Assessing the impact of social Determinants of Health on multiple binary outcomes using multivariate statistical methods

Abstract

Social Determinants of Health (SDOH) are the non-medical factors for example; housing, employment, education, food security, etc., influencing health conditions/outcomes including chronic respiratory diseases and cardiovascular diseases. It is important to understand the social factors to reduce SDOH gaps across different social groups and improve public health outcomes. The objectives examined the impact of SDOH, modeling the interaction between SDOH and health outcomes, and assessing covariate distribution. This research employed multivariate statistical methods; Multivariate Logistic Regression (MLR) and Canonical Correlation Analysis (CCA). Simulations of different covariate shapes of the Wishart distribution was run to understand how they affect model fit. The data utilized is a multidimensional survey conducted in Costa Rica’s provinces of San Jose, Alajuela, Cartago, and Heredia between February 2019 and December 2022. The findings showed that significant associations between SDOH and health outcomes. Middle-income and unemployed individuals showed higher odds of developing diabetes, mental health disorders, obesity, and CRD. Smoking and physical exercise are strongly associated with increased odds of obesity, CVD, and diabetes. CCA revealed a strong canonical correlation between SDOH and health outcomes. Smoking, informal housing, income, and employment influenced health outcomes, CVD, CKD, and CRD. Covariate assessment showed robustness of logistic regression models under different covariance structure of Wishart-distribution, with higher degrees of freedom improving model performance. It highlighted insight into application of multivariate statistical methods in public health policymakers. Health-care practitioners should focus on lifestyle factors and socioeconomic factors to help reduce health disparities. There is also need for targeted health interventions and policy addressing to promote healthier lifestyle. Further research should explore longitudinal data to understand SDOH impacts on population health.