Objectives. We evaluated the effects of an individual intervention versus a network intervention on HIV-related injection and sexual risk behaviors among street-recruited opiate injection drug users in 5 Ukraine cities.
Methods. Between 2004 and 2006, 722 opiate injection drug users were recruited to participate in interventions that were either individually based or based on a social network model in which peer educators intervened with their network members. Audio computer-assisted self-interview techniques were used to interview participants at baseline and follow-up.
Results. Multiple logistic analyses controlling for baseline injection and sexual risks revealed that 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 and that 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.
Conclusions. Social network interventions may be more effective than individually based interventions in changing injection risk behaviors among both peer educators and network members. The effectiveness of network interventions in changing sexual risk behaviors is less clear, probably owing to network composition and inhibitions regarding discussing sexual risk behaviors. (Am J Public Health. 2011;101:336-343.)
Ukraine has experienced a dramatic and rapid rise in HIV infections, fueled initially by injection drug users. Yet, as recently as 1995 the World Health Organization characterized Ukraine as a low-prevalence country.[
In 2007 it was estimated that 1.63% of the adult population (440000 residents of Ukraine) was living with HIV/AIDS, an increase from 1.46% in 2005.[
Over the past decade, the epidemic has begun to spread through heterosexual transmission.[
A number of factors were probably responsible for the HIV epidemic among injection drug users in Ukraine. Following the collapse of the Soviet Union in December 1991, there was widespread social and economic disintegration throughout Eastern Europe, but particularly in Ukraine.[
The number of registered drug users increased between 1990 and 1996 from 30000 to 63000, and new cases of drug dependence grew from an estimated 4544 in 1991 to 11443 in 2001.[
The injecting practices of Ukraine drug users also probably played a critical role in the epidemic.[
Pseudoephedrine (known as boltushka, vint, and jeff) is typically obtained in over-the-counter cold medicines and also involves the use of shared drug-mixing containers.[
We compared 2 intervention projects we conducted with injection drug users in Ukraine. The first was an individually focused intervention implemented between 2004 and 2006 in Kiev, Odessa, and Donetsk. The second was a network intervention conducted in 2005 and 2006 in Nikolayev and Simferopol, 2 cities in the Crimea. HIV prevalence at baseline ranged from 20% (Donetsk) to 65% (Nikolayev). In each location, notable for their high concentrations of injection drug users, interventions were delivered by nongovernmental agencies (NGOs) selected as a result of their experience in working with injection drug users and their expressed interest in HIV prevention.
The 2 interventions (both guided via instruction manuals) were implemented in their respective cities to reduce injection and sexual risk behaviors. The sampling plan was similar across all sites, as were the study methodology, participant eligibility criteria, and measures used, including audio computer-assisted self-interview (ACASI) techniques. Also, intervention training was conducted by the same staff, including the principal investigator, data manager, and intervention trainers.
Five hundred fifty injection drug users took part in the assessment of the individually focused intervention, and 172 injection drug users (54 peer educators and 118 network members, as described subsequently) participated in the network intervention. Peer educators and those in the individual intervention were recruited by recovering drug users serving as outreach workers, an approach that has been shown to provide appropriate samples for this type of research.[
To obtain a diverse sample, we extended recruitment throughout all of the districts in each city, although specific areas within districts were targeted on the basis of the NGO staff's knowledge of where injection drug users congregated. Prospective participants were referred to NGO offices, where eligibility was finalized, a further description of the study was provided, and informed consent was obtained. As a means of minimizing social desirability, ACASIs were used to conduct structured interviews.
Following the interview, HIV testing was performed with a rapid test approved in Ukraine, the HIV I+II One-Step Test (Orgenics Ltd., Yavne, Israel), and counseling was conducted with a modified version (based on the injection practices of Ukraine drug users[
In addition, peer educators in the network intervention, as described subsequently, were required to recruit up to 3 members of their injecting network and express a willingness to participate in 5 training sessions designed to educate and empower them to teach their network members about HIV prevention. Because only 2 injection drug users in the network study did not report injecting opiates, the sample in the individually focused intervention was also restricted to opiate injectors, either alone or in combination with other drugs. Participants were compensated the equivalent of US $5 for their initial interview and US $6 for their 6-month follow-up interview.
Before the commencement of either intervention project, a 5-day training session for all staff led by the principal investigator and US trainers was held. The first day of training was devoted to good research practices for all staff. Next, separate training sessions were held for interviewers, outreach workers, and HIV testers and counselors. Outreach workers were indigenous members of the target population hired and trained to recruit individual intervention participants and peer educators and deliver the interventions. Outreach worker training was led by US outreach workers familiar with the interventions.
