Abstract
OBJECTIVES: This study is aimed to evaluate the prognostic ability of EQ-5D questionnaire for hospitalization and mortality in an elderly Italian population. Although many studies show that Health-Related Quality of Life is a significant predictor of survival and morbidity, even after controlling for socio-demographic covariates and physical health status, to date there are few studies that prove the efficacy of EQ-5D questionnaire to predict mortality and hospitalization on general population.
METHODS: A retrospective cohort study on 5256 subjects aged 65 years or more, recruited in 2003 in Italy. EQ-5D Index and VAS, socio-demographics variables and some medical covariates (collected with a postal questionnaire) were used to predict hospitalization and mortality. Mortality and hospitalization data during the 12-months period after the completion of the questionnaire were obtained by record linkage with administrative mortality and discharge datasets. Hospitalization was defined as any hospital admission for natural causes (ICD-9 CM diagnosis codes: 0–799) that included at least one overnight stay in the hospital. Kaplan-Meier’s method in univariate analysis and Cox proportional
hazards model with robust variance estimator to calculate the relative risks of mortality and first hospitalization were used.
RESULTS: The EQ-5D questionnaire is a significant predictor of mortality. The curves estimated with Kaplan-Meier’s method show significant differences in survival time at one year (Log rank test, p-value < 005) among individuals reporting different health status. The Cox proportional hazard model performed controlling for sex, age, BMI, physical activity and functional activity shows increasing risk for people perceiving worse health status. Hospitalization results are not jet available as the record linkage procedure between survey and discharge data is in progress.
CONCLUSIONS: Evaluation of EQ-5D instrument as a mortality and morbidity predictor will make it a proxy variable of prognostic mortality score systems, including various laboratory parameters, seldom available in surveys not specifically designed for epidemiological study.