Expanding the definition of clinical differences: from minimally clinically important differences to really important differences. Analyses in 8931 patients with rheumatoid arthritis

Abstract

Objective: Minimally clinically important differences (MCID) have become an important way to interpret data of randomized clinical trials (RCT), but do not reflect the degree of improvement consistent with a "really important difference" (RID). To define RID, we compared mean and/or least desirable clinical states with best and/or most desirable states.

Methods: In total, 8931 patients with rheumatoid arthritis (RA) < 65 years of age completed the Health Assessment Questionnaire (HAQ) and Medical Outcomes Survey Short Form 36 Physical Component Score (PCS). Definitions of RID were based on values for HAQ and PCS corresponding with the best and worst category of the following conditions: disabled vs not disabled: joint replacement vs no joint replacement; < poverty level vs > poverty level; very satisfied with health vs not; and independent in participation activities vs not.

Results: In contrast to published MCID values for the HAQ of approximately 0.22, RID was as high as 0.76 using objective reference conditions and 0.87 using the subjective measure of dependence vs independence. The HAQ score of independent RA patients was 0.38 (SD 0.45), and was 0.42 (SD 0.53) for those very satisfied with their health. The difference in HAQ scores between disabled and working patients was approximately 0.75. PCS differences were similarly increased.

Conclusion: RID values are 3 to 4 times greater than MCID values. Although MCID are meaningful statistics for RCT, the RID percentage achieved [(actual improvement/RID) 100%] is a simple way to put the results of RCT in a broader perspective that gives an idea of how much additional treatment effect is needed.