Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2008 Jul 10.
Published in final edited form as: J Subst Abuse Treat. 2006 Jul 13;31(2):163–171. doi: 10.1016/j.jsat.2006.04.002

Behavioral HIV Risk Reduction among People Who Inject Drugs: Meta-Analytic Evidence of Efficacy

Michael M Copenhaver a,✉, Blair T Johnson a, I-Ching Lee a, Jennifer J Harman b, Michael P Carey c; SHARP Research Teama,d
PMCID: PMC2452992  NIHMSID: NIHMS52201  PMID: 16919744

Abstract

We conducted a meta-analysis of randomized controlled trials (RCTs) to evaluate behavioral HIV-risk reduction interventions targeting people who inject drugs. We included 37 RCTs evaluating 49 independent HIV-risk reduction interventions with 10,190 participants. Compared to controls, intervention participants reduced injection-and non-injection drug use, increased drug treatment entry, increased condom use, and decreased trading sex for drugs. Interventions were more successful at reducing injection drug use when participants were non-Caucasians, when content focused equivalently on drug- and sex-related risk, and when content included interpersonal skills training specific to safer needle use. Condom use outcomes improved when two intervention facilitators were used instead of one. Injection drug use outcomes did not decay, but condom use outcomes did. Behavioral interventions do reduce risk behaviors among people who inject drugs, especially when interventions target both drug- and sexual-risk behavior, and when they include certain behavioral skills components. Implications for future interventions are presented.

Keywords: HIV prevention, risk reduction, behavioral intervention, injection drug use, meta-analysis, substance abuse

1. Introduction

Despite a wide array of HIV prevention approaches and related research, there is a decade-long trend of 40,000 new HIV infections per year in the U.S. (CDC, 2004). Injection drug use (IDU) remains an important vector for the transmission of new HIV infections (Avants et al., 2004; Margolin et al., 2003), and people who inject drugs experience increased risk through both drug- and sexual behaviors. Because people who inject drugs exhibit diverse risk behaviors, this group presents a formidable prevention challenge. Specifically, investigators report considerable variability over time in the risk profiles of people who inject drugs owing predominantly to their current amount of HIV risk behavior, HIV-serostatus, and the recency of their exposure to a HIV risk reduction intervention. For example, investigators have described observing different individuals who inject drugs as circulating in and out of risky behavior states over time rather than exhibiting steady HIV risk behavior states (Celantano et al., 2001).

A number of HIV prevention interventions have now been identified as demonstrating evidence of effectiveness (Semaan et al., 2002; Strathdee et al., 1998; Vlahov et al., 1997). Researchers have also begun to evaluate strategies to facilitate the integration of evidence-based interventions into clinical settings (Fisher et al., 2004; Kalichman et al., 2001). Even so, individual studies have yielded little evidence about the intervention components that are most helpful for risk reduction interventions for people who inject drugs. Therefore, we conducted a meta-analysis of the randomized controlled trials (RCTs) of behavior change interventions targeting people who inject drugs. We strove to specify the intervention components that could be used for the future development and deployment of more effective strategies for reducing drug- and sex-related HIV risk among those who inject drugs.

2. Materials and methods

2.1 Sample of Studies

Reports were gathered using four overlapping strategies: (1) searches with terms including “HIV,” “AIDS,” “human and immun* and virus,” “acquired and immun* and syndrome” and “interven* and prevent* paired with (needle* or inject* or idu or IDU or IVDU or ivdu”) in electronic databases MEDLINE, PsycINFO, AIDSLINE, CINAHL, Dissertation Abstracts Online, and ERIC; (2) requests for papers sent to authors and to e-mail list serves, (3) searching reference sections of articles already obtained, (4) manually searching conference proceedings and professional journals not available electronically. Studies that were available as of March 30, 2004 and that matched the selection criteria (discussed below) were included.

2.2 Study Selection Criteria

To qualify for the meta-analysis, a study had to (1) evaluate a group or individual behavioral HIV prevention intervention, (2) include a sample of at least 50% of people who report injecting drugs, (3) include IDU-related behavioral outcomes, (4) use an RCT design, and (5) provide statistical information for the calculation of effect size (Figure 1). After reviewing 713 abstracts that contained one or more keywords, we identified 129 studies that reported evaluating a behavioral intervention with IDU-related outcomes. Studies were excluded if they did not implement a control group (k = 41) or if data reported in one study overlapped with data reported in another study or when a study presented follow-up data that were included in another analysis (k = 20). The primary instance was the public domain report that included the individual site data from the National AIDS Demonstration Research project (NADR; NIDA, 1994). Seventeen other studies were excluded because they had insufficient statistics and 14 studies were excluded because less than 50% of their samples were characterized as people who inject drugs or because they included a specialized sample (e.g., prisoners). Studies reporting behavioral measures were included even if measures were indirectly or distally related to HIV-risk reduction (e.g., communication skills pertaining to needle sharing; discussing safer sex with a partner). Thirty-seven studies ultimately qualified based on our selection criteria; of these, 26 studies evaluated a single intervention, 10 evaluated 2 interventions, and 1 evaluated 3 interventions. In total, there were 49 independent interventions using a total of 10,190 participants following attrition, representing a retention rate of approximately 65% (SD = 25%) of participants.

