This review found that individual-, group- and community-level interventions are effective in reducing the odds of unprotected anal intercourse in men who have sex with men. There was no evidence for an effect on biological or clinical outcomes. Overall, the findings and recommendations for further research reflect the evidence presented and appear appropriate.

To assess the effectiveness of and economic efficiency of behavioural interventions to reduce the risk of acquiring or transmitting human immunodeficiency virus (HIV) infection in men who have sex with men (MSM).

EMBASE, MEDLINE, PsycINFO and Sociofile were searched from 1985 to August 2004. Additional searches were carried out using the Internet, references from reviewed articles and referrals from specialists. Thirty-five key journals were handsearched for articles published between August 2004 and June 2005. Published conference abstracts, the Cochrane Controlled Trials Register and the National Institute of Health’s CRISP database were also searched, and principal investigators were contacted for information on current and ongoing research. The searches were limited to English language publications.

Study designs of evaluations included in the review

Studies that compared groups exposed to an intervention with non-exposed, less exposed or delayed exposure groups were eligible for the review. The majority of included studies were randomised controlled trials (RCTs).

Specific interventions included in the review

Studies of interventions intended to modify the sexual behaviour of MSM, aimed at individuals, small groups or communities, were eligible for the review. Individual interventions in the included studies included counselling and encouraging individuals to keep a sex diary. The comparator groups were normally given standard HIV testing and counselling, or were on a waiting list. Group-level interventions varied in content and group size, and ranged from 1 to 32 sessions. The comparator groups were on a waiting list or received a different intervention. Community interventions involved the advocacy of behaviour change by opinion leaders, compared with waiting list or educational materials. Follow-up ranged from immediately after the intervention to 18 months.

Participants included in the review

Studies of adult MSM (20 years or older) were eligible for the review. The majority of participants in most studies were white, but some studies had exclusively African-American or Hispanic participants. Where reported, the proportion of participants who were HIV-positive at baseline ranged from 0 to 61%.

Outcomes assessed in the review

Studies were required to assess outcomes related to sexual behaviour (unprotected sex, condom use and number of sexual partners) or associated biological outcomes (incidence of sexually transmitted diseases and HIV infection). The majority of included studies assessed behavioural outcomes.

How were decisions on the relevance of primary studies made?

At least two reviewers assessed studies that met preliminary inclusion criteria; any disagreements were resolved by consensus among the review team.

At least two independent reviewers assessed validity; any disagreements were resolved by consensus among the review team. Studies were assessed on the basis of design suitability (greatest, moderate or least) and quality of execution. The latter was classified as good, fair or limited based on six categories of threats to validity (study population and intervention descriptions, sampling, exposure and outcome measurement, data analysis, interpretation of results, and other biases). Studies with greatest or moderate design suitability and good or fair quality of execution were included in the review.

At least two reviewers extracted the data; any disagreements were resolved by consensus among the review team. Odds ratios (ORs) were calculated for outcomes reported as categorical data. For outcomes reported as means and standard deviations, the standardised mean difference effect size was calculated and converted to an OR.

How were the studies combined?

Within each level of intervention (individual, group or community), studies were combined by meta-analysis using a random-effects model weighted by the inverse of lnOR. Publication bias was assessed by inspection of a funnel plot of standard error estimates versus effect size estimates.

How were differences between studies investigated?

Statistical heterogeneity was assessed using the Q statistic. Various sensitivity analyses were performed (details reported in the paper), and stratified analyses were conducted to assess whether intervention effectiveness was influenced by characteristics of the samples, intervention features, study design and length of follow-up.

Nineteen studies were included: 18 RCTs with over 8,000 participants and one non-randomised trial with 129 participants.

The funnel plot showed no evidence of publication bias.

Individual-level interventions.

Intervention significantly reduced the odds of unprotected anal intercourse (UAI) compared with controls (OR 0.57, 95% confidence interval, CI: 0.37, 0.87; based on 6 intervention arms in 4 studies). Statistical heterogeneity was significant and was not explained by sensitivity analyses. One study reported on incident HIV infection: the effect of the intervention was not statistically significant (OR 0.62, 95% CI: 0.36, 1.06). Other outcomes were reported.

Group-level interventions.

Intervention significantly reduced the odds of UAI compared with controls (OR 0.73, 95% CI: 0.61, 0.88; based on 15 intervention arms in 13 studies). Interventions with multiple sessions, delivered by MSM and having a skills-building component reduced the odds of UAI, whilst those lacking these features did not. Statistical heterogeneity was not significant. Intervention significantly increased the odds of condom use with anal intercourse (OR 1.81, 95% CI: 1.12, 2.93; based on 5 intervention arms in 4 studies). One study reported on incident sexually transmitted diseases: the effect of the intervention was not statistically significant (OR 1.84, 95% CI: 0.85, 3.99). Other outcomes were reported.

Community-level interventions.

Intervention significantly reduced the odds of UAI compared with controls (OR 0.65, 95% CI: 0.48, 0.89; based on 3 intervention arms in 3 studies). Other behavioural outcomes were reported.

Two economic evaluations of group-level interventions reported that the interventions were cost-saving in terms of cost per quality-adjusted life-year (QALY). Two evaluations of one community-level intervention found that the cost per HIV infection averted was $76,764 in one evaluation and $14,073 in the other; both evaluations found a negative cost per QALY saved. An evaluation of a different community-level intervention estimated costs per HIV infection averted that ranged from $49,580 in 1 year to $7,373 in 20 years, depending on assumptions about the HIV prevalence rate.

Individual-, group- and community-level interventions are effective in reducing the odds of unprotected anal intercourse, while group-level interventions increase the odds of condom use. The findings are applicable across a range of settings and populations.

This review addressed a clear question and had clear but broad inclusion criteria. The authors searched a wide range of relevant sources. The search was limited to English language material, which means that some relevant studies could have been missed. Publication bias was assessed using a standard method. Validity was assessed and only studies meeting certain quality criteria were included in the review. The study selection, validity assessment and data extraction were performed by two independent reviewers, thereby minimising the risk of reviewer error or bias in the review process.

Full details of the included studies were presented in the paper and appendix. The studies were combined by meta-analysis. Statistical heterogeneity was assessed and various sensitivity analyses were performed. Unexplained heterogeneity was detected in the main analysis for individual-level interventions, which suggests that the decision to pool these studies might not have been appropriate. Overall, however, the authors’ conclusions and research recommendations reflect the evidence presented and appear appropriate.

Practice: The authors did not state any implications for practice.

Research: The authors stated that further research is needed to identify effective interventions for under-studied subgroups of MSM; to assess effects on biological outcomes; to improve assessment of behavioural outcomes; and to identify the core components of effective interventions.

Oak Ridge Institute for Scientific Education.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.