In sub-Saharan Africa, HIV infection rates remain disproportionately high among young women . In 2006, HIV prevalence was nearly 30% among women attending public antenatal clinics in South Africa, with 90% of all infections in those less than 35-years old . Although conditions of poverty and gender inequality continue to pose major challenges to HIV prevention efforts [3–8], there has been limited experience in the design and testing of interventions that ask whether and in what contexts poverty reduction and gender empowerment programs might contribute to HIV risk reduction.
We recently conducted the Intervention with Microfinance for AIDS and Gender Equity (IMAGE) study, a cluster-randomized trial which assessed the effect of a structural intervention combining group-based microfinance with a gender and HIV training curriculum on HIV risk behavior and intimate partner violence (IPV). Over a 2-year period, we observed improvements in economic well being and multiple dimensions of empowerment among program participants . Furthermore, levels of IPV were reduced by 55% .
Intervention effects on HIV risk were more complex to evaluate, as microfinance participants were generally older women (median age 42 years) and outside the high-risk age group for HIV infection. For statistical reasons, a previous per-protocol analysis reported only indirect intervention effects on HIV risk behavior among young people (aged 14–35 years) residing in the households and communities where the program was offered. In these groups, we observed only modest improvements in household communication, and no effect on sexual behavior or HIV incidence .
The question remains as to whether the intervention might have influenced HIV risk among intervention participants themselves. To address this, we analyzed quantitative data on HIV risk behavior collected from young women who were direct participants in the IMAGE intervention. To help contextualize our findings we also assessed complementary qualitative data from this group.
The study was conducted in a densely settled rural area of South Africa's Limpopo Province. Villages were between 2 and 20 km from a main trading center and major sources of income included government grants, local public sector employment, and migrant remittances. Subsistence agriculture is not a viable option for most households in the area.
Key components of the intervention and the study design are described elsewhere [9–11]. The IMAGE intervention consisted of two components. The first was group-based microfinance, in which groups of five women received loans to establish small businesses. Further credit was offered when all women in these ‘solidarity groups’ repaid their loans. The second component consisted of a gender and HIV training curriculum, which was integrated into established meetings of 40 women that took place every 2 weeks for approximately 1 year.
A cluster-randomized design was used to assess intervention effects. Briefly, eight villages were pair matched by size and accessibility, with one from each pair randomly selected to receive the microfinance and training intervention. For each woman joining the intervention, a woman of similar age and poverty status was randomly selected from comparison villages into the control group. All intervention participants were women. The trial was registered with ClinicalTrials.gov (number NCT00242957) and received ethical approval at the London School of Hygiene and Tropical Medicine and the University of the Witwatersrand.
Surveys were conducted at baseline and after 2 years of follow-up. This analysis compared HIV risk behavior reported by the subgroup of women aged 14–35 years who were direct intervention participants with women of the same age and poverty status from comparison villages. In this paper, we examine the same set of indicators previously assessed among young people in the households and communities of intervention participants as part of a per-protocol analysis strategy .
Intervention effects were assessed using a cluster level analysis to compare the intervention group to the comparison group. Crude measures of effect (prevalence or risk ratios, identified as RR) were calculated by entering log village level summaries, weighted by village denominator, into an analysis of variance model that included terms for intervention and village pair. In order to control for possible baseline imbalances between groups, adjusted measures of effect [adjusted risk ratio (aRR)] were calculated using a two-stage process. First, using a logistic regression model fitted to individual level data from control villages, expected outcomes were derived for each village based on the age, marital status, and baseline measure of the outcome indicator of each respondent . Standardized village level summaries of the ratio of observed to expected outcomes were then entered into an analysis of variance model as described above. All statistical analyses were performed using Stata version 9.0 (StataCorp, College Station, Texas, USA).
Thematic content analysis of qualitative data collected during the study allowed further assessment of intervention effects on young women. One hundred and five transcripts from multiple sources were analyzed, including nonparticipant observation of microfinance loan center meetings (160 women followed over 1 year); focus group discussions (FGD) (conducted at two points in time with eight loan groups, each consisting of five participants); key informant interviews (conducted with eight loan recipients over a 3-year period); and, diaries of IMAGE training facilitators kept over the duration of the study. All data were translated, transcribed, and entered into a qualitative database (Nud*ist version 6.0, QSR International, Doncaster, Australia).
