Evolving Strategies, Opportunistic Implementation: HIV Risk Reduction in Tanzania in the Context of an Incentive-Based HIV Prevention Intervention 102451 Laura Packel1*, Ann Keller2, William H. Dow2, Damien de Walque3, Rose Nathan4, Sally Mtenga4 1 Global Health Sciences, University of California, San Francisco, San Francisco, CA, USA, 2 School of Public Health, University of California, Berkeley, Berkeley, CA, USA, 3 World Bank Development Research Group, The World Bank, Washington DC, USA, 4 Ifakara Health Institute, Dar es Salaam, Tanzania Abstract Background: Behavior change communication (BCC) interventions, while still a necessary component of HIV prevention, have not on their own been shown to be sufficient to stem the tide of the epidemic. The shortcomings of BCC interventions are partly due to barriers arising from structural or economic constraints. Arguments are being made for combination prevention packages that include behavior change, biomedical, and structural interventions to address the complex set of risk factors that may lead to HIV infection. Methods: In 2009/2010 we conducted 216 in-depth interviews with a subset of study participants enrolled in the RESPECT study - an HIV prevention trial in Tanzania that used cash awards to incentivize safer sexual behaviors. We analyzed community diaries to understand how the study was perceived in the community. We drew on these data to enhance our understanding of how the intervention influenced strategies for risk reduction. Results: We found that certain situations provide increased leverage for sexual negotiation, and these situations facilitated opportunistic implementation of risk reduction strategies. Opportunities enabled by the RESPECT intervention included leveraging conditional cash awards, but participants also emphasized the importance of exploiting new health status knowledge from regular STI testing. Risk reduction strategies included condom use within partnerships and/or with other partners, and an unexpected emphasis on temporary abstinence. Conclusions: Our results highlight the importance of increasing opportunities for implementing risk reduction strategies. We found that an incentive-based intervention could be effective in part by creating such opportunities, particularly among groups such as women with limited sexual agency. The results provide new evidence that expanding regular testing of STIs is another important mechanism for providing opportunities for negotiating behavior change, beyond the direct benefits of testing. Exploiting the latent demand for STI testing should receive renewed attention as part of innovative new combination interventions for HIV prevention. Citation: Packel L, Keller A, Dow WH, de Walque D, Nathan R, et al. (2012) Evolving Strategies, Opportunistic Implementation: HIV Risk Reduction in Tanzania in the Context of an Incentive-Based HIV Prevention Intervention. PLoS ONE 7(8): e44058. doi:10.1371/journal.pone.0044058 Editor: Patrick S Sullivan, Rollins School of Public Health, Emory University, United States of America Received January 29, 2012; Accepted July 30, 2012; Published August 27, 2012 Copyright: ß 2012 Packel et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The study was funded by the World Bank Research Committee, the Spanish Impact Evaluation Fund and the Knowledge for Change Program managed by the World Bank, the William and Flora Hewlett Foundation through the Population Reference Bureau, the Berkeley Population Center (R21 HD056581), and a CDC dissertation grant. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: ljpackel@gmail.com Introduction now being made for combination prevention packages that include behavior change, biomedical, and structural interventions in an Traditional behavior change communication interventions, while effort to address these barriers along with the complex, multilay- still a necessary component of HIV prevention, have not in and of ered set of vulnerabilities and risk factors that may lead to HIV themselves been shown to be sufficient to stem the tide of the epidemic infection [1,2,5,6]. Additionally, incentive-based interventions that [1]. In light of recent research findings showing the effectiveness of stimulate demand for HIV testing and knowledge of HIV status male circumcision and early initiation of anti-retroviral therapy (ART) among both individuals and couples, potentially increasing in preventing HIV, the prevention landscape has shifted away from referrals into male circumcision and/or treatment, are increasingly behavioral interventions toward biomedical interventions, and has being explored [7,8]. Interventions that address the structural placed a central importance on HIV testing as a gateway to more barriers to behavior change and have the potential to increase the evidence-based HIV prevention interventions [2] . effectiveness of behavioral change interventions are also on the rise The shortcomings of behavioral change interventions are at [8,9]. least in part due to barriers arising as a result of structural or Qualitative data from one such structural intervention in economic constraints on behavior change [3,4]. Arguments are Tanzania, the RESPECT trial [9], provides a unique opportunity PLOS ONE | www.plosone.org 1 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania to learn more about sexual behavior change strategies and sexual behaviors, and if so, how. Our investigation also helps to implementation in the context of a monetary incentive to engage understand how the structural cash intervention and the testing in safer sex strategies. Our hypothesis is that an economic component could work synergistically as part of a combination incentive to avoid unsafe sex, along with regular testing for prevention package that can assist men and women with sexually transmitted infections (STIs), may work in combination opportunities to better act on behavioral change intentions, thus with traditional behavior change messages to overcome some of potentially increasing the effectiveness of traditional behavior the barriers inherent in sexual behavior change, especially for change interventions. women. In particular, we explore how different components of the RESPECT intervention may have created both new leverage in Methods sexual negotiations, as well as increased opportunities for exercising that leverage. The data reported on in this paper come from the qualitative We know from previous research that individuals are frequently component of the RESPECT study. Detailed methods of the employing risk reduction strategies, but often not in the ways that RESPECT study have been reported elsewhere [9], but briefly, dovetail with the traditional sexual behavior change messages [10– the study was a three-arm randomized trial testing the effectiveness 13]. For example, strategies of married women often center of a cash incentive conditional on testing negative for a panel of around convincing their husbands to leave outside partners or to curable STIs. The intervention arm was divided into two sub-arms use condoms with outside partners, while male strategies are – a low-value award arm eligible for up to $30 over the course of focused on careful partner selection and condom use with non- the study (approximately $10 per testing round), and a high-value primary partners [11]. award arm eligible for up to $60 (approximately $20 per testing Risk reduction strategies such as condom use and reducing the round). Those amounts were determined based on focus-group number of sexual partners may be implemented inconsistently as a discussions in neighboring villages conducted before the interven- result of structural and cultural factors including gender inequality tion started, balancing sufficient incentive levels against concerns [12,14–16], women’s economic dependence on men, norms of about scalability and potential coercion. All participants were marriage [17–20], [21], desire to have children [22–25], difficulty tested for STIs at