Examining the Multilevel Predictors of Condom Use in the US South

Background and Purpose

Individuals ages 13-24 are disproportionately impacted by HIV as they comprise a quarter of new HIV infections and over half of them are unaware that they have HIV. The social determinants of health continue to drive HIV inequities. While inconsistent condom use continues to be the main mode of HIV transmission, the drivers of it are intersectional in nature. Sex educational policies, community norms, and sexual behaviors are factors that have been found to be associated with condom use. There is a need to understand the multilevel predictors of condom use to then develop multi-level interventions to promote sexual health and change the conditions, rather than solely individual behaviors. This study used a multilevel framework to understand the predictors of condom use among ages 18-24 in Houston, Texas.

Methods

Data were collected using a cross-sectional survey conducted with young adults (18-24; N=563) attending a college or university between 2015-2018 in Houston as part of a SAMHSA-funded study to promote sexual and mental health awareness and engagement in HIV testing. The multilevel predictor variables were gender (individual), access to care (mezo), and health insurance status (macro). The dependent variable was condom use during the last sexual activity. We conducted a chi-square (model 1) to better understand the associations, however, only gender and health insurance was significant. Therefore, a binary logistic regression (model 2) analysis was performed to understand whether gender and/ or health insurance status predicted participants’ condom use at last sexual activity.

Results

The demographic breakdown was as follows: 28.9% African American, 21% Asian, 41.7% White, and 38.3% Hispanic). Fifty-four percent were male (46% female) and the majority of all participants attended high school in the US South (82.4%). Their mean age was 22.25, all were attending a local community college or university, and 67.6% had an annual household income below national average of $57,617.

We found that model 2, consisting of one predictor variable-health insurance status, was statistically significant (X2=7.675, df=2, N=563, p<.022). The odds (OR=1.779) of using a condom at last sexual activity increased if the participant had health coverage. The model accounts for 55.8% of the variance in condom use in college students. There were no significant interaction effects. The key finding is that some structural barriers, such as health coverage, are important correlates of condom use among college students.

Conclusion and Implications

The findings indicate that having health insurance predicted condom use at last sexual intercourse. Understanding the impact of multilevel factors, in this case health insurance, can have implications for developing multi-level interventions. Findings such as these can play an important role in health insurance advocacy and sexual health advocacy particularly in states that typically have not favored holistic, sex positive sexual health education (most states in the US South).