Concept Architecture
Concept
Theoretically, Minimally Important Difference is the smallest change in an outcome measure that patients or informed stakeholders perceive as important and that would justify a change in clinical management in the absence of excessive costs or adverse effects. It is founded on the concepts of clinical importance and responsiveness rather than statistical significance. The concept exists to distinguish changes that are meaningful to patients from changes that are merely detectable statistically. The term is commonly used interchangeably with minimal important difference in the health economics and outcomes research literature.
Mathematically, minimally important difference has no universally recognised canonical mathematical formula. It is estimated using recognised anchor-based or distribution-based methods, with anchor-based approaches generally regarded as the preferred method because they relate observed score changes to an independent measure of meaningful change.
In practice, minimally important difference is estimated by comparing changes in health-related quality of life or clinical outcome scores with external anchors such as patient global ratings of change, clinician assessments or other validated reference measures. Distribution-based methods, including fractions of the standard deviation or standard error of measurement, may be used to support interpretation but should not replace anchor-based estimation where suitable anchors are available.
Purpose
Used to determine the smallest change in an outcome measure that is considered clinically or practically meaningful, supporting interpretation of treatment effects, sample size determination and health economic evaluation.
Mathematical Formulae
Primary Formula
There is no universally recognised canonical mathematical formula.
Supporting Formulae
Common distribution-based estimators include:
MID � 0.5 ? SD
and
SEM = SD�(1 ? r)
where:
- SD = standard deviation of the outcome measure
- r = reliability coefficient
- SEM = standard error of measurement
Related Mathematical Methods
- Anchor-based estimation
- Distribution-based estimation
- Receiver operating characteristic (ROC) analysis
- Standard error of measurement
- Effect size estimation
Example
A health-related quality of life questionnaire has a baseline standard deviation of 12 points. An anchor-based analysis indicates that patients reporting themselves as ""a little better"" improve by an average of 6 points. The estimated minimally important difference is therefore 6 points. The distribution-based estimate of 0.5 ? SD also equals 6 points, providing supportive evidence for the estimate.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| AVERAGE | =AVERAGE(B2:B101) | Calculates the mean change score for an anchor group. |
| STDEV.S | =STDEV.S(B2:B101) | Calculates the standard deviation used in distribution-based estimation. |
| IF | =IF(B2>=MID_Value,""Clinically Important"",""Not Clinically Important"") | Classifies observed changes using the estimated minimally important difference. |
| SQRT | =B2*SQRT(1-C2) | Calculates the standard error of measurement from the standard deviation and reliability coefficient. |
VBA (Optional)
Automate the estimation and reporting of minimally important differences using anchor-based and distribution-based methods across multiple outcome measures.
Sources
- Jaeschke R, Singer J, Guyatt GH. Measurement of Health Status: Ascertaining the Minimal Clinically Important Difference. Controlled Clinical Trials. 1989.
- Revicki D, Hays RD, Cella D, Sloan J. Recommended Methods for Determining Responsiveness and Minimal Important Differences for Patient-Reported Outcomes. Journal of Clinical Epidemiology. 2008.
- Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. Oxford University Press.
- Brazier J, Ratcliffe J, Salomon JA, Tsuchiya A. Measuring and Valuing Health Benefits for Economic Evaluation. Oxford University Press.
- NICE. Health Technology Evaluation Manual.
Related Concepts (2)
Library
Publications
1
NICE DSU Technical Support Document 8: An Introduction to the Measurement and Valuation of Health for NICE Submissions — Brazier, Rowen, TSD 8 ed., 2011 (NICE Decision Support Unit (University of Sheffield))
An introduction to the measurement and valuation of health for NICE submissions — the QALY, health-state utility values, generic preference-based measures, and the requirements of the NICE reference case.
Frequently Asked Questions (6)
What is the minimally important difference?
The threshold change in a health outcome score considered meaningful, estimated using either anchor-based or distribution-based methods.
Source: Jaeschke, Singer & Guyatt 1989
How does the minimally important difference differ from statistical significance?
A difference can be statistically significant yet too small for any patient to notice, and a real, worthwhile change can fail to reach significance in a small study. The minimally important difference addresses the first problem by asking not whether a change is unlikely to be chance but whether it is large enough to matter to those experiencing it. It concerns the magnitude and meaning of a change rather than the confidence that it is not zero. Jaeschke and colleagues (1989) drew this distinction when introducing the concept.
Source: Jaeschke et al. 1989
How is the minimally important difference estimated?
It is estimated by two families of method. Anchor-based methods relate changes in the score to an external indicator of meaningful change, such as a patient's global rating of improvement, taking the score change that corresponds to a small but important change on the anchor. Distribution-based methods derive the threshold from the statistical distribution of scores, such as around half a standard deviation. The two address importance and statistical size respectively, and are often used together.
Source: Jaeschke, Singer & Guyatt 1989
Why are both anchor-based and distribution-based methods used for the minimally important difference?
Both are used because each has a weakness the other offsets: anchor-based methods address whether a change is important but depend on the choice and interpretation of the anchor, while distribution-based methods are easy to compute and independent of an anchor but do not establish that a change matters. Using them together, and seeing whether they converge, gives a more defensible estimate of the threshold than either alone, since agreement lends confidence and divergence signals caution.
Source: Jaeschke, Singer & Guyatt 1989
How is the minimally important difference applied?
The minimally important difference is applied to judge whether a treatment effect or an individual's change is large enough to be meaningful, to interpret trial results beyond statistical significance, and to set the target difference in sample-size calculations. It provides the benchmark that turns a change in the units of a measure into a statement about clinical or patient-relevant benefit, so that a result can be described as important rather than merely detectable.
Source: Jaeschke, Singer & Guyatt 1989
What are the limitations of the minimally important difference?
The threshold is not fixed, since it depends on the estimation method, the anchor, the population, and the baseline, so reported values vary and no single figure is definitive. It is generally an average that may not fit every patient, and it embeds a judgement about what counts as important. Because of this variability, it is treated as an approximate guide, and the method and context of its estimation are reported when it is used.
Source: Jaeschke, Singer & Guyatt 1989
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 1 Sep 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-EE-HU-050
Stable URI · Machine-readable · Resolvable · CC BY 4.0