|Home | About | Journals | Submit | Contact Us | Français|
Peer education for HIV prevention has been widely implemented in developing countries, yet the effectiveness of this intervention has not been systematically evaluated.
We conducted a systematic review and meta-analysis of peer education interventions in developing countries published between January 1990 and November 2006. Standardized methods of searching and data abstraction were utilized. Merged effect sizes were calculated using random effects models.
Thirty studies were identified. In meta-analysis, peer education interventions were significantly associated with increased HIV knowledge (OR:2.28; 95% CI:1.88, 2.75), reduced equipment sharing among injection drug users (OR:0.37; 95% CI:0.20, 0.67), and increased condom use (OR:1.92; 95% CI:1.59, 2.33). Peer education programs had a non-significant effect on STI infection (OR: 1.22; 95% CI:0.88, 1.71).
Meta-analysis indicates that peer education programs in developing countries are moderately effective at improving behavioral outcomes, but show no significant impact on biological outcomes. Further research is needed to determine factors that maximize the likelihood of program success.
Peer education interventions are a frequently utilized strategy for preventing HIV and other sexually transmitted infections (STIs) worldwide. Such interventions select individuals who share demographic characteristics (e.g. age or gender) or risk behaviors with a target group (e.g. commercial sex work or intravenous drug use) and train them to increase awareness, impart knowledge and encourage behavior change among members of that same group. Peer education can be delivered formally in highly structured settings (such as classrooms) or informally during the course of everyday interactions.
Peer education programs are based on the rationale that peers have a strong influence on individual behavior (Population Council, 2000). As members of the target group, peer educators are assumed to have a level of trust and comfort with their peers that allows for more open discussions of sensitive topics (Campbell & MacPhail, 2002). Similarly, peer educators are thought to have good access to hidden populations that may have limited interaction with more traditional health programs (Sergeyev et al., 1999). Peer education programs may be empowering to both the educator (Milburn, 1995; Strange, Forrest & Oakley, 2002) and to the target group by creating a sense of solidarity and collective action (Population Council, 2000; Campbell & Mzaidume, 2001). Interventions using peers can also be more cost-effective than interventions that rely on highly trained professional staff (Hutton, Wyss & N’Diekhor, 2003; Pinkerton, Holtgrave, DiFranceisco, Stevenson & Kelly, 1998; Tao & Remafedi, 1998), although the costs of these interventions are often underestimated (Population Council, 2000).
Peer education interventions have been used with a number of target populations in developing countries, including youth (Agha & van Rossem, 2004; Brieger, Delano, Lane, Oladepo & Oyediran, 2001; Merati, Ekstrand, Hudes, Suarmiartha & Mandel, 1997), commercial sex workers (Ford, Wirawan, Suastina, Reed & Muliawan, 2000; Basu et al., 2003; Morisky, Stein & Chaio, 2006), and injection drug users (Broadhead et al., 2006; Hammett et al., 2006; Li, Luo & Yang, 2001). Yet to date, there has been no systematic evaluation of the effectiveness of these interventions in changing HIV related knowledge, attitudes, and behaviors in these settings. To address this gap, we conducted a systematic review and meta-analysis to assess the effect of peer education interventions on HIV knowledge, injection drug equipment sharing, condom use, and STI infection in developing country settings.
This review is part of a larger series of systematic reviews of HIV behavioral interventions in developing countries conducted jointly by the Medical University of South Carolina and the World Health Organization. Other interventions that have been systematically reviewed under this project include mass media (Bertrand, O'Reilly, Denison, Anhang & Sweat, 2006), psychosocial support (Sweat, O'Reilly, Kennedy & Medley, 2007), treatment as prevention (Kennedy, O'Reilly, Medley, & Sweat, 2007), and voluntary counseling and testing (Denison, O’Reilly, Schmid, Kennedy, & Sweat, 2008). We use standardized methods across all reviews. In this review, we sought to answer the question, what is the impact of peer education on HIV/AIDS-related outcomes? We utilized methods consistent with the “Meta-Analysis of Observational Studies in Epidemiology” (MOOSE) guidelines (Stroup et al., 2007).