Figure 1.

Figure 1

Open in a new tab

Selection process for study inclusion in the meta-analysis

2.3 Extraction of Effect Size Information

A minimum of two trained raters independently coded each study to describe the studies and to determine in the stratified analyses whether variation in effect sizes (ESs) could be linked to features of the studies. Across the study- and intervention-level categorical dimensions, coders agreed on 82% to 100% of the judgments, resulting in an average Cohen’s κ value of .80. Disagreements were resolved through discussion. Effective reliability for continuous variables was calculated with the Spearman-Brown result, which takes into account the mean inter-rater correlation as well as the number of raters, and ranged from .85 to 1.00 (M = .99).

ES estimates were based on the information provided in a report or from data provided by the authors. ES estimates were calculated in standard deviation units because the selected studies typically used parametric statistical tests with continuous measures of intervention outcomes (odds ratios are the ES of choice when the outcomes are dichotomous in nature; Lipsey & Wilson, 2001). Specifically, the ES was calculated as the difference on risk measures between the means of the intervention group and the comparison group divided by the pooled standard deviation; ESs were corrected for sample size bias. Signs of ESs were set so that positive values indicated improvement (i.e., reduced risk). If more than one comparison group was included in a given study, which occurred in a minority of cases (k = 6), we used the comparison group that was most similar to the modal comparison group in the literature, typically a brief HIV intervention or HIV/AIDS education.

Multiple ES estimates were calculated from individual studies when more than one behavioral measure was reported, including these behavioral outcomes: (1) frequency of injection drug use, (2) frequency of non-injection drug use, (3) frequency of sharing needles/equipment, (4) frequency of bleaching needles/equipment, (5) entry into drug treatment, (6) frequency of condom use, (7) frequency of sex-related HIV risk behavior, and (8) frequency of trading sex for drugs. No studies provided data on HIV incidence, precluding examination of this outcome. Due to sample size constraints, ESs were calculated using only data reported at baseline and post-intervention measurement points although a minority of the studies reported data at follow-up points beyond post-intervention. ESs were corrected for any significant between-group differences on behavioral outcomes at baseline.

We calculated weighted mean ESs and 95% confidence intervals around mean ESs using conventional fixed-effects meta-analytic procedures and estimated the homogeneity (Q) of the ESs that comprise each mean. In order to avoid violating the assumption of independent observations, analyses proceeded by examining the studies’ outcomes in separate analyses. For the two outcomes with significant overall Q values (i.e., IDU and condom use), we determined whether coded features of studies related significantly to the variability in the magnitude of the ESs. The IDU set of ES outcomes was large enough to apply more than one feature at a time but the condom use set was not. In these models, we controlled for Type I error by using the Bonferroni correction for each set of ESs (effectively, .0008 for IDU outcomes and .0009 for condom use outcomes). Study dimensions that were significantly related to IDU variability were further pursued in a series of multiple-regression models that controlled for inter-correlations among study dimensions that were maintained. Such models determined whether variation attributed to one study dimension would be better attributed to another study dimension. The number of studies was too small to permit all study dimensions to be included simultaneously, but did permit model testing with three predictors at one time. The univariate predictor with the best fit was paired sequentially against the other univariate predictors. The end result was that the final multiple regression model with the remaining significant study dimensions was correctly specified, meaning that the variation remaining unexplained was within the limits of sampling error and that there was no need to incorporate random effects in this analysis, consistent with meta-analytic convention (e.g., Johnson et al., 2003; Johnson & Eagly, 2000; Lipsey & Wilson, 2001). Supporting this conclusion, an analysis incorporating random-effects assumptions did not change the pattern of results on the regression analysis. To maintain cases in these models, we used the multiple-imputation method to estimate missing values because it has proven superior for this purpose (Pigott, 2001; Sinharay et al., 2001). Of variables pertaining to methodology, sample characteristics, and control group characteristics, 32% of the variables had missing values; these were estimated by multiple imputation method using NORM 2.03 (Schafer & Graham, 2002), whereas 68% of those variables required no imputation. NORM assigns a value from a set of multiple imputation values created from conceptually meaningful variables provided by each study under an unstructured normal model. Imputation was not used if less than 50% of the observations were available.

3. Results

3.1 Description of Studies and Their Interventions

Seventy-eight percent of the studies were in articles published in peer-reviewed journals; 20% were studies from individual sites that were published in the public domain report of the National AIDS Demonstration Research project (NADR; NIDA, 1994); and 2% were obtained from unpublished documents. Approximately 49% of the studies advocated a specific behavioral theory (or a blend of theoretical perspectives) in their approach to risk reduction whereas the remainder (about 51%) did not. The majority of investigations (88%) occurred in the U. S. and involved participants within medium to large cities. Slightly less than a half of the studies (47%) recruited community participants who were not enrolled in drug treatment while the remainder recruited participants enrolled in treatment (e.g., methadone maintenance, outpatient drug treatment, inpatient drug treatment, detoxification).