Individually focused intervention. This preventive intervention, based on the Indigenous Leader Outreach Model,[
Network intervention. This intervention, tested in Nikolayev and Simferopol, was based on Latkin's model,[
Each highly scripted session included role-playing and other interactive learning techniques and exercises. Outreach workers from the NGOs were expected to conduct these 5-session training programs with peer educators, who were recruited in 3 waves, with 10 members in each wave. Network members typically accompanied their peer educator to the NGOs in groups. No interventions with peer educators occurred after the fifth session was completed, and there were no intervention sessions directly involving network members. Network members did, however, receive HIV testing and counseling, as did peer educators.
The primary interview schedule was a modification of the risk behavior assessment developed by a grantee consortium as part of the 1990 cooperative agreement sponsored by the National Institute on Drug Abuse. It assesses demographic characteristics, health history, drug use and injection behaviors, and sex-related risk behaviors. Reliability and validity assessments of the risk behavior component support its use with injection drug users.[
Drug use, needle risk, and sexual risk behaviors were assessed at baseline and 6 months, with a focus on the 30-day period preceding each interview. Needle risk behaviors were number of times injecting, frontloading and backloading, drawing drugs from a common container used by other injection drug users, injecting with a needle or syringe after its use by another injection drug user, using works (cotton, cooker, or water) previously used by other injectors, and a composite binary risk measure defined as a report of at least 1 of these behaviors. Sexual risk behaviors included multiple sex partners, vaginal or anal sex without a condom, sex with a partner who was an injection drug user, sex with a partner who was HIV positive or whose HIV status was unknown, and a sexual risk composite measure defined as a report of at least 1 of these risk behaviors.
Given that the purpose of this study was to evaluate the outcomes of an individual versus a network intervention, comparisons were made separately between participants in the individual intervention and peer educators and between those in the individual intervention and network members. Initially, the χ
Finally, we computed 2 sets of logistic regression equations for each of the drug use, needle risk, and sexual risk behaviors at follow-up. One set compared individual intervention participants with peer educators, and the other compared individual intervention participants with network members. A variable contrasting the 2 samples was included in each logistic regression equation (the individual sample was assigned a value of 0 and the network sample a value of 1). The baseline measure of the outcome was included in each equation as a control variable. We report odds ratios (ORs) for contrast variables along with 95% confidence intervals (CIs) and probability levels. ORs greater than 1 indicated that the individual intervention sample had more positive outcomes after control for baseline differences, whereas ORs less than 1 indicated that the network samples had more positive outcomes. Least squares regression was used for number of times injecting, and beta weights were used for contrast variables and probability levels.
Findings related to sample characteristics and the various risk behaviors assessed are described in the sections to follow.
Overall, 628 opiate injectors or injectors of a mix of opiates and sedatives were recruited for the individual intervention sample along with 61 peer educators and 145 network members. At 6 months, 550 (88%) participants in the individual intervention were successfully reinterviewed, as were 54 peer educators (84%) and 118 network members (81%). Of those not contacted at follow-up, 7% of the individual intervention participants were located and found to be incarcerated, hospitalized, or deceased or to have moved out of the area. In the network sample, 7% of peer educators were in prison or hiding from the police, and 12% of network members were in prison, were deceased, or had moved to another city. Thus, follow-up locator information was obtained for 91% of individual intervention participants, 95% of peer educators, and 93% of network members.
Baseline, needle use, and sexual risk variables were used in conducting attrition analyses. One significant finding was noted: those in the network cohort who were not reinterviewed were more likely than were those who were reinterviewed to have injected with a used needle or syringe (29% vs 13%; χ
At baseline, 73% of individual intervention participants reported injecting opiates, 56% reported injecting a mix of opiates and sedatives, and 35% reported injecting stimulants. The corresponding percentages were 78%, 48%, and 19% among peer educators, respectively, and 84%, 36%, and 12% among network members, respectively. Demographic and background data for those who completed the follow-up interview are presented in Table 1.
The average age of the sample members was 30.[
The 3 cities where the individual intervention was implemented are much larger and less remote, with higher employment and sexually transmitted disease rates, than those in the Crimea. Overall, 39% of the sample tested positive for HIV, with rates ranging from 32% among peer educators to 45% among network members.
