Colin Green1,2, Elizabeth Goodwin1, Annie Hawton1,2. 1. Health Economics Group, Institute of Health Research, University of Exeter, Exeter, UK (CG, EG, AH). 2. South West Collaboration for Leadership in Applied Health Research and Care (CLAHRC), University of Exeter Medical School, University of Exeter, Exeter, UK (CG, AH).
Abstract
INTRODUCTION: Health state valuation is a key input in many economic evaluations that inform resource allocation across competing healthcare interventions. Empirical evidence has shown that, in preference elicitation surveys, respondents may value a health state differently if they are aware of the condition causing it ('labeling effects'). This study investigates the impact of including a multiple sclerosis (MS) label for valuation of MS health states. METHODS: Health state values for MS were elicited using two internet-based surveys in representative samples of the UK population ( n = 1702; n = 1788). In one survey respondents were not informed that health states were caused by MS. The second survey included a condition label for MS. Surveys were identical in all other ways. Health states were described using a MS-specific eight-dimensional classification system (MSIS-8D), and the time trade-off valuation technique was used. Differences between values for labeled and unlabeled states were assessed using descriptive statistics and multivariate regression methods. RESULTS: Adding a MS condition label had a statistically significant effect on mean health state values, resulting in lower values for labeled MS states v. unlabeled states. The data suggest that the MS label had a more significant effect on values for less severe states, and no significant effect on values for the most severe states. The inclusion of the MS label had a differential impact across the dimensions of the MSIS-8D. Across the MSIS-8D, predicted values ranged from 0.079 to 0.883 for unlabeled states, and 0.066 to 0.861 for labeled states. CONCLUSION: Differences reported in health state values, using labeled and unlabeled states, demonstrate that condition labels affect the results of valuation studies, and can have important implications in decision-analytic modelling and in economic evaluations.
INTRODUCTION: Health state valuation is a key input in many economic evaluations that inform resource allocation across competing healthcare interventions. Empirical evidence has shown that, in preference elicitation surveys, respondents may value a health state differently if they are aware of the condition causing it ('labeling effects'). This study investigates the impact of including a multiple sclerosis (MS) label for valuation of MS health states. METHODS: Health state values for MS were elicited using two internet-based surveys in representative samples of the UK population ( n = 1702; n = 1788). In one survey respondents were not informed that health states were caused by MS. The second survey included a condition label for MS. Surveys were identical in all other ways. Health states were described using a MS-specific eight-dimensional classification system (MSIS-8D), and the time trade-off valuation technique was used. Differences between values for labeled and unlabeled states were assessed using descriptive statistics and multivariate regression methods. RESULTS: Adding a MS condition label had a statistically significant effect on mean health state values, resulting in lower values for labeled MS states v. unlabeled states. The data suggest that the MS label had a more significant effect on values for less severe states, and no significant effect on values for the most severe states. The inclusion of the MS label had a differential impact across the dimensions of the MSIS-8D. Across the MSIS-8D, predicted values ranged from 0.079 to 0.883 for unlabeled states, and 0.066 to 0.861 for labeled states. CONCLUSION: Differences reported in health state values, using labeled and unlabeled states, demonstrate that condition labels affect the results of valuation studies, and can have important implications in decision-analytic modelling and in economic evaluations.