Graduation Year

2016

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Community and Family Health

Major Professor

William M. Sappenfield, M.D., M.P.H., C.P.H.

Co-Major Professor

Bruce Lubotski Levin, Dr.P.H., M.P.H.

Committee Member

Celia Maria Lescano, Ph.D.

Committee Member

Arlene Estela Calvo, Ph.D., M.P.H., C.H.E.S.

Committee Member

Wei Wang, Ph.D.

Keywords

alcohol use, risky sexual behaviors, syndemics, men who have sex with men (MSM), meta-analysis, structural equation modeling

Abstract

Since the early 1990s with the AIDS pandemic, there has been an increasing interest on the importance of risky sexual behaviors, especially among men who have sex with men (MSM). An important antecedent for these behaviors is alcohol use. Studies consistently show an increased frequency of both alcohol use and risky sexual behaviors in MSM populations. However, to date, there has not been a precise estimate of the effect size in these diverse populations and a consistent way to measure it. More so, the importance of context is often cited as a source of variability, but is rarely measured in these studies.

Contextual factors are different and specific for MSM, as they have been approached by two theories, both of which will be guiding this dissertation: Singer’s Syndemic Theory and Meyer’s Minority Stress Theory. Chapter 1, then presents a comprehensive review of both theories as they apply to alcohol use, risky sexual behaviors, and contextual factors driving them.

Chapter 2 answers the first research question about effect sizes through a systematic literature review. The effect sizes or measures of association of these contextual risk and protective factors were summarized using meta-analytic techniques. Using five electronic databases, we identified 26 studies in 26 years (1990–2015), all diverse in terms of sampling techniques, assessment of sexual orientation, operationalization of alcohol use and risky sexual behaviors, contextual factors included, and measurement of effect sizes. Despite this diversity, studies reporting an effect size were pooled and summarized using both descriptive and meta-analytic techniques, as appropriate. Meta-analyses were conducted using Cochrane’s guidelines for generic inverse variance outcomes with random effects. The pooled effects of alcohol use on condomless anal intercourse (CAI; OR 1.73 [95% CI 1.43, 2.10], I2 0%), heavy episodic drinking on CAI (OR 1.88 [95% CI 1.25, 2.81], I2 32%), and heavy episodic drinking on condomless oral sex (OR: 8.00; 95% CI 2.48, 25.81), as well as the effects of substance use, mental health status, violence and victimization, and self-reported HIV status as contextual factors in the pathway between alcohol use and risky sexual behaviors were calculated, reported, and discussed along with study limitations and implications for public health.

Chapter 3 answers the second research question regarding a standardized measurement model for effect sizes and the multiple mediation of contextual factors. We used the male subset of Wave IV of the AddHealth dataset to test our hypotheses through structural equation modeling approaches, including measurement analysis with invariance testing, path analysis for direct effects, and multiple mediation analysis through bootstrapping for indirect effects. The AUD scale was invariant between MSM and MSW, but the risky sexual behavior scale was not. For MSM, the standardized direct effect of AUD onto risky sexual behaviors was –1.25 and the standardized total indirect effect of the multiple mediation model was 1.58, 95% CI [1.42, 1.73]. Among the mediators, the strongest indirect effect for any measured or latent mediator was the mental health construct (2.09). We conclude that even though AUD has the same measurement structure for MSM and MSW, its effect on risky sexual behaviors does not operate the same way for these two populations, supporting both causal and contextual behavioral theories.

Conclusions are individually discussed, respectively, in Chapters 2 and 3. However, Chapter 4 puts both manuscript conclusions in context and further discusses future implications for public health research, practice, and policy.

Download

Share

COinS