We defined peer education interventions as the sharing of HIV/AIDS information in small groups or one-to-one by a peer matched, either demographically or through risk behavior, to the target population. This definition distinguishes peer education from mass media programs that may be hosted by a peer, but where no interpersonal interaction occurs and information flows in only one direction.
Studies were included in the review if they met the following criteria: (1) a peer education intervention meeting our definition was implemented; (2) the intervention was conducted in a developing country, defined as the World Bank categories of low-income, lower-middle income, or upper-middle income economies (World Bank, 2008); (3) an evaluation design was employed that compared post-intervention outcomes using either a pre/post or multi-arm study design (including post-only exposure analysis); (4) behavioral, psychological, social, care, or biological outcome(s) related to HIV prevention were presented; and (5) the article was published in a peer-reviewed journal from January 1990 through November 2006. We restricted the review to developing country settings for three reasons: (1) the epidemic is generally more severe in these settings; (2) the factors affecting the success or failure of these interventions may differ based on resource availability; and (3) there is a dearth of information on the effectiveness of these interventions at changing HIV-related attitudes and behaviors in these settings. No language restrictions were used; English translations were conducted when necessary. If two articles presented data for the same project and target population, the article with the longest follow-up was retained for analysis.
We conducted searches using five electronic databases: the U.S. National Library of Medicine's (NLM) Gateway system, PsycINFO, Sociological Abstracts, EMBASE, and the Cumulative Index to Nursing & Allied Health Literature (CINAHL). A complete list of search terms can be found in the Appendix. To identify articles not obtained from electronic database searching, we hand-searched the table of contents of four journals: AIDS, AIDS and Behavior, AIDS Care, and AIDS Education and Prevention. We also examined the reference lists of included articles to further identify articles we may have missed. This process was iterated until no new articles were found.
Initial inclusion/exclusion of studies was conducted by a member of the study staff, who excluded clearly non-relevant citations based on titles and abstracts. The remaining citations were then screened by two senior study staff according to the inclusion criteria above. The results of these two independent screenings were merged for comparison, and discrepancies were discussed to establish consensus. Final inclusion/exclusion of studies was based on a thorough reading of the full-text article.
Each article meeting the inclusion criteria underwent data extraction by two independent reviewers. Data were entered into a systematic coding form that included detailed questions on intervention, study design, methods, and outcomes. The two completed coding forms were compared and discrepancies were resolved by a third reviewer.
The rigor of the study design of included articles was assessed using an eight-point scale. This scale was developed for the larger series of systematic reviews and is additive, with one point awarded for each of the following items: (1) prospective cohort; (2) control or comparison group; (3) pre/post intervention data; (4) random assignment of participants to the intervention; (5) random selection of subjects for assessment; (6) follow-up rate of 80% or more; (7) comparison groups equivalent on socio-demographic measures; and (8) comparison groups equivalent at baseline on outcome measures.
We converted effect size estimates to the common metric of an odds ratio since all studies compared two groups and reported dichotomous outcomes. We utilized standard meta-analytic methods to derive standardized effect size estimates (Cooper & Hedges, 1994) and used the software package Comprehensive Meta-Analysis V.2.2 to conduct statistical analyses. For each outcome, we entered the odds ratio directly into the program or calculated the odds ratio from the percentages reported in the article. When other statistics were presented such as chi-square or mean differences, we converted them to the standardized mean difference, “d”, and then converted “d” to an odds ratio using readily available and widely accepted formulas (Cooper & Hedges, 1994). Odds ratios were pooled using random effects models. We attempted to contact authors when published articles provided insufficient information to conduct these calculations.
