The predictive validity of health-related quality of life measures: mortality in a longitudinal population-based study

Objective This study examined the association between health-related quality of life (HRQL) and mortality risk, and compared the predictive ability of Health Utilities Index Mark 3 (HUI3) with self-rated health (SRH). Methods Data were from the 1994/95 Canadian National Population Health Survey, consisting of 12,375 women and men aged 18 and older. Cox proportional hazards regression models were performed to estimate mortality risk over eight years. Results Mortality risks for people reporting good, fair, and poor health at baseline were, respectively, 1.44 (95% confidence interval [CI] 1.04, 2.00), 1.97 (1.35, 2.88), and 3.21 (2.08, 4.95) times greater than those who reported excellent health. In a model excluding SRH, the effect of HUI3 on mortality was strong and significant (HR = 0.47; 95%, 0.33, 0.67) when adjusted for possible confounders. When HUI3 and SRH were considered simultaneously, the effect of the HUI3 on mortality was somewhat attenuated, but still significant (HR = 0.61, 0.42, 0.89) after adjusting for potential confounders. Conclusions Although SRH is a modestly stronger predictor of mortality than HUI3, HUI3 adds to the mortality prediction ability of SRH.