HIV Prevention Programming among Transgenders: Lessons from Laos

Transcription

1 HIV Prevention Programming among Transgenders: Lessons from Laos Kim Longfield Xouchai Panyanouvong Judy Chen Megan Kays PSI Research & Metrics Working Paper No. 76 December 2008

2 Population Services International PSI is a Washington, DC-based nonprofit organization that harnesses the vitality of the private sector to address the health problems of low-income and vulnerable populations in more than 60 developing countries. PSI has programs in malaria, reproductive health, child survival, and HIV/AIDS, and deploys commercial marketing strategies to promote products, services and healthy behaviors that enable people to lead healthier lives. PSI is the leading nonprofit social marketing organization in the world. Research Division Population Services International 1120 Nineteenth Street NW, Suite 600 Washington, DC Authors Kim Longfield: Population Services International Xouchai Panyanouvong: Population Services International* Judy Chen: Population Services International* Megan Kays: Population Services International * With PSI/Laos at the time of the study Recommended Citation Longfield, K., Panyanouvong, X., Chen, J., & Kays, M. (2008). HIV prevention programming among transgenders: Lessons from Laos. (Working Paper No. 76). Washington, DC. Population Services International. Acknowledgments This research was supported by the United States Agency for International Development (USAID) and the Department for International Development (DFID). The study would not have been possible without the permission of the Center for HIV/AIDS and STI (CHAS) of Lao PDR and the Center s Director, Dr. Chansy Phimphachanh, as well as the support of John Deidrick, Country Representative for PSI/Laos at the time of this study. The study was designed by Dr. Xouchai Panyanouvong (at the time of the study, Research Manager for PSI/Laos), Ms. Kongchay Vongsaiya (Research Assistant, PSI/Laos), and Ms. Judy Chen, MHS (at the time of the study, Technical Advisor for PSI/Laos), with help from Dr. Kim Longfield (PSI Senior Researcher, Southeast and Central Asia) and Dr. Kathryn O Connell (Technical Advisor, PSI/Cambodia). Data collection was conducted by PSI/Laos fieldworkers and supervised by Dr. Panyanouvong and Ms. Vongsaiya. Data input and cleaning were completed by Mr. Bounyang Ratsamee (PSI/Laos Research Assistant) and Dr. Panyanouvong. Ms. Chen completed data analysis and writing for the original project report. Dr. Longfield reviewed the analysis and final report, and prepared the manuscript for this article. Dr. Hongmei Yang reviewed the evaluation syntax and provided suggestions for improving the next study round. Finally, Ms. Ditte Hjorth and Mr. Rob Gray provided comments on preliminary versions of this article.

3 ABSTRACT Purpose: In Asia, men who have sex with men (MSM) are at an elevated risk for HIV infection due to low levels of condom use, low perceived threat of HIV infection, and high levels of stigma. PSI launched an HIV prevention intervention in late 2002 for a subgroup of Lao MSM, male-to-female transgenders (hereafter referred to as transgenders ). PSI initiated peer-led interpersonal communication (IPC) activities and launched a brand extension of the popular Number One Deluxe condom brand, distributed in Laos since The new product, Number One Deluxe Plus (two condoms prepackaged with a sachet of water-based lubricant) was intended to appeal to transgenders and their partners. This study utilizes data from 2004 and 2006 tracking surveys to identify changes in key HIV preventionrelated indicators over time in the population of interest and to evaluate the impact of PSI s transgender-specific interventions. Methods: Time-location sampling was used to collect survey data from transgenders in November 2004 (n=288) and June 2006 (n=415). Respondents were aged and were from three urban centers in Laos: Vientiane, Luang Prabang, and Savannakhet. For inclusion in the study, respondents must have had oral or anal sex with a man in the past six months. UNIANOVA tests were used to compare baseline and follow-up survey data and to evaluate the impact of PSI s transgender-specific interventions on items that changed significantly over time. Results: Between 2004 and 2006, there were significant increases in the following behaviors: condom use at last anal sex with casual partners (phoubao) (p<.001), condom use at last sex with boyfriends (fan) (p<.001), and use of water-based lubricant during anal sex (p<.001). More transgenders in 2006 reported that condoms are available in locations where they spend time during the day (p<.001). There were also significant increases in perceived availability of Number One Deluxe Plus (NODP) condoms in places where transgenders meet sex partners (p<.001) and spend time during the day (p<.001). Knowledge about the role of consistent and correct condom use with regular partners in preventing STIs improved over time (p<.001). Unfortunately, there were significant declines in three knowledge items: the role of oil-based lubricants in promoting condom breakage (p<.001), the increased risk of contracting HIV when infected with STIs (p<.001), and the idea that a healthy-looking person can be infected with HIV (p<.001). The attitude it is very important to use condoms with regular partners to prevent HIV improved over time (p<.001); however, the intention to use condoms with casual partners if water-based lubricant is available declined significantly (p<.001). Evaluation analysis showed that high exposure to transgender-specific interventions had a significant positive impact on condom use at last anal sex with casual partners (p<.05). For ever use of water-based lube during anal sex (p<.01), low and high exposure were both associated with comparable increases from baseline. While increases in perceived NODP availability and knowledge about consistent and correct condom use may be related to exposure, they are not associated with exposure to transgender-specific interventions: they may be attributed to exposure to media, another organization s interventions, or seeing products available in outlets. There was no indication that PSI s transgender-specific interventions had a positive effect on condom use with boyfriends. Likewise, it appears that - 1 -

