AIM GlossaryBrowse the glossary:
minimal clinically important difference (MCID)
— The smallest change in score perceived as meaningful by the patient.
Full explanation:
MCID reflects change that patients consider important, even if it may not reflect a change in diagnosis.
Examples of MCID include:
Anchor-based MCID: Comparison to external assessment. This could be a separate item (e.g., respondents global rating of change), or a related measure.
Distribution-based MCID: SEM: Standard error of measurement. The SEM looks at score variation and indicates how much scores may vary over repeated administrations of a measure. The formula for the SEM is Standard Deviation *square root of (1- reliability usually Cronbach alpha). If the measure has high reliability, scores are likely to be consistent. Conversely, if a measure has low reliability, scores may vary across administrations of a measure. MCID is calculated using change scores. If change scores (e.g., baseline to Time 1 or pre-post intervention) are greater than SEM then the change is likely to reflect a ‘true’ change (not due to error).
Standard deviation (SD): MCID can also be calculated using SD. This is typically 0.5 SDs but does vary.
Effect Size (ES): Commonly used effect sizes to indicate meaningful change are: 0.20 = small; 0.50 = moderate; 0.80 = large.
Examples of MCID include:
Anchor-based MCID: Comparison to external assessment. This could be a separate item (e.g., respondents global rating of change), or a related measure.
Distribution-based MCID: SEM: Standard error of measurement. The SEM looks at score variation and indicates how much scores may vary over repeated administrations of a measure. The formula for the SEM is Standard Deviation *square root of (1- reliability usually Cronbach alpha). If the measure has high reliability, scores are likely to be consistent. Conversely, if a measure has low reliability, scores may vary across administrations of a measure. MCID is calculated using change scores. If change scores (e.g., baseline to Time 1 or pre-post intervention) are greater than SEM then the change is likely to reflect a ‘true’ change (not due to error).
Standard deviation (SD): MCID can also be calculated using SD. This is typically 0.5 SDs but does vary.
Effect Size (ES): Commonly used effect sizes to indicate meaningful change are: 0.20 = small; 0.50 = moderate; 0.80 = large.
Still don't get it?
If you feel that this definition hasn't helped you to understand the term, click on our monkey to let us know.
← meta-analysis minimization →