Meta-analysis was conducted on four outcomes that were reported across multiple studies: HIV knowledge, injection drug equipment sharing, condom use, and STI infection. HIV knowledge included variables measuring correct and incorrect information regarding modes of HIV transmission and prevention. Intravenous drug equipment sharing included reported episodes of sharing needles/syringes, rinse water and/or cookers. Condom use was the dichotomous proportion of respondents who either (a) did or did not use condoms, or (b) did or did not have unprotected sex. STI measures included STI incidence, current prevalence, and lifetime prevalence, and were measured by self-report, chart review, and clinical diagnosis. All STI measures were dichotomous proportions of respondents who reported or who were diagnosed with an STI using these measures.
For all outcomes, our selection of the measure to be included in meta-analysis prioritized the comparison with the longest follow-up. When articles presented multiple measures of the same outcome (e.g. condom use at last sex and ever condom use), we calculated an average effect size across these various measures within each study. The specific measures contributed by each study are listed in Table 1.
Out of 271 potentially relevant articles discovered in initial searching, thirty-four articles met our predefined inclusion criteria (Figure 1). Six of these thirty-four articles were not included in the meta-analysis because they either included the same intervention and study population as another article (Agha, 2002; Morisky, Chiao, Stein, & Malow, 2005; Morisky, Nguyen, Ang, & Tiglao, 2005), provided insufficient information for the meta-analysis (Basu, Jana, Rotheram-Borus, Swendeman, Lee, Newman et al., 2004; Sloan & Myers, 2005), or did not include one of the four outcomes of interest (Brieger et al., 2000). One article presented the results of three different studies (Ngugi, Wilson, Sebstad, Plummer & Moses, 1996). Each of these three studies was treated separately in the meta-analysis, giving a total of 30 included studies from 28 articles. The characteristics of each study are detailed in Table 1.
Of these thirty studies, thirteen were conducted in sub-Saharan Africa, ten in East and Southeast Asia, five in Central Asia, and two in Latin America and the Caribbean. Target populations included youth (N=8), commercial sex workers (CSWs) (N=12), injection drug users (IDUs) (N=4), transport workers (N=3), heterosexual adults (N=6), prisoners (N=2), and miners (N=1).
Study design rigor was mixed; rigor scores ranged from 1 to 6, with a mean score of 2.8 out of 8, which is the most rigorous (Table 2). Only three studies employed a randomized controlled design. The remaining studies included twelve serial cross-sectional studies, two post-only cross-sectional studies, ten before/after studies, and three non-randomized trials.
Table 3 presents a summary of the random-effects pooled effect sizes for the four outcomes, both overall and stratified by the seven target populations.
Eighteen studies (Agha & van Rossem, 2004; Asamoah, 1999; Merati et al., 1997; Ford et al., 2000; Morisky et al., 2006; Broadhead et al., 2006; Ergene, Cok, Tumer & Unal, 2005; Gao et al., 2002; Ozcebe, Akin & Aslan, 2004; Kinsler, Sneed, Morisky & Ang, 2004; Leonard et al., 2000; Morisky, Ang, Coly & Tiglao, 2004; Norr, Norr, McElmurry, Tlou, & Moeti, 2004; Wang & Keats, 2005; Williams et al., 2003; Vaz, Gloyd, & Trindade, 1996; Laukamm-Josten et al., 2000; Walden, Mwangulube & Makhumula-Nkhoma, 1999), with a combined study population of 15,989, generated twenty-six discrete effect size estimates on HIV knowledge (Table 4). Pooled, these data show that peer education had a moderate but positive effect on this outcome (OR: 2.28, 95% CI: 1.88, 2.75). The Q statistic for heterogeneity of 514.29 was statistically significant (p<0.0001) indicating variation across studies. Stratifying the discrete random effect size estimates by target population, peer education was significantly associated with an increase in HIV knowledge among all populations except transport workers (Table 3).
