Estimating utility weights for the vision related quality of life index

Purpose The VisQoL instrument was constructed as a vision-specific addition to the AQoL-6D multi-attribute utility instrument. The composite instrument, the AQoL-7D, was assigned utility scores that are the basis for now estimating VisQoL utilities when it is used as a stand-alone instrument. This study aimed to construct mapping functions that allow utility scores to be assigned to the Vision Related Quality of Life (VisQoL) instrument, a stand-alone vision-specific quality-of-life measure. Methods A sample of 164 patients completed the AQoL-7D, which includes the VisQoL. Mapping algorithms between VisQoL and AQoL-7D were then derived using two econometric methods, ordinary least squares estimator and generalized linear model (GLM). Two model specifications were considered with either six VisQoL raw item values or VisQoL overall dimension value as the key independent variables. The predictive performance of each method on each model specification was assessed using the mean absolute error (MAE) and intraclass correlation coefficient (ICC). Both internal and external validation tests (using a second, independent sample of 164 patients) were performed. Results The mapping algorithms derived from the GLM had superior properties to the ordinary least squares-based algorithms in both internal and external validation tests. The ICC values ranged from 0.851 to 0.913, and the MAE ranged from 0.043 to 0.052 for two model specifications, based on two econometric methods. However, predicted utilities tend to over-predict/under-predict the lowest/highest observed utility. Conclusions Mapping algorithms predicting AQoL-7D utility based on six VisQoL items or VisQoL dimension value have been developed. The algorithm can be used to estimate quality adjusted life years. This allows the VisQoL to be used in cost utility analyses.