4 factors unrelated to the intervention evaluated in this study improved the perception that condoms are available in places where transgenders spend time during the day and the attitude that it is important to use condoms with regular partners. No significant change in STI knowledge was apparent over time, with knowledge hovering at about 60%. Of greatest concern are the three knowledge items and the one intention item that were better at baseline than for any exposure group in 2006: oil-based lubricant can promote condom breakage; having STIs can increase the likelihood of contracting HIV; healthy-looking people can be HIV-positive, and the intention to use condoms when water-based lubricant is available. Conclusion: Current intervention strategies have been successful at increasing condom use with casual partners, but modified strategies are required for achieving a further increase and for influencing condom use with boyfriends. It may be necessary to strengthen existing peer education and outreach programs, particularly the training of peer educators, so that complete and accurate information is disseminated to transgenders. Messages related to consistent condom use during anal sex with all types of partners should be emphasized and efforts should be made to improve knowledge and correct misconceptions about HIV and AIDS, STIs, condom use, and lubricant use among transgenders and their partners. Improving perceptions of NODP and condom availability does not require sophisticated interventions: media promotion and ensuring product availability and accessibility in outlets appear to be sufficient. While the uptake of water-based lubricant is positive, the importance of using it in conjunction with condoms for HIV prevention must be emphasized. Program implementers are hampered by the dearth of information about transgenders and other Asian MSM populations. MSM-specific messaging may not resonate with people who selfidentify as female or a third gender; therefore more formative work is necessary to understand how transgenders define themselves and their HIV prevention needs

5 INTRODUCTION In Asia, men who have sex with men (MSM) are at an elevated risk for HIV infection, due to low levels of condom use, low perceived threat of HIV infection, and high levels of stigma (MSM and HIV/AIDS Risk in Asia, 2006). Laos has one of the lowest national HIV prevalence rates in Southeast Asia, with less than 0.2% of its general population aged 15 to 49, or an estimated 6,000 persons, living with HIV (UNAIDS/WHO, 2008). However, Laos remains at a significant risk for increased HIV transmission due to high levels of STIs among at-risk populations and cross-border migration with its higher prevalence neighbors, Cambodia and Thailand (Phimphachanh & Sayabounthavong, 2004). MSM, due to elevated risk factors, could potentially be among the highest-risk groups in Laos. There are little data on Laos MSM populations, which may be due to the sensitive nature of MSM behavior or a previous lack of recognition of MSM among authorities. Recently, however, the Laos government included MSM as a target group for HIV prevention in the National Strategic and Action Plan on HIV/AIDS/STI Among those targeted for HIV prevention measures in Laos are transgendered individuals and their partners. Transgenders are biological males who self-identify as female, but may dress and act as either males or females. Although transgenders self-identify as female, they can be considered a subclassification of MSM, as the term refers to biological behavior rather than identity. Although there is a lack of data about HIV prevalence among transgenders, similar transgender populations in neighboring countries report high levels of HIV infection. Studies of Thai transgendered men (katoey) and Indonesian transgendered men (warias) found HIV rates ranging from 12% to 22% (van Griensven, 2006; Pisani, et al., 2004). These data suggest that transgenders in Laos may also be at an elevated risk for HIV infection. Population Services International (PSI) began an HIV prevention program for Lao transgenders in 2002, and scaled up its activities in PSI administered tracking surveys - 3 -

6 in 2004 and 2006 to acquire information about HIV risk behaviors among transgenders, and about associated behavioral determinants. This paper utilizes data from the surveys to identify changes in key indicators over time and to evaluate the impact of PSI s transgenderspecific interventions. It is anticipated that findings will contribute to a greater understanding of the HIV prevention needs of transgenders in Laos and similar populations in other countries. LITERATURE REVIEW Asian Men Who Have Sex with Men: Identities and Practices In Asia, MSM encompass a diverse set of gender identities and sexual practices. Many MSM do not self-identify as gay or bisexual, nor do they consider having sex with a man, particularly a transgendered man, a homosexual act (MSM and HIV/AIDS Risk in Asia, 2006). A survey of Lao men found that 18.5% of men reported having had sex with other men, including transgendered men, and that most of this group also reported having sex with women. Married men were more likely than single men to visit male sex workers, and most sex for money was negotiated in non-brothel settings (Toole, et al., 2006). Qualitative research suggests that the majority of Lao transgenders engage in some form of commercial sex work (Doussantousse & Keovongchith, 2004). Taken together, these findings indicate that partners of Lao transgenders may serve as a bridging population, transmitting HIV to the general population. Although a small minority of Lao transgenders report having female partners, the majority have only male partners (Doussantousse & Keovongchith, 2004). Transgenders have two main types of partners: boyfriends and casual partners. Boyfriends are defined as regular partners to whom transgenders are emotionally committed. Casual partners are defined as men with whom transgenders are not in an emotionally committed relationship, or men they might have only had sex with once or a few times. The partners of transgenders are typically young men who self-identify as heterosexual and have sexual relations with women, but also have sex with transgenders for sexual experience, pleasure, experimentation or money. They are commonly students, army recruits and laborers such as security guards, waiters and hotel - 4 -

