Table 3.

Summary of outcome measures, major findings, and intervention related limitations of Level III studies (K=13)

Author	Outcome measures	Important Results	Major Findings	Limitations
Broadhead, RS et al. 43	HIV prevention education; HIV-risk behaviors (sharing syringes, sharing cookers and filters, sharing rinse water, frequency of injection, safer sex)	PDI outperformed the TOI with respect to the number of PWIDs recruited, the ethnic and geographic representativeness of the recruits, and the effectiveness of HIV prevention education. The costs of recruiting PWIDs into the intervention was one- thirtieth as much in the PDI as in the TOI.	In contrast to TOI, the PDI reached a larger and more diverse set of PWIDs at much less expense.	Participants were not randomly assigned to two groups; the lack of randomization in a quasi- experiment introduces the potential for bias resulting from differences among the study groups and sites.
Heckathorn, DD et al. 44	HIV prevention education; HIV-risk behaviors (sharing syringes, sharing cookers and filters, sharing rinse water, frequency of injection, safer sex)	PDI outperforms the TOI with respect to number of people accessed, reductions in self- reported levels of HIV risk behavior and cost.	Network features such as structure, composition, and relations pay a dual role of both HIV transmission and increase the effectiveness of network-based HIV- prevention interventions.	Not mentioned
Servegev, B et al. 46	HIV prevention education; HIV-risk behaviors (sharing syringes, sharing cookers and filters, sharing rinse water, frequency of injection, safer sex); and harm reduction	The rate of drug injection among the participants remained stable during the intervention until the third follow-up visit, when the rate dropped to 53% (p = 0.005). Results also suggested large and highly significant reductions in sharing of syringes, cookers/filters, and rinse water over time, which could be linked to the client’s continuing exposures to the intervention.		Not mentioned.
Broadhead, RS et al. 49	Retention in care (%of appointments kept), adherence to medication (pill- count)	Results support the feasibility of HIV positive drug users willingness and ability to play active roles in helping one another keep up with their medical treatments. The participants kept 95% of their appointments. The peers succeeded in keeping 80% of their health advocate’s appointments. The overall adherence score for all participants was 90%. 75% of participants enrolled in drug treatment by the end of the study.	The results suggest that an alternative social support structure to drug treatment is feasible for increasing active drug users’ adherence to medical care.	Single dyads may not be especially feasible for projects that work with larger numbers of individuals. Instead, use of “chains” consisting between five and eight participants was recommended for larger projects. This would allow subjects to substitute for no-shows at the weekly meeting by serving as both a peer and an advocate for those in attendance. A collective approach that relies on chains of participants can also pool and divide the rewards that participants- in-attendance collectively earn in any given week. Such a reward arrangement may further enhance the pressure that peers exert on one another to maintain high rates of adherence for the good of the group overall. Another challenge to replicate this project on a larger scale will be the recruitment of HDUs who are not receiving any medical care.
Latkin, CA et al. 54	HIV-risk behaviors (Injection and sexual risk behaviors)	Experimental group were 3 times more likely to report reduction of injection risk behaviors and 4 times more likely to report increased condom use with casual sex partners compared to the control group. Participants in the experimental condition, compared with those in the control condition, were more likely to report talking about HIV with family members, sex partners, and drug users at the 6-month follow-up.	Results suggest that psychosocial intervention emphasizing social roles and incorporating peer outreach strategies, can reduce HIV risk in low- income and drug-using communities.	Not mentioned.
