Eighteen articles (19 studies) were included in the review. Average quality score was 3.9 (range 1 to 8). Seventeen studies were reported to be prospective cohorts. Four studies used some form of participant randomisation. Eight studies had a control/comparison group. All reported pre- and post-intervention data. Twelve studies reported at least 80% follow-up.
Differential effect of interventions by serostatus (nine studies): Pooled analysis showed a statistically significant differential effect between HIV positive and HIV negative individuals in terms of condom use (four studies, n=4,322 participants). In HIV positive individuals, condom use was significantly increased (OR 3.61, 95% CI 2.61 to 4.99, I2=0%). The result for HIV negative individuals was not statistically significant (I2=85%). There was no statistically significant difference between individual and couples counselling. A very limited intervention effect was reported for contraceptive use (two studies).
Interventions that targeted HIV positive individuals (10 studies): Pooled analysis showed a statistically significant effect on condom use (OR 7.84, 95% CI 2.82 to 21.79, I2=96%; seven studies, n=1,801). Differences between HIV positive heterosexual individuals and HIV serodiscordant couples were significantly different; the latter population group showed a substantially increased effect (OR 67.38, 95% CI 36.17 to 125.52, I2=0%, three studies, n=312). Two studies showed a modest decrease in the number of participants who reported sexual activity with non-primary partners. Two studies showed increases in HIV status disclosure.