Study participants tended to be middle-aged males (M = 35 years, range: 26 to 41 years; 61% male). The samples were ethnically diverse, with 51% African American, 29% Caucasian, and 20% Latino. The vast majority (88%) reported recent IDU (i.e., within the past 3 months), 46% reported recent needle sharing or borrowing, and 29% reported recent sex trading. Those who reported current IDU were most likely to report using heroine (54%) and cocaine (54%). Approximately 82% reported being sexually active, with 50% reported having sex with another person who injects drugs, and 28% being aware of that they were HIV-infected.

All studies adopted randomized controlled trials. Among them, fifty-one percent of the studies implemented a brief HIV-risk reduction intervention as the control condition whereas 22% used HIV/AIDS education only, 20% implemented a wait-list/no treatment, 4% used substance abuse treatment only, and 2% used an intervention with non-HIV content as the control. All studies used a pre-test/post-test design in which data were obtained from the same participants at distinct time points. Outcomes were assessed at baseline and post-test for 49% of studies, at baseline, post-test, and one follow-up for 45%; and at baseline, post-test, and two follow-ups for 6% of studies.

Group interventions were the most common (51%), with individual interventions used in 44% of studies; the remaining 5% used a combination of group and individual modes. On average, interventions consisted of 8 sessions of approximately 87 minutes each (range = 33 to 300 minutes) over the course of 29 days (range = 1 to 113 days). Studies evaluating a group-level intervention included approximately 4 participants per group and were most commonly facilitated by either 1 or 2 group leaders. Group facilitators included individuals described as paraprofessionals (without an advanced degree or license; 63% of studies), professionals (with an advanced degree or license; 27% of studies), both paraprofessionals and professionals (8% of studies), and neither paraprofessionals nor professionals (e.g., peers; 2% of studies).

Most interventions included the following components: HIV/AIDS education (90% of interventions), condom use skills (69%), self-management skills (e.g., coping with drug cravings; 57%), and included both drug- and sex-related risk reduction (70%). Less commonly, interventions contained drug treatment (35%), provided bleach (35%), and provided condoms (35%). No intervention included the provision of clean syringes to participants.

3.2 How Efficacious Were the HIV-Risk Reduction Interventions?

First, we evaluated how successful interventions were relative to control conditions on risk reduction outcomes (see Table 1). The outcomes included IDU (30 studies); condom use (16); non-injection drug use (11); and entry into drug treatment (6). Using data from the first available assessment point following the completion of the intervention, participants significantly: (1) reduced IDU, d = 0.08 (95% CI = 0.03, 0.13), (2) decreased non-injection drug use, d = 0.18 (95% CI = 0.06, 0.30), (3) increased entry into drug treatment, d = 0.11 (95% CI = 0.02, 0.21), (4) increased condom use, d = 0.19 (95% CI = 0.11, 0.26), and (5) reduced the frequency of trading sex for drugs, d = 0.33 (95% CI = 0.10, 0.57). Although the ES for trading sex was statistically significant, it was based on only four studies and thus should be interpreted cautiously.

Table 1.

Summary of intervention efficacy outcomes (ESs), grouped as to whether the outcome measure was primarily drug-risk related, or sexual-risk related.

Homogeneity of ESs
Outcome (ESs) k of studies Weighted mean ES (and 95% CI)
 Q P**
Drug-Risk Related Outcomes
     Decreased injection drug use 30 0.078 (0.028, 0.127) 83.95 .000
     Decreased non-injection drug use 11 0.181 (0.063, 0.299) 12.12 .280
     Entered drug treatment 6 0.114 (0.020, 0.209) 8.45 .130
     Frequency of sharing needles/equipment 16 0.033 (−0.037, 0.104) 24.22 .062
     Frequency of bleaching needles/equipment 10 0.052 (−0.028, 0.132) 14.42 .108
Sexual-Risk Related Outcomes
     Increased condom use 16 0.186 (0.115, 0.258) 28.91 .017
     Frequency of trading sex for drugs 4 0.334 (0.097, 0.570) 2.26 .520
     Frequency of unprotected sex 15 0.042 (−0.041, 0.124) 18.87 .170
Open in a new tab

Note. Effect sizes (ESs) are positive for differences of lowered risk in the treatment group relative to the control group; all measures were from the first follow-up assessments available in each study. CI = 95% Confidence Interval.

**

Significance implies rejection of the hypothesis of homogeneity. Significant ES outcomes appear in bold.

