Mechanisms of behavior change in a brief dual-target motivational intervention: Reduction in alcohol use mediates intervention effects on risky sex

Abstract

To examine the mechanisms underlying the efficacy of a dual-target motivational intervention (MI) to reduce heavy drinking and risky sex. A priori hypotheses were that: increases in alcohol-related readiness to change (RTC) and self-efficacy would mediate the effect of MI on alcohol use; increases in sex-related RTC and self-efficacy would mediate the effect of MI on risky sex; and reductions in alcohol use would mediate reductions in risky sex. Patients in Emergency Departments who screened positive for heavy drinking and risky sex were randomly assigned to receive MI or brief advice. RTC and self-efficacy were assessed at baseline and immediately postintervention. Alcohol use and sexual behavior was assessed at baseline, 3-, 6-, and 9-month follow up. Single- and serial-mediation models were tested. Patients who received MI had higher postintervention RTC and self-efficacy, but neither mechanism mediated the effects of MI on behavioral outcomes. Reduction in heavy drinking mediated the effect of MI on frequency of sex under the influence (SUI). Further, the effect of MI on condomless sex was mediated by an indirect path in which reductions in heavy drinking at 3 months predicted less SUI at 6 months, which in turn predicted reduction in condomless sex at 9-months. Although some effect of dual-target MI on risky sex is independent of drinking, treatment-related reduction in heavy drinking does account for a significant portion of reduction in risky sex, providing support for the utility of this intervention in patient populations where heavy drinking and risky sex co-occur. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Figure 1.

Mediator model illustrating the direct and indirect associations of treatment (BA=0, MI=1), number of heavy drinking days, and sex under the influence (SUI), controlling for gender, site, and baseline levels of the mediator and outcome variables. The c path represents the total effect, and c’ represents the direct effect after accounting for the effect of the indirect path. Path estimates are reported as unstandardized regression coefficients. Only individuals with complete data (n = 293) were included in this analysis. *p < .05; **p < .01; ***p < .001.

Figure 2.

Serial-mediation model illustrating the direct and indirect associations of treatment (BA=0, MI=1), number of heavy drinking days, sex under the influence (SUI), and condomless sex, controlling for gender, site, and baseline levels of the mediator and outcome variables. The c path represents the total effect, and c’ represents the direct effect after accounting for the effect of the indirect path. Path estimates are reported as unstandardized regression coefficients. Only individuals with complete data (n = 293) were included in this analysis. *p < .05; **p < .01; ***p < .001.

Figure 3.

Time-lagged serial-mediation model illustrating the direct and indirect associations of treatment (BA=0, MI=1), number of heavy drinking days, sex under the influence (SUI), and condomless sex, controlling for gender, site, and baseline levels of the mediator and outcome variables. The c path represents the total effect, and c’ represents the direct effect after accounting for the effect of the indirect path. Path estimates are reported as unstandardized regression coefficients. Only individuals with complete data (n = 283) were included in this analysis. *p < .05; **p < .01; ***p < .001.