baseline and then every 4 months for one year, in assessing risk of partners [10–13] and fear of gender-based and provided free treatment if they tested positive. Participants in violence [26,27]. In fact, studies done in Tanzania and Kenya the two intervention arms were eligible to receive award incentive revealed that married women sometimes fail to disclose their HIV payments if they tested negative for curable STIs at the 4, 8, and status to their partners due to fear of abandonment, accusations, 12-month testing rounds. Inclusion criteria consisted of males and physical violence and loss of confidentiality, potentially leading to females, aged 18–30 (and spouses ages 16 or over), residing in one HIV transmission within marriage [28,29]. In two studies from of 10 study villages within the Kilombero/Ulanga districts of the Tanzania, only 17% and 40% of the women testing positive for Ifakara Health and Demographic Surveillance System (IHDSS) in HIV, respectively, had disclosed their HIV status to their partner, south-west Tanzania [33]. even after a considerable follow-up period [30,31]. The main The use of cash incentives to encourage sexual behavioral reasons for not disclosing HIV status were fear of stigma and change is an innovative approach, but not without controversy. divorce, fear of losing confidentiality, women’s lack of decision- The background and justification of the RESPECT design are making power, communication patterns between partners and discussed in more detail [34]. For a more general discussion on the male partners’ attitudes to HIV voluntary counseling and testing ethics of incentives for health promotion, particularly in low- [30,31]. income settings, see London et al [35]. The one-year RESPECT Less research has been conducted on behavior change barriers intervention resulted in a 27% reduction in STIs in the high-value that might exist for men, in large part because the epidemiology cash award arm as compared to the control arm [9] thus appeared shows higher HIV prevalence among women. Notably, one recent to impact sexual behavior. The accompanying qualitative inter- exception from the comparative anthropology perspective discuss- views were designed to elucidate the constraints on sexual es in detail the ‘‘opportunity structures’’ in place that perpetuate behavior change and the strategies used by participants to the benefits that men derive from having multiple and extramarital effectuate change. partners [32]. While barriers to behavior change for men The qualitative study participants were recruited from four of specifically is not discussed here, it is important to note that for the ten villages that were participating in the RESPECT study. men, as for women, benefits derived from engaging in risky The four qualitative villages represented a range of semi-urban to behaviors may outweigh the perceived risks of doing so. more rural, and ranged from 15 minutes to a 2-hour drive to Using qualitative data from a trial of economic incentives for Ifakara, the main urban center in the district. We used stratified testing STI negative, in this paper we explore how a cash random sampling to select the qualitative study participants at incentive, in combination with regular STI testing, influenced baseline. In each village, the strata of interest were gender, marital strategies for risk reduction. This is important to understanding status, and intervention/control group. We over-sampled from the behavioral mechanisms through which the RESPECT trial led to treatment group as we were interested in hearing more reduced STIs. The cash awards are a type of structural experiences relating to how the money did or did not motivate intervention, which we hypothesized could lead to behavioral sexual behavior change. This analysis utilizes data from in-depth change by both men and women, but it was a priori unclear what interviews conducted at the baseline, 4-month and 8-month study type of behavioral change strategies would be adopted and how. In visits. At baseline, we randomly sampled 92 trial enrollees (43 men, addition, the study’s provision of regular STI testing (in an 49 women) from four of the ten study villages to be interviewed. Of environment in which STI testing was generally not available on these 92, 80 showed up at the study site to pick up their STI results demand) allowed analysis of the mechanisms through which STI at the next study visit three weeks later, at which time the testing could operate. STI testing is understood to reduce risks qualitative interviews were conducted. Those who did not show up through epidemiological pathways by identifying treatment needs, to the study site were more likely to be male (p = 0.08), but there but in this paper we explore whether the expanded health status were no other significant differences. Of these 80 that were knowledge from STI testing led RESPECT participants to change interviewed, 66 transcripts were received (14 transcripts were lost PLOS ONE | www.plosone.org 2 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania through data management error–either the recordings were participants. Differences in demographic and other characteristics inadvertently deleted or the electronic version of the transcript noted here were examined using a Chi Square test, and differences was inadvertently deleted). were considered significant if p,0.05. At the 4 and 8-month study visits new qualitative participants The study was approved by the Institutional Review Boards of were purposively sampled in order to increase the number of the Ifakara Health Institute, the University of California at participants interviewed who had tested positive for an STI. In Berkeley, and by the Tanzania National Institute for Medical addition, we intentionally conducted fewer overall interviews at 8 Research. All participants provided written informed consent. months because we had reached saturation with the number of participants we had already enrolled. As a result, some participants Results who were interviewed in 4 months were purposely not interviewed in 8 months. The decision of which participants to drop was not Qualitative Study Population, for a total across the whole study of random; transcripts were reviewed to screen out the least 216 interviews, representing 111 individuals. Table 1 shows the cooperative respondents who were not deemed to be particularly same demographics and other characteristics for the qualitative candid, and consideration was given also to gender, intervention study population at baseline, 4 months, and 8 months, in addition group, and STI status of the participant. Transcripts were not to the overall study population at baseline. Notable differences in reviewed for content when making the decision to keep or drop a the qualitative and quantitative populations included the distribu- respondent; rather the text was reviewed for coherency of tion by intervention group (oversampling of those in the treatment responses and a willingness to talk at relative length about the groups was intentional), and marital status (the proportion of subject matter. No participants who were recruited for the married participants was lower in the qualitative population). qualitative study from those coming to the study station to pick The qualitative coding resulted in several categories of strategies up test results refused to be interviewed, however, some targeted to avoid unsafe sex, mentioned both at baseline and at the follow- qualitative respondents did not return to the study station for the up interviews. These included abstinence or periodic abstinence, follow up visits at months 4 (4.6%) and 8 (1.8%). having one partner who has tested and not using condoms, having Qualitative participants received a small cash payment equal to one partner only and using condoms with this partner, convincing approximately $3 USD at the end of each interview to reimburse your spouse to use condoms outside of the marriage, convincing them for transport