Results were similar with fixed- or random-effects assumptions. Outcomes associated with treatment entry, non-injection drug use, and trading sex for drugs showed a high degree of homogeneity; thus, they were well characterized by their respective means (Table 1) and further (moderator) analyses were not warranted. ES arrays for condom use and IDU were not well-characterized by their mean ESs, Q(15) = 28.91 and Q(29) = 83.95, respectively, Ps < .05 (Figure 2), but further analyses explained these heterogeneous effects. Finally, no difference between the HIV risk reduction and conditions were observed for three outcomes: (1) frequency of sharing needles/equipment, (2) frequency of bleaching needles/equipment, and (3) frequency of unprotected sex (Table 1).

Figure 2.

Figure 2

Figure 2

Open in a new tab

Forest plots for injection-drug use (upper panel) and condom use (lower panel) effect sizes, in which delta points to the right of the zero point reflect risk-reduction and those to the left reflect risk increase, relative to the control group. Zero values indicate exactly no difference between the two groups. The symbols for statistically significant effects are filled whereas those that are not significant are open. The size of the delta symbol for each effect size reflects its weight in the analyses. The confidence interval for the mean effect size is displayed by the width of its diamond.

3.3 Intervention Features that Influenced IDU and Condom Use Outcomes

Reduced injection drug use

Considered individually, nine sample and intervention features were associated with reduced IDU: (1) more non-Caucasian participants, (2) more Latina/Latino participants; (3) more participants who reported recent cocaine use; (4) more participants who reported recently injecting “speedball” (a mixture of heroin and cocaine); (5) participants recruited from self-selection in the community or clinical contact; (6) using an individual intervention approach; (7) not providing bleach; (8) providing interpersonal skills training specific to safer needle use; and (9) providing equivalent intervention content on drug- and sexual-related HIV risk. Follow-up multiple-predictor models indicated that six of these associations became non-significant when controlling for other features. IDU ESs were modeled successfully on the basis of the remaining three features, included simultaneously: (1) more non-Caucasian participants, (2) dedicating equivalent intervention content to drug- and sexual-related HIV risk, and (3) providing interpersonal skills training specific to safer needle use (Table 2). This model explained 63% of the variation in ESs and had good model fit, QE(26) = 31.21, P = .22.

Table 2.

Final models of study dimensions that significantly related to interventions’ efficacy.

Patterns
 Model statistics
Study dimension Level Weighted mean (95% CI) k β alone† β‡ P
Reduced Injection Drug Use
Percentage of Caucasians intervention participants 0% in the intervention 0.213 (0.138, 0.286) 30 −.397 −0.299 <.01
30% in the intervention 0.133 (0.076, 0.189)
60% in the intervention 0.053 (−0.037, 0.144)
Equal emphasis on injection- and sexual-risk behaviors Equal emphasis 0.298 (0.203, 0.393) 30 .649 0.626 <.001
Unequal emphasis −0.027 (−0.087, 0.032)
Intervention provides interpersonal skills training on safer needle use Present 0.196 (0.119, 0.273) 30 .396 0.261 <.05
Absent 0.074 (−0.026, 0.151)
Increased Condom Use
Number of intervention facilitators 1 facilitator −0.005 (−0.197, 0.186) 10a .834 .834 <.001
2 facilitators 0.344 (0.172, 0.516)
Open in a new tab

Note. Analyses follow fixed-effects assumptions with each effect size (d) weighted by the inverse of its variance. The results for injection drug-risk outcomes result from one model with the 30 IDU studies, following fixed-effects assumptions in which all three study dimensions appeared. Each displayed mean is adjusted for the presence of the other study dimension, with that dimension held constant at zero. The full regression equation, with each study dimension contrast coded (+1, −1) was: ES = 0.2118 –0.2643 (% Caucasian) + 0.0608 (interpersonal skills) + 0.1626 (equal emphasis); Multiple R2 = .628. Analyses on increase of condom use are a series of models for each of the study dimensions, based on fixed-effects assumptions, where each effect size (d) was weighted by the inverse of its variance.

a

the number of studies reduced to 10 due to missing data.

CI = confidence interval.

k = number of studies.

P = probability.

β = standardized regression coefficient.

QE = Model specification statistic (distributed as a χ2 with k – 2 degrees of freedom); when statistically significant, variation remaining unexplained exceeds that expected by sampling error alone.

†

Without controlling for the other study dimension (i.e., univariate analysis).

‡

Controlling for the other study dimensions within the drug-use effect sizes.

Because decay of intervention effects is an important concern, we also examined whether injection drug use ESs related to the time that elapsed between the end of the intervention and the post-test and follow-up measurement points. Based on all thirty interventions, no decay was found in IDU outcomes following the intervention to the post-test point, β = .08, P = .46. Based on six of the interventions that gathered both post-test and follow-up data, no decay was found in the IDU outcome up to 52 weeks following the intervention, β = −.06, P = .77.

Increased condom use

Only one study feature was associated with variability in the magnitude of condom use ESs (Table 2): Interventions were more successful when they included two facilitators, as evidenced by a model that fit the data correctly, QE=4.90, P = .77. No evidence was found that facilitators’ qualifications (e.g., professionals vs. non-professionals) played a role in this outcome, QB(1) = 0.01, P = .91.

