The Minimal Clinically Important Difference: A Review of Clinical Significance

Background: The minimal clinically important difference (MCID) is a term synonymous with orthopaedic clinical research over the past decade. The term represents the smallest change in a patient-reported outcome measure that is of genuine clinical value to patients. It has been derived in a myriad of ways in existing orthopaedic literature. Purpose: To describe the various modalities for deriving the MCID. Study Design: Narrative review; Level of evidence, 4. Methods: The definitions of common MCID determinations were first identified. These were then evaluated by their clinical and statistical merits and limitations. Results: There are 3 primary ways for determining the MCID: anchor-based analysis, distribution-based analysis, and sensitivity- and specificity-based analysis. Each has unique strengths and weaknesses with respect to its ability to evaluate the patient’s clinical status change from baseline to posttreatment. Anchor-based analyses are inherently tied to clinical status yet lack standardization. Distribution-based analyses are the opposite, with strong foundations in statistics, yet they fail to adequately address the clinical status change. Sensitivity and specificity analyses offer a compromise of the other methodologies but still rely on a somewhat arbitrarily defined global transition question. Conclusion: This current concepts review demonstrates the need for (1) better standardization in the establishment of MCIDs for orthopaedic patient-reported outcome measures and (2) better study design—namely, until a universally accepted MCID derivation exists, studies attempting to derive the MCID should utilize the anchor-based within-cohort design based on Food and Drug Administration recommendations. Ideally, large studies reporting the MCID as an outcome will also derive the value for their populations. It is important to consider that there may be reasonable replacements for current derivations of the MCID. As such, future research should consider an alternative threshold score with a more universal method of derivation.