7 boys. Contrary to expectations, transgenders reported often paying their young partners for sex, whereas older men paid transgenders for sex (PSI, 2002). HIV Risk Behaviors among Transgendered MSM in Southeast Asia Few studies have been conducted on the HIV risk behaviors of the transgender population in Laos. However, early studies of transgendered communities in other Southeast Asian countries revealed high levels of HIV risk behaviors. Among Indonesian warias, the most common sexual practice was receptive anal sex, with study participants reporting that they used condoms for an average of 1.2 of their last 5 sexual acts (Lubis et al., 1994; Lubis, et al., 1997). Subsequent studies with warias have shown varying rates of condom use, with 12% to 41% reporting consistent condom use (Joesoef, et al., 2003; Pisani, et al., 2004). Cambodian MSM populations that included transgenders appeared to have higher levels of condom use, with only 13% reporting unprotected penetrative sex in the last month; however, consistent condom use with male clients was reported by fewer than half of MSM (Family Health International, 2000). Inconsistent condom use and other risky sexual behaviors among MSM have been associated with poor HIV knowledge and low perceived threat of HIV infection. A study of Thai MSM, including transgenders, found that believing myths about the source of HIV infection and feeling that HIV is not very serious were associated with inconsistent condom use (Mansergh, et al., 2006). Among Indonesian warias, 40% of those who felt that they were not at risk for HIV had eight or more sexual partners per week (Lubis, et al., 1994). A qualitative study of Lao transgenders found that while most were aware of HIV, some held contradictory beliefs about its transmission. Also, knowledge about HIV did not appear to translate into protective behaviors (Doussantousse & Keovongchith, 2004). PSI s own findings about Lao transgenders correspond to this body of evidence. Transgenders reported low levels of both recent and general condom use in a qualitative study conducted by PSI in 2002, and also expressed misperceptions about the riskiness of sexual behaviors and about their own personal risk of HIV infection (PSI, 2002). The study presented in this paper uses data from a 2004 tracking survey along with data from a 2006 follow-up survey to identify changes in key indicators and evaluate the effectiveness of PSI s - 5 -

8 transgender-specific HIV prevention activities. Data from the 2004 tracking survey that are not included in the study analysis provide additional information about HIV risk behavior among transgenders. The 2004 survey of 288 transgenders in Laos found that 76% reported ever using a condom during anal sex and that 50% reported using condoms during any sex in the last week. However, only 5% indicated that they usually use condoms with all sexual partners (PSI, 2004). Unlike other studies, the PSI survey did not find significant associations between knowledge and condom use or between perceived threat and condom use. Respondents exhibited a relatively high level of knowledge about HIV for the most part, with a moderate level of knowledge about the use of lubricants and the risk of HIV transmission via oral sex. The majority reported concern about contracting STIs or HIV. Two opportunity/ ability/motivation-related (OAM) determinants and one behavior of interest were found to be significantly associated with having used a condom at last anal sex with a casual partner. Condom users were more likely to report condom availability at the places where they spend their days, to have higher self-efficacy for condom negotiation, and to have ever used a water-based lubricant for anal sex (PSI, 2004). Addressing the HIV Prevention Needs of Transgendered MSM in Southeast Asia Only 2% of the total MSM population in Asia is estimated to have access to an HIV prevention program (UNAIDS, 2004). Transgenders appear to be particularly neglected; our literature review identified only one evaluated intervention that included this population as part of its target audience. Condom use among male sex workers (including warias) in Indonesia showed an increase of 37% in consistent condom use from 2002 to 2004 following the implementation of an FHI-sponsored outreach program (FHI, 2004). The lack of evaluated interventions may be due to challenges faced by implementers trying to reach transgender populations. In addition to political disenfranchisement, many transgendered individuals are subject to varying levels of stigma given their unique position in society (MSM and HIV/AIDS Risk in Asia, 2006). Although one qualitative study of transgenders found that there appears to be general acceptance of transgenders and their lifestyle in Laos, there is not sufficient evidence to determine how discrimination may - 6 -

9 affect Lao transgenders and their willingness to seek HIV services (Doussantousse & Keovongchith, 2004). PSI S TRANSGENDER-SPECIFIC INTERVENTIONS IN LAOS PSI began conducting peer-led interpersonal communication (IPC) activities in a limited number of locations frequented by transgenders in late PSI also launched a brand extension of the popular Number One Deluxe condom brand, distributed in Laos since The new product, Number One Deluxe Plus (NODP), is comprised of two condoms prepackaged with a sachet of water-based lubricant. In addition to the new product, a brochure with the NODP logo was developed, which contained condom and lubricant use information (explaining that a condom should be used every time with every partner, and that lubricant should be used with a condom to prevent condom breakage) and instructions. Between 2004 and 2006, PSI scaled up its IPC activities for transgenders and their partners. The key component of the scaled-up program included two drop-in centers, one in Vientiane and one in Savannakhet, which was opened near the end of the study period. The drop-in centers served as safe spaces where Lao transgenders and their partners could access quality health information and services; they provided transgender-specific health information and referrals to transgender-friendly doctors, while also serving as launching points for peer education outreach through group discussions, team-building activities, and peer education training sessions and workshops. During the same period, the Lao Youth Action AIDS Prevention Program (LYAP) and the Burnet Institute sponsored HIV prevention activities for MSM, including transgenders. LYAP worked in Vientiane, Champasak, and Luang Prabang on peer education and IPC programs. Burnet s activities were concentrated in Vientiane, where the organization distributed printed materials and conducted peer education and IPC activities. METHODS Theoretical Framework The design for this study was guided by the Performance Framework for Social Marketing (PERForM) and the PSI Behavior Change Framework, which together propose a set of - 7 -