Finally, an exploratory analysis revealed that condom use ESs tended to decrease as the interval between the end of the intervention and the measurement point increased, β = −.51, P = .006. Seven of the interventions gathered both post-test and follow-up data pertaining to condom use; a similar trend of decay appeared, but did not reach conventional levels of significance, β = −.30, P = .19.

4. Discussion

This meta-analysis examined both sexual- and drug-related HIV risk reduction outcomes in studies that included persons who inject street drugs, evaluating only the strongest evidence, randomized controlled trials (RCTs). The sample of 37 studies included data from both participants who were already enrolled in drug treatment and persons “out-of-treatment” and living in the community. Thus, we could evaluate the extent to which interventions were successful at prompting out-of-treatment drug users to enroll in treatment.

The synthesis relied on 49 independent interventions from the 37 studies and supported the conclusion that, overall, the interventions facilitated condom use and promoted entry into drug treatment, and helped to reduce IDU, non-injection drug use, and sex trading. These effects were consistent with regard to non-injection drug use, drug treatment entry, and sex trading outcomes; that is, intervention participants tended to enter drug treatment and to reduce both their use of non-injected drugs and the practice of trading sex for drugs regardless of other intervention or personal characteristics. The enhanced drug treatment-entry outcome is a particularly encouraging outcome given the multiple benefits that accrue from successful drug treatment (Metzger et al., 1998; NIH, 1997; Prendergast et al., 2001).

The observed effects were not consistent for two key outcomes: IDU and condom use. The tendency for the interventions to enhance condom use and reduce IDU relative to controls was modified by specific features of the interventions as well as characteristics of the participants. Interventions tended to be more successful at reducing IDU for non-Caucasians. With respect to HIV, this result is encouraging because ethnic/racial minorities are disproportionately vulnerable to HIV (CDC, 2004).

Injection drug use was reduced to the extent that interventions were characterized by an equivalent focus on drug- and sexual behaviors as opposed to an exclusive focus on either type of risk behavior. Further, reduced injection drug use was more likely when interventions provided interpersonal skills training specific to safer needle use. Importantly, this intervention outcome appears to be durable; we found no evidence of decay up to a year following the interventions.

Frequency of condom use improved to the extent that the interventions were led by more than a single intervention facilitator. To date, no primary-level study with an IDU sample has compared the effects of interventions using one vs. two facilitators. Future studies could examine whether use of two facilitators increases intervention fidelity, affords more enriched opportunities for modeling (and shaping) interpersonal skills, improves the ability to manage complex group processes, or enhances facilitator-participant matching (e.g., on gender, race/ ethnicity, and sexual orientation).

Of some concern, the interventions’ effects on condom use tended to decay over time. Maintaining consistent condom use may require additional strategies (e.g., booster sessions) in order to address emergent challenges and maintain behavior change. Considered together, these findings suggest an HIV risk reduction program led by two facilitators that targets a broad range of drug- and sexual-risk behaviors and provides periodic booster sessions.

Overall, like previous reviews, we observed modest effect sizes on risk behavior outcomes. As previous reviews have noted (Semaan et al., 2002; Prendergast et al., 2001; Strathdee et al., 1998), these effects may reflect the use of control conditions that were similar, though briefer, to the intervention being evaluated. This design feature, which is ethically necessary, inadvertently restricts the range of ESs that might be observed from the literature. Moreover, behavior change is generally difficult as demonstrated in numerous health-behavior domains (e.g., smoking, weight loss, substance abuse; Pigott, 2001). Behavior change interventions, in general, and behavioral HIV risk reduction interventions in particular, tend to produce small increments of behavior change. Nonetheless, when extrapolated to high risk populations, the modest changes reported here and in prior reviews, have the potential to improve public health (e.g., Semaan et al., 2002; Glasgow et al., 1999).

Meta-analyses, like primary level studies, have limitations that need to be acknowledged so that findings can be interpreted appropriately. One limitation of the current analyses involved the decision to include only RCTs. This decision allowed us to examine the strongest evidence, but it also constrained the range of potential analyses that we could conduct. For example, because only a minority of the included studies reported outcomes assessed at a distant follow-up occasion, our analyses could not provide extensive detail on long-term outcomes. Similarly, the variables that we could analyze were limited by the outcomes reported, most of which were based on self-reported data. Thus, although other outcomes (e.g., sero-conversion rates) are of keen interest, such data were not available. Third, the size of the available literature constrained the number of predictors that could be included in the multivariate analyses we conducted. Despite these limitations, this meta-analysis provides a valuable summary of the extant research. Our results provide evidence that behavioral interventions do reduce behaviors that place persons who inject drugs at risk for contracting and transmitting HIV. Enhancing the magnitude of HIV risk reduction is facilitated by the identification of specific intervention features associated with positive outcomes. In addition, these results identify participants who are less likely to profit from behavioral interventions as currently constructed, and point to the need for continued refinement of these interventions so that they benefit all who are in need.

