behavior (men)	Absolute change: 9.0% point increase**
	Females 15–24	Celentano and Kerrigan33	2012–2015	Tanzania	Population based	Community cluster randomized controlled trial	12,000 individuals in larger trial 24 clusters	Iringa Combination HIV Prevention Trial	Cash transfer for parent/guardian and child	Unconditional	HIV incidence	Study ongoing
											Sexual behavior 1,800 15–24 women/girls for CT behavioral outcomes
Adults	Females and males 18–30 years of age, plus all spouses 16 and over	de Walque50	2009–2010; ongoing follow-up through Spring 2011	Tanzania	Population based	Randomized controlled trial	2,399 individuals	RESPECT (Rewarding STI Prevention and Control in Tanzania)	Cash transfer to individual tested	Conditional Negative STI test	STI incidence (Chlamydia trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis, Mycoplasma genitalium)	Adjusted RR high-value transfer arm to control: 0.73 (95% CI: 0.47,0.99)** Adjusted RR high-value transfer arm to low-value transfer arm: 0.69 (95% CI: 0.45,0.92)**

¹The referent in the adjusted OR (AOR) are girls not exposed to the cash transfer

²No significant difference in effect among conditional versus unconditional intervention groups, or between individuals enrolled in school at baseline and individuals who had already dropped out.

^*P value ≤ 0.10

^**P value ≤ 0.05

^***P value ≤ 0.01

^†No P value given

Discussion

Programmatic considerations of cash payment interventions

Despite the promise of interventions that provide cash to reduce HIV risk, it is not clear that there is a one-size-fits-all cash payment intervention or whether such interventions will be effective in reducing HIV infection across populations. To date only one study has found a decrease in HIV prevalence related to cash payments (primarily because few studies to date have had biological endpoints), although the vast majority of studies have found positive impacts on sexual behaviors.

The mechanism whereby cash reduces risk depends heavily on the target population and how cash may (or may not) influence HIV risk factors for that population. Thus, how a cash payment intervention should be structured—whether it should be conditioned and on what, the frequency of payments, the amount of the payment, and to whom it is made—depends heavily on these factors. For example, conditioning payments on school attendance may only be relevant in settings where there are financial barriers to schooling and where schooling appears to be protective against HIV. Formative research on incentives for HIV prevention conducted in Malawi has confirmed theory from behavioral economics and contingency management that smaller payments made more frequently and closer to the behavior being observed are more effective than larger payments in the future.^24,34

As we describe, there are two main types of cash/incentive for behavior change, one that aims to address upstream drivers of risk, such as poverty and education, and the other that provides cash for the downstream behavior change itself. The majority of studies aim to address upstream, structural barriers that increase HIV risk such as education or poverty. These studies hypothesize that improving the socio-economic situation of vulnerable populations or providing cash payments conditioned on social goods, such as school attendance, will reduce HIV risk (category 1 interventions). Other studies hypothesize that providing cash for specific outcomes like a negative STI test will serve as an incentive for individuals not to engage in high risk behavior (category 2 interventions). However, it is unclear whether interventions premised on rewarding specific HIV-related outcomes actually address factors that place individuals at risk. While there is the assumption that cash payments will serve as motivation for HIV/STI reduction behaviors, there is some evidence that the relationship between cash incentives and behavior might be more complex. The Malawi Incentives Project found payments for negative HIV tests increased risk in men, suggesting that giving cash to individuals may have unintended consequences.³⁰

The amount of the payment likely will matter in determining whether behavior change occurs or not. It is still unclear what payment amount is necessary for desired behavior change and how the type (cash vs. in-kind payment) and frequency of payment influence outcomes. Research conducted in Malawi to date suggests that even a small incentive can encourage uptake of interventions and behavior change; however, the applicability of these findings to wealthier and more urban settings is unknown. At the end of the day what matters is which intervention type is more effective in preventing new infections—likely this will not be black and white and different designs may work differently for different populations. This complexity speaks to the importance of formative and ethnographic research and pilot studies in potential study populations to understand pathways that lead to risk and how cash transfers or incentives might best be structured to reduce risk.

Who gets the payment also likely matters with regard to cash payments. For young women who are at high risk of HIV infection in sub-Saharan Africa, recent evidence from the SIHR trial suggests that addressing structural factors by providing cash to young women reduces their risk.³¹ It appears that providing young women with access to their own income was important in reducing HIV risk, as it enabled them to make safer choices in sex partners—young women who received cash irrespective of whether or not it was conditional, were less likely to have older partners or exchange sex for money. Many existing cash transfer programs do not provide cash payments to minors; rather the parent/guardian receives the payment. Given the potential impact on HIV and sexual behavior of direct payment to adolescents, the results of the CAPRISA 007 and HPTN 068 studies, which provide cash payments to adolescents, may have implications for future recipients of cash transfer programs.

While some cash transfer interventions do not address up-stream factors that affect HIV risk (category 2 interventions), in the new prevention landscape, there are circumstances where these interventions may play a significant role. Treatment as prevention,⁵ Pre-exposure Prophylaxis (PrEP),^1,4 and Prevention of Mother to Child Transmission³⁵ all require individuals to test for HIV as a necessary first step. Cash incentives may be one way to increase testing numbers. Evidence from Malawi suggests that cash incentives are effective in encouraging individuals to receive their HIV test results.³⁶ Adherence to drugs is also a critical component of the efficacy of these interventions; studies have been conducted looking at the role of contingency management in improving ART adherence, including HPTN 065/TLC PLUS,³⁷ which is currently testing the impact of using cash transfers for HIV-related health visits and adherence to ART.^38–40 Clearly, there are important behavioral and structural barriers to individuals testing for HIV or to adhering to medication that need to be addressed; cash incentive programs may not address these factors but may still have an impact on behavior.

While providing cash to individuals is hypothesized to reduce risk, concerns have been raised that individuals receiving the cash could be placed at risk for negative outcomes such as violence, bullying, coercion, or using the cash for high-risk activities. Similar concerns were raised when government social welfare programs provided payments to female instead of male heads of households. Concerns were raised that women would suffer negative consequences as a result of receiving the payment, however, to date, the evidence shows that payments made to female heads of have not resulted in increased intimate partner violence. In fact, women are the more effective beneficiaries with regard to multiple health and education outcomes compared to men.^41,42 Though there is no evidence to date of social harm to individuals as a result of providing cash transfers, existing programs should carefully monitor changes in intergenerational and gender relationships for such events. Infusing cash into communities can create complex dynamics; thus, the need for transparency about selection criteria and intended purpose for receiving the cash is paramount.

Conclusion

Preliminary data from cash payment interventions to reduce HIV risk suggest that they might be effective, particularly among young women. As always, concerns over scalability come into play with implementing cash transfer programs. However, with large social welfare programs in place in many countries, including those hard hit by the HIV epidemic, if cash payment programs are found to be effective these programs could be tailored to address scale up and cost of implementation. Ongoing research from randomized controlled studies will provide information on whether cash payment programs are a cost-effective strategy for preventing new HIV infections and what role they may play in the larger prevention agenda.