Four studies (Broadhead et al., 2006; Sergeyev et al., 1999; Hammett et al., 2006; Li et al., 2001), with a combined study population of 3,240, generated six discrete effect sizes on equipment sharing (Table 5). Three of the four studies showed statistically significant positive effects on equipment use (Sergeyev et al., 1999; Broadhead et al., 2006; Hammett et al., 2006). In one such study, receptive needle sharing during the past 6 months decreased significantly between baseline and the 24-month follow-up (47% versus 11%) (Hammett et al., 2006). However, another study showed non-significant changes in needle sharing before and after the intervention (68.3% versus 62.0%) (Li et al., 2001). This study was carried out in a drug rehabilitation center in China and the authors attribute the insignificant results to the high attrition rate among peer educators. The meta-analysis of these four studies (Table 5) showed a statistically significant reduction in equipment sharing (OR: 0.37; 95% CI: 0.20, 0.67).
Nineteen studies (Agha & van Rossem, 2004; Ford et al., 2000; Morisky et al., 2006; Broadhead et al., 2006; Li et al., 2001; Ngugi et al., 1996; Kinsler et al., 2004; Speizer, Tambashe & Tegang, 2001; Asamoah et al., 1994; Thomsen et al., 2006; van Griensven, Limanonda, Ngaokeow, Ayuthaya, & Poshyachinda, 1998; Welsh, Puello, Meade, Kome, & Nutley, 2001; Leonard et al., 2000; Morisky et al., 2004; Norr et al., 2004; Wang & Keats, 2005; Williams et al., 2003; Laukamm-Josten et al., 2000; Walden et al., 1999) with a combined study population of 17,916 individuals, generated twenty-nine discrete effect size estimates for condom use (Table 6). Results across these studies were mixed and varied by target population (Table 3). For example, only one of three studies among youth showed positive effects on condom use (Kinsler et al., 2004). The other two studies showed no intervention effects on reported condom use (Agha & van Rossem, 2004; Speizer et al., 2001), resulting in a non-significant pooled effect size for condom use among youth (OR: 1.12; 95% CI: 0.85, 1.48). However, positive intervention effects were observed among IDUs, CSWs, transport workers, heterosexual adults, and miners. When results for all target populations were combined, there was a statistically significant positive impact of peer education on condom use (OR: 1.92; 95% CI: 1.59, 2.33) (Table 6).
We also stratified the meta-analysis results for condom use by partner type (Table 3) and found moderate increases with both casual (OR: 2.23; 95% CI: 1.70, 3.09) and regular (OR: 1.94; 95% CI: 1.28, 2.94) partners. Again, there were no intervention effects among youth for either regular or casual partners. Among CSWs, there was a small intervention effect on condom use with regular partners (OR: 1.66; 95% CI: 1.06, 2.60), but no observed effect among casual partners (OR: 1.25; 95% CI: 0.38, 4.14). In contrast, among transport workers, there was a positive intervention effect on condom use with casual partners (OR: 2.93; 95%CI: 2.39, 3.60), but no effect with regular partners (OR: 2.23; 95%CI: 0.96, 5.17). Among heterosexual adults, positive intervention effects were observed for both casual and regular partners.
Seven studies (Ford et al., 2000; Morisky et al., 2006; Ngugi et al., 1996; Morisky et al., 2004; Williams et al., 2003; Laukamm-Josten et al., 2000; Dolan, Bijl & White, 2004) with a combined study population of 11,105 yielded eleven discrete effect size estimates for STI infection (Table 7). Pooled, these data reveal a non-significant increase in STI infection following the intervention (OR: 1.22; 95% CI: 0.88, 1.71). This finding was driven by three studies that reported an increase in STI infection following peer education programs (Williams et al., 2003; Laukamm-Josten et al., 2000; Dolan et al., 2004).
Differences in effectiveness across studies may be attributable to evaluation methods, context-specific factors, or different ways in which interventions are implemented. These implementation issues are analyzed below.
