How is the anchor-based approach used to establish the minimal important change (MIC) for a patient-reported outcome measure?

The anchor-based approach establishes the MIC by administering the patient-reported outcome measure on two occasions to people expected to change and, at the second administration, collecting an external anchor such as a global rating of change. The MIC is the change score on the measure that corresponds to the anchor category considered meaningful, such as the average change for patients who say they are "a little better" or "much better." An MIC produced this way is population-specific and can also be derived statistically using receiver operating characteristic curve analysis.

To use the anchor-based approach, researchers first administer the focal measure on two separate occasions to a sample in which at least some participants are expected to change. At the second administration, participants also provide information on an anchor, usually a single-item global rating scale, such as a 7-point rating of change from much worse to much better. Researchers then link change scores on the focal measure to each response category of the anchor. For example, if patients who rated themselves "a little better" had a mean physical-function change of 2.17, researchers who emphasize the smallest meaningful improvement would set the MIC at 2.17. If instead they use the "much better" group, whose mean change might be 3.89, the MIC would be higher. Either way, the MIC becomes the threshold used to classify individual patients as having or not having a clinically important change. Anchor-based methods can also use receiver operating characteristic curve analysis to derive the MIC. The anchor does not have to be a patient global rating; it could be a performance test or a clinician assessment. Establishing an MIC with this method requires a large sample whose changes over time vary, and the resulting MIC is specific to the population being studied.