Broadhead, RS et al. 58	HIV prevention education, HIV-risk behaviors (sharing syringes, sharing cookers and filters, sharing rinse water, frequency of injection, safer sex)	Both PDIs achieved high baseline recruitment rates, although the Standard-PDI out-performed the Simplified-PDI by approximately 35% (493 recruits versus 365 recruits, respectively). However, the IDU-recruiters in the Simplified-PDI did a significantly better job educating their recruits at both baseline (an average knowledge test score of 5.19 versus 4.07 on an 8-point scale) and at follow-up 6 months later (an average knowledge test score of 7.21 versus 5.56 on an 8-point scale). Both PDIs demonstrated about equal and significant efficacy in reducing respondents’ injection frequency, the sharing of syringes and other equipment, and rates of unprotected sex.	Holding all costs constant between the two interventions except for the different rewards offered to recruiters for educating peers and recruiting them to the project, the Simplified-PDI is approximately 50% less costly in respondent fees than the Standard-PDI although the latter results in a 35% higher recruitment rate. The study appears to have demonstrated that intervention projects get what they pay for.	This is a quasi- experiment, unable to control many variables that may have influenced the results. The cities had similar characteristics but were not identical. There may be other factors at work within the two cities, or in one but not the other, which are producing the differences.
Booth, RE et al. 66	HIV-related injection and sexual risk behaviors	Both peer educators and network members in the network intervention reduced injection-related risk behaviors significantly more than did those in the individually based intervention. Peer educators increased condom use significantly more than did those in the individual intervention. Individual intervention participants, however, showed significantly greater improvements than did network members with respect to reductions in sexual risk behaviors.	Social network interventions may be more effective than individually based interventions in changing injection risk behaviors among both peer educators and network members	Although there were many similarities between the two studies, possible differences cannot be eliminated.
Deering, KN et al. 67	Pharmacy records (PR) and indirectly with self-report adherence and viral load (VL) outcomes, risk behaviors including drug use and unstable housing	Overall self-reported adherence was high (92%) and most women (11) reported increased adherence from the first to the last 13 PDI meetings attended (average increase = 18%). The number of viral load tests ≤ 50 copies/mL increased by 40% from the pre-PDI period (1 year before enrollment), to the PDI period (duration enrolled). PR adherence and improvements in VL outcomes were higher among participants with greater housing instability and frequency of injecting/smoking drugs.	The study suggests that the PDI may have had a positive impact on adherence and HIV treatment outcomes. Although this would not predict long-term treatment success, the PDI approach to HIV treatment support is a promising program for women who might otherwise be excluded from treatment altogether.	Did not have a comparison group for the PDI attendees.
Sherman, SG et al. 68	Self-reported methamphetamine use; condom use; incidence of STI	Over time, participants in both conditions showed a significant and dramatic decline in self-reported methamphetamine use (99% at baseline versus 53% at 12- months, p <0.0001) and significant increase in consistent condom use (32% baseline versus 44% at 12 months, p <0.0001). Incident STIs were common, with no differences between arms. Chlamydia had the highest incidence rate, 9.85/100 person-years and HIV had a low incidence rate of 0.71/100 person-years.	The study found that a peer educator intervention was associated with reductions in methamphetamine use, increases in condom use, and reductions in incident STIs over 12 months. Parallel reductions with the life-skills condition were also significant. Small group interventions are an effective means of reducing methamphetamine use and sexual risk among Thai youth.	Firstly, the study utilized nonrandom sampling recruitment methods and had inclusion criteria regarding regular sexual behavior and methamphetamine use; therefore generalizability of the study’s results could be limited. Secondly, condoms were available to study participants who were underwent HIV/STI counseling and testing as well as upon request. This could have affected condom use in both arms but we do not know the extent that this effected condom use over time. Lastly, there is the possibility that tight social networks were randomized to both control and intervention arms, leading to a high degree of contamination that resulted in a bias towards the null.
Booth, RE et al. 73	Needle risks, perceived risks, injection related risk behaviors	Peer leaders recruited an average of 2.4 network members; two-thirds attended at least four of the five training sessions; and a positive relationship was observed between greater session attendance by peer leaders and increased communication with network members about HIV prevention. Leaders who did not engage in high-risk behaviors at follow-up were much more likely to have had network members who did not engage in high-risk activities compared to leaders who continued high- risk behaviors.	The findings from this study suggest that a peer leader intervention approach to reducing HIV risk behaviors among IDUs in Ukraine may be effective.	The peer leaders were identified as leaders either by outreach workers or by the members of their network. It is not possible to know how representative the samples were of IDU leaders or their standing within their network. The absence of a comparison condition of IDUs not receiving the peer leader intervention prevented from making causal inferences.