10 pathways through which social marketing interventions seek to influence behavior and consequently individuals health and well-being (Figure 1). According to PERForM, the social marketing intervention acts on the population of interest, improving determinants of health behavior and subsequently leading at-risk individuals to increase their use of protective products or services and/or to increase their risk-reducing behavior, with the ultimate goal of improving health status or quality of life. Figure 1: The Performance Framework for Social Marketing (PERForM), with the PSI Behavior Change Framework as a Component of the Second Level The second level of PERForM identifies two different sets of factors influencing the behavior of the population of interest: population characteristics and mutable behavioral determinants. The social marketing intervention is designed to influence the behavioral determinants theorized to have the greatest effect on whether or not the protective product, service or behavior is adopted. The PSI Behavior Change Framework, a component of the second level of PERForM, categorizes behavioral determinants into 16 summary constructs, which are classified as either opportunity, ability, or motivation determinants. Opportunity determinants encompass institutional or structural factors that influence an individual s - 8 -

11 chance of performing a desired behavior. Ability determinants relate to an individual s skill or proficiencies needed to perform a promoted behavior. Motivation determinants include factors associated with an individual s desire to perform the behavior in question. This study focuses on six behavioral determinants thought to be relevant to condom use among transgenders and their partners: two (availability and self-efficacy) were found to be significant in the baseline tracking study; the others were from the project logframe (knowledge, attitudes, intention, and threat). Sample and Design Two quantitative surveys were conducted among transgenders in November 2004 (n=288) and June 2006 (n=415). Respondents were aged and were from three urban centers in Laos: Vientiane, Luang Prabang, and Savannakhet. Transgenders were eligible for inclusion in the study if they reported having oral or anal sex with a man in the past six months. Time-location sampling (TLS) was used to collect data during both study rounds. Researchers conducted a mapping exercise of venues where transgenders congregated in each of the three urban centers. In 2004, a total of 144 venues were identified (94 in Vientiane, 35 in Savannakhet, and 15 in Luang Prabang). In 2006, 139 venues were identified (102 in Vientiane, 20 in Savannakhet, and 17 in Luang Prabang). Researchers then created a list of TLS clusters that identified the time slots with the greatest number of transgenders present on weekdays and weekends for the sampling framework. Probability proportional to size sampling was then used to select 36 clusters in 2004 and 67 clusters in Fieldworkers visited each cluster at a peak time for the target population, staying for two hours on each occasion and interviewing every eligible transgender who agreed to participate in the study. A screening questionnaire established eligibility and explained confidentiality before requesting verbal confirmation from the respondent for participation. Interviews took place at the venues or in locations close by. Study participants received modest monetary incentives for completing the interview

12 Measures and Data Collection Eight transgendered fieldworkers and two female fieldworkers collected data from study respondents with a paper/pen survey form. Respondents were asked about boyfriends and casual partners. The study questionnaire included questions on background characteristics; condoms and lubricants; sexual history; sexual partners; receiving or giving money in exchange for sex; and exposure to PSI interventions. Multi-item scales were used to measure most behavioral determinants; true/false questions were used to measure knowledge about HIV transmission. The questionnaire was developed in English and then translated into Lao. It was pretested with a sample of 24 respondents at venues not selected for the survey. Transgenders who participated in the pretesting exercise were not eligible to participate in the survey. All data were collected in the Lao language and each interview took approximately 40 minutes to complete. Analysis SPSS 13.0 was used for all data entry and analyses. While additional data analyses were performed to provide programmatic direction, only the monitoring and evaluation results are presented in this paper. Reliability analysis was conducted on scaled items; however, none of the constructs achieved a Chronbach s alpha of.70 or higher, so none were retained as constructs. Each scaled item was analyzed separately. Monitoring analysis was conducted by combining the data from the first and second rounds and running UNIANOVA tests to identify significant differences between study years. This analysis included variables from the project logframe and the three significant determinants of condom use identified during the 2004 survey: condom availability in the places where transgenders spend their days, selfefficacy for condom negotiation, and ever use of water-based lubricant for anal sex. The following sociodemographic characteristics were controlled for when testing whether there was a significant change over time for each item: age, highest level of education completed, occupation, average monthly income, and current living arrangements. Evaluation analysis was conducted by combining the data from the first and second study rounds and running UNIANOVA tests to identify the impact of exposure to PSI programs on items found in the monitoring analysis to have changed significantly between 2004 and The same aforementioned sociodemographic controls for confounding were used

