This presentation is part of: Poster Session - Utility Theory; Health Economics; Patient & Physician Preferences; Simulation; Technology Assessment
A SUMMARY MEASURE OF POPULATION HEALTH
Susan T Stewart, PhD1, Rebecca M Woodward, PhD1, and David M Cutler, PhD2. (1) Harvard Interfaculty Program for Health Systems Improvement, Harvard University and the National Bureau of Economic Research, Cambridge, MA, (2) Harvard University and the National Bureau of Economic Research, Interfaculty Program for Health Systems Improvement and Department of Economics, Cambridge, MA
Purpose: We propose a method for quantifying non-fatal health that addresses concerns raised about existing measures by: 1) incorporating a range of symptoms and impairments from mild to severe, 2) basing disutility weights on current health ratings rather than counterfactual scenarios, and 3) accounting for multiplicative relationships between health problems. We then quantify the effects of diseases such as diabetes and heart diseases on health-related quality of life, based on how the diseases affect specific symptoms and problems. Methods: Alternative disutility weights for health problems from the Quality of Well-Being Scale (QWB) are derived by examining their effects on global self-rated health status (SRHS) and time-tradeoff (TTO) ratings of current health. Data are from 1420 respondents age 45 to 89 in the Beaver Dam Health Outcomes Study. Ordered probit and OLS regression are used, with interaction terms testing for non-additive relationships. Disease weights are based on our weights derived for symptoms/problems, and on probit equations that examine the effect of each disease on each symptom/problem. Results: Health problems with the greatest impact on quality of life were: limited ability to work, physical activity limitations, pain, and taking medications or following a prescribed diet. Other problems, such as sensory impairments, speech problems, sexual problems, and problems with weight or appearance, had little independent effect. Disutilities across eight domains were similar to comparable QWB scores, and mean overall scores were similar, with worse scores on our measure among those with problems in some domains. The TTO distribution was truncated (with 59% accepting no trade-offs), and SRHS was more closely related to objective measures of health and yielded higher disutility weights. Diseases with the worst disutility included heart and respiratory conditions, mood disorders, pain, sleep problems, ulcers, and type I diabetes. Conclusions: Our method yields weights and scores similar to those from the QWB scale, with some important differences. Deriving disutility weights for diseases via their impact on specific symptoms/impairments allows examination of the mechanisms through which changes in mean quality of life with a specific disease occur over time. Replication of these methods in a larger, more representative sample is recommended as the basis for a summary measure of population health.