The intervention reached 10% of poor households in the study villages in line with program targets and standard microfinance practice in sub-Saharan Africa . A detailed process evaluation conducted alongside the trial suggested high levels of participation and program retention among loan recipients .
A total of 262 women were under 35 years at study onset and eligible for surveys on HIV risk behaviors. Of these, 83% (108/130) of the intervention group and 85% (112/132) of the control group were successfully interviewed. Two-year follow-up rates among those interviewed at baseline were 92 and 79%, respectively. There were no significant differences in baseline characteristics between the groups, or between those retained and lost to follow-up, and the mean age of respondents was 29 years. Summary of quantitative and qualitative results are presented in Tables 1 and 2, respectively.
Effects on HIV-related knowledge and communication
At follow-up, both intervention and control groups demonstrated an increase in HIV-related knowledge. There was some evidence to suggest that women participating in the intervention felt more comfortable discussing sexual matters at home, although this was not statistically significant.
Qualitative data suggested that open discussions about sexuality and HIV were initially taboo in many households, and that women struggled to find ways to raise these issues with family members. Attempts to communicate with children rather than sexual partners appeared more frequently in the data, suggesting that these were easier to initiate. However, over time, as participants internalized the real threat that HIV posed to their families, women began to overcome this discomfort, and tried to communicate in concrete ways about the importance of condom use and voluntary counseling and testing for HIV (VCT). Quantitative data note significant increases in communication about sex or HIV within the homes of IMAGE participants compared with controls [aRR 1.46 95% confidence interval (CI) 1.01–2.12].
Effects on uptake of voluntary counseling and testing for HIV
Between both the intervention and comparison groups, VCT uptake at baseline was low (11%) despite widespread availability of rapid testing at the primary healthcare level. At follow-up, quantitative findings documented a significantly higher proportion of intervention participants reporting they had undergone VCT relative to those in the comparison group (aRR 1.64, 95% CI 1.06–2.56).
Qualitative data reflected the general fear surrounding VCT, which often centered upon the social stigma and emotional distress associated with a positive result. Although there was little evidence that this fear diminished over time, findings suggest that participants nonetheless began to find ways to motivate both themselves and others to go for VCT.
Effects on sexual behavior
At baseline, only 77/220 (35%) women were married, and few women (3%) in either group reported having had more than one sexual partner in the last year. There was no difference in numbers of reported partnerships between intervention and comparison groups at follow-up. Qualitative data suggested that whereas multiple partnerships for both men and women might be part of the broader social context, few personal accounts of changes in such relationships emerged.
However, though levels of unprotected sex at last intercourse with a nonspousal partner were overall high, they were significantly lower among young women in the intervention group relative to the comparison group at follow-up (aRR 0.76, 95% CI 0.60–0.96). Qualitative data collected during loan center meetings suggested women readily acknowledged the challenges they faced when using condoms with sexual partners. Typical reasons for resistance to their use included the association of condoms with mistrust between partners, questions regarding their effectiveness, and complaints of reduced sexual pleasure and intimacy. Despite these challenges, data from focus group discussions and key informant interviews (Table 2) indicated a sense of enhanced bargaining power among intervention participants, which in a number of instances was expressed as increased confidence in negotiating safer sex and the successful introduction of condom use with male partners.
Previous research has suggested that an intervention combining microfinance with a gender and HIV training curriculum can lead to improvements in household economic well being, women's empowerment, and to reductions in levels of IPV [9,10]. This analysis provides further evidence that the intervention may also have influenced HIV risk behavior among younger women who received the intervention, in whom we observed increases in HIV-related communication and VCT uptake alongside reductions in levels of unprotected sex.
A number of factors are important to consider in interpreting these results. Although we cannot exclude response bias in the context of an intervention, qualitative data suggest participants openly admitted to realistic obstacles as well as opportunities for behavior change, providing plausible narratives that complement quantitative measures of intervention effect. We also note the encouraging potential for synergy between several of the outcomes assessed, as previous research has highlighted the protective effect of VCT on sexual risk behavior in some settings [13–16], and the importance of greater communication about sex in facilitating behavior change [17–19].