Needle and sexual risk behaviors at baseline and 6 months are shown in Table 2. Change scores and 95% CIs between baseline and 6-month follow-up are reported in Table 3. Results showed that peer educators injected more frequently than did participants in the individual intervention (53 vs 29 times; P<.05). Both peer educators and network members were more likely to use a common container than were individual intervention participants (51%, 44%, and 28%, respectively; peer educators vs individual intervention participants, P<.001; network members vs individual intervention participants, P<.01); they were also more likely to use dirty works (39%, 33%, and 25%, respectively; peer educators vs individual intervention participants, P<.05). However, a higher percentage of individual intervention participants (88%) than peer educators (80%) or network members (72%; P<.001) frontloaded or backloaded.
Participants in the individual intervention were also more likely to report at least 1 of the behaviors included in the needle risk composite (92%) than were peer educators (80%; P<.01) or network members (79%; P<.001). However, higher percentages of peer educators and network members than individual intervention participants reported 1 or more risk behaviors on the sex risk composite (98%, 88%, and 74%, respectively; P<.001 for both comparisons).
Change at follow-up. Changes from baseline to follow-up (and 95% CIs) were computed for needle and sexual risk behaviors by subtracting follow-up scores from baseline scores. Thus, a positive change actually reflected a reduction in risk behavior from baseline to follow-up (Table 3). Findings showed a strong trend for greater change in needle-related risk behaviors in the 2 network cohorts than among the individual intervention participants. Peer educators exhibited greater levels of change than did the individual intervention participants with respect to injection frequency (reduction of 28 times vs 8 times), frontloading and backloading (reduction of 36% vs 16%), using a common container (reduction of 32% vs 6%), using dirty works (reduction of 27% vs 11%), and the drug risk composite (reduction of 33% vs 17%). Relative to individual intervention participants, network members exhibited greater reductions in injection frequency (9 times vs 8 times), use of a common container (18% vs 6%), and use of dirty works (20% vs 11%). By contrast, individual intervention participants showed greater change than network members in use of a dirty syringe (11% vs 1%). Results for sexual risk behaviors were more mixed. Peer educators exhibited greater reductions than individual intervention participants with respect to sex without a condom (26% vs 13%) and the sex risk composite (17% vs 9%); the decrease in sex with injection drug users was greater among individual intervention participants (8% vs 2%). In addition, there was a decrease among individual intervention participants in sex with an HIV-positive partner or a partner of unknown status (8%), whereas an increase was observed among peer educators (11%). Sexual risk effect sizes indicated more positive changes among individual intervention participants than among network members.
We also examined change within each of the 5 cities included (Makeyevka, Kiev, and Odessa for the individual intervention and Simferopol and Nikolayev for the social network samples; data not shown) to assess whether there were different patterns of results within the different cities. These results showed that overall the pattern of change was very similar to that observed for the combined samples. There was greater change in injection risk behaviors in the 2 social network cities than in the 3 individual intervention cities, and the patterns for sexual risk behaviors within the cities generally corresponded to those shown in Table 3.
Logistic regression findings. Table 4 shows ORs for contrast variables along with 95% CIs and probability levels. As mentioned, ORs less than 1 reflect outcomes more favorable to peer educators and network members, whereas ORs greater than 1 indicate outcomes more favorable to participants in the individual intervention. Logistic regression equations were computed for each 6-month outcome measure separately between individual intervention participants and peer educators and between individual intervention participants and network members. Dichotomous contrast variables for comparisons between individual intervention participants (coded as 0) and peer educators (coded as 1) and for comparisons between individual intervention participants (0) and network members (
Reductions in needle risk behaviors were greater in the 2 network samples than they were among individual intervention participants. Peer educators (OR=0.35; P<.001) and network members (OR=0.67; P<.001) exhibited significantly reduced frontloading and backloading at follow-up relative to individual intervention participants. They also showed significantly greater change on the needle risk composite measure (peer educators: OR=0.32; P<.001; network members: OR=0.67; P<.001).
Sexual risk behavior findings were less consistent. Reductions in sex without a condom were greater among peer educators than they were among individual intervention participants (OR=0.42; P<.05). By contrast, individual intervention participants had more favorable outcomes than did peer educators in terms of sex with an HIV-positive partner or a partner of unknown status (OR=2.01; P<.05) and more favorable outcomes than did network members in terms of sex with multiple partners (OR=1.53;P<.001), sex with an HIV-positive partner or a partner of unknown status (OR=1.31; P<.02), and the sex risk composite (OR=2.17; P<.001).
