Mapping chronic liver disease questionnaire scores onto SF-6D utility values in patients with primary sclerosing cholangitis.
Kalaitzakis E., Benito de Valle M., Rahman M., Lindkvist B., Björnsson E., Chapman R., Kontodimopoulos N.
PURPOSE: The chronic liver disease questionnaire (CLDQ) is a frequently used liver-specific quality of life instrument, but it does not provide information on preference-adjusted health status, which is essential for cost-utility analysis. We aimed to develop a mapping function deriving utilities from the CLDQ in primary sclerosing cholangitis (PSC). METHODS: Short form-6D (SF-6D) utilities were calculated from SF-36 data collected in a recent prospective study in which unselected patients with PSC also completed the CLDQ. Ordinary least squares (OLS), generalized linear, median, and kernel regression analyses were employed to devise a mapping function predicting utilities. This was validated in three random subsamples of the cohort and in a separate sample of PSC patients following liver transplantation. Adjusted R (2) and root-mean-square error (RMSE) as well as Pearson's r coefficients and mean absolute errors between predicted and observed values were used to determine model performance. RESULTS: Decompensated liver disease and fatigue, systemic symptoms, and emotional distress, assessed with the CLDQ, were related to worse SF-6D utilities. The final OLS prediction model explained 66.3 % of the variance in the derivation sample. Predicted and observed utilities were strongly correlated (r = 0.807, p < 0.001), but the mean absolute error (0.0604) and adjusted RMSE (10.6 %) were of intermediate size. Similar model characteristics were observed after employment of generalized linear and median regression models and at validation. CONCLUSIONS: A model has been constructed, showing good validity predicting SF-6D utilities from CLDQ scores at the group level in PSC. Further testing is required to externally validate the model.