OBJECTIVES: The Visual Functioning-14 (VF-14) and EQ-5D-5L are two important questionnaires for estimating health-related quality of life. The EQ-5D-5L is widely recommended and used by regulatory bodies such National Institute for Health and Care Excellence (NICE) but not always collected. We have service evaluation data from patients who completed both questionnaires, and aim to develop a model mapping the VF-14 onto EQ-5D-5L scores for estimation of utility values for ophthalmology interventions in the UK.

METHODS: We will evaluate different modelling approaches—linear ordinary least squares (OLS), Tobit, censored least absolute deviation (CLAD), and major axis regression. Performance will be evaluated based on i) predictive ability using graphical plots, and root mean squared error (RMSE), mean absolute error (MAE), and Pearson and Spearman correlation coefficients; ii) goodness of fit using pseudo R², AIC, and BIC. We will divide our dataset into two random samples—one for developing the model and the other for validation. We will explore sensitivity by mapping the VF-14 onto both the index and individual dimension responses.

RESULTS: We will identify the best performing model based on predictive ability and goodness of fit. We will also report the proportion of EQ-5D values that fall within the model’s 95% confidence interval. We will report properties of the regression datasets as the EQ-5D index has been shown to exhibit different characteristics such as ceiling effects and bimodal distribution.

CONCLUSIONS: We are not aware of UK studies that have mapped VF-14 scores on to EQ-5D utility scores. The closest study mapped ophthalmology-specific 25-item National Eye Institute Visual Functioning Questionnaire (NEI VFQ-25) to EQ-5D utility scores. We anticipate that the model can help regulators make better resource allocation decisions even in the absence of generic preference scores. Unlike previous studies that have predominantly used <500 data points, we have >3200 observations.