As our findings reflect, the behaviors of Ukraine injection drug users place them at great risk for HIV. The members of our sample began injecting in their late teens to early 20s on average, and most had injected for 10 or more years at the time of the study. Overall, in the 30 days before the baseline interview, 85% had frontloaded or backloaded, 32% had injected drugs from a common container, nearly 20% had injected with a used needle or syringe, 27% had used dirty works, and 89% had engaged in at least 1 of these needle-related risks. In terms of sexual risks, 25% of the participants reported multiple partners, 53% reported vaginal or anal sex without a condom, 37% reported sex with another injection drug user, 36% reported sex with someone who was HIV infected or whose HIV status was unknown, and 78% reported at least 1 of these sex risk behaviors. Not surprisingly, 39% were HIV positive. Given the relatively young age of the injection drug users in our study, these figures are disturbing.
Risk behaviors were prevalent across all 3 cohorts. Injection frequency was highest among peer educators, who were also the most likely to inject drugs from a common container, use dirty works, and report at least 1 sexual risk behavior. Participants in the individual intervention, in contrast, were the most likely to frontload or backload and to report at least 1 risk on the needle risk composite measure. On both of the composite measures, as well as use of a common container and dirty works, network members tended to be more similar to peer educators than they were to individual intervention participants.
Several noteworthy findings regarding the interventions and changes from baseline to follow-up emerged. For the most part, injection and sexual risks decreased over time. Reductions in each of the needle-related risk behaviors assessed were reported by all 3 cohorts. Findings relative to sexual risks were more uneven. Vaginal and anal sex without a condom and sex with an injection drug user decreased in all 3 samples, but more network members reported having multiple sex partners at follow-up; in addition, both network and peer educators reported higher frequencies of sex with a partner who was HIV positive or whose HIV status was unknown.
In the comparison of the 2 interventions, peer educators and network members reported significantly greater risk reduction than did those in the individual intervention with respect to frontloading or backloading and the needle risk composite score, and peer educators exhibited more reductions than did individual intervention participants in unprotected vaginal and anal sex. However, individual intervention participants decreased their sexual activity with HIV-positive partners or partners of unknown status more than did peer educators and network members; also, their reports of multiple sex partners decreased more than did those of peer educators, as did their scores on the sexual risk composite measure.
These findings appear to indicate that drug and needle behaviors are more amenable than are sexual behaviors to discussion within networks, given that drug users often talk about acquiring and using drugs. Moreover, as the networks tended to be composed of more drug users than sexual partners, it is likely that there was more discussion about injection risk behaviors, which injection drug users perceive as their greatest risk factor. Talking about drug use may also be more normative than talking about sex. In their social network study in Thailand and Philadelphia, Latkin et al. also found greater decreases in injection risk behaviors in the network condition than in the control condition, whereas there were no between-group differences in sexual risk reduction.43
One of the considerations in comparing the 2 interventions is their cost. Although a thorough cost-effectiveness analysis is beyond the scope of this article, some comparisons can be made. NGO directors, project managers, outreach workers, interviewers, and HIV testers and counselors were paid approximately the same in all locations, depending on the local economy. However, the individual intervention required 4 outreach workers in each city to recruit participants and perform interventions, whereas the network intervention required only 2. In the individual intervention, 10 participants were recruited monthly at each site. In the network arm, in each of the 3 phases 4 weeks were devoted to recruiting peer educators and their network members followed by 2 weeks of training peer educators. This resulted in a monthly average of 26 participants recruited over the 18 weeks of recruitment in each of the 2 cities conducting the network intervention. Thus, with one half the number of outreach workers recruiting for the network intervention as for the individual intervention, 2.6 as many participants (peer educators and network members) were brought into the network intervention as compared with the individual intervention.
There are a number of limitations to consider when drawing inferences from our data. For example, the findings reported were derived from 2 studies. Although there were many similarities between the studies, including training and oversight by the same staff, identical measures, use of an ACASI, eligibility criteria, recruitment methods, and time periods, possible differences cannot be eliminated. Conducting international research also presents challenges; however, quarterly site visits by the US investigators and bimonthly visits by our Ukraine colleagues, as well as training in good research practices, likely minimized this problem.
Findings were derived from self-reported information, which is subject to both recall error and social desirability bias. The 30-day time period participants were asked to remember should have reduced recall error, and although social desirability could not be eliminated, it would have been present in all cohorts. Finally, our findings may not be generalizable to all injection drug users, given that our sample was composed of individuals who were willing to spend the time required to participate and who were motivated by the modest stipend offered.