13 Baseline levels from 2004 were compared to three exposure categories in 2006: none, low, and high. None was the designation for respondents who had never heard of PSI s transgender-specific activities but may have participated in other MSM HIV prevention activities, such as those sponsored by LYAP and the Burnet Institute. Those who had heard of PSI s transgender-specific activities, ever read PSI s brochure for transgenders or heard of PSI s drop-in center were classified as having low exposure. Transgenders who had ever been contacted by a peer educator, attended a PSI discussion group, or gone on a camping/training trip were categorized as having high exposure. In the evaluation table (Table 3), proportions with the same superscript letter do not differ significantly from one another, whereas those with different superscript letters differ significantly from one another. All findings are significant at p<.05 or lower. The protocol for this study was reviewed by the Lao Ministry of Health (MOH) through the National Center for the Control of HIV/AIDS. RESULTS Baseline and Follow-up Sample Descriptions Table 1 presents population characteristics for transgenders who participated in the two surveys. In both study years, slightly more respondents self-identified as female than male (55.5% vs. 45.5% in 2004; and 52.5% vs. 47.5% in 2006). The majority were aged 25 or younger (76.0% in 2004; 79.3% in 2006) and had attended at least some secondary school (91.7% in 2004; 95.0% in 2006). More than 80% reported living with their families in both study years and only a small percentage reported living alone or in a dormitory (8.0% and 3.8%, respectively, in 2004; 8.4% and 4.6%, respectively, in 2006). Reported incomes were higher in 2006 than they were in In 2004, 27.8% of transgenders fell into the lowest income category, reporting an income of US$ 200 or less per month; in 2006 only 13.7% of transgender fell into this category. In 2004, 44.4% of transgenders reported earning between US$ 201 and US$ 600 per month, an income considered high by Lao standards; in % reported the same. The majority of transgenders in both study years were either students or employed as hairdressers or shop owners

14 Table 1: Population Characteristics of Transgenders Surveyed in Luang Prabang, Vientiane, and Savannakhet, Laos (November 2004 and June 2006) 2004 Total (n=288) 2006 Total (n=415) Type of Transgender n % n % Transgendered MM 1 Transgendered MW Age n % n % years old years old years old years old Mean=22.4 years old Mean=21.8 years old Education n % n % Never attended school Primary school Lower secondary school Upper secondary school Higher than secondary school Living arrangements n % n % Married, living with spouse Married, living with other sexual partner Not married, living with female sexual partner Not married, living with male sexual partner Not married, living with family Not married, living at army barracks Not married, living alone Not married, living in dormitory Not married, living with other relatives Not married, living with friends Average monthly income 3 n % n % 200,000 kip or less 200,001 kip to 600,000 kip 600,001 kip to 1,000,000 kip 1,000,001 kip to 3,000,000 kip 3,000,001 kip or higher Mean=741,194 kip Mean=812,246 kip Occupation n % n % Student Hairdresser/beauty salon Merchant/small shop owner Construction worker Waiter Guesthouse/hotel employee Teacher Government worker Army soldier Other Unemployed Transgendered MM are defined as biological males who identify as female, but in appearance are male. Their reported partners are young men, and they have a sexual preference for men. In general, transgendered MM perform oral sex and receive anal sex. 2 Transgendered MW are defined as biological males who identify as female, and in appearance are female. Their reported partners are young men, and they have a sexual preference for men. In general, transgendered MW perform oral sex and receive anal sex. 3 10,000 kip is equivalent to approximately US$

15 Trends in Behavior, Behavioral Determinants, and Exposure to the Intervention Table 2 presents 2004 and 2006 data for protective behaviors and behavioral determinants and shows changes between the two survey rounds. All three protective behaviors assessed in the study increased over time. Around three-quarters of transgenders in 2006 reported condom use at last anal sex with both casual partners and boyfriends, significant increases from 2004 (75.5% vs. 57.5% and 76.7% vs. 23.5%, respectively, both p<.001). The proportion of transgenders reporting water-based lubricant use for anal sex, which was found to be significantly associated with condom use in the baseline study, also increased significantly, from 56.6% in 2004 to 81.1% in 2006 (p<.001). Regarding opportunity-related behavioral determinants, all three items measuring perceptions of availability increased significantly. More transgenders in 2006 than in 2004 reported that condoms were available in locations where they spend time during the day (84.6% vs. 55.0% p<.001). This item was a significant predictor of condom use in the baseline study. There were also significant increases in perceived availability of NODP condoms in places where transgenders met sex partners (26.5% in 2006 vs. 5.8% in 2004, p<.001) and spent time during the day (22.3% in 2006 vs. 8.3% in 2004, p<.001). In terms of ability, one knowledge item showed a significant improvement over time. More transgenders in 2006 than in 2004 could identify the role of consistent and correct condom use with regular partners in preventing STIs (97.3% vs. 89.3%, p<.001). Unfortunately, there were significant declines in other knowledge items over time. Smaller proportions of 2006 respondents knew that oil-based lubricants increase condom breakage (33.6% in 2004 vs. 9.1% in 2006, p<.001); that having an STI can increase the likelihood of contracting HIV (87.9% in 2004 vs. 75.3% in 2006, p<.001); and that a healthy-looking person can have HIV (90.4% in 2004 vs. 49.5% in 2006, p<.001). There was no significant change in the proportion of transgenders able to identify at least two STI symptoms, with roughly 60% of respondents in both study rounds being able to do so. There was also no significant change in transgenders reported self-efficacy for condom negotiation with casual partners, which the baseline study identified as a predictor of condom use; about one-third of respondents in both study rounds said they could insist on condom use

