Health Economics Response Shift and Adaptation in Chronically III Patients

The purpose was to explore the impact of response shift on quality-of-life measurement and consequently on cost-effectiveness analysis. This was done by using data from an earlier study. Adaptation to illness provides a probable explanation for the occur-rence of response shift. In the case of proxy respondents, it is likely that the adaptation effect is absent. Therefore, two hypotheses were tested: 1) retrospective quality-of-life measurement does not differ from ratings given by proxies; and 2) proxy ratings differ from prospectively assessed pretreatment quality-of-life scores. Quality of life was as-sessed using a visual analog scale (VAS), time tradeoff (TTO), and the standard gam-ble (SG). Analyses revealed that neither hypothesis could be rejected, indicating that prospectively assessed pretreatment quality of life is enhanced by adaptation to the imperfect health state during the pretreatment period. Consequently, the cost-effec-tiveness ratio is different when using proxy measures or retrospective assessments of pretreatment quality of life, compared with using assessments of the quality of life in currently ill patients. Key words: quality-of-life measurement; adaptation; response shift; proxy measurement; cost-effectiveness. (Med Decis Making 2000;20:186-193) In recent years there has been increasing interest in assessing the health consequences of interventions in terms of quality of life. 1 One reason is the in-creased prevalence of chronic diseases in an aging society.3 Especially for chronic conditions, impact on quality of life might be the primary measure of the effectiveness of therapy. For these irreversible diseases, the goal of medical interventions is not to cure, but rather to improve the well-being of pa-tients.4 When it is accepted that quality of life is a major outcome parameter in the health care sector, for both clinical decisions and policy making, there is a need to measure health-related quality of life and to assess preferences accurately.5-8 This holds particularly true since quality of life appears to vary depending on the way it is measured. 4,8-12 For ex-ample, it was found that retrospectively assessed pretreatment quality of life significantly differed from prospectively assessed quality of life in chron-ically ill patients.