Do Differences in Self-Rated Health Really Reflect Difference In Health?

Presenter: Yukiko Asada, Dalhousie University

Abstract

Rationale: Self-rated health (SRH), typically measured as excellent, very good, good, fair, and poor health, is a popular measure of general health status. While previous studies have shown that SRH is correlated with morbidity and mortality, others have suggested otherwise. For example, Newfoundlanders regularly report the best SRH among Canadians but exhibit the worst rates of mortality and certain morbidities. To understand such divergence, researchers have begun to aknowledge that SRH may capture latent health, the true state of health if it could be measured without bias, and reporting behaviour, the systematic and measureable variation in SRH away from latent health.

Objectives: The overall goal of this study was to understand what SRH measures from the perspectives of latent health and reporting behaviour. To meet this goal, we developed an operational definition of reporting behaviour and assessed the degree to which reporting behaviour varied systematically with age, gender, income, education, unemployment, language, health behaviour, and province.

Methodology: We used the 2005 Canadian Community Health Survey, a cross-sectional survey of non-institutionalized Canadians. Included were all those who had answered questions on SRH and the Health Utilities Index (HUI) Mark 3 and lived in one of the 10 provinces (sample size: 24736). We operationally defined reporting behaviour as the systematic residual variation that could not be explained by latent health. Considering the HUI as a measure of latent health, we regressed HUI on the five levels of SRH. To determine whether reporting behaviour systematically varied by different factors, we employed regression models with reporting behaviour as the dependent variable and age, gender, income, education, unemployment, language, health behaviour, and province as independent variables. We first ran bivariate models with each of these independent variables, and then, additively included age in each of these models. Positive coefficients in our models suggested optimism (perceiving own health better than latent health), while negative coefficients indicated pessimism (perceiving own health worse than latent health).

Results: We observed the strongest magnitude of reporting behaviour by the oldest age group (80 +). This group showed a statistically significant, strong optimism about their health (0.12 increase in the HUI, p<0.001). Men and women had the opposite perception of their health with men being optimistic and women being pessimistic. Those who ate plenty of fruits and vegetables and exercised regularly were pessimistic about their health, as were those in the highest income brackets and with the most education. Daily smokers were optimistic. Across Canada, Nova Scotians and Newfoundlanders were optimistic while the Quebecois were pessimistic. This pessimisism was also reflected among Francophones across Canada while Anglophones were optimisitic. Examining people with each of these characteristics by age group, we found that the older they were the more optimistic they were.

Conclusions: SRH should be interpreted with caution. Several factors are expected to have a predictable and statistically significant effect on reporting behaviour which will influence SRH results.