Intention to use condom, cusp modeling, and evaluation of an HIV prevention intervention trial.

Adolescents are at particularly high risk to acquire HIV infection; increasing the likelihood of condom use is an effective measure to reduce the risk of such infections. Challenges in assessing actual condom use behavior among early adolescents render the precursor measure, intention to use condoms, an appealing alternative. While analyzing data from a randomized controlled trial to evaluate a theory-based intervention program to promote condom use among early adolescents, we observed a modest effect with regard to condom use intention when the linear analytical approach was used. If intention, as a measure of the readiness to perform a behavior, also contains a nonlinear discrete component, it would be more appropriately modeled using a non-linear approach. In this study, data from a randomized controlled trial (N=1360) were analyzed using the cusp catastrophe method with HIV knowledge and condom skills as the asymmetry variables and condom use self-efficacy as the bifurcation variables. Findings from concurrent and longitudinal modeling analyses indicated a much better fit of the cusp model (R 2 = 0.85 and 4+ times smaller AIC and BIC) than the linear (R 2 <=0.10 and 4+ times larger AIC and BIC) or the logistic model (R 2 <0.15, also 4+ times larger AIC and BIC). Receipt of the intervention as an asymmetry variable was significantly predicted condom use intention but did not as a bifurcation variable. In conclusion, adolescent intentions to use a condom contain both a continuous process and a discrete process and can better be modeled with cusp methods. A much greater program effect is likely from the same prevention intervention if additional measures are taken to foster sudden changes in condom intention.