Using specification curve analysis to explore prospective associations between dimensions of positive psychological well-being and cardiometabolic disease.

Abstract

Objective: Literature suggests that higher positive psychological well-being (PPWB) is associated with reduced risk of cardiometabolic disease. However, PPWB is multidimensional. Most models do not distinguish between dimensions of PPWB in relation to cardiometabolic disease. This study demonstrated how specification curve analysis can be used to explore if the association between PPWB and incident cardiometabolic disease is influenced by the dimension of PPWB, cardiometabolic disease, and covariates under investigation.

Method: Data came from the Midlife in the United States study (n = 2,895). Nine dimensions of PPWB and covariates were measured at baseline (2004-2005) and five cardiometabolic diseases were self-reported at follow-up (2013-2014). One hundred eighty model specifications, each containing one dimension of PPWB, one cardiometabolic disease, and one set of covariates, were generated. Standardized odds ratios from corresponding logistic regression models, in which PPWB predicted incident cardiometabolic disease, were then plotted on a specification curve.

Results: The median standardized odds ratio across models was 0.94. PPWB was inversely associated with incident cardiometabolic disease in 18% of models. Significant associations depended upon the dimension of PPWB, the outcome, and covariates.

Conclusion: Researcher decisions about the dimension of PPWB, cardiometabolic disease outcome, and covariates under investigation appear to be consequential. Specification curve analysis can be used to develop an evidence base that starts to distinguish between dimensions of PPWB in relation to cardiometabolic disease. Thinking carefully about if and how specific dimensions of PPWB are associated with particular health outcomes is an avenue for theory refinement. (PsycInfo Database Record (c) 2025 APA, all rights reserved).