and extra time spent at the study station. your spouse not to go outside the marriage, using condoms with Interviews took place at the study station in tents or secluded areas. your spouse, avoiding situations and circumstances that might lead All interviews were audio-recorded, transcribed in Kiswahili, to unsafe sexual behavior, filling time with other activities, and then translated into English. The interviews were conducted reducing the number of sexual partners overall, having less risky using an interview guide. The main topic areas covered during the partners, using the money to help convince a partner to stay safe, interviews were opinions about the study, community perceptions and separating or divorcing a current partner. about the study, strategies and/or steps for avoiding STI and Within these categories of risk reduction strategies that were getting the cash award, perceptions of the cash incentive, and coded, three prominent themes emerged from the data as future plans generally and for use of the cash incentive if received. participants discussed how they adapted these strategies to avoid The guide was revised for each follow-up visit (months 4, 8 and 12) risk in the context of their daily lives, and in the context of the and included questions about respondent’s experiences being RESPECT study. First, we introduce the centrality of regular STI enrolled in the study over the previous four months, what testing as a reliable method of overcoming the barrier of risk strategies they tried, and why or why not these strategies were assessment and discuss the combination of targeted condom use, successful. STI testing and treatment as a three-pronged approach to Following a methodology developed by Watkins and Swidler decreasing risk of infection. Second, we describe how certain termed conversational journals, we also hired ten diarists who situations provide increased leverage for sexual negotiation. Risk were ‘‘cultural insiders’’ in all of the communities in which the reduction strategies are often opportunistically implemented in RESPECT trial took place [36]. Swidler and Watkins have very these situations. Third, we explore the use of temporary abstinence successfully used this method as part of their HIV research in as a frequently mentioned risk reduction strategy. What emerges Malawi [36]. Diarists were paid $30 per filled notebook (limit one from the data is that the barriers described in the previous section notebook per month). Data collected from the conversational are addressed through innovative means of risk reduction. There journals does not include names or identifying information. The are difficulties in assessing risk, but frequent and strategic STI data analyzed for this manuscript includes diaries from February testing can be used to ease those difficulties. Furthermore, women through July 2009, from all ten study villages. who otherwise lack agency in sexual decision-making as a result of Qualitative analysis was conducted using a phenomenological marital and/or economic constraints can take advantage of new approach – relying on in-depth descriptions by study participants opportunities where they have leverage (enabled by both the cash to derive meaning and understanding of experience [37]. English awards and the new information about their sexual health from transcripts of interviews and diary entries were imported into the STI testing) to minimize risk at certain points in time. TAMS Analyzer Qualitative Coding Software. We developed codes iteratively for strategies used by study participants or Importance of STI Testing discussed by diarists to avoid unsafe sex or to protect themselves The difficulty of assessing risk of a potential or current partner, from HIV or STI infection based on descriptions from study in conjunction with the generally negative attitudes toward participants. Codes were then grouped into clusters by similarity of condom use creates a situation in which testing for HIV or STIs strategy, and all transcripts were then coded with TAMS Analyzer becomes an important tool in minimizing exposure to infection. using this set of strategy codes. We then used the software to For people who are in the early stages of a relationship, compare and analyze strategies discussed by intervention group determining when a partner is safe, and when condoms are no and by gender. The demographic and other quantitative data longer needed is not straightforward. Testing has helped included here came from the structured questionnaire conducted individuals in the RESPECT trial in making this determination. at the baseline study visit, and administered to all study In fact, one strategy mentioned frequently during the interviews PLOS ONE | www.plosone.org 3 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania Table 1. Characteristics of total study population at baseline and qualitative study population. All Study Participants Qualitative Participants at Qualitative Participants at Qualitative Participants at Variable at Baseline (%) Baseline (%) 4-month (%) 8-month (%) (N = 2399) (N = 66) (N = 95)a (N = 55)b Female 50.3 57.6 61.1 56.4 Age (mean) 26.4 (se: 0.12) 25.4 (se: 0.51) 25.5 (se: 0.43) 26.6 (se: 0.53) Marital Status Single 21.0 30.0 26.9 18.2 Married 63.5 48.5 48.4 54.5 Living Together 11.7 12.1 15.1 16.4 Divorced 4.1 6.1 7.5 7.3 Widowed 0.2 3.0 2.2 3.6 Education Level No Education 11.0 9.1 11.6 14.5 Some Primary School 77.2 80.3 74.7 81.8 Some Secondary School 11.9 10.6 13.7 3.6 Religion Muslim 38.2 47.0 41.9 38.2 Catholic 43.7 43.9 48.4 45.5 Other/None 18.1 9.1 9.7 16.4 Intervention Group High Value 25.6 31.8 30.9 18.2 Low Value 27.5 33.3 33.0 41.8 Control 46.9 34.9 36.2 40.0 HIV/STI status at baseline HIV-pos at baseline 3.5 3.0 6.3 3.6 STI-pos at baselinec 16.1 7.6 12.6 16.4 a 59 of the 95 qualitative participants interviewed at 4 months were also interviewed at baseline b 37 of the 55 qualitative participants interviewed at 8 months were also interviewed at baseline and 4 months, 9 were also interviewed only at 4 months c Based on STI tests for Chlamydia, Gonorrhea, Syphilis and Trichomonas doi:10.1371/journal.pone.0044058.t001 was a combination strategy involving condom use and testing for tests were performed only at the baseline visit and the 12-month HIV and/or STIs. For those who are married or in long-term visit of the study while STI tests were performed at all study visits relationships for whom condoms are no longer a realistic option, (4 times in total). What emerged as particularly important to the regular testing provides an opportunity to assess risk and to bring study participants was the regular opportunity to learn about their temporary security (or insecurity, depending on the results) to health status. Every four months all study participants received relationships that are often plagued with uncertainty. This 29 year- information about their own health and their partner’s health if old divorced woman (R for respondent) describes to the their partner was enrolled in the study and willing to share their interviewer (I) how she and her partner negotiate risk by relying results. These results, as the data illustrate, translated into on the regular STI testing that the RESPECT study provided. information about the risks participants face in their relationships. So while HIV testing is important, and in some ways the ultimate I: Do you think you will continue using condom until in round four or test of health status and risk exposure, the opportunity to you will use another strategy to make sure that you receive award? repeatedly check health status using a proxy measure for sexual R: We can use condom with him until there is time to examine his risk was of paramount importance. health… The qualitative data also reveal that condom use is frequently I: Therefore, do you think that you will continue using condom or you viewed as a temporary strategy for risk reduction until some preferred strategy is made available. Condom use is often