Purpose: Most patient-reported outcome measures used in ophthalmology show floor effects in a very low vision population, which limits their use in vision restoration trials. The Impact of Vision Impairment-Very Low Vision scale (IVI-VLV) was developed to specifically target a very low vision population, but its test-retest reliability has not been investigated yet.
Methods: The German version of the IVI-VLV was administered twice to patients with stable disease of a low vision clinic. Test and retest person measures of the IVI-VLV subscales were obtained from Rasch analysis. Test-retest reliability was investigated by intraclass correlation coefficients and Bland-Altman plots.
Results: We included 134 patients (72 women, 62 men) at a mean age of 62 ± 15 years. The intraclass correlation coefficients were 0.920 (95% confidence interval, 0.888-0.944) for the activities of daily living and mobility subscale of the IVI-VLV and 0.929 (95% confidence interval, 0.899-0.949) for the emotional well-being subscale. Bland-Altman plots did not indicate any systematic bias. In linear regression analysis, test-retest differences were not significantly associated with visual acuity or administration interval.
Conclusions: Both subscales of the IVI-VLV showed excellent repeatability independent of visual acuity and length of repeat interval. Further validation steps including an assessment of the patient-reported outcome measure's responsiveness are required for use in vision restoration trials.
Translational relevance: The results support repeated use of the IVI-VLV as a patient-reported end point in future studies in very low and ultralow vision populations.