Abstract
Utility score is a preference-based measure of general health state – where 0 is equal to death, and 1 is equal to perfect health. To understand a patient's smallest perceptible change in utility score, the minimal clinically important difference (MCID) can be calculated. However, there are multiple methods to calculate MCID with no consensus about which method is most appropriate. The aim of this study is to calculate MCID values for the Veterans-RAND 12 (VR12) utility score using varying methods. Our hypothesis is that different methods will yield different MCID values.
A tertiary institutional registry (SMART) was used as the study cohort. Patients who underwent unilateral TKA for osteoarthritis from January 2012 to January 2020 were included. Utility score was calculated from VR12 responses using the standardised Brazier's method. Distribution and anchor methods were used for the MCID calculation. For distribution methods, 0.5 standard deviations of the baseline and change scores were used. For anchor methods, the physical and emotional anchor questions in the VR12 survey were used to benchmark utility score outcomes. Anchor methods included mean difference in change score, mean difference in 12 month score, and receiver operating characteristics (ROC) analysis with the Youden index.
Complete case analysis of 1735 out of 1809 eligible patients was performed. Significant variation in the MCID estimates for VR12 utility score were reported dependent on the calculation method used. The MCID estimate from 0.5 standard deviations of the change score was 0.083. The MCID estimate from the ROC analysis method using physical or emotional anchor question improvement was 0.115 (CI95 0.08-0.14; AUC 0.656).
Different MCID calculation methods yielded different MCID values. Our results suggest that MCID is not an umbrella concept but rather many distinct concepts. A general consensus is required to standardise how MCID is defined, calculated, and applied in clinical practice.