Despite the alarming rate of HIV among injection drug users in Ukraine, interventions other than needle exchanges, which have limited coverage, are lacking. Our work is the first and the most rigorously conducted to date with street-recruited injectors in the Ukraine. The 5 cities included in our study represent the areas of Ukraine with the highest concentrations of injection drug users and HIV infection.[
Our aim was to compare 2 intervention efforts with similar methodologies to determine which was more effective in reducing the risk behaviors known to transmit HIV. Although both interventions were generally found to be beneficial, the network intervention was more effective than was the individualized intervention in reducing needle-related risks, the primary HIV transmission factor in Ukraine. Among peer educators, the network intervention was also more effective than was the individual intervention in reducing unprotected sex, perhaps the single most important behavior change that can be made regarding sexual risk, particularly among those who are already infected.[
The fact that the network-focused intervention was more effective with peer educators than it was with network members -- which is not surprising given that peer educators were directly trained by project staff and that information about safer sex and injection practices was transmitted to network members by their peer educators -- helps validate our findings (i.e., network members indirectly received the intervention through their peer educator).[
Reprints can be ordered at
This article was accepted September 15, 2009.
This study was supported by the National Institute on Drug Abuse (grant RO1 DA17620).
We acknowledge the dedicated staff and directors who participated in this project including Elena Teryayeva with Health of Nation in Makeyevka/Donetsk; Olga Kostyuk and Tatiana Semikop with Faith, Hope and Love in Odessa; Oxana Gorbach with the Substance Abuse and AIDS Prevention Foundation in Kiev; Tatyana Safonova with Lotus in Simferopol; and Elena Goryacheva with the Charity Foundation Vykhod in Nikolayev. Their commitment to preventing the further spread of HIV in their country is inspiring. We are also indebted to the drug users who agreed to participate and gave their time, without which we could not have conducted this study. We thank Christina Schmitt for her editorial assistance. Finally, we acknowledge the thoughtful suggestions of the anonymous reviewers.
This study, including procedures for informed consent, was approved by the Colorado Multiple Institutional Review Board at the University of Colorado Denver.
Note. GED = general equivalency diploma.
[a]For comparison between individual intervention participants and peer educators.
[b]For comparison between individual intervention participants and network members.
*P<.05; ***P<.001.
[a]For comparison between individual intervention participants and peer educators.
[b]To squirt a drug solution into the front or back of the user's syringe using the dealer's syringe.
[c]For comparison between individual intervention participants and network members.
[d]Cotton, cooker, or water previously used by another injection drug user.
[e]A binary risk measure defined as a report of at least 1 of these needle risk behaviors.
[f]A binary risk measure defined as a report of at least 1 of these sexual risk behaviors.
*P<.05; **P<.01; *** P<.001.
Note. CI = confidence interval. Changes from baseline to follow-up were computed for needle and sexual risk behaviors by subtracting follow-up scores from baseline scores. Thus, a positive change actually reflected a reduction in risk behavior from baseline to follow-up.
[a]To squirt a drug solution into the front or back of the user's syringe using the dealer's syringe.
[b]Cotton, cooker, or water previously used by another injection drug user.
[c]A binary risk measure defined as a report of at least 1 of these needle risk behaviors.
[d]A binary risk measure defined as a report of at least 1 of these sexual risk behaviors.
Note. CI = confidence interval; 0R=odds ratio.
[a]Beta weight from least squares regression.
[b]To squirt a drug solution into the front or back of the user's syringe using the dealer's syringe.
[c]Cotton, cooker, or water previously used by another injection drug user.
[d]A binary risk measure defined as a report of at least 1 of these needle risk behaviors.
[e]A binary risk measure defined as a report of at least 1 of these sexual risk behaviors.
By Robert E. Booth, PhD, Robert E. Booth is with the Division of Substance Dependence, Department of Psychiatry, School of Medicine, University of Colorado Denver. 1741 Vine St, Denver, CO 80206 (e-mail: robert.booth@ uchsc.edu).; Wayne E. K. Lehman, PhD, At the time of the study, Wayne E.K. Lehman was with the Division of Substance Dependence, Department of Psychiatry, School of Medicine, University of Colorado Denver.; Carl A. Latkin, PhD, Carl A. Latkin is with the Department of Health, Behavior and Society, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD. Sergey Dvoryak is with the Ukrainian Institute on Public Policy, Kyiv, Ukraine.; Sergey Dvoryak, MD, PhD; John T. Brewster, LCSW, John T. Brewster is with the Division of Substance Dependence, Department of Psychiatry, School of Medicine, University of Colorado Denver.; Mark S. Royer, MS, Mark S. Royer is with the Division of Substance Dependence, Department of Psychiatry, School of Medicine, University of Colorado Denver. and Larisa Sinitsyna, MA, Larisa Sinitsyna is with Counterpart International, Kyiv, Ukraine.