will tell him that you are OK and hence there is no need to use condom? situational, sometimes based on objective evidence of the risk R: I will trust him if he will be faithful and if he will protect his health. level of the partner, and sometimes based on a general feeling of I can do sex without using condom when I am assured that we have trust. This man in the cash award group discussed how he and his tested and we are OK. partner use condoms between opportunities to test, especially -Divorced woman, cash award group, 8-month visit when one of them has been away from the home for some time. It should be noted here that repeat testing in the context of the RESPECT study and as mentioned by the study participants I: When your partner went for testing she was negative. You still use during the in-depth interviews refers mostly to STI testing. HIV condom even you are both negative? PLOS ONE | www.plosone.org 4 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania R: We still use condom because of the opinion of the woman. When one making existed—testing is important both for risk assessment as a goes away you have to use condom until when you test again and bargaining chip in sexual negotiations. confirm that you are both OK. -Unmarried man, cash award group, 4-month visit Situational Leverage and Opportunistic Implementation A second insight that came out of the qualitative data and the Some participants discussed how they find it difficult to trust discussions about strategies to avoid unsafe sex related not to a potential partners when they say they have tested negative, and specific strategy, but rather to how and when strategies were show concern if the most recent test was several months ago. implemented. Risk avoidance was often practiced inconsistently Others mentioned that they were fine with just a verbally and episodically. This was especially true for women who faced the communicated result. . To show how study participants negotiate barriers discussed in the Introductory section; women who much risk, we illustrate with several examples. The first comes from a 28 of the time lacked sexual decision-making power in their year old unmarried man in the low-value cash award group who relationships. During their interviews, these women talked about used condoms with his partner until they could both get tested how they took measures to reduce their risk of infection when they with good results. felt they could–if an opportunity arose that temporarily gave them increased agency, they took that opportunity to protect themselves. These opportunities included circumstances or situations that I told her to use condoms because before we married she was at her place provided women with increased knowledge, and as a result, added and I don’t trust her. Therefore, before I paid the dowry I wanted to use leverage with which to negotiate with their partners. Specifically, a condom until we shall go for medical check up. After the check up if added leverage arose within the RESPECT study from the cash our results are OK then we can continue doing sex without using incentive and STI testing status knowledge. More broadly outside condom. It was good because we came here for medical check up and the context of RESPECT, other situations could provide since we are both OK we continued without using a condom. opportunities for women to refuse sex or enforce condom use— –Unmarried man, cash award group, 4 month visit such as after having a baby, or when a husband or partner felt guilty about being with another woman. One 19 year-old married Testing to assess risk is not a strategy limited to partner woman explained how she is sometimes able to convince her selection, but is also used to assess the risk level of a current husband to use condoms. partner, as is illustrated above, or to prove to a partner that he or she is in fact risky. This interview excerpt is from a 29 year-old I: Did you try to avoid unsafe sex? married woman enrolled in the RESPECT study. R: Yes I: What did you do? I: Did you talk with him when you came here for the first time and got R: I asked my husband to use condom the days when he was not in good your results? mood R: Yes, I talked with him that my results were good and I also advised I: Did you use it throughout or in dangerous days only him to go for check up because my results were good. I advised him but R: During dangerous days only he refused. I: What are these dangerous days? I: Now that you have received your results today will you talk with him R: When my husband comes back at 2 am in the night and he needs to about the kind of results you have got? stay with me I ask him to use R: I will not tell him because if I am infected it means that he is the one I: Is this because you don’t trust him? who transmitted this to me. This is because maybe he is the one infected R: Yes and had he come for tests this disease would have been treated already. I: So, you think this time you can get STI? Therefore, when I am going he must go for test and if he doesn’t want to R: Yes come here I will tell him to go to the hospital I: But when he comes back early you continue as usual? I: Do you think he will accept when you tell him to go for testing? R: Yes R: This time he will accept to go to test I: Can’t you get STI when he comes early? What do you think? -Married woman, cash award group, 4 month visit R: I can –Married woman, control group, 4 month visit This passage highlights the centrality of testing as a risk reduction strategy within a marriage, where condom use is not an One possible interpretation of this passage is that this woman is option. This woman discusses the importance of having the actual taking action when she feels she has leverage to take action. She test results in hand so that she can use them as leverage to get her understands that she can still be infected if she uses a condom husband to test, and as a result of his test results, she can convince today but not tomorrow, but she also understands that she needs him to change his behavior. The power of persuasion in this case to be strategic about when she can implement her prevention lies in the test results and the ability to show a partner that he or efforts. Another possible interpretation is that she is angry about she is risky with authority. her husband staying out late with other women and when this Our data reveal that regular testing alone and in combination situation presents itself, it gives her an opportunity to confront him with condom use is a risk reduction strategy used to facilitate the and perhaps to discourage his infidelities. problematic issue of understanding the risk level of a current or Some women used the RESPECT award cash to help extract potential partner. The information that test results provide brings themselves from risky sexual behaviors and relationships that had some certainty to relationships that, as regards risk of infection, are been driven partly by economic constraints. This woman who lives often filled with ambiguity. The new information and knowledge with her partner talked specifically about how the cash award that comes from test results can also create opportunities for associated with the study has allowed her to leave other men negotiating safer sex where previously little agency in decision- because the money from the study will help her in her life. PLOS ONE | www.plosone.org 5 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania R: I have come back for round two of this study because when I come convince husbands that there could be real consequences to his back I learn more about my health…another thing is the award, when I own risky behavior during her postpartum abstention. came I was given award and I know what to do with it so as to take care of my children. The prostitution behavior I had before, I can stop it Temporary Abstinence completely Temporary abstinence is a risk reduction strategy that, I: You were a prostitute…can you explain to me what kind