Examination of Age Variations in the Predictive Validity of Self-Rated Health

Anna Zajacova, University of Wyoming
Hyeyoung Woo, Portland State University

The high validity of self-rated health (SRH) is a major strength of this widely used population health measure. Recent studies, however, noted systematic differences across gender, race, and socioeconomic status in the predictive validity of SRH. We add the examination of age as a potential moderator of SRH predictive validity to the literature while carefully considering confounding by covariates including other time dimensions: cohort, period, and duration of mortality followup. Using data from the NHIS-LMF 1986-2006, we estimate Cox proportional hazard models of mortality for adults 45-84 years old (N=574,008). The findings indicate that for older birth cohorts, the power of SRH to predict mortality attenuates significantly with age. For more recent cohorts, there is no age attenuation. The findings imply that more recent birth cohorts suffer little attenuation in the predictive validity of SRH with age, making this measure particularly useful for capturing age-related health changes in the population.