Utility-weighted modified Rankin Scale: Still too crude to be a truly patient-centric primary outcome measure?

Background The utility-weighted modified Rankin Scale (UW-mRS) is an outcome measure recently proposed to improve statistical efficiency and interpretability of the mRS. Statistical properties of the UW-mRS have been well investigated, but construct validity has yet to be established. Aims To investigate the construct validity of the UW-mRS as a primary outcome measure by assessing variability in utility values within and between mRS categories, over time post-stroke, and by different derivation methods. Methods UW-mRS was derived using assessment of quality of life (AQoL-4D) and mRS scores at 3 and 12 months (n = 2030) from a large randomized controlled trial, A Very Early Rehabilitation Trial (AVERT). Receiver operator characteristic (ROC) analysis of AQoL-4D was conducted to differentiate between sequential mRS categories. Intraclass correlation was used to explore variability in utility values over time post-stroke, UW-mRS values, and derivation methods from multiple studies. Results UW-mRS values for mRS categories 0-6 at three months were 0.80, 0.78, 0.63, 0.37, 0.11, 0.03, and 0. Based on AQoL-4D utility values, areas under the ROC curve varied from 0.54 to 0.87. Time post-stroke explained 42%-56% of variability in AQoL-4D utility values in patients with no change in mRS between 3 and 12 months. The choice of the derivation method contributed to 25% of the variability in UW-mRS values. Conclusions The high variability in utility values between and within mRS categories, over time post-stroke, and using different derivation methods is not adequately reflected in the UW-mRS. These threats to construct validity warrant caution when using UW-mRS as a primary outcome measure.