Descriptive regression tree analysis of intersecting predictors of adult self-rated health: Does gender matter? A cross-sectional study of Canadian adults

BACKGROUND: While self-rated health (SRH) is a well-validated indicator, its alignment with objective health is inconsistent, particularly among women and older adults. This may reflect group-based differences in characteristics considered when rating health. Using a combination of SRH and satisfaction with health (SH) could capture lived realities for all, thus enabling a more accurate search for predictors of subjective health. With the combined measure of SRH and SH as the outcome we explore a range of characteristics that predict high SRH/SH compared with predictors of a low rating for either SRH or SH. METHODS: Data were from the Canadian General Social Survey 2016 which includes participants 15 years of age and older. We performed classification and regression tree (CRT) analyses to identify the best combination of socioeconomic, behavioural, and mental health predictors of good SRH and health satisfaction. RESULTS: Almost 85% of the population rated their health as good; however, 19% of those had low SH. Conversely, about 20% of those reporting poor SRH were, none-the-less, satisfied. CRT identified healthy eating, absence of a psychological disability, no work disability from long-term illness, and high resilience as the main predictors of good SRH/SH. Living with a spouse or children, higher social class and healthy behaviours also aligned with high scores in both self-perceived health measures. Sex was not a predictor. CONCLUSIONS: Combining SRH and SH eliminated sex as a predictor of subjective health, and identified characteristics, particularly resilience, that align with high health and well-being and that are malleable.