Authors‘ Response to Letter to the Editor Regarding “Factors Associated With Self-Efficacy for Condom Use and Sexual Negotiation Among South African Youth”

Abstract

Listen

Translate article

Reply: Alencastro et al raise concern about the inclusion of appropriate variables in the multivariate model, which includes the outcome of self-efficacy in this analysis. We would like to clarify that the multivariate model as displayed in Tables 3 and 4 of the paper included all candidate variables that had been previously identified in each domain of the conceptual model, in addition to sociodemographic characteristics. We agree with Alencastro et al that it is important to take into account the relation between variables included in a conceptual model within the context of a multivariate analysis. To address this in our study, collinearity was examined between independent variables and we tested for interactions that we hypothesized may exist in our model. No significant collinearity or interactions between covariates were identified. Finally, a sensitivity analysis was performed and showed that the significant factors associated with the outcome of self-efficacy were stable and did not change when other covariates were included or omitted from the final model. The details of these additional analyses were not included in the original paper for clarity of presentation and because they were not found to have a significant impact on the analysis. With regard to the self-efficacy for condom use and sexual negotiation index, we agree with Alencastro et al that the ideal measure would be one whose validity has been tested among South African youth. Such a measurement was not available at the time the National Youth Survey was conducted, however. The self-efficacy index that was used in our analysis was derived from an often used measure with proven reliability and high validity among US adolescents and youth in the developing world.1,2 Thus, we attempted to include a shorter version of the best measure available to us at the time of the survey. The intermediate coefficient α of our scale suggests that our 5-item index may not be as sensitive a measure of self-efficacy in South African youth. This means that we may be underestimating the magnitude of relations between key factors and self-efficacy-in effect, biasing our results toward the null. In this context, relations between factors in our conceptual model and self-efficacy may be stronger than we were able to detect in this analysis. Nevertheless, this measure was strongly associated with condom use at last sex3 and condom use consistency4 in these data. Future work on a valid and reliable measure of self-efficacy for condom use and sexual negotiation among South African youth is needed to characterize these relations better. Finally, we would like to clarify that the goal of this particular analysis was not to examine the association between HIV infection and other factors such as first sexual experience under the age of 15 years, for which Alencastro et al reference 3 studies. The focus of the paper was on identifying factors associated with self-efficacy for condom use and sexual negotiation. Any limitations in measurement of self-efficacy certainly would not apply to the lack of finding a significant relation between HIV infection and early age of sexual initiation in the overall results of the National Youth Survey, as is incorrectly suggested by Alencastro et al. This paper did find a number of factors to be significantly associated with self-efficacy for condom use and sexual negotiation, and this information may prove to be beneficial for individuals designing HIV prevention programs among youth in South Africa. Jennifer N. Sayles, MD* Audrey E. Pettifor, PhD† Mitchell D. Wong, MD, PhD* Thomas J. Coates, PhD* *University of California Los Angeles, CA †University of North Carolina Chapel Hill, NC