Linking health-related quality-of-life indicators to large national data sets

Abstract

Objective: This study investigated the feasibility and usefulness of linking algorithms for well known quality-of-life (QOL) indicators to large nationally representative databases. Design and Setting: The National Medical Expenditure Survey (NMES) was utilised. We developed an algorithm to match the EuroQOL health indicator and drew on a previous match of the Health Utilities Index (HUI) in a companion paper. This process allowed the sensitivity and detail of health-related quality-of-life (HR-QOL) indicators to be combined with the benefits of large, nationally representative data sets. Patients and participants: A total of 19 525 individuals aged 18 years and older (constituting a nationally representative sample of the noninstitutionalised civilian population of the US contained within the 1987 NMES database) were investigated. Interventions: Sensitivity analyses using several related specifications of each indicator were performed. We analysed the correlations of these alternatives for both the HUI and EuroQOL measures. Correlations between the HUI and EuroQOL measures were also examined. We investigated the construct validity by examining the performance of the HUI and EuroQOLs in empirical situations in which we had knowledge about the relationships (e.g. health decreases with age). Main outcome measures and results: The benefits of HR-QOL measures can be achieved relatively cheaply and efficiently via linking rather than developing a large scale QOL survey. Although the NMES data allowed a good match with the EuroQOL and the HUI, the matches were not perfect. By examining the within-domain correlations and the between-domain correlations, we found that the alternate specifications within-domain were very similar and that the 2 HRQOL indicators were comparable in many (but not all) aspects. Conclusions: The results of this study suggested good construct validity. Thus, linked HR-QOL measures of the types derived in this study may be useful in characterising the health of large populations, and in investigating the causes and consequences of health.