of a depending on the circumstances, is likely to be limited in its prostitute were you? effectiveness in preventing HIV infection. However, what emerges R: Changing men from time to time. All this was because of from the data is that given the range of strategies available, problems individuals in this setting are at least as likely to rely on temporary I: Did you have the behavior of having many men before you were abstinence as they are on condom use. Temporary abstinence enrolled to this study? could include divorce (in its most extreme form), temporary R: Yes, I had this behavior before I was enrolled to this study physical separation from a partner, enforcing no sex for several months after the birth of a child, enforcing no sex after a positive I: Over the period of four months ago have you had multiple STI test, or enforcing no sex because of the recognition that a sex partners? husband has been out with another woman. R: I didn’t have this behavior; I left them completely…since I knew my If discussions of condom use, either within the marriage or with health status and got the award I decided to continue with my life and external partners, were ineffective, one option is to physically stop this behavior; I have known that my health is OK…Therefore, separate themselves from their husbands temporarily to lower risk. when I get here there will be money for inconvenience to use with my In the case of this 27 year-old married woman, simply the threat of children. So, what is the importance of continuing with these men? separating their beds convinced her husband of the importance of –Married woman, cash award group, 4 month visit avoiding having external partners. Others used their enrollment in the study as a leverage point to achieve goals that they had previously (such as getting their I didn’t know what to do because at times he was coming back home at husbands to be more faithful), and were able to successfully avoid 11 pm in the night and when I asked him where he was he said that he sex or enforce condom use as a result. Joint strategizing between was working. I warn him if he was moving around but he told me that couples enrolled in the study was also mentioned by both men and he was still working. I told him that if it doesn’t work out we shall women in the cash award groups. If a woman’s partner or spouse separate our beds so that everyone sleeps alone. He listened to me and is enrolled in the study, the couple might discuss staying safe that is why I tested negative today together so that both of them can receive the award, and the –Married woman, cash award group, baseline visit woman might have more leverage, backed by the money that they will both receive, to convince her husband to leave outside The negative test results were enough to convince this woman partners or use condoms with outside partners. This example from that her husband had listened to her and was not ‘‘moving a woman in the cash award group illustrates how enrollment in the around’’ (engaging in sexual relationships with other women). This study and the promise of the reward provided an opportunity for example again illustrates the importance of the test results in her to discuss the issue of safer sex with her husband. establishing trust with a partner—either for partner selection, or building trust with a current partner. For others though, a test R: I told him we are supposed to go for check up, at the first round we result is not enough to convince a partner to change his or her behavior, or even to get tested. In this case, physical separation shall check blood and STI and after four months we shall come again from a partner is sometimes the only option. for testing and if we test negative we will get reward. This passage from one of the diaries in May of 2009 documents I: Therefore, you discussed together about what you should do in order to a conversation between two women, one of whom is convinced she get reward? has an STI and wants her husband to get tested so they can both R: yes we discussed together…I told him after four months if I test take treatment. negative for STI we shall get reward. Therefore, I told him to take care and he was ready I: Did he listen to you? When I go to the working place on foot, I saw two girls about 30 years R: He listened to me so attentively they were in their own talks. After a time first girl said to her fellow that I: Did he accepted to take care? ‘‘aiseeh! Everyday I talked to your relative about my health, but he R: Yes didn’t want to understand me, please go and advise him, I am ready -Married woman, cash award group, 4 month visit start to use a dose but not him. And when I advise him to go and to check it [his health], he didn’t. When I want to use condom during sex Outside of the RESPECT intervention mechanisms, other he refuse to do it. So it’s your time to educate him. Nevertheless I will go points of leverage for women to gain some agency in sexual back home even without permission from him.’’ The second girl agreed decision-making include recently having had a child, or being and adds that, ‘‘even me I amazed from him, now is the world of truth infected with an STI. Because it is common not to resume sex for in marriage, ok I will try to advice him and if he refused! Even me I will some time after a baby is born, women who have just given birth support you to go back home– that is not life.’’ (Diary, after baseline are in a position of having the ability to refuse sex without visit) consequences, and can then attempt to extend this abstinence— though only temporarily. Having this temporary power to refuse This example uncovers two alternative strategies–physical sex with husbands may be especially important for married separation from an uncooperative husband, but also using social women who know that their husbands have other partners, but are networks and connections to help with convincing a partner to not sure if he uses condoms with his external partners. Adding the understand the consequences of his actions and perhaps change as RESPECT STI testing to this situation could further help women a result. PLOS ONE | www.plosone.org 6 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania The next step after temporary separation is permanent with someone who has proven through testing or through verbal separation or divorce–another form of temporary abstinence, in acknowledgment that they are safe. many cases a more realistic strategy for risk reduction, for both the Another woman discussed her difficulty in trying to convince wife and the husband, than convincing a partner to use condoms her husband to use condoms even though she knew that he had within the marriage. Work done in Malawi has shown that divorce other women. Part of the difficulty, as she mentions, had to do has increasingly become more common overall, and marital with the expectations around having children and how using a dissolution has increasingly been implemented as a strategy to condom when more children were expected was not an option. protect oneself from an unfaithful spouse and from HIV [11,38,39]. In fact, as HIV/AIDS became more prevalent and I: How do you feel now that you that you do not have STIs or HIV? was perceived as more of a threat, divorce as a response to R: I feel so much by knowing this than just staying without knowing my infidelity was also steadily increasing in Malawi [39]. Over time, the proportion of women who agreed that divorcing a husband health status. who was unfaithful or was suspected of having HIV was justified I: Will this make you change your behavior? also increased significantly based on these data from Malawi [11]. R: This has made me change my behavior quite much to the extent that Both women and men enrolled in the RESPECT study it made me leave my husband because he was just talking about these discussed having used this strategy in the past to separate things but he was not ready to change his behavior. I decided to leave themselves from a