Author: Lawrence Singerman (United States)
Co-authors: Cristian Gugiu, Elizabeth Tschosik, Helen Doll, Faye Drawnel, Burton Singerman, Brittany Gentile
Purpose
The NEI VFQ-25 is widely used to assess vision-related functioning in clinical trials across a variety of retinal diseases, but the length of the questionnaire may make it cumbersome for use in clinical practice settings. The purpose of this study was to derive and psychometrically validate a short form of the NEI VFQ-25 that could be more easily integrated with routine clinical practice and amenable to digital assessment in patients with neovascular age-related macular degeneration (nAMD).
Setting/Venue
Study included secondary analyses of ANCHOR, MARINA, and PIER clinical trials data of intravitreal ranibizumab, a recombinant, humanized antibody antigen-binding fragment (Fab), designed for intraocular use in nAMD, macular edema following retinal vein occlusion (RVO), myopic choroidal neovascularization (mCNV), diabetic retinopathy (DR) and diabetic macular edema (DME) patients. ANCHOR was a phase 3 randomized, multicenter, double-masked, sham injection-controlled study of ranibizumab in nAMD (N = 716). MARINA was a phase 3 randomized, multicenter, double-masked, active treatment-controlled study of ranibizumab in nAMD (N = 432). PIER was a phase 3b multicenter, randomized, double-masked, sham injection-controlled study in nAMD (N = 184).
Methods
Statistical analyses were conducted on pooled data (total N = 1318) from the three clinical trials. A combination of Rasch modeling and principal component analysis (PCA) was employed to reduce the NEI VFQ-25 to a minimum number of items representing the composite score. Psychometric analyses of the short form were then conducted to ensure internal consistency, test-retest reliability, construct (known-groups and convergent/divergent) validity, and responsiveness to change. Distribution- and anchor-based methods were used to estimate thresholds of clinically meaningful change using best-corrected visual acuity (BCVA).
Results
Quantitative item reduction resulted in a seven-item short form (VFQ-SF) containing items from the near activities, distance activities, general vision, peripheral vision, and role difficulties scales. Psychometric analyses on the VFQ-SF demonstrated the questionnaire had excellent internal consistency (α = 0.88) and test-retest reliability (ICCs ranged 0.79 to 0.91), and correlated highly with the NEI VFQ-25 (NEI VFQ-25 composite r = 0.95; domain rs ranged 0.49 to 0.90). The VFQ-SF demonstrated moderate correlations with BCVA [r = 0.46 (worst seeing eye) and r = 0.68 (best seeing eye)] and with the exception of the HUI3 vision score (r = 0.72), smaller correlations with health-related quality of life questionnaires [HUI3 r range: -0.23 to 0.37; SF-36 r range: 0.10 to 0.30). Correlations between the full NEI VFQ-25 and each of these constructs were highly similar. The VFQ-SF differentiated between BCVA severity groups and demonstrated responsiveness to change from baseline to 12- and 24-month follow-ups. Triangulating distribution- and anchor-based estimates determined a threshold for clinically meaningful improvement of 7 points.
Conlusions
This study applied quantitative item reduction methodology to reduce the NEI VFQ-25 from 25 to 7 items and validated the resulting questionnaire, the VFQ-SF. The VFQ-SF demonstrated robust psychometric properties that may support its use in place of the NEI VFQ-25 to generate a composite vision-related functioning score in nAMD patients when time is limited. This brief questionnaire could be suitable for use in clinical practice settings or remote monitoring of vision-related functioning, via electronic platforms such as web- or device-based apps.
Financial Disclosure
Alcon Laboratories- Grant Support Alkeus- Grant Support Appellis-Grant Support Chengdu- Grant Support Genentech/Roche- Grant Support NEI- Grant Support Novartis- Grant Support
Comments
-
