A research team based at the Wilmer Eye Institute, Johns Hopkins University School of Medicine, have reported a new NEI VFQ-25C aimed at calibrating a number of items in the National Eye Institute Visual Function Questionnaire-25. The researchers aimed to improve the well-known NEI VFQ-25 due to some instrument-specific and scoring methodology flaws, including multidimensionality in the scoring – e.g., inclusion of conceptually distinct visual function [VF] and socioemotional [SE] constructs. This revised patient reported outcome measure questionnaire expects to enable the estimation of measures on an invariant scale and comparisons between patients and across studies.
The current study collected datasets of baseline NEI VFQ-25 responses from nine studies – seven (7) retina randomized trials (n = 2770) and two (2) low vision studies (n = 572) – ranging in age from 19 to 103 years, and with a mean (SD) age of 69.3 (11) years and a mean logMAR visual acuity of 0.30 (Snellen 20/40). The psychometric model, based on Rasch analysis, reported the paper on the item measure estimates with a high precision (standard error range, 0.026–0.085 logit), and person measure estimates had lower precision (standard error range, 0.108–0.499 logit). The research team commented that the “applied modern psychometric techniques to calibrate items in a modified version of the NEI VFQ-25 called the NEI VFQ-25C and resolved the problem of multidimensionality by providing domain-specific versions: NEI VFQ-VF and NEI VFQ-SE”. As demonstrated in their paper (Transl Vis Sci Technol. 2022;11(5):10), the logit scale summarised of the NEI VFQ-25C shows the “more positive item measures (items to the right) require greater visual ability, whereas more negative item measures require less visual ability”, on the left:
In conclusion, the researchers commented that: “[t]he calibrations using Rasch analysis provided for the NEI VFQ-25 enable estimation of visual function (NEI VFQ-VF), socioemotional function (NEI VFQ-SE), and function combining both VF and SE domains (NEI VFQ-25C) for individual patients on an invariant scale. This recommended approach rectifies known problems with composite scores, subscale scores, missing data, and multidimensionality with the NEI VFQ-25”.