Validation of a novel diabetic retinopathy utility index using discrete choice experiments. Issue 2 (16th May 2019)

• Record Type: Journal Article
• Authors: Fenwick, Eva K; Bansback, Nick; Gan, Alfred Tau Liang; Ratcliffe, Julie; Burgess, Leonie; Wong, Tien Yin; Lamoureux, Ecosse Luc
• Is Part Of: British journal of ophthalmology. Volume 104:Issue 2(2020)
• Page Start: 188
• Page End: 193
• Publication Date: 2019-05-16
• Subjects: diabetic retinopathy -- quality of life -- utility -- discrete choice experiment -- vision impairment -- health economics; Ophthalmology -- Periodicals; 617.7
• DOI: 10.1136/bjophthalmol-2019-313899
• Languages: English
• ISSNs: 0007-1161

Abstract

Abstract : Background/aims: To validate a preference-based Diabetic Retinopathy Utility Index (DRU-I) using discrete choice experiment (DCE) methods and assess disutilities associated with vision-threatening DR (VTDR: severe non-proliferative DR, proliferative DR and clinically significant macular oedema) and associated vision impairment. Methods: The DRU-I comprises five quality-of-life dimensions, including Visual symptoms, Activity limitation/mobility, Lighting and glare, Socio-emotional well-being and Inconvenience, each rated as no, some, or a lot of difficulty. The DRU-I was developed using a DCE comprising six blocks of nine choice sets which, alongside the EuroQoL-5D (EQ-5D-3L) and Vision and Quality of Life (VisQoL) utility instruments, were interviewer-administered to participants. To ensure the DRU-I was sensitive to severe disease, we oversampled patients with VTDR. Data were analysed using conditional logit regression. Results: Of the 220 participants (mean±SD age 60.1±11.3 years; 70.9% men), 57 (29.1%) and 139 (70.9%) had non-VTDR and VTDR, respectively, while 157 (71.4%), 20 (9.4%) and 37 (17.3%) had no, mild or moderate/severe vision impairment, respectively. Regression coefficients for all dimensions were ordered as expected, with worsening levels in each dimension being less preferred (theoretical validity). DRU-I utilities decreased as DR severity (non-VTDR=0.87; VTDR=0.80; p=0.021) and better eye vision impairment (none=0.84; mild=0.78; moderate/severe=0.72; p=0.012) increased. DRU-I utilities had low (r=0.39) and moderate (r=0.58) correlation with EQ-5D and VisQoL utilities, respectively (convergent validity). Discussion: The DRU-I can estimate utilities associated with vision-threatening DR and associated vision impairment. It has the potential to assess the cost-effectiveness of DR interventions from a patient perspective and inform policies on resource allocation relating to DR.