Peer education interventions are dependent upon the individuals who are chosen to be the peer educators, and the proper selection of peer educators is therefore key to program success. However, of the 30 studies included in this review, 16 did not report how peer educators were recruited. Of those that did report this information, three studies used self-nominated volunteers (Li et al., 2001; Ergene et al., 2005; Welsh et al., 2001). In six studies the peer educators were nominated by the target audience or through snowball recruitment (Sergeyev et al., 1999; Broadhead et al., 2006; Ngugi et al., 1996; Leonard et al., 2000; Norr et al., 2004; Laukamm-Josten et al., 2000), while in five studies the peer educators were nominated by other individuals (Merati et al., 1997; Morisky et al., 2006; Hammett et al., 2006; Kinsler et al., 2004; Dolan et al., 2004), usually program staff or supervisors of the target audience, such as prison staff (Dolan et al., 2004), brothel owners/managers (Morisky et al., 2006), and youth group leaders (Merati et al., 1997).
Similarly, how peer educators are recruited can determine how they are perceived by the target population. For example, peer educators chosen by their peers might be expected to be more popular, but less motivated than volunteers, or less skilled than peer educators chosen by program staff. Unfortunately, only 3 of 30 articles discussed how their recruitment strategies affected the implementation or success of their interventions. One study reported that their program worked “because the peer counselors were carefully selected, and were considered influential leaders among the different groups” (Morisky et al., 2004). On the other hand, another article noted that because they used a snowball recruitment approach, they did not expect all recruits to be good peer educators (Broadhead et al., 2006).
Stratifying the meta-analysis by recruitment method was possible for two outcomes: HIV knowledge and condom use. All three recruitment strategies yielded significant increases in both HIV knowledge (self-nomination: OR=7.95, 95%CI: 5.06, 12.49; target audience nomination: OR=1.38, 95%CI: 1.06, 1.79; other nomination: OR=2.36, 95%CI: 2.33, 2.39) and condom use (self-nomination: OR=2.50, 95%CI: 1.17, 5.33; target audience nomination: OR=2.84, 95%CI: 1.89, 4.27; other nomination: OR=2.01, 95%CI: 1.63, 2.49).
Training and supervision of peer educators is also likely to be an important factor in intervention effectiveness. However, seven studies did not describe what, if any, training peer educators received (Sergeyev et al., 1999; Li et al., 2001; Agha & van Rossem, 2004; Ngugi et al., 1996; Asamoah et al., 1994; Thomsen et al., 2006; Williams et al., 2003), while three studies reported that peer educators were trained but did not provide details of that training (Ngugi et al., 1996; Gao et al., 2002; Welsh et al., 2001). Fifteen studies reported that peer educators received a one-time training course that ranged in length from a few days to two months (Merati et al., 1997; Ford et al., 2000; Morisky et al., 2006; Broadhead et al., 2006; Asamoah, 1999; Ergene et al., 2005; Ozcebe et al., 2004; Kinsler et al., 2004; van Griensven et al., 1998; Morisky et al., 2004; Norr et al., 2004; Wang & Keats, 2005; Vaz et al., 1996; Walden et al., 1999; Dolan et al., 2004). Only five studies reported that peer educators were provided refresher training beyond the initial training program (Hammett et al., 2006; Ngugi et al., 1996; Speizer et al., 2001; Leonard et al., 2000; Laukamm-Josten et al., 2000). Stratifying the meta-analysis by one-time training versus on-going refresher training was again possible for two outcomes: HIV knowledge and condom use. For HIV knowledge, one-time training was associated with increased knowledge, whereas refresher training was not associated with a change in HIV knowledge (one-time training: OR=2.52, 95%CI: 1.95, 3.27; refresher training: OR=0.92, 95%CI: 0.47, 1.80). For condom use, both one-time and on-going refresher training were associated with increased condom use (one-time training: OR=1.66, 95%CI: 1.29, 2.13; refresher training: OR=2.49, 95%CI: 1.67, 3.72).