Acknowledgment

Funding to support the design and conduct of this research and preparation of this manuscript was provided as follows: NIH grants K23-DA017015 to Michael M. Copenhaver; R01-MH58563 to Blair T. Johnson; F31-MH069079 to Jennifer J. Harman; and K02-MH01582 to Michael P. Carey.

References

  1. Avants SK, Margolin A, Usubiaga MH, Doebrick C. Targeting HIV-Related outcomes with intravenous drug users maintained on methadone: A randomized clinical trial of a harm reduction group therapy. Journal of Substance Abuse Treatment. 2004;26:67–78. doi: 10.1016/S0740-5472(03)00159-4. [DOI] [PubMed] [Google Scholar]
  2. Celantano DD, Muñoz A, Cohn S, Vlahov D. Dynamics of behavioral risk factors for HIV/AIDS: A 6-year prospective study of injection drug users. Drug and Alcohol Dependence. 2001;61:315–322. doi: 10.1016/s0376-8716(00)00154-x. [DOI] [PubMed] [Google Scholar]
  3. Centers for Disease Control and Prevention. [Retrieved on August 18,2004];Special! data request. 2004 from http://www.statehealthfacts.kff.org.
  4. Fisher JD, Cornman DH, Osborn CY, Amico KR, Fisher WA, Friedland GA. Clinician-initiated HIV risk reduction intervention for HIV-positive persons: Formative research, acceptability, and fidelity of the OPTIONS project. Journal of Acquired Immune Deficiency Syndrome. 2004;37(2):S78–S87. doi: 10.1097/01.qai.0000140605.51640.5c. [DOI] [PubMed] [Google Scholar]
  5. Glasgow RE, Vogt TM, Boles SM. Evaluating the public health impact of health promotion interventions: The RE-AIM framework. American Journal of Public Health. 1999;89:1322–1327. doi: 10.2105/ajph.89.9.1322. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Johnson BT, Carey MP, Marsh KL, Levin KD, Scott-Sheldon LAJ. Interventions to reduce sexual risk for the human immunodeficiency virus in adolescents, 1985–2000: A research synthesis. Archives of Pediatric Adolescent Medicine. 2003;157:381–388. doi: 10.1001/archpedi.157.4.381. [DOI] [PubMed] [Google Scholar]
  7. Johnson BT, Eagly AH. Quantitiative synthesis of social psychological research. In: Reis HT, Judd CM, editors. Handbook of Research Methods in Social and Personality Psychology. London: Cambridge University Press; 2000. pp. 496–528. [Google Scholar]
  8. Kalichman SC, Rompa D, Cage M, DiFonzo K, Simpson D, Austin J. Effectiveness of an intervention to reduce HIV transmission risks in HIV-positive people. American Journal of Prevention Medicine. 2001;21:84–92. doi: 10.1016/s0749-3797(01)00324-5. [DOI] [PubMed] [Google Scholar]
  9. Lipsey MW, Wilson DB. The way in which intervention studies have “personality” and why it is important to meta-analysis. Evaluation and the Health Professions. 2001;24:236–254. doi: 10.1177/016327870102400302. [DOI] [PubMed] [Google Scholar]
  10. Margolin A, Avants SK, Warburton LA, Hawkins KA, Shi J. A randomized clinical trial of a manual-guided risk reduction intervention for HIV-positive injection drug users. Health Psychology. 2003;22(2):223–228. [PubMed] [Google Scholar]
  11. Metzger DS, Navaline H, Woody G. Drug abuse treatment as AIDS prevention. Public Health Reports. 1998;113:97–106. [PMC free article] [PubMed] [Google Scholar]
  12. National Institutes of Health. Interventions to prevent HIV-risk behaviors. Bethesda, MD: National Institutes of Health; National Institutes of Health Consensus Development Conference Statement. 1997 [PubMed]
  13. National Institute on Drug Abuse. Outreach/risk reduction strategies for changing HIV-related risk behaviors among injection drug users: The National AIDS Demonstration Research (NADR) Project. Washington, DC: National Institute on Drug Abuse; NIH Publication No. 94-3726. 1994
  14. Pigott TD. Missing predictors in models of effect size. Evaluation & the Health Professions. 2001;24:277–307. doi: 10.1177/01632780122034920. [DOI] [PubMed] [Google Scholar]
  15. Prendergast ML, Urada D, Podus D. Meta-analysis of HIV risk-reduction interventions within drug abuse treatment programs. Journal of Consulting and Clinical Psychology. 2001;69:389–405. doi: 10.1037//0022-006x.69.3.389. [DOI] [PubMed] [Google Scholar]
  16. Semaan S, Des Jarlais DC, Sogolow E, Johnson WD, Hedges LV, Ramirez G, Flores SA, Norman L, Sweat MD, Needle R. A meta-analysis of the effect of HIV prevention interventions on the sex behaviors of drug users in the United States. Journal of Acquired Immune Deficiency Syndrome. 2002;30:S73–S93. [PubMed] [Google Scholar]
  17. Schafer JL, Graham JW. Missing data: Our view of the state of the art. Psychological Methods. 2002;7:147–177. [PubMed] [Google Scholar]
  18. Sinharay S, Stern HS, Russell D. The use of multiple imputation for the analysis of missing data. Psychological Methods. 2001;6:317–329. [PubMed] [Google Scholar]
  19. Strathdee SA, Hogg RS, Martindale SL, Cornelisse PGA, Craib KJP, Montaner JSG. Determinants of sexual risk-taking among young HIV-negative gay men. Journal of Acquired Immune Deficiency Syndrome. 1998;19:61–66. doi: 10.1097/00042560-199809010-00010. [DOI] [PubMed] [Google Scholar]
  20. Vlahov D, Junge B, Brookmeyer R, Cohn S, Riley E, Armenian H, Beilenson P. Reductions in high-risk drug use behaviors among participants in the Baltimore needle exchange program. Journal of Acquired Immune Deficiency Syndrome. 1997;16:400–406. doi: 10.1097/00042560-199712150-00014. [DOI] [PubMed] [Google Scholar]