partner they perceived as risky. Female him because I advised him so much but he didn’t want to change. Since participants in the cash award groups also sometimes discussed we are a husband and wife I cannot tell him to use condom because we divorce as an extreme strategy they might need to implement in are still procreating. We have to do sex without using condom. Since you order to stay safe within the context of the RESPECT trial, and to do not want to change then you can cause death to me (diseases). This is ensure that they would be eligible to receive the award—though what made me leave him. this seems unlikely given that participants were only enrolled in -Married woman, cash award group, 8-month visit RESPECT for one year. Other women are clear that they do not see divorce as an option at all, and are resigned to the reality that For other women, divorce or separation is not part of the menu facing risk from their husband is part of the marriage experience. of available strategies. They may be financially constrained from This 29 year-old woman left her first husband (in the past, prior to leaving their husbands, or because of the emotional attachment the RESPECT trial) because she was worried that he would bring leaving may be too difficult. infection into their marriage. I: Do you think this time your husband will accept to use condom? In the past I would fear but since last year I was not worried to say that R: I don’t know if he will accept, if he refuses that is it I have infections because I divorced my husband. I stayed for 5 years I: What will you do if he refuses? and I got another husband. I refused to accept him for quite long time R: What will I do; I am in the marriage but he told me that he was OK. We decided to go for testing and we I: Do you have freedom to tell your husband that if he does not put on found that both were OK. I stayed with him and I gave birth to one condom you will not do sex with him? child who is 2 years and 2 months old. Therefore, since I gave birth to R: Yes, I will refuse…because I will refuse today…tomorrow I will this child I stayed without doing sex and I didn’t do sex even with the refuse…the day after tomorrow I will refuse… at the end I have to one I have this child. I decided to leave him because his service was not agree, I have to do it without condom…this man is just like that, you good. Therefore, I saw there was no meaning to stay with such a person. can tell him this today and he can accept, but he can refuse tomorrow. –Divorced woman, cash award group, baseline visit –Married woman, cash award group, 4-month visit This next woman also appears serious in considering divorce as Younger, single men often discussed exercising sexual control an option as she exploits the new opportunity arising from through avoidance—either avoidance of sex or avoidance of RESPECT9s STI testing program to convince her husband to situations that might lead to risky sex, for example those with either leave his external partners, use condoms with his external alcohol. Another path of avoidance mentioned frequently was partners, go for testing, or use condoms within the marriage. This keeping oneself occupied with other activities, such as exercise, married woman discusses her frustration in trying to convince her studying, and or working on their farm, so that little time remained husband to either use condoms with her in the marriage or to go to focus on meeting women and sex. Such avoidance was for testing. preferable to consistently using condoms with partners–these men found it more realistic to avoid sex than to trust themselves R: The new strategy is to use condom and if he doesn’t want…But the that they would use a condom every time. This excerpt from a first strategy is to tell him to go test and I should make sure that I go single man enrolled in the RESPECT study illustrates these with him if he accepts to go for testing. If it will be OK we will do safe strategies. sex. I: Do you think he will agree with this? I: What strategies did you use in order to get the reward? R: He will accept…I don’t know what I will do if he refuses; it is R: The first thing is to come back home early, not staying with people better to divorce than to get diseases…I will continue advising him to with bad behavior…some people want to stay with ladies all the time, come for testing in round three. If he refuses to come I am ready to when they sit somewhere they are talking about sex only. Now I am divorce him rather than getting STI avoiding such people…because when you join their company anything –Married woman, cash award group, 4 month visit can happen. I: Will it be easy for you to avoid these temptations? It is notable that safe sex in this excerpt, and often in the R: Yes, it is easy…when I think that my friends are about to come I can interview transcripts, refers not to sex using a condom, but sex go somewhere, I can decide to go to shamba [the farm] to look at my PLOS ONE | www.plosone.org 7 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania maize. Then I decide to leave, when they come they will be told that I The present analysis was not designed to definitively establish am not around the relative importance of cash awards versus STI testing in –Unmarried man, cash award group, 4 month visit providing opportunities and tools for behavioral change. The quantitative evaluation results previously published indicate that What emerges from these data are that in this context, while RESPECT9s high value cash award arm did experience 27% episodic use of condoms and abstinence are likely similar in their fewer STIs than the control arm; since testing was available to limited effectiveness in HIV prevention, individuals in this setting both arms, this difference is likely attributable to the cash awards are at least as likely to rely on temporary abstinence than they are [9]. That prior analysis was not able to quantify the effects of the on temporary condom use. Temporary abstinence is not a reliable STI testing regimen per se (the study was not designed to directly means of preventing infection. However, refusing or avoiding sex do so), thus the present analysis is particularly important for at certain opportune moments or under specific circumstances is uncovering a strong latent demand for STI testing. It is also an intermediate strategy that over time might lead to more reasonable to speculate that the testing demand was enhanced by permanent strategies for risk reduction. If, for example, enforcing the cash awards, and that conversely the effects of the cash awards temporary abstinence by sleeping in separate beds or refusing sex may also have been enhanced by the behavioral change if a partner comes home late at night is perceived as punishment, opportunities afforded through the testing. Thus cash awards this intermediate strategy could eventually lead to a partner and STI testing may well be synergistic components of the changing his behavior. Sexual behavior change is slow at best, and RESPECT package – just as these may have enhanced (and in implementation of intermediate strategies such as temporary terms been enhanced by) the effectiveness of traditional behavioral abstinence is both a means of gaining transient control, and change communication campaigns. perhaps a method of pushing for change through increased Lessons for adopting STI testing interventions beyond a control. RESPECT-style intervention are more difficult to ascertain. Regular, comprehensive STI testing is a costly intervention; Discussion alternatively, using a less comprehensive testing regimen could undermine women’s ability to use the testing for the purposes of By exploring how study participants responded to an economic risk assessment (and could even raise risks due to inaccurate incentive to remain STI negative, this analysis adds to our current infection information). STI testing has often been considered an understanding of approaches to risk reduction by highlighting the