16 Regarding motivation-related determinants, there was significantly more agreement over time with the attitudinal item, It is very important to use condoms with regular partners to prevent HIV (58.7% in 2004 vs. 91.7% in 2006, p<.001). On the other hand, the proportion of respondents agreeing that they were much more likely to use condoms with casual partners if water-based lubricant is available declined significantly (71.3% in 2004 vs. 56.1% in 2006, p<.001). There was no significant change in the level of perceived threat from 2004 to 2006: 71% of transgenders in both study rounds reported that they were very concerned about getting HIV/AIDS. One measure of exposure to the PSI intervention was included in both study rounds, and six other measures of exposure were added in There was no significant difference between transgenders in 2004 and transgenders in 2006 who reported having heard of NODP; approximately half of respondents in both rounds had heard of the product. As for other 2006 exposure variables, nearly 80% of transgenders reported that they had ever heard of PSI s transgender-specific interventions. Slightly more than half had read the PSI brochure for transgenders (56.6%) and had heard of the New Friends (Peuan Mai) drop-in center in Vientiane (59.5%). Nearly half of the respondents (47.7%) had been contacted by a PSI peer educator, and among them, 85.9% had attended a PSI group discussion and 48.0% had attended a PSI camping trip/training session

17 Table 2: Trends in Behavior, Behavioral Determinants, and Exposure to the Intervention among Transgenders Aged in Luang Prabang, Vientiane, and Savannakhet, Laos (November 2004 and June 2006) INDICATORS November 2004 N=288 June 2006 N=415 Sig. BEHAVIOR/USE 4 % % Used condom with casual partner at last anal sex *** Used condom with boyfriends at last anal sex *** Ever used water-based lubricant for anal sex *** OPPORTUNITY % % Availability Condoms are available any of the places where I spend time during the day *** Number One Deluxe Plus is available in any of the places where I meet sex partners *** Number One Deluxe Plus is available in any of the places where I spend time during the day *** ABILITY % % Knowledge Knows that oil-based lubricants increase condom breakage *** Knows that having an STI can increase the likelihood of contracting HIV *** Can identify at least two STI symptoms n.s. Knows that a healthy-looking person can still be infected with HIV *** Knows that consistent and correct condom use with regular partner can prevent STIs *** Self Efficacy Always able to insist on condom use with casual partners in the past 6 months n.s. 4 Control variables for each item included in the table are as follows: province where the interview was conducted, age, highest level of education completed, occupation, average monthly income in kip and current living arrangements. 5 Among respondents who had ever used a condom and had casual partners (n=612). In round 1, condom use at last sex with casual partners was defined as transgenders who had used a condom the last time they had anal sex with casual partners. However, in round 2, condom use at last sex with casual partners had a more stringent definition and was defined as transgenders who responded that they had used a condom the last time they had anal sex from the moment of penetration to the end of intercourse with casual partners. 6 Among respondents who had ever used a condom and had a boyfriend (n=455). 7 Among respondents who had ever heard of Number One Deluxe Plus (NODP) (n=485). At the time of the study, NODP was the only water-based lubricant available on the Lao market. 8 This item includes respondents who answered all places, most places, or some places to the following question, Are condoms available in or near the places you spend time during the day? 9 This item includes respondents who answered all or some places to the following question, Would you say NODP is available in or near the places where you meet your sex partners? 10 This item includes respondents who answered all or some places to the following question, Would you say NODP is available in or near the places where you spend time during the day? 11 The item knows that consistent and correct condom use with regular partner can prevent HIV was asked in 2006, but not in Therefore, no finding for this HIV indicator is included in the above table

18 INDICATORS November 2004 N=288 June 2006 N=324 Sig. MOTIVATION % % Attitudes Believes it is very important to use condoms with regular partners to prevent HIV *** Intention Much more likely to use condoms with casual partners if water-based lubricant available *** Threat Very concerned about getting HIV/AIDS n.s. EXPOSURE % % Has heard of Number One Deluxe Plus Has ever heard of PSI transgender activities Has ever read the PSI transgender brochure Has ever heard of the Peuan Mai drop in center Has ever been contacted by a PSI peer educator Has ever attended a PSI group discussion 12 Has ever attended a PSI camping/training session n/a n/a n/a n/a n/a n/a n.s. * <.05 p-value; ** <.01 p-value; *** <.001 p-value n.s.= non-significant 12 Among those who had ever been contacted by a PSI peer educator (n=198). 13 Among those who had ever been contacted by a PSI peer educator (n=198). Evaluation Results Table 3 shows associations between exposure to PSI transgender-specific interventions and changes in behaviors and behavioral determinants from 2004 to Exposure had a significant overall effect on condom use at last anal sex with casual partners (p<.001). All three exposure groups reported significantly higher levels of condom use than the baseline group (77.9% vs. 57.5%, p<.05 for no exposure; 68.0% vs. 57.5%, p<.05 for low exposure; and 80.3% vs. 57.5%, p<.001 for high exposure). Furthermore, transgenders with high exposure were more likely to use condoms with a casual partners than those with low exposure (80.3% vs. 68.0%, p<.05). For the item condom use with a boyfriend at last anal sex, a significant overall correlation was observed between exposure and behavior (p<.001). In all three exposure groups, the