Sources of Data for Meta-analysis (in alphabetical order)

  1. Avants SK, Margolin A, Usubiaga MH, Doebrick C. Targeting HIV-Related outcomes with intravenous drug users maintained on methadone: A randomized clinical trial of a harm reduction group therapy. Journal of Substance Abuse Treatment. 2004;26:67–78. doi: 10.1016/S0740-5472(03)00159-4. [DOI] [PubMed] [Google Scholar]
  2. Avants SK, Margolin A, Sindelar JL, Rounsaville BJ, Schottenfeld R, Stine S, Cooney NL, Rosenheck RA, Li SH, Kosten TR. Day treatment versus enhanced standard methadone services for opioid-dependent patients: A comparison of clinical efficacy and cost. American Journal of Psychiatry. 1999;156:27–33. doi: 10.1176/ajp.156.1.27. [DOI] [PubMed] [Google Scholar]
  3. Baker A, Heather N, Wodak A, Dixon J, Holt P. Evaluation of a cognitive-behavioural intervention for HIV prevention among injecting drug users. AIDS. 1993;7:247–256. doi: 10.1097/00002030-199302000-00014. [DOI] [PubMed] [Google Scholar]
  4. Baker A, Boggs TG, Lewin TJ. Randomized controlled trial of brief cognitive-behavioral interventions among regular users of Amphetamine. Addiction. 2001;96:1279–1287. doi: 10.1046/j.1360-0443.2001.96912797.x. [DOI] [PubMed] [Google Scholar]
  5. Baxter S. AIDS education in the jail setting. Crime and Delinquency. 1991;37:48–63. [Google Scholar]
  6. Calsyn DA, Saxon AJ, Freeman G, Jr, Whittaker S. Ineffectiveness of AIDS education and HIV antibody testing in reducing high-risk behaviors among injection drug users. American Journal of Public Health. 1992;82:573–575. doi: 10.2105/ajph.82.4.573. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Celantano DD, Muñoz A, Cohn S, Vlahov D. Dynamics of behavioral risk factors for HIV/AIDS: A 6-year prospective study of injection drug users. Drug and Alcohol Dependence. 2001;61:315–322. doi: 10.1016/s0376-8716(00)00154-x. [DOI] [PubMed] [Google Scholar]
  8. Deren S, Davis WR, Beardsley M, Stephanie T, Clatts M. Outcomes of a risk-reduction intervention with high-risk populations: The Harlem Aids Project. AIDS Education and Prevention. 1995;7:379–390. [PubMed] [Google Scholar]
  9. Dushay RA, Singer M, Weeks MR, Rohera L, Gruber R. Lowering HIV risk among ethnic minority drug users: Comparing culturally targeted intervention to a standard intervention. American Journal of Drug and Alcohol Abuse. 2001;27:501–524. doi: 10.1081/ada-100104515. [DOI] [PubMed] [Google Scholar]
  10. Gibson DR, Lovelle-Drache J, Young M, Hudes ES, Sorensen JL. Effectiveness of brief counseling in Reducing HIV risk behavior in injecting drug users: Final results of randomized trials of counseling with and Without HIV Testing. AIDS and Behavior. 1999;3:3–12. [Google Scholar]
  11. Harris RM, Bausell RB, Scott DE, Hetherington SE, Kavanagh KH. An intervention for changing high-risk HIV behaviors of African American drug-dependent women. Research in Nursing and Health. 1998;21:239–250. doi: 10.1002/(sici)1098-240x(199806)21:3<239::aid-nur7>3.0.co;2-i. [DOI] [PubMed] [Google Scholar]
  12. Latkin CA, Sherman S, Knowlton A. HIV prevention among drug users: Outcome of a network-oriented peer outreach intervention. Health Psychology. 2003;22:332–339. doi: 10.1037/0278-6133.22.4.332. [DOI] [PubMed] [Google Scholar]
  13. Latkin CA, Knowlton A, Hoover D, Mandell W. Drug network characteristics as a predictor of Cessation of drug use among adult injection drug users: A prospective study. American Journal of Drug and Alcohol Abuse. 1999;25:463–473. doi: 10.1081/ada-100101873. [DOI] [PubMed] [Google Scholar]
  14. Latkin CA, Mandell W, Vlahov D, Oziemkowska M, Celentano DD. The long-term outcome of a personal network-oriented HIV prevention intervention for injection drug users: The SAFE study. American Journal of Community Psychology. 1996;24:341–364. doi: 10.1007/BF02512026. [DOI] [PubMed] [Google Scholar]
