|
Meta-analysis of the effects of behavioral HIV prevention interventions on the sexual risk behavior of sexually experienced adolescents in controlled studies in the United States |
Mullen P D, Ramirez G, Strouse D, Hedges L V, Sogolow E |
|
|
Authors' objectives To determine the effectiveness of behavioural and social interventions on the sexual risk behaviour of sexually experienced adolescents in the United States, and to assess factors associated with variations in outcome.
Searching The HIV/AIDS Prevention Research Synthesis (PRS) database (see Other Publications of Related Interest) was searched for studies that were reported from 1988 to 1998. This database is based on searches of 8 electronic databases (AIDSLINE, CINAHL, DAI, ERIC, HealthSTAR, MEDLINE, PsycLIT and Sociofile), handsearches of 53 journals, conference proceedings, reference lists, and contact with researchers and agencies working in the field.
Study selection Study designs of evaluations included in the reviewStudies using control or comparison groups with random assignment or non-random methods such as matching, which appeared to be unbiased, were eligible for inclusion (see Other Publications of Related Interest).
Specific interventions included in the reviewSocial and behavioural interventions, conducted in the United States in both in-school and out-of-school settings, with the purpose of reducing sexual risk behaviours for human immunodeficiency virus (HIV) were eligible. A behavioural intervention was defined as an approach that emphasised individuals, even if the intervention was implemented in a group setting. A social intervention was defined as a direct or indirect attempt to change peer or community norms or structures through natural social networks. One of the included studies evaluated a social intervention, while the remainder evaluated a behavioural intervention. The content of the interventions included interpersonal skills, technical skills and personal skills, with and without practice, as well as the provision of condoms, responsibility enhancement and content designed to increase the perceived risk of HIV/AIDS.
Participants included in the reviewParticipants aged 13 to 19 years in the United States were eligible for inclusion. Three of the included studies were of adolescents 16 years and over, while the remaining studies were of under 16-year-olds. The proportion of participants with prior sexual activity ranged from 8 to 100% in the included studies.
Outcomes assessed in the reviewStudies measuring a change in sexual risk behaviour or a biologic indicator were eligible for inclusion. The outcomes assessed in the included studies were sex without condoms, the number of sexual partners, a mixed behavioural risk index, a composite risk behaviour outcome and the incidence of sexually transmitted diseases (STDs).
How were decisions on the relevance of primary studies made?The authors do not state how the papers were selected for the review, or how many of the reviewers performed the selection. However, trained librarians selected the studies for the PRS database from which these papers were drawn.
Assessment of study quality The authors do not state that they assessed quality, although methodological criteria were used to include or exclude studies from the PRS database on which this review was based.
Data extraction The data extraction was carried out in duplicate by two reviewers with any inconsistencies resolved by a third reviewer or an expert panel. When baseline data were available on the outcome measures and there were between-group differences, the authors adjusted for the baseline differences. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated for each outcome construct. When an outcome was reported as continuous data, an OR and CI were derived. Where the studies reported multiple measures of a single construct, the ORs were calculated for each measure and averaged. Where the studies used a nested design with analysis at the individual level, overestimation was adjusted for by using estimates of intra-class correlation. An overall risk behaviour index was calculated using a single outcome measure from each study and the following order of priority: reduction of sex without condoms, number of partners, and mixed risk behaviour index.
Methods of synthesis How were the studies combined?The studies were stratified according to outcome measure and pooled ORs with 95% CIs were calculated using fixed-effect and random-effects models. The studies were weighted by study precision as reflected by the CI. The results for the random-effects model were reported when the Q statistic indicated significantly heterogeneous studies. Publication bias was investigated using funnel plots, the Begg-Mazumdar adjusted rank correlation test and the Egger regression asymmetry test.
How were differences between studies investigated?In addition to the stratified analysis based on outcome measure, the Q statistic was used to investigate heterogeneity between the studies (p=0.1). Some subgroup analyses were also pre-specified in relation to the composite outcome measure (study design, intervention, setting and population characteristics).
Results of the review Twenty studies met the inclusion criteria and 16 were included in the pooled analysis (10 used randomised assignment and 6 used non-random methods).
Sex without condoms was less likely in the sexual risk reduction intervention group than the control/comparison group (13 studies): OR 0.66 (95% CI: 0.55, 0.79, p<0.001). Intervention also had a positive protective effect on the mixed behavioural risk index (2 studies)and the composite behavioural risk outcome (16 studies); the ORs were 0.66 (95% CI: 0.50, 0.88, p<0.01) and 0.65 (95% CI: 0.50, 0.85, p<0.01), respectively. Intervention was not associated with having fewer sexual partners in comparison with the control/comparison group (OR 0.89, 95% CI: 0.76, 1.05) or with reduced STD incidence (2 studies; OR 1.18, 95% CI: 0.48, 2.86).
Statistically-significant heterogeneity was found for sex without condoms and the composite behavioural risk outcome. Many of the subgroup analyses resulted in heterogeneous subgroups without significant differences between the subgroups.
The formal tests for publication bias were not statistically significant, though the authors indicated there was graphical evidence of possible publication bias.
Authors' conclusions There was a statistically-significant protective effect of sexual risk-reduction interventions, both in and out of the classroom, on sexual risk behaviour, primarily the risk of having sex without condoms.
CRD commentary The review question was clear in terms of the intervention, participants, study design and outcomes. However, some studies were excluded from the analysis although they met the eligibility criteria. A number of relevant electronic databases were searched and unpublished data were included. The data extraction was carried out in duplicate, which helps minimise errors and bias, though details of other review processes were not reported. Although the studies were required to meet specified methodological criteria for entry into the database from which the studies were selected, a quality assessment was not reported. There was also limited discussion of the results in the context of study quality. Individual study details were reported.
It is unlikely that it was appropriate to pool the data for the outcome of sex without condoms and the composite behavioural risk outcome, given the evidence of heterogeneity. Possible sources of heterogeneity were investigated, though this was limited by the data available to the authors: for example, they were unable to investigate the effects of gender on the outcome. The pooled estimates may be imprecise given the inappropriate pooling of the studies and the unexplained heterogeneity. In addition, the applicability to other settings is unclear since only U.S. studies were included.
Implications of the review for practice and research Practice: The authors did not state any implications for practice.
Research: The authors state that there should be more complete reporting of the data in journal articles so that appropriate subgroup analyses can be carried out as part of a meta-analysis.
Bibliographic details Mullen P D, Ramirez G, Strouse D, Hedges L V, Sogolow E. Meta-analysis of the effects of behavioral HIV prevention interventions on the sexual risk behavior of sexually experienced adolescents in controlled studies in the United States. Journal of Acquired Immune Deficiency Syndromes 2002; 30(Supplement 1): S94-S105 Other publications of related interest Sogolow E, Peersman G, Semaan S, Strouse D, Lyles CM. The HIV/AIDS Prevention Research (PRS) Project: scope, methods and initial findings. J Acquir Immune Defic Syndr 2002;30 Suppl 1:S15-S29.
Indexing Status Subject indexing assigned by NLM MeSH Acquired Immunodeficiency Syndrome /prevention & Adolescent; Behavior Therapy /methods /standards; Female; Humans; Male; Risk-Taking; Sexual Behavior; United States; control AccessionNumber 12002001770 Date bibliographic record published 30/11/2003 Date abstract record published 30/11/2003 Record Status 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. |
|
|
|