Baseline risk behavior interviews impact RCT outcomes of an HIV risk reduction intervention: Experimental analysis of assessment reactivity

Background: The potential for self-report assessments of behavior to change behavior and to interact with experimental interventions is a concern in randomized controlled trials (RCT) of brief behavioral interventions. Methods: We tested the effects of a brief HIV risk reduction intervention among 1,728 young STD clinic patients in a black township of Cape Town, South Africa, and examined whether baseline sexual behavior assessments improved intervention outcomes. We tested for main intervention effects (Phaphama, a 60-minute HIV risk reduction intervention found effective in previous studies) and for intervention x assessment interaction effects using the Solomon four-group design: 2 (intervention: Phaphama vs. no intervention) x 2 (baseline assessment: no baseline vs. sexual behavior interview) at the three-month follow-up. Outcomes included condom use with main partner and with secondary partners, 100% condom use with secondary partners, and number of sex partners. Results: As hypothesized, the intervention was effective and resulted in significantly greater condom use and fewer partners compared to control groups (condom use ps <.01 main, p <.05 secondary, 100% condom use = p < .01, N partners p <.05). However, contrary to hypotheses, intervention alone resulted in lower risk behavior than intervention with baseline (Phaphama vs. control: condom use p< .001 and .01; 100% condom use p<.001; Phaphama + pretest vs. control = >.05, except for N partners, p<.05). Conclusions: Findings from RCTs of brief behavioral interventions that include baseline behavior interviews may underestimate intervention effects achieved in non-research settings where baseline interviews are not conducted.