Foundations of the minimal clinically important difference for imaging

This article develops a generic conceptual framework for defining and valid ating the concept of minimal clinically important difference. We propose 3 approaches. The first uses statistical descriptions of the population ("dis tribution based"), the second relies on experts ("opinion based"), and a th ird is based on sequential hypothesis formation and testing ("predictive/da ta driven based"). The first 2 approaches serve as proxies for the third, w hich is an experimentally driven approach, asking such questions as "What c arries the least penalty?" or "What imparts the greatest gain?" As an exper imental approach, it has the expected drawbacks, including the need for gre ater resources, and the need to tolerate trial and error en route. compared to the other 2 models.