Developing a preference-based glaucoma utility index using a discrete choice experiment

Purpose. To estimate a utility-based glaucoma health outcome measure, known as the Glaucoma Utility Index. Methods. Based on focus group studies, involving people with glaucoma, existing profile measures relevant to glaucoma were modified and a six-dimensional profile instrument was developed. Dimensions were: central and near vision; lighting and glare; mobility; activities of daily living; eye discomfort and other effects. Each dimension was assigned four levels (no difficulty; some difficulty; quite a lot of difficulty; and severe difficulty). The discrete choice experiment (DCE) approach was employed to move from this profile instrument to a preference-based utility measure. Experimental design techniques were used to derive a sample of health states for which preferences were elicited using the DCE. Four hundred and seventy-three people with glaucoma received the choice questionnaire. Results. The regression analysis was based on 286 consistent responses to the DCE. The regression coefficients for three of the dimensions ("central and near vision," "mobility," and "activities of daily living") moved as expected. Moving from "no difficulty" to "severe difficulty" for central and near vision resulted in the most loss of utility, followed by activities of daily living and mobility. Systemic ("other effects") and local side effects were considered the least important. Utility weights were related to self-reported glaucoma state. Utility estimates moved in line with generic measures of health outcome. Conclusions. This study developed a preference-based utility measure (Glaucoma Utility Index) using the DCE approach. The index, estimated on the basis of 286 respondents, demonstrated both theoretical and convergent validity with other generic health outcome measures and measures of glaucoma severity. Further research investigating preferences by clinically defined glaucoma health status is indicated. Methodological research should focus on alternative methods of scaling for use within a generic Quality Adjusted Life Year framework.