The minimum clinically important difference is fundamental to all clinical trials

Editor, We commend Walder et al. on their recent Editorial1 and endorse their recommendations about choosing a minimum clinically important difference (MCID) from a patient's point of view in perioperative medicine trials. We also commend their advice to consider both clinical importance and statistical significance when interpreting clinical trials. However, we would like to clarify that the concept of a MCID, whether patient focussed or based on some other justified metric, is not limited to trials in perioperative medicine. The MCID has a fundamental and central role in all clinical trials (although it may be referred to in other terms such as the ‘minimum effect size of interest’ or the ‘minimum clinical worthwhile difference).2–52–52–52–5 This is because for any preset α (type I error), sample variability and desired level of power, the MCID chosen by the investigators determines the requisite sample size for the clinical trial.2–52–52–52–5 It also determines the threshold for clinical importance for any observed effect size once the trial is completed. In other words, if the observed effect size is less than the MCID, it must, be by the authors’ own definition, not be clinically important, whether or not it is statistically significant. The investigators cannot change their mind about the MCID after completion of the trial. Reducing the MCID posthoc (i.e. arguing that an observed effect size less than MCID is still clinically important) would in effect reduce the power of the trial and thereby the replicability of the statistically significant finding.5,65,6 The MCID is also the minimum difference that can reliably be detected in a trial for any given level of power.2–52–52–52–5 In other words, nonstatistically significant results pertain only to differences equal to or larger than the MCID, and no inferences with the same level of power can be made about the presence or absence of smaller true differences. This information about the relationship between MCID, sample size and power is available in most standard statistical texts. Unfortunately, the importance of the MCID, whether patient focussed or not, has often been overlooked in anaesthesia trials,5,75,7 so we applaud Walder et al.1 for the points made in their editorial. Acknowledgements relating to this article Assistance with the letter: none. Financial support and sponsorship: none. Conflicts of interest: none.