Improved adjusted minimal important change took reliability of transition ratings into account

Objectives: The anchor-based minimal important change (MIC), based on the receiver operating characteristic (ROC) analysis or predictive modeling, is biased by the proportion of improved patients. The adjusted MIC, published in 2017, adjusts the predictive MIC for this bias but does not take the reliability of the transition ratings (i.e., the anchor) into account. The aim of this study was to examine whether the transition ratings reliability affects the accuracy of the adjusted MIC and, if so, whether the adjustment can be improved. Study Design and Setting: Multiple simulations of patient samples involved in anchor-based MIC studies with different characteristics of patient-reported outcome scores were used to determine the impact of reliability of the transition ratings on the MIC estimate. An improved adjustment formula was derived in an exploration set of simulated samples (number of samples = 19,440) and validated in a different set of simulated samples (number of samples = 12,960). The effect of sample size (100–1,000) was also evaluated in simulated datasets. Results: Reliability of the transition ratings biased the MIC estimate if the proportion improved was different from 0.5. The improved adjustment formula performed well, especially if the proportion improved was between 0.3 and 0.7. Smaller sample sizes were at the expense of the precision of the MIC estimates. Conclusion: We provide an improved formula for calculating the adjusted MIC, taking into account the proportion of improved patients and the reliability of the transition ratings.