intervention targeted at identifying positive individuals so as to often opportunistic and episodic implementation of strategies by provide them treatment (or more controversially, to reduce HIV women who face behavior change constraints [40]. Women who transmission probabilities [41]), which may be difficult to justify on may in typical circumstances lack sexual decision-making ability direct cost-benefit grounds. But the present study does suggest that can and often do take advantage of situations that present added beyond this direct epidemiological value, STI testing may have leverage with which to negotiate. important behavioral implications as well, through its role in The analysis is particularly helpful for understanding mecha- providing opportunities and leverage for behavioral change. This nisms through which behavior change occurred in the RESPECT behavioral pathway merits careful future research. trial. Both the cash awards and the receipt of STI test results A deeper understanding of how cash awards and STI testing through RESPECT provided opportunities for attempting behav- was perceived and acted on would help in structuring related new ior change. Simply being enrolled in a study that repeatedly intervention models. The opportunities provided by the RE- provided the ability to check one’s health status, and the promise SPECT study may have altered perceived behavioral control and of the cash incentive for testing negative for those in the cash self-efficacy among women enrolled in the trial, thus temporarily award groups, provided participants with added negotiating power facilitating behavior change [42,43]. Our results demonstrate the in their sexual relationships. In fact, the repeated testing became fact that individuals in this setting are willing to adopt temporary part of the strategy in some cases as the combination of targeted risk reduction strategies even if they fall short of consistent condom use, testing and treatment was sometimes implemented as behavior change. Future work would also be useful to further a three-pronged approach to decreasing risk of infection. Having a explore how these opportunities and strategies may have either back-up for condom use is important in a society where condom been stymied among women facing gender-based violence, or use carries with it such strong meaning about the type of alternatively helped to overcome such barriers. Related analysis relationship [12]. In addition, having the ability to use testing as a has documented a decline in gender-based violence during the milestone after which condoms are no longer necessary provides course of the 1-year RESPECT study [44]; future research some leverage for convincing a partner to use condoms. examining varying risk change strategies by baseline violence level Temporary abstinence, in its many forms, also emerged as a would be illuminating. more prevalent strategy than expected. The apparent popularity of This study brings with it some limitations. As with any study this approach may be a result of the negative associations with and that includes self-reported sexual behavior as a data source, there low levels of use of condoms within long-term serious partnerships. is the possibility that unsafe sexual behaviors were under-reported. The data presented here suggest that abstinence is a favored Plummer et al report on the validity of the collection of sexual strategy over condom use in certain situations–among young single behavior data using five different methods from a study done in people, among women who have just had children, and among Northern Tanzania [16]. While this study was conducted amongst women who can argue that they have been put at risk by their adolescents, and results among adults may be slightly more husbands. Married women seem to be more likely to be able to congruous across the five methods, they find striking inconsisten- abstain from sex with their husbands in order to at least cies in reports of sexual behavior from self-administered surveys, temporarily reduce their risk of infection than they are to insist face-to-face surveys, in-depth interviews, participant observation on condom use. This is of course an imperfect strategy for and biological markers. Social desirability bias is common in these protecting their own sexual health, thus it will be important for types of studies, and in this case may have been exacerbated by the future educational efforts to ensure that women understand the extended counseling on safer sex practices that participants were limited situations in which this can protect them from infections. receiving throughout the study. PLOS ONE | www.plosone.org 8 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania An additional limitation relates to the translation and strategies over and above other strategies. This may have transcription of audio-taped interviews. There is the potential for systematically influenced our findings, although we have no the content of the interview to lose meaning and nuance during evidence of this happening. the transcription and the translation process. We addressed this Our data suggest that when opportunities to implement risk limitation to some degree by having one of our study interviewers reduction strategies present themselves, they are regularly taken. (who is bi-lingual) review the transcripts and translations within The qualitative data point to the importance of not only the cash two weeks of the actual interview, and revise the transcripts as incentive, but also the access to regular, reliable testing and necessary. This method still suffers from a secondary but related knowledge of health status in opening opportunities to discuss risk limitation; that is the revised translated transcripts are an reduction strategies with partners and as leverage in negotiating interpretive process, wholly dependent on the perception of the and implementing risk reduction strategies. With the understand- interviewer. Limitations related to the use of the village diaries ing that behavioral change strategies are a necessary but not should also be noted. The pay for the diarists may have motivated sufficient means of preventing HIV, the RESPECT approach may them to seek out situations in which HIV and/or the study is being have acted as a combination HIV prevention intervention. Our discussed [36]. Bias may also result from the diarist’s individual results suggest that RESPECT9s structural cash intervention and perceptions of what he/she is hearing, and bias resulting from the testing component worked synergistically to assist men and potentially inexact recall of the situation may also arise. However, women with opportunities to better act on behavioral change it is also important to emphasize the purpose of this data is to intentions, thus potentially increasing the effectiveness of tradi- provide a window into what is going on in the community from tional behavior change interventions as well. the community perspective, fully understanding that we are gaining this information through the lens of one of the members of the community. Acknowledgments Loss of the 14 of the qualitative transcripts during the first round We thank Carol Medlin, Ann Swidler, the RESPECT Study Team, and is an additional limitation of the study. The recordings were many others for their contributions to the project. The findings, inadvertently deleted while the team was out in the field and could interpretations, and conclusions expressed in this paper are entirely those not be recovered. While this does not pose a problem in relation to of the authors. They do not necessarily represent the views of the bias since there was no systematic loss of data, it is a limitation in International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World that we lost data that would have contributed to our findings and Bank or the governments they represent. results, but the limitation is minor given the large number of qualitative interviews analyzed. The purposive sampling to select more candid respondents in the subsequent follow-up visits is an Author Contributions additional limitation in that this type of selection may have Conceived and designed the experiments: WHD DD RN. Performed the introduced a bias. By selecting those respondents who were more experiments: LP SM. Analyzed the data: LP AK WHD DD SM. Wrote the open to discuss their strategies for avoiding unsafe sex, we may paper: LP AK WHD DD. have also selected respondents utilizing particular types of References 1. Coates TJ, Richter L, Caceres C (2008) Behavioural strategies to reduce HIV 15. de Walque D, Kline R (2011) Variations in Condom Use by Type of Partner in transmission: how to make them work better. The Lancet 372: 669–684. 