Anchored Minimal Clinically Important Difference Metrics are Biased by Regression-to-the-Mean

ABSTRACT Minimal clinically important differences (MCID) are used to understand clinical relevance. However, repeated observations produce biased analyses unless accounting for baseline observation, called Regression-to-the-Mean (RTM). Using an International Knee Documentation Committee survey data set, the effect of RTM on MCID is demonstrated by 1) MCID estimate dependence on baseline observation, and 2) MCID estimate bias being higher when post-pre data correlation is lower. Ten IKDC datasets were created with 5,000 patients and a specific correlation under both equal and unequal variances. For every 10 points increase in baseline IKDC, MCID decreased by 3.5, 2.7, 1.9, 1.2, and 0.7 points where post-pre correlations were 0.10, 0.25, 0.50, 0.75, and 0.90 under equal variances. Failing to account for RTM results in a static 20 point MCID. MCID estimates may be unreliable. MCID calculations should report the correlation and variances between post-pre data and consider a baseline covariate-adjusted ROC analysis to calculate MCID. KEY POINTS