Li, J. et al. 78	Adoption of innovative peer intervention delivery (self reported ‘gave’ intervention ties and self reported “received intervention ties); reduction of HIV transmission risk behaviors	The RAP innovation of peer intervention delivery and modeling had clearly diffused from PHAs to their network members and to the broader drug using community. At 6- months follow-up, more than 90% of trained PHAs had become active peer interventionists, and more than 2/3 of all study participants had adopted the peer intervention delivery and modeling innovation. Innovation diffusion is likely to reach a “critical mass” and will be efficiently adopted by the rest of the population. Sociometric network analysis showed that adoption of and exposure to an intervention was associated with proximity to a PHA, being directly linked to multiple PHAs, and being located in a network sector where multiple PHAs were clustered.	The network analysis of the RAP intervention diffusion process demonstrated that training active drug users as peer interventionist can be successful, has the potential to reach a “critical mass”, cost- effective than RCTs, and is delivered and adopted at multiple levels (individual, network, and community).	Network analysis for this study continued for 3- year long pre- and post- intervention design with only two time-point measures for each participant. It was possible that fine differences and trends in effects were not measured between the two-points. For example, network ties used in these analyses reflected relations at two time points for each participant (his/her own baseline and 6-month surveys); some of these ties may not have been continuous over the 3- year period of data collection. Another limitation of the study is the lack of a control group.
Smyrnov, P et al. 80	Number and composition: age, gender) of new recruits; knowledge test score	PDIs recruited 455 new respondents on average in each city during their six months of operation, indicating that the PDI was 6.3 times more powerful as a recruitment mechanism. Compared to TOI, the PDIs resulted in significant increases in the recruitment of women- and young- injectors, and IDUs who injected a more diverse variety of drugs.	The PDI can have a rejuvenating effect when added to HR projects. There was an increase in the number and diversity of new IDU-respondents, thereby reaching hard-to- reach critical population.	The PDI model did not perform equally in all sites, which called for more detailed analyses of contextual or geographical effects of the sites on the model’s performance. There was no pre/post test comparison of respondents’ knowledge scores, which limited the ability to conclude with confidence about the PDI’s effectiveness in increasing respondents’ overall knowledge levels. Nor can respondents’ knowledge test scores be viewed as a simple measure of recruiters’ educational effectiveness. Another limitation of the study was the inability to measure respondents’ response to the PDI’s reward structure by demographics or type of drug used. Although the PDI demonstrated statistically significant increases in the number and proportion of new recruits by age, gender and types of drug used, because we cannot exclude other possible explanations for these increases.
Latkin, CA et al. 87	Injection related HIV risk behaviors (sharing needles, sharing cookers, sharing cotton, front/back-loaded); social norm of these 4 risk behaviors	There was a statistically significant intervention effect on all four social norms of injection behaviors, with participants in the intervention reporting less risky social norms compared with controls. There was statistically significant bidirectional association with social norms predicting injection risk behaviors at the next assessment and risk behaviors predicting social norms at the subsequent visits. Through social network interventions it is feasible to change both injection risk behaviors and associated social norms.	It is critical that social network interventions focus on publically highlighting behavior changes, as changing social norms without awareness of behaviors change may lead to relapse of risk behaviors.	The assessment of social norms may have overlap with some of the risk behaviors. Since these behaviors are social behaviors, the basis of making judgments about social norms may be due in part to observations of one’s own behaviors. In the data analyses, collapsing norms and behaviors into dichotomous variables reduced investigators ability to examine linear trends.

Notes: TOI = Traditional outreach intervention; PDI = Peer driven intervention; PWIDs = People who injects drugs; PWUDs = People who uses drugs; HDUs = HIV-positive drug users; HR = Harm reduction