13 Sub-Saharan African Countries. Studies in family planning 42: 1–10. 2. Padian NS, McCoy SI, Karim SSA, Hasen N, Kim J, et al. (2011) HIV 16. Plummer ML, Ross DA, Wight D, Changalucha J, Mshana G, et al. (2004) "A prevention transformed: the new prevention research agenda. The Lancet 378: bit more truthful": the validity of adolescent sexual behaviour data collected in 269–278. rural northern Tanzania using five methods. Sexually transmitted infections 80 3. Gupta P, Ogden Aggleton, Mahal (2008) Structural approaches to HIV Suppl 2: ii49–56. prevention. Lancet 6736: 60887–60889. 17. Bongaarts J (2007) Late marriage and the HIV epidemic in sub-Saharan Africa. 4. Krishnan S, Dunbar MS, Minnis AM, Medlin CA, Gerdts CE, et al. (2008) Popul Stud (Camb) 61: 73–83. Poverty, Gender Inequities, and Women’s Risk of Human Immunodeficiency 18. Carpenter L, Kamali A, Ruberantwari A, Malamba S, Whitworth J (1999) Rates Virus/AIDS. Annals of the New York Academy of Sciences 1136: 101–110. of HIV-1 transmission within marriage in rural Uganda in relation to the HIV 5. Auerbach JD, Parkhurst JO, Ca ´ ceres CF (2011) Addressing social drivers of sero-status of the partners. AIDS 13: 1083. HIV/AIDS for the long-term response: Conceptual and methodological 19. Clark S (2004) Early marriage and HIV risks in sub-Saharan Africa. Studies in considerations. Global Public Health 6: 293–309. family planning 35: 149–160. 6. Gupta GR, Ogden J, Warner A (2011) Moving forward on women’s gender- 20. Glynn J, Carall M, Buve A, Musonda R, Kahindo M (2003) HIV risk in relation related HIV vulnerability: The good news, the bad news and what to do about it. to marriage in areas with high prevalence of HIV infection. JAIDS Journal of Global Public Health 6: 370–382. Acquired Immune Deficiency Syndromes 33: 526. 7. Kohler HP, Thornton RL (2011) Conditional Cash Transfers and HIV/AIDS 21. de Walque D (2007) Sero-discordant couples in five African countries: Prevention: Unconditionally Promising? The World Bank Economic Review: implications for prevention strategies. Population and Development Review doi: 10.1093/wber/lhr1041. 33: 501. 8. Thornton RL (2008) The demand for, and impact of, learning HIV status. The 22. Bankole A, Biddlecom AE, Dzekedzeke K (2011) Women’s and Men’s Fertility American economic review: 1829–1863. Preferences and Contraceptive Behaviors by HIV Status in 10 Sub-Saharan 9. de Walque D, Dow WH, Nathan R, Abdul R, Abilahi F, et al. (2012) African Countries. AIDS Education and Prevention 23: 313–328. Incentivising safe sex: a randomised trial of conditional cash transfers for HIV 23. Beyeza-Kashesya J, Kaharuza F, Mirembe F, Neema S, Ekstrom AM, et al. and sexually transmitted infection prevention in rural Tanzania. BMJ open 2: (2009) The dilemma of safe sex and having children: challenges facing HIV sero- e000747. doi:000710.001136/bmjopen-002011–000747. discordant couples in Uganda. African health sciences 9: 2–12. 10. Poulin M (2007) Sex, money, and premarital partnerships in southern Malawi. 24. Harrison A, O9Sullivan LF (2010) In the absence of marriage: long-term Social science & medicine 65: 2383–2393. concurrent partnerships, pregnancy, and HIV risk dynamics among South 11. Smith K, Watkins S (2005) Perceptions of risk and strategies for prevention: African young adults. AIDS and behavior 14: 991–1000. responses to HIV/AIDS in rural Malawi. Social science & medicine 60: 649– 25. Nakayiwa S, Abang B, Packel L, Lifshay J, Purcell DW, et al. (2006) Desire for 660. children and pregnancy risk behavior among HIV-infected men and women in 12. Tavory I, Swidler A (2009) Condom Semiotics: Meaning and Condom Use in Uganda. AIDS and behavior 10: S95–104. Rural Malawi. American Sociological Review 74: 171–189. 26. Jewkes RK, Dunkle K, Nduna M, Shai N (2010) Intimate partner violence, 13. Watkins S (2004) Navigating the AIDS epidemic in rural Malawi. Population relationship power inequity, and incidence of HIV infection in young women in and Development Review: 673–705. South Africa: a cohort study. The Lancet 376: 41–48. 14. Chimbiri A (2007) The condom is an intruder in marriage: Evidence from rural 27. (2010) Addressing Violence Against Women and HIV/AIDS: What Works? Malawi. Social science & medicine 64: 1102–1115. Geneva, Switzerland: World Health Organization. PLOS ONE | www.plosone.org 9 August 2012 | Volume 7 | Issue 8 | e44058 HIV Risk Reduction Behaviors in Tanzania 28. Maman S, Mbwambo J, Hogan N, Kilonzo G, Sweat M (2001) Women’s 36. Watkins S, Swidler A (2009) Hearsay ethnography: Conversational journals as a barriers to HIV-1 testing and disclosure: challenges for HIV-1 voluntary method for studying culture in action. Poetics 37: 162–184. counselling and testing. AIDS Care 13: 595–603. 37. Moustakas CE (1994) Phenomenological research methods: Sage Publications, 29. Temmerman M, Ndinya-Achola J, Ambani J, Piot P (1995) The right not to Inc. know HIV-test results. Lancet 345: 969. 38. Kaler A (2004) AIDS-talk in everyday life: the presence of HIV/AIDS in men’s 30. Antelman G, Smith Fawzi MC, Kaaya S, Mbwambo J, Msamanga GI, et al. informal conversation in southern Malawi. Social science & medicine 59: 285– (2001) Predictors of HIV-1 serostatus disclosure: a prospective study among 297. HIV-infected pregnant women in Dar es Salaam, Tanzania. AIDS 15: 1865. 39. Reniers G (2008) Marital strategies for regulating exposure to HIV. 31. Kilewo C, Massawe A, Lyamuya E, Semali I, Kalokola F, et al. (2001) HIV Demography 45: 417–438. counseling and testing of pregnant women in sub-Saharan Africa: experiences 40. Schatz E (2005) ‘Take your mat and go!’: Rural Malawian women’s strategies in from a study on prevention of mother-to-child HIV-1 transmission in Dar es the HIV/AIDS era. Culture, Health & Sexuality 7: 479–492. Salaam, Tanzania. JAIDS Journal of Acquired Immune Deficiency Syndromes 41. Hayes R, Watson-Jones D, Celum C, van de Wijgert J, Wasserheit J (2010) 28: 458. Treatment of sexually transmitted infections for HIV prevention: end of the road 32. Hirsch J, Holly W, Wardlow H, Smith D, Phinney H (2009) The Secret: Love, or new beginning? AIDS 24: S15. Marriage, and HIV: Vanderbilt Univ Pr. 42. Ajzen I, Madden TJ (1986) Prediction of goal-directed behavior: Attitudes, 33. Armstrong Schellenberg J, Mukasa O, Abdulla S, Marchant T, Lengeler C, et intentions, and perceived behavioral control. Journal of experimental social al. (2002) Chapter 11. Ifakara DSS, Tanzania. Population and Health in Developing Countries 1: 159–164. psychology 22: 453–474. 34. De Walque D, Dow WH, Medlin C, Nathan R (2012) Stimulating demand for 43. Bandura A (1990) Perceived self-efficacy in the exercise of control over AIDS AIDS prevention: lessons from the RESPECT trial. National Bureau of infection. Evaluation and program planning 13: 9–17. Economic Research. 44. Krishnan S, Vohra D, Nathan R (2011) Couples’ Perspectives on Gender and 35. London AJ, Borasky DA, Bhan A (2012) Improving Ethical Review of Research Intimate Partner Violence: Insights from the RESPECT Study, Tanzania. Involving Incentives for Health Promotion. PLoS medicine 9: e1001193. Population Association of America Conference. Washington, D.C. PLOS ONE | www.plosone.org 10 August 2012 | Volume 7 | Issue 8 | e44058