Reducing respondent burden: validation of the Brief Impact of Vision Impairment questionnaire.

Abstract

To develop a psychometrically sound and valid Brief Impact of Vision Impairment (IVI) questionnaire. Cross-sectional data from four prospective studies (2001-2008) were pooled and randomly divided into development/validation sets (n=416) each. Items with suboptimal psychometric properties were iteratively removed in the development set to form the Brief IVI. Psychometric properties of the Brief IVI were independently tested in the validation sample. Correlation between person measures from the original and Brief IVI was assessed [Pearson r and intraclass correlation coefficient (ICC)]. Criterion validity was determined by testing the Brief IVI's ability to discriminate levels of vision impairment (analysis of variance, ANOVA). Responsiveness was tested by comparing the ICC of the original and Brief IVI data obtained pre-/post-intervention. The 15-item Brief IVI, and its 9-item Visual Functioning and 6-item Emotional Well-being subscales had ordered thresholds, good precision and targeting, unidimensionality, and minimal item misfit (replicated in the validation sample). Brief and original IVI person measures were highly correlated (r=0.97 and ICC=0.98, p<0.001), indicating the Brief IVI provides statistically similar measurement of vision-related quality of life (VRQoL). Brief IVI mean logit scores declined as vision impairment worsened (p=0.001) demonstrating criterion validity. ICC of the original versus Brief IVI pre-/post-intervention was excellent (0.98), establishing that the Brief IVI was as responsive to changes in VRQoL as the original. The Brief 15-item IVI can obtain valid and responsive measurement of VRQoL with half the items in the original and has potential to reduce respondent burden in QoL studies.