The analysis was also subject to several limitations. Although we attempted to ensure women in the two arms were similar in terms of age and poverty, there may have been important unmeasured differences affecting both the response to the intervention and the generalizability of the findings. There was also a higher level of nonresponse among the comparison group at follow-up, which could potentially bias the results, though there were no significant differences between this group and those retained in the study. Finally, though data on HIV infection was collected as part of the main study, it was not possible to examine differences in HIV incidence due to low numbers of new infections (n = 8) in this sub-group.
These findings suggesting promising effects on HIV risk behavior among young program participants contrast with prior analyses in which more modest indirect effects were observed among young people living in the households or communities when the intervention was offered. There are a number of potential reasons for this. First, indirect effects in the latter group would have to occur through diffusion from those receiving the intervention to the wider community via mentorship, education, or participation in community activities. Because the time for recruitment and for participants to receive the full intervention package was on an average 18 months, the opportunity for such diffusion to take place over the 2–3 year study duration was limited. Second, though the intervention reached 10% of eligible households, this may have been insufficient to generate wider effects. Third, as the program targeted the poorest, it may not have reached key opinion leaders in target communities. Finally, though the intervention effects may have been evident among direct intervention participants, social mobilization may have been insufficiently robust to stimulate wider community level effects on HIV risk behavior.
The present research highlights the potential for structural interventions that address the economic and social vulnerability of women to contribute to measurable health gains, including reductions in levels of IPV and high-risk sexual behavior. Although the relative contributions of the economic and educational dimensions of our intervention remain the subject of further study, it is clear that addressing women's immediate financial needs provided an important incentive for maintaining sustained contact with a gender-focused HIV prevention program in an area where few such opportunities exist.
However, how to best deliver integrated health and development interventions is not always straightforward. Recent research from southern Africa where microfinance-based programs target exclusively younger clients as a means of addressing HIV risk have met with mixed success . Young women are often more mobile, less socially rooted, and less experienced in establishing income-generating enterprises than the usual profile of older microfinance clients. The economic viability of these pilot initiatives has suffered, and in such settings, integrating health components can be a tremendous challenge . When viewed alongside the results of our study, addressing HIV risk behaviors may be better achieved as a result of partnerships with well established microfinance programs working with diverse age groups in vulnerable communities, rather than specifically tailoring novel interventions to reach high-risk groups.
Finally, though microfinance may be one strategic entry point for integrating economic and health interventions, there are likely many others– from schools and workplace programs, to incentive-based initiatives linking cash transfers to participation in health programs [22–24]. Our findings raise intriguing questions about the potential synergy of such combined approaches and highlight the need for further innovation and operational research.
P.M.P. was the principal investigator of the study and project leader in South Africa, led the drafting of this manuscript and contributed to all aspects of the study. J.C.K. was responsible for the development and implementation of the training component of the intervention, the design and validation of survey instruments for gender violence, and contributed to the analysis. T.A. performed the statistical analysis. G.P. was the team leader in South Africa for the qualitative component of the study and conducted the qualitative analysis for this study. J.R.H. was responsible for the study design, field management of survey teams, data management, overall quantitative analysis, and assessment of process indicators. L.A.M. contributed to the overall design of the study and provided major support for the statistical analysis. C.W. provided support to the training intervention, the design of survey tools, and the analysis. J.B. provided technical support to the qualitative research team. J.D.H.P. participated in the initial conceptualization of the intervention, the trial design, and advised on most aspects of the study. All authors contributed to the drafting of this manuscript.
We would like to thank Prof. John Gear for his support throughout, the Managing Director of Small Enterprise Foundation, John de Wit, and the many staff who have made this work possible.
We declare that we have no conflict of interest.
We have received support from AngloAmerican Chairman's Educational Trust, AngloPlatinum, Department for International Development (UK), The Ford Foundation, The Henry J. Kaiser Family Foundation, HIVOS, South African Department of Health and Welfare, The Swedish International Development Agency.
1. UNAIDS. AIDS epidemic update. Geneva: UNAIDS; 2006.
2. Department of Health. National HIV and Syphilis prevalence survey: South Africa 2006. Pretoria: Department of health, South Africa; 2007.
3. Piot P, Greener R, Russell S. Squaring the circle: AIDS, poverty and human development. PLoS Med 2007; 4:1571–1575.
4. Wellings K, Collumbien M, Slaymaker E, et al
. Sexual behavior in context: a global perspective. Lancet 2006; 368:1706–1728.