Similarly, eleven studies did not mention the level of supervision peer educators received, though one of these studies did note that peer educators wanted more supervision (Asamoah et al., 1994). Three studies reported that no supervision was given after the initial training session (Sergeyev et al., 1999; Broadhead et al., 2006; Wang & Keats, 2005). Four other studies mentioned that peer educators were supervised “regularly” or “throughout the project”, but did not provide specifics (Merati et al., 1997; Ngugi et al., 1996; Welsh et al., 2001; Ergene et al., 2005). Ten studies noted the frequency of supervision as every session (Vaz et al., 1996), twice weekly (Ford et al., 2000), weekly (Agha & van Rossem, 2004; Hammett et al., 2006; Leonard et al., 2000; Morisky et al., 2004), biweekly (van Griensven et al., 1998), monthly (Morisky et al., 2006; Laukamm-Josten et al., 2000), or quarterly (Speizer et al., 2001). Both on-going supervision and no supervision beyond training were associated with positive changes in HIV knowledge (on-going supervision: OR=2.22, 95%CI: 1.52, 3.23; no supervision: OR=5.66, 95%CI: 2.60, 12.31) and condom use (on-going supervision: OR=1.94, 95%CI: 1.41, 2.65; no supervision: OR=2.87, 95%CI: 1.40, 5.91) in stratified meta-analysis.
Peer education is often believed to be more cost-effective than other interventions because it uses volunteers, or minimally paid peers to deliver information (Milburn, 1995). Compensation of peer educators, therefore, varies widely, and the effect of compensation on intervention efficacy is unknown. Of the 8 studies that discussed compensation, only one provided no financial compensation; the school-aged peer educators in this intervention later received course credit for participation (Gao et al., 2002). The remaining seven studies generally provided small amounts of money to reimburse time or travel expenses (Morisky et al., 2006; Ergene et al., 2005; Ozcebe et al., 2004; Wang & Keats, 2005; Laukamm-Josten et al., 2000) or give incentives for educating others (Sergeyev et al., 1999; Broadhead et al., 2006). All of these studies showed positive intervention effects on condom use (Morisky et al., 2006; Wang & Keats, 2005; Laukamm-Josten et al., 2000), HIV knowledge (Ozcebe et al., 2004; Morisky et al., 2006; Broadhead et al., 2006; Ergene et al., 2005; Wang & Keats, 2005), and equipment sharing (Sergeyev et al., 1999; Broadhead et al., 2006).
Retention of trained peer educators is crucial to program effectiveness and sustainability. However, retention may be difficult since peer education interventions are often conducted with marginalized or hidden populations such as CSWs or IDUs. Twenty studies did not discuss retention of peer educators. Of the ten studies that reported retention of peer educators, one reported high retention rates among in-school youth (Speizer et al., 2001). The remainder reported mediocre to poor retention, either of previous trainees (Williams et al., 2003) or current peer educators (Ford et al., 2000; Hammett et al., 2006; Li et al., 2001; Welsh et al., 2001; Laukamm-Josten et al., 2000; Walden et al., 1999). As expected, retention seemed particularly difficult in marginalized or hidden populations. For example, two interventions with CSWs found dropout rates of 50% in the first month (Ford et al., 2000) and four months (Walden et al., 1999) respectively. Similarly, two studies attributed high turnover in IDU peer educators to a cycle of drug use and rehabilitation (Hammett et al., 2006; Li et al., 2001).
The results of this systematic review and meta-analysis yielded thirty studies that examined the effectiveness of peer education programs for HIV prevention in developing countries. These thirty studies covered a broad range of both countries and target populations. Study designs were mostly cross-sectional in nature or before-after studies with no comparison. Despite generally weak study designs, combined data from these studies showed an overall positive effect on behavioral outcomes. Peer education interventions were associated with increased HIV knowledge, reduced equipment sharing among IDUs, and increased condom use. These findings are encouraging and support the use of peer education programs in these settings. They also suggest that peer education can be an effective strategy for changing behavior among hard-to-reach, hidden populations such as CSWs and IDUs. However, while statistically significant, the effect sizes were moderate. Given limited resources, public health practitioners need to consider what effect sizes are programmatically significant.
