Dimensional Sensitivity of the 15D Multiattribute Utility Instrument in Diabetic Retinopathy

The purpose of this study was to further investigate the previously suggested high sensitivity of the 15D instrument to diabetic retinopathy (DR) by examining: i) its dimensional sensitivity in patients differing in DR severity, and ii) its dimensional relationship with a well-known vision-targeted health status instrument. Eighty five diabetes patients with non-proliferative or proliferative DR completed the 15D and the National Eye Institute Visual Function Questionnaire 25-item (NEI-VFQ-25) instruments. Dimensional response distributions were examined with chi-square, and mean VFQ25 and 15D dimension score differences with ANCOVA. Dimensional relationships between the instruments were assessed with correlations and ordinary least squares (OLS) regressions. Patients with proliferative DR scored significantly lower (P?<?0.05 or better) on 9/15 and 10/12 of the 15D and VFQ25 dimensions respectively. Correlations were mostly strong (>0.50) and significant (P?<?0.001). Most OLS models with 15D dimensions as predictors of VFQ25 subscales showed high explanatory power (adjusted R ²?>?0.85 in 10/13 models), with ocular pain being the only exception (R ²?=?0.44). The 15D’s seeing dimension was a significant predictor of 10 VFQ25 scales, whereas speech was not significant in any model. 15D utilities discriminated between the DR subsamples, supporting its appropriateness in this group. Its dimensions (seeing in particular), explained large portions of variance in NEI-VFQ25 scales, implying a strong association with this vision-targeted instrument. These findings might help to explain why 15D performs well in DR and support its use for QALY calculations. However, the need remains for further testing in larger and more diverse samples.