RESULTS FROM A NEW VISUAL ANALOGUE SCALE PROTOCOL FOR EQ-5D VALUATIONS
Author(s)
Bailey H1, La Foucade A1, Kind P21The University of the West Indies, St. Augustine, Trinidad and Tobago, 2University of York, York, United Kingdom
OBJECTIVES: The MVH protocol uses a valuation subset of 43 EQ-5D states. Smaller valuation subsets would allow smaller samples if all respondents can value one set of states. This study tests the performance of valuation subsets based on orthogonal experiment designs. VAS values obtained for EQ-5D states are affected by the identity of adjacent states in the elicitation. This study also tests a VAS protocol developed to minimize such effects. METHODS: 230 respondents were each randomly assigned to one of two orthogonal EQ-5D valuation subsets. EQ-5D states were printed on 9cm x 4cm cards with rhomboid edges. Respondents first ranked a shuffled deck of 23 cards. A 1 meter VAS was then placed next to the ranked cards. Respondents transferred the cards on to the VAS with the rhomboid edges pointing to the respective values. They could see all of the cards in place at the same time. Ties were permitted and respondents were free to change the order of states in moving the cards from the ranked order to the VAS. RESULTS: OLS regression of VAS Scores produced internally valid models for both valuation subsets with r2 of 0.54 and 0.56. Within sample comparisons of model versus observed VAS values produced residuals of MAD < 3% with individual values crossing 6 VAS points in only 2 cases. Rank correlation coefficients between model and observed states =0.996. Comparisons across the two subsets/subgroups produced residuals of MAD=4 VAS points and rank correlation coefficients of 0.98 and 0.96. CONCLUSIONS: This study suggests that orthogonal experiment designs may be suitable as the basis for the construction of EQ-5D valuation subsets, which can allow for smaller respondent samples. The statistical properties of the models arising in this study suggest that some concerns with VAS methods for EQ-5D elicitations may be reduced using this protocol.
Conference/Value in Health Info
2011-05, ISPOR 2011, Baltimore, MD, USA
Value in Health, Vol. 14, No. 3 (May 2011)
Code
PRM30
Topic
Patient-Centered Research
Topic Subcategory
Patient-reported Outcomes & Quality of Life Outcomes
Disease
Multiple Diseases