  15. Malow RM, Corrigan SA, Pena JM, Calkins AM, Bannister TM. Effectiveness of a psychoeducation approach to HIV risk behavior reduction. Psychology of Addictive Behaviors. 1992;6:120–125. [Google Scholar]
  16. Malow RM, West JA, Corrigan SA, Pena JM, Cunningham SC. Outcome of psychoeducation for HIV risk reduction. AIDS Education and Prevention. 1994;6:113–125. [PubMed] [Google Scholar]
  17. Mandell W, Vlahov D, Latkin CA, Carran D, Oziemkowska MJ, Reedt L. Changes in HIV risk behaviors among counseled injecting drug users. Journal of Drug Issues. 1994;24:555–567. [Google Scholar]
  18. Margolin A, Avants SK, Warburton LA, Hawkins KA, Shi J. A randomized clinical trial of a manual-guided risk reduction intervention for HIV-positive injection drug users. Health Psychology. 2003;22(2):223–228. [PubMed] [Google Scholar]
  19. National Institute on Drug Abuse. Outreach/risk reduction strategies for changing HIV-related risk behaviors among injection drug users: The National AIDS Demonstration Research (NADR) Project. Washington, DC: National Institute on Drug Abuse; NIH Publication No. 94-3726. 1994
  20. O'Neill K, Baker A, Cooke M, Collins E, Heather N, Wodak A. Evaluation of a cognitive-behavioural intervention for pregnant injecting drug users at risk of HIV infection. Addiction. 1996;91:1115–1125. doi: 10.1046/j.1360-0443.1996.91811154.x. [DOI] [PubMed] [Google Scholar]
  21. Robles RR, Feliciano C, Colón HM, Aponte L. Washington, DC: National Institute on Drug Abuse; Puerto Rico AIDS Prevention Project (PRAPP; Final report submitted to the National Institute on Drug Abuse) 1993
  22. Schilling RF, El-Bassel N, Schinke SP, Gordon K, Nichols S. Building skills of recovering women drug users to reduce heterosexual AIDS transmission. Public Health Reports. 1991;106:297–304. [PMC free article] [PubMed] [Google Scholar]
  23. Siegal HA, Falck RS, Carlson RG, Wang J. Reducing HIV needle risk behaviors among injection-drug users in the Midwest: An evaluation of the efficacy of standard and enhanced interventions. AIDS Education and Prevention. 1995;7:308–319. [PubMed] [Google Scholar]
  24. Simpson DD, Camacho LM, Vogtsberger KN, Williams ML, Stephens RC, Jones A, Watson DD. Reducing AIDS risks through community outreach interventions for drug injectors. Psychology of Addictive Behaviors. 1994;8:86–101. [Google Scholar]
  25. Sorenson JL, London J, Heitzmann C, Gibson DR, Morales ES, Dumontet R. Psychoeducational group approach: HIV risk reduction in drug users. AIDS Education and Prevention. 1994;6:95–112. [PubMed] [Google Scholar]
  26. St. Lawrence JS, Eldridge GD, Shelby MC, Little CE, Brasfield TL, O'Bannon RE., III HIV risk reduction for incarcerated women: A comparison of brief interventions based on two theoretical models. Journal of Consulting and Clinical Psychology. 1997;65:504–509. doi: 10.1037//0022-006x.65.3.504. [DOI] [PubMed] [Google Scholar]
  27. Stein MD, Anderson B, Charuvastra A, Maksad J, Friedmann PD. A brief intervention for hazardous drinkers in a needle exchange program. Journal of Substance Abuse Treatment. 2002;22:23–31. doi: 10.1016/s0740-5472(01)00207-0. [DOI] [PubMed] [Google Scholar]
  28. Sterk CE, Theall KP, Elifson KW, Kidder D. HIV risk reduction among African-American women who inject drugs: A randomized controlled trial. AIDS and Behavior. 2003;7:73–86. doi: 10.1023/a:1022565524508. [DOI] [PubMed] [Google Scholar]
  29. Yancovitz SR, Des Jarlais DC, Peyser NP, Drew E, Friedmann P, Trigg HL, Robinson JW. A randomized trial of an interim methadone maintenance clinic. American Journal of Public Health. 1991;81:1185–1191. doi: 10.2105/ajph.81.9.1185. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES