The minimal clinically important difference
Why it is needed
Statistical significance says a difference is unlikely to be chance; it says nothing about whether the difference is worth having. In a large trial a change of no consequence to anyone will be significant, and in a small trial an important one will not be. The MCID supplies the missing scale.
How it is derived
- ANCHOR-based: relate the change in score to an external judgement, usually the patient's own global rating of whether they are better, and read off the score change corresponding to a small but noticeable improvement.
- DISTRIBUTION-based: use a fraction of the standard deviation or the standard error of measurement. These describe what change can be reliably detected, not what change matters.
- Consensus among clinicians or patients, which is transparent and subjective.
- The best evidence combines an anchor-based estimate with a distribution-based floor, since a difference smaller than the measurement error cannot be claimed anyway.
- The value should come from a population resembling the study's, and be cited.
Where it is used
In sample size calculation, as the difference the study should be able to detect. In interpreting a confidence interval: an interval lying entirely below the MCID rules out an important effect, which is a positive finding rather than a null one. And in noninferiority trials, where the margin is a version of the same judgement.
Group and individual are different
The MCID was defined for individual change, and a group mean difference of that size means something else: it can arise from most people unchanged and a minority improving a lot. Reporting the proportion of patients exceeding the threshold, alongside the mean difference, addresses this directly.
What to watch out for
There is no single MCID for an instrument: published values for the same scale differ by a factor of two or more, because they depend on the population, the direction of change, the baseline severity and the method used to derive them. Distribution-based values, half a standard deviation and the like, are statements about measurement precision rather than about what patients care about, and using one as if it were an anchor-based value confuses the two.
Clinical example
A trial finds a mean improvement of 2.1 points on a symptom scale, p equal to 0.01, where the anchor-based MCID is 5 points. The result is real and too small to matter, and the confidence interval excluding 5 is the sentence worth writing.
- Measurement of health status. Ascertaining the minimal clinically important difference (opens in a new tab) · Control Clin Trials · doi 10.1016/0197-2456(89)90005-6
- Reporting of patient-reported outcomes in randomized trials: the CONSORT PRO extension (opens in a new tab) · JAMA · doi 10.1001/jama.2013.879