Multivariable Regression Model of the EuroQol 5-Dimension Questionnaire in Patients Seeking Outpatient Low Vision Rehabilitation

Purpose: To understand the source of between-person variance in baseline health utilities estimated from EuroQol 5-dimension questionnaire (EQ-5D) responses of a representative sample of the US low vision outpatient population prior to rehabilitation. Methods: A prospective, observational study of 779 new low vision patients at 28 clinic centers in the US. The EQ-5D, Activity Inventory (AI), Telephone Interview for Cognitive Status (TICS), Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36) physical functioning component, and Geriatric Depression Scale (GDS) were administered by telephone interview prior to rehabilitation. EQ-5D responses were transformed into health utilities, which served as the dependent variable in all analyses. Data were then analyzed to determine how much overall visual ability, functional domains of visual ability, and comorbidities (e.g. physical functioning, depression, cognition) independently contribute to the EQ-5D-based health utility index. Results: Multivariable regression analyses showed that the GDS and SF-36 physical account for nearly 40% of the variance observed in health utilities estimated from EQ-5D responses of low vision patients. Age was also a significant predictor of health utilities, but accounted for very little variance. None of the other variables were significant predictors. Conclusions: Health utilities of low vision patients estimated from the EQ-5D primarily are associated with comorbid factors that are not likely to be responsive to low vision rehabilitation, thereby rendering the EQ-5D an unsuitable outcome measure for this population. However, because the EQ-5D is responsive to comorbid states, it could be a useful tool for evaluating the impact of comorbidities on low vision patient quality of life.