In the OMERACT framework, Discrimination refers to an outcome measure’s ability to distinguish between situations of interest. This includes detecting differences:
1. Between groups (e.g., treatment vs. control/placebo groups).
2. Over time (e.g., responsiveness to change within individuals).
3. Within individuals (e.g., change in a single patient’s status).
It encompasses reliability (consistency across raters or occasions), sensitivity to change (responsiveness), and thresholds of meaning (the smallest change that is clinically important).
Example
An instrument measuring fatigue in rheumatoid arthritis is tested in a clinical trial. It successfully detects:
A difference between groups (patients receiving the new therapy report significantly less fatigue than those on placebo).
A change over time (fatigue scores decrease within the same patients as treatment progresses).
This demonstrates good discrimination, meaning the instrument can reliably distinguish between situations of interest.