Revisiting the Concept of Minimal Detectable Change for Patient-Reported Outcome Measures

Abstract

Abstract Interpreting change is a requisite component of clinical decision making for physical therapists. Physical therapists often interpret change using minimal detectable change (MDC) values. Current MDC formulas are informed by classical test theory and calculated with group-level error data. This approach assumes that measurement error is the same across a measure’s scale and confines the MDC value to the sample characteristics of the study. Alternatively, an item response theory (IRT) approach calculates separate estimates of measurement error for different locations on a measure’s scale. This generates a conditional measurement error for someone with a low, middle, or high score. Error estimates at the measure-level can then be used to determine a conditional MDC (cMDC) value for individual patients based on their unique pre- and post-score combination. cMDC values can supply clinicians with a means for using individual score data to interpret change scores while providing a personalized approach that should lower the threshold for change compared with the MDC and enhance the precision of care decisions by preventing misclassification of patients. The purpose of this Perspective is to present how IRT can address the limitations of MDCs for informing clinical practice. This Perspective demonstrates how cMDC values can be generated from item-level psychometrics derived from an IRT model using the patient-reported Activities-specific Balance Scale (ABC) commonly used in stroke rehabilitation and also illustrates how the cMDC compares to the MDC when accounting for changes in measurement error across a scale. Theoretical patient examples highlight how reliance on the MDC value can result in misclassification of patient change and how cMDC values can help prevent this from occurring. This personalized approach for interpreting change can be used by physical therapists to enhance the precision of care decisions.