Abstract
Introduction: In the U.S., HIV acquisition risk is highest among men who have sex with men (MSM). Syndemics, psychosocial problems that interact with each other and poor health behavior, are additively associated with increased condomless anal sex, HIV incidence, and HIV prevalence among MSM. However, information about how these syndemics interrelate with each other is under-explored. Methods: To examine the associations between nine syndemics in 194 MSM at high risk of HIV acquisition, we examined bivariate polychoric correlations, and then compared exploratory factor analyses (EFA) to a network analysis. Syndemics were assessed by self-report measures. Network analyses (presented graphically) consist of variables, called nodes, and the partial associations between variables, called edges. This regularized network employed the graphical LASSO algorithm and Extended Bayesian Information Criteria with a hyperparameter of γ=.5 to produce a graphically parsimonious yet sensitive network. Results: Correlation analyses revealed many associations between syndemics, but this was not always the case. EFA suggested that the 1-, 2-, and 3-factor solutions were possible based on various methods for determining potential factor solutions. However, the 1-factor solution appeared to be the most appropriate for these data, in which suicidal ideation (SI), injection drug use (IDU), depression, social anxiety, intimate partner violence, substance use, and sexual compulsivity had substantial loadings (≥.30). A pattern of interconnectedness emerged in this network analysis, which revealed that the relationships (absolute edge weights) between SI and IDU (b=.52, SD=.22, 95% CI[.17, .93]), IDU and substance use (b=.41, SD=.18, 95% CI[.18, .85]), social anxiety and SI (b=.30, SD=.36, 95% CI[.06, .77]), and depression and SI (b=.27, SD=.14, 95% CI[.05, .62]), were all significant. The most central nodes were SI, IDU, substance use, and depression. Discussion: This is the first study to compare network analysis to EFA in syndemics. It is also one of the earliest studies to conduct a network analysis of syndemics to describe the interrelatedness of conditions, beyond additive associations. Network analysis may be methodologically preferable to EFA in examining the interrelatedness of syndemics because it provides measures of centrality, which can potentially indicate the conditions that drive increased HIV acquisition risk, and lends itself to a visual presentation of the interrelatedness among these intertwined problems. This study therefore has the potential to provide better understanding of the ways in which syndemics, among high-risk MSM, interrelate and, subsequently, affect health behavior.