19 proportion of transgenders who used condoms with boyfriends increased significantly from baseline (83.7% vs. 23.5% for none; 70.2% vs. 23.5% for low; and 79.8% vs. 23.5% for high; p<.001 for all). However, there were no significant differences between exposure groups, which means that the increase in the proportion of condom users at last sex with a boyfriend cannot be attributed to exposure to PSI s transgender-specific interventions. Exposure had a significant overall correlation with the behavioral determinant ever used water-based lubricant during anal sex (p<.001). Respondents reporting low and high exposure to PSI s transgender-specific interventions had significantly higher levels of waterbased lubricant use than baseline respondents (76.0% vs. 54.6% for low exposure; and 86.4% vs. 54.6% for high exposure; p<.001 for both). The low and high exposure groups did not differ significantly from each other, or from the unexposed group. There was also a significant overall correlation with exposure for the opportunity item condoms available any of the places where spend time during the day (p<.001), which was identified as a determinant of condom use in the baseline study. The proportions of transgenders agreeing with this statement were significantly higher for all three exposure categories than the proportion observed at baseline (83.2% vs. 55.0% for no exposure; 81.2% vs. 55.0% for low exposure; and 87.5% vs. 55.0% for high exposure; p<.001 for all). There were no significant differences between the no exposure, low exposure and high exposure groups. Exposure also had a significant overall correlation with the opportunity item, NODP is available in any of the places where transgenders meet their sex partners (p<.001). Agreement with this statement was significantly higher in all three exposure groups than in the baseline group (55.4% vs. 5.7%, p<.001 for no exposure; 23.6% vs. 5.7%, p<.01 for low exposure; and 25.4% vs. 5.7%, p<.001 for high exposure). Oddly enough, the no exposure group demonstrated a significantly higher level of perceived availability than either the low exposure group (55.4% vs. 23.6%, p<.05) or high exposure group (55.4% vs. 25.4%, p<.05). Findings for the last availability item, NODP available in any of the places where transgenders spend time during the day, were similar. A significant overall correlation was

20 found (p<.01). The no exposure and high exposure groups both had significantly higher levels of agreement than did the baseline group (44.4% vs. 8.3% for no exposure; and 22.9% vs. 8.3% for high exposure; both p<.01). Curiously, transgenders with no exposure reported significantly higher perceived availability than those with low exposure (44.4% vs. 17.5%, p<.05), but no significant differences were detected between no exposure and high exposure, or between low exposure and high exposure. In terms of ability, three knowledge items were significantly affected by exposure, but demonstrated declines in knowledge levels (p<.001 for all overall correlations). For the first item, about oil-based lubricants increasing condom breakage, the baseline proportion of respondents who demonstrated correct knowledge (33.6%) was significantly higher than the proportions found for all three 2006 exposure groups (no exposure, 1.6%; low exposure, 9.3%; and high exposure, 11.0%, p<.001 for all). A similar trend occurred for the item knows that having an STI can increase the likelihood of contracting HIV. More baseline respondents demonstrated correct knowledge than did respondents in any of the three exposure categories (87.9% vs. 66.9% for no exposure, p<.001; 87.9% vs. 75.0% for low exposure, p<.01; and 87.9% vs. 77.8% for high exposure, p<.01). The trend was repeated with a knowledge item about healthy-looking people being infected with HIV. The vast majority of transgenders at baseline (90.5%) knew that healthy-looking people can still be infected with HIV. This proportion was significantly higher than those for all three 2006 exposure categories (no exposure, 43.5%; low exposure, 46.3%; high exposure, 53.8%, p<.001 for all). Exposure to PSI s transgender-specific interventions had an overall influence on the final knowledge item about correct and consistent condom use (p<.01). Knows that consistent and correct condom use with a regular partner can prevent STIs was significantly higher for the low and high exposure groups than for the baseline group (97.6% vs. 89.3%, p<.01 for low exposure; 97.4% vs. 89.3%, p<.001 for high exposure). There was no difference between transgenders at baseline and the non-exposed 2006 group. There was also no

21 difference between no exposure and low exposure, or between no exposure and high exposure. For all three exposure groups, over 90% of respondents were aware that consistent and correct condom use with a regular partner can prevent STIs. In the motivation category, exposure showed an overall significant effect on the attitudinal item, It is very important to use condoms with regular partners to prevent HIV (p<.001). All three exposure groups had significantly higher levels of agreement than the baseline group with this statement (94.7% vs. 58.8% for no exposure; 87.9% vs. 58.8% for low exposure; and 93.9% vs. 58.8% for high exposure, p<.001 for all). The three exposure groups did not differ significantly from each other. The other motivation item in the evaluation table also demonstrated an overall correlation with exposure (p<.01). This item concerned the likelihood of using condoms with casual partners if water-based lubricant is available. Transgenders at baseline were more likely to state this intention than transgenders in 2006, regardless of exposure level (71.4% vs. 46.5% for no exposure; 71.4% vs. 56.3% for low exposure; and 71.4% vs. 58.3% for high exposure, p<.01 for all). The three exposure groups did not differ significantly from each other