The only reported biological outcome was STI infection, which in meta-analysis showed a null effect. This may indicate that while peer education programs can lead to positive changes in knowledge and behaviors, these changes may not result in biological impact. However, of the seven studies that measured STI infection, none used a randomized control design and only one included a cohort. Specifically, the three studies with negative results that drove the meta-analysis were all serial cross-sectional studies with an average rigor of 1.7. Weak study designs may have limited the ability of these studies to detect differences in this outcome. Further research with more rigorous designs will be needed to examine the long-term impact of peer education programs on distal outcomes such as STI incidence/prevalence and ultimately HIV incidence/prevalence.
To understand how issues such as peer educator recruitment, supervision, and training affect program success we stratified the analysis by key implementation issues. These stratified analyses did not reveal any important differences on the effectiveness of these interventions in changing behavioral and biological outcomes. Unfortunately, many articles did not report on these implementation issues. This reduced the sample size for the stratified analyses and may have limited our ability to detect important differences in program success. Moreover, even when articles did discuss implementation issues, there was no consistent way to compare across interventions. Operational research is needed to identify the factors that maximize program success. Additionally, future articles that report on peer education interventions should include more detailed descriptions of their programs so that best practices can be identified and the impact of different strategies on outcomes can be examined.
Limitations to the meta-analysis and the included studies should be considered when interpreting these findings. In addition to the limitations described above, the measurement of outcome variables varied greatly across studies. For example, condom use was alternatively defined as ever use, use at last sex, or consistent use at every sex act. Moreover, articles often stratified condom use by partner type, further complicating comparisons across studies. In order to fully utilize all of these data, we averaged the results of all definitions for each outcome across the different sub-groups within studies, and used that result in meta-analysis. However, so as not to hide important differences in effects among sub-groups, we also reported condom use stratified by partner type and by target population. Similarly, STI infection was variously reported as STI incidence, current prevalence, and lifetime prevalence, and was measured by self-report, chart review, and/or clinical diagnosis. A widely used, standardized set of measures for condom use, HIV knowledge, and other behavioral and biological outcomes would facilitate future comparisons across studies.
In summary, this is the first systematic review and meta-analysis of the effectiveness of peer education programs as an HIV prevention strategy in developing countries. The findings provide evidence that peer education programs are effective at improving knowledge and behavioral outcomes. While study designs were frequently weak, there were consistent positive effects with moderate effect sizes. However, peer education programs had no significant impact on STI infection. Unfortunately, there has been no comparable review of these interventions in developed countries, so we are unable to compare whether these interventions are more or less efficacious in developing versus developed country settings. Evaluation and critical examination of peer education programs, including exploration of implementation issues that may affect program effectiveness, are needed in order to strengthen the impact that these interventions have on changing HIV-related behavior and ultimately in reducing the incidence of HIV infections globally.
The authors wish to thank Anne Palaia, Amy Gregowski, Priya Emmart, Rae Hower, Elizabeth Jere, Sarah Mauch, Carolyn Pleisca, Andrea Ippel, Morgan Philbin, Andrea Wirtz, Devaki Nambiar, Jennifer Gonyea, Sidney Callahan, Lisa Fiol Powers, Alexandra Melby, and Lauren Tingey for their coding work.
Support: This research was supported by the World Health Organization, Department of HIV/AIDS, the US National Institute of Mental Health, grant number R01 MH071204, and the Horizons Program. The Horizons Program is funded by the US Agency for International Development under the terms of HRN-A-00-97-00012-00.
The following search terms were used when searching electronic databases: peer education and HIV; peer counseling and HIV; peer teaching and HIV; peer interventions and HIV; peer approaches and HIV; peer outreach and HIV; peer meetings and HIV; peer education and AIDS; peer counseling and AIDS; peer approaches and AIDS; peer outreach and AIDS; peer meetings and AIDS; peer evaluation and HIV; peer and HIV and intervention; peer leaders and HIV; peer networks and HIV