5. Fenton L. Preventing HIV/AIDS through poverty reduction: the only sustainable solution. Lancet 2004; 364:1186–1187.
6. Parker RG, Easton D, Klein CH. Structural barriers and facilitators in HIV prevention: a review of international research. AIDS 2000; 14(suppl 1):S22–S32.
7. Buve A, Bishikwabo-Nsarhaza K, Mutangadura G. The spread and effect of HIV-1 infection in sub-Saharan Africa. Lancet 2002; 359:2011–2017.
8. Weiser SD, Leiter K, Bangsburg DR, Butler LM, Percy-de Korte F, Hlanze Z, et al
. Food insufficiency is associated with HIV-risk sexual behavior among women in Botswana and Swaziland. PLoS Med 2007; 4:1589–1598.
9. Kim JC, Watts CH, Hargreaves JR, Ndhlovu LX, Phetla G, Morison LA, et al
. Understanding the impact of a microfinance-based intervention on women's empowerment and the reduction of intimate partner violence in the IMAGE Study, South Africa. Am J Public Health 2007; 97:1794–1802.
10. 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 randomized trial. Lancet 2006; 368:1973–1983.
11. RADAR. Social Interventions for HIV/AIDS: intervention with microfinance for AIDS and gender equity. IMAGE Study Intervention Monograph No. 2. In: Rural AIDS and Development Action Research Programme, School of Public Health, University of the Witwatersrand. http://web.wits.ac.za/NR/rdonlyres/3C2A3B30-DE20-40E0-8A0AA14C98D0AB38/0/Intervention_monograph_picspdf.pdf
; 2002. [Accessed 30 January 2008].
12. Grosskurth H, Mosha F, Todd J, Mwijarubi E, Klokke A, Senkoro K, et al
. Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania: randomized control trial. Lancet 1995; 346:530–536.
13. Voluntary counselling and testing efficacy group. Efficacy of voluntary HIV-1 counselling and testing in individuals and couples in Kenya, Tanzania, and Trinidad: a randomised trial
. The Voluntary HIV-1 Counseling and Testing Efficacy Study Group. Lancet
14. van der Straten A, King B, Grinstead O, Serofilira A, Allen S. Couple communication, sexual coercion and HIV risk reduction in Kigali, Rwanda. AIDS 1995; 9:935–944.
15. Kamenga C, Ryder RW, Jingu M, Mbuyi N, Mbu L, Behets F, et al
. Evidence of marked sexual behavior change associated with low HIV-1 seroconversion in 149 married coupled with discordant HIV-1 serostatus: experience at an HIV counseling centre in Zaire. AIDS 1991; 5:61–67.
16. Allen S, Tice J, Van der Perre P, Serufilira A, Hudes E, Nsengumuremyi F, et al
. Effect of serotesting with counseling on condom use and seroconversion among HIV discordant couples in Africa. BMJ 1992; 304:1605–1609.
17. McBride DC, Freier MC, Hopkins GL, Babikian T, Richardson L, Helm H, et al
. Quality of parent-child relationship and adolescent HIV risk behavior in St. Maarten. AIDS Care 2005; 17:S45–S54.
18. DiClemente RJ, Wingood G, Crosby R, Cobb BK, Harrington K, Davies SL. Parent-adolescent communication and sexual risk behaviors among African American adolescent females. J Pediatr 2001; 139:407–412.
19. Halpern-Felsher BL, Kropp RY, Boyer CB, Tschann JM, Ellen JM. Adolescents' self efficacy to communicate about sex: its role in condom attitudes, commitment, and use. Adolescence 2004; 39:443–456.
20. Epstein H. The Invisible Cure: Africa, the West and the Fight Against AIDS. London: Viking/Penguin publishers; 2007.
21. Pronyk PM, Hargreaves JR, Morduch J. Microfinance and better health: prospects for sub-Saharan Africa. JAMA 2007; 298:1925–1927.
22. IFPRI. PROGRESA: breaking the cycle of poverty. Washington DC: International Food Policy Research Institute; 2002.
23. Morris SS, Flores R, Olinto P, Medina JM. Monetary incentives in primary healthcare and effects on use and coverage of preventive healthcare interventions in rural Honduras: cluster randomised trial. Lancet 2004; 364:2030–2037.
24. Lagarde M, Haines A, Palmer N. Conditional cash transfers for improving uptake of health interventions in low- and middle-income countries. JAMA 2007; 298:1900–1910.