22 Table 3 : Impact of PSI s Transgender-specific Interventions on Behavior and Behavioral Determinants among Transgenders Aged in Luang Prabang, Vientiane, and Savannakhet, Laos (November 2004 and June 2006) INDICATORS Exposure to PSI s Transgender Interventions 12 Ref. (n=288) 41.0% None (n=58) 8.3% Low (n=159) 22.6% High (n=198) 28.2% BEHAVIOR/USE 14 % % % % Sig 13 Used condom with casual partners at last anal sex 57.5 a 77.9 bd 68.0 b 80.3 cd *** Used condom with boyfriends at last anal sex 23.5 a 83.7 b 70.2 b 79.8 b *** Ever used water-based lubricant for anal sex 54.6 a 64.6 a, b 76.0 b 86.4 b *** OPPORTUNITY % % % % Availability Condoms available any of the places where spend time during the day Number One Deluxe Plus is available in any of the places where meet sex partners Number One Deluxe Plus is available in any of the places where spend time during the day 55.0 a 83.2 b 81.2 b 87.5 b *** 5.7 a 55.4 b 23.6 c 25.4 c *** 8.3 a 44.4 b 17.5 a,b 22.9 b ** ABILITY % % % % Knowledge Knows that oil-based lubricants increase condom breakage 33.6 a 1.6 b 9.3 b 11.0 b *** Knows that h 15 aving an STI can increase the likelihood of contracting HIV 87.9 a 66.9 b 75.0 b 77.8 b *** Knows that a healthy-looking person can still be infected with HIV 90.5 a 43.5 b 46.3 b 53.8 b *** Knows that consistent and correct condom use with regular partner can prevent STI 89.3 a 96.2 a,b 97.6 b 97.4 b ** MOTIVATION % % % % Attitudes Believes it is very important to use condom with regular partners to prevent HIV Intention Much more likely to use condoms with casual partners if lubricant available 58.8 a 94.7 b 87.9 b 93.9 b *** 71.4 a 46.5 b 56.3 b 58.3 b ** * <.05 p-value, ** <.01 p-value, *** <.001 p-value n.s.= non-significant 12 The baseline category contains respondents who were surveyed in The none exposure category contains respondents who had never heard of PSI s transgender-specific activities but may have participated in activities for men-who-have-sex-with-men, including transgenders, sponsored by other organizations, including LYAP or the Burnet Institute. The low exposure category contains respondents who had ever heard of PSI transgender-specific activities, read the PSI transgender-specific brochure, or heard of the Peuan Mai transgender drop-in center. High exposure contains respondents who had ever been contacted by a PSI transgender peer educator, attended a PSI group discussion, or attended a PSI camping/training session. 13 Proportions with the same letter in their superscripts do not differ significantly from one another according to the least significance difference (LSD) test. The stars in the last column represent the overall effect of exposure on the measured item. Pairwise comparisons identified as significant are at the p<.05 level or less. 14 Percentages are adjusted for province, age, education, occupation, average monthly income, and living situation

23 DISCUSSION The intent of this study was to identify changes in key indicators over time and to evaluate the impact of PSI s transgender-specific interventions on behavior change as well as transgenders opportunity, ability, and motivation to adopt condom use. Most importantly, the study found that exposure to transgender-specific interventions was associated with a higher level of condom use at last anal sex with casual partners, and with the use of waterbased lubricant, a behavior significantly associated with condom use in the baseline study. Unfortunately, no significant improvement was seen across the study years for the selfefficacy item always able to insist on condom use with boyfriends, which was identified as a behavioral predictor of condom use in the baseline study. The other baseline behavioral predictor, condoms available where transgenders spend time during the day showed improvement over time, which can likely be attributed to wider product availability and seeing the product or point-of-purchase material in retail outlets. Findings for other indicators from the project logframe showed mixed results. This study had some limitations. First, after recruiting participants at sites where transgenders congregate socially, fieldworkers were not always able to find appropriate locations to conduct the interviews, and the lack of privacy may have resulted in some information bias. In 11.3% of the 2006 interviews, a third party (for example, a friend or colleague) was present during the interview, and fieldworkers reported that the majority of these people appeared to distract respondents. Second, the use of transgendered interviewers could have biased study results and led to some over-reporting of desirable behaviors. While precautions were taken to ensure that the most appropriate interviewers were hired, and transgenders had said during formative research and pretesting exercises that they would prefer being interviewed by women and transgenders, they may have actually felt less comfortable providing sensitive information to a peer (transgender). Third, the proportion of 2006 respondents who reported never having been exposed to PSI interventions was only 8.3%. This presented challenges for the construction of exposure variables, and some spurious findings may have resulted, particularly for measures of perceived availability of NODP. Fourth, the study questionnaire did not ask about exposure to LYAP and Burnet activities, so no control measure could be used during evaluation analyses. Finally, there were