Predicting self-assessed health status: a multivariate approach

Abstract

Two-stage multivariate analysis was used to examine factors affecting personal perception of health status. In the first stage, sociodemographic variables were used as independent variables in Automatic Interaction Detector (AID) analysis in order to partition the study sample (11,153 civilian noninstitutionalized adults aged 58-63) into subgroups. In the second stage, binary multiple regression analysis was performed on each AID subgroup and on the total sample. Predictors used were indicators of psychological, socioeconomic, and sociomedical well-being. Finally the applicability of these indicators in classifying persons in one of the two categories of perceived health status was examined by discriminant function analysis. Sociomedical health indicators were better explanatory variables of self-assessed health status than socioeconomic or psychological indicators of well-being.