Purpose: This study examines the importance of clinical risk factors for predicting self-ratings of general health status, with and without controls for a broad range of self-reported indicators of physical and psychological well-being.
Methods: Ordered probit models, estimated on 928 respondents aged 54 years and older who participated in an ongoing national survey in Taiwan, are used to examine the correlates of self-rated health. The model is estimated in two stages, first testing the association between clinical risk factors and self-rated health and second, examining whether those clinical measures remain significant after controlling for a large number of other factors hypothesized to affect self-ratings.
Results: Most of the clinical variables are significantly associated with self-rated health, even in the presence of control variables. The largest effects pertain to the BMI, ratio of total to HDL cholesterol (among men) and presence of the epsilon4 allele of the APOE gene (among women).
Conclusions: The results suggest a variety of pathways linking clinical measures to the self-ratings. The findings also suggest that the clinical measures are less powerful predictors than self-reports about diverse aspects of well-being.