Clinical library · Outcome measures
Six-Minute Walk Test
A corridor, a stopwatch and six minutes. The most portable measure of functional exercise capacity there is — and the one where the published minimal clinically important difference varies more than fivefold depending on which paper you read.
In one line. The distance a patient walks in six minutes at their own pace along a flat corridor, with rests permitted. It is submaximal, self-paced, and reflects the exercise level of daily activity rather than peak capacity.
It is the standard functional capacity measure in cardiorespiratory rehabilitation and is used far beyond it. The trap is the minimal clinically important difference: quoting one number without naming its population is the single most common error made with this test.
Which MCID? A systematic review restricted to values derived by receiver operating characteristic analysis found MCIDs ranging from 14.0 to 30.5 metres across COPD, lung cancer, coronary artery disease, diffuse parenchymal lung disease, non-cystic-fibrosis bronchiectasis and adults with fear of falling. [4] An earlier and widely quoted review in COPD gave a conservative estimate of 54 to 80 metres, and noted that an individual patient would need to change by about 86 metres for statistical confidence. [3] Both are correctly reported. They are not interchangeable, and the difference is method, population and era — not error.
The numbers you actually need
| Property | Value | Source and caveat |
|---|---|---|
| MCID, adults with pathology | 14.0 to 30.5 metres | Six studies, all ROC-derived with area under the curve of at least 0.70; mean baseline distances 295 to 551 m [4] |
| MCID, COPD (older estimate) | 54 to 80 metres | Conservative estimate using distributional and discriminative methods; about 86 metres needed for confidence in an individual [3] |
| Healthy adults, 40 to 80 years | Median 576 m (men), 494 m (women) | 117 healthy men and 173 healthy women. Gender-specific regression equations explained about 40% of the variance [2] |
| Test-retest reliability, stroke | ICC 0.98 (95% CI 0.98 to 0.99) | Five studies pooled from six meeting criteria; other reliability types, SEM and MDC were rarely reported [5] |
| Multiple sclerosis versus controls | -177.2 ± 19.1 m (d = -1.87) | 34 studies, 3,204 people. Mild versus moderate-to-severe disability differed by 185.19 ± 9.2 m (d = 1.83) [6] |
| Protocol | Standardised by the American Thoracic Society | Course length, encouragement phrasing and rest rules are specified; deviations make results non-comparable [1] |
What it measures
Submaximal functional exercise capacity: the integrated performance of the pulmonary, cardiovascular, peripheral circulatory and neuromuscular systems during a self-paced walk. It is not a measure of any one of them, which is its strength as a functional outcome and its weakness as a diagnostic test.
Because it is self-paced and submaximal, it reflects the intensity at which patients actually live rather than the ceiling they could reach under supervision. That is usually the more relevant question in rehabilitation.
Where it misleads
1. Quoting an MCID without its population
This is the central problem. A change of 25 metres is clinically important in some populations [4] and well within noise in others, where the estimate ran to 54 to 80 metres. [3] Any statement of the form "the MCID for the six-minute walk test is X" with no population attached should be treated as unusable. Report the population and the derivation method with the number, every time.
2. Encouragement and course layout change the result
These are not trivial procedural details. In the multiple sclerosis meta-analysis, the provision of encouragement and the course layout both moderated the effect of the disease on performance, and course layout also moderated the effect of disability status. [6] A protocol variation can therefore mimic or mask a real change. The American Thoracic Society statement exists precisely to remove that variance. [1]
3. Reference equations are not clinical targets
The gender-specific equations for healthy adults explain only about 40% of the variance in distance walked, and are intended for computing percent-predicted values in adults performing the test for the first time under the standardised protocol. [2] Treating predicted distance as a rehabilitation goal misreads what the equation is. Note also that an erratum to that paper was subsequently published; anyone applying the equations should work from the corrected source rather than from a figure copied out of a textbook.
4. Reliability in neurological populations is less established than it looks
Test-retest reliability in stroke is genuinely excellent at ICC 0.98. But the same review found that only six studies met inclusion criteria, only one investigated inter-rater and intra-rater reliability, and standard error of measurement and minimal detectable change were rarely reported. [5] Without an MDC you cannot say whether an individual patient's change is real.
What the evidence supports — and what it does not
Supported
- Measuring functional exercise capacity under a standardised protocol. [1]
- Detecting change of 14 to 30.5 metres in the pathologies where that has been derived. [4]
- Repeat measurement after stroke, with excellent test-retest reliability. [5]
- Comparison against healthy reference distances, interpreted as percent predicted. [2]
- Quantifying disease burden in multiple sclerosis, with a large effect against controls. [6]
Not supported
- A single universal MCID. Published values span 14 to 80 metres. [3][4]
- Comparing results across different protocols, courses or encouragement practices. [1][6]
- Using predicted distance as a treatment target. The equations explain about 40% of variance. [2]
- Interpreting individual change in stroke against a published MDC. Those values are rarely reported. [5]
- Any inference about maximal exercise capacity. The test is submaximal by design.
How certain is this?
Evidence grade: Moderate.
The protocol itself is the most secure element: a society guideline that has standardised practice internationally for two decades. [1] It is cited here for the protocol only; no numerical claim on this page rests on it.
The MCID evidence is where the uncertainty sits, and it is uncertainty of a specific kind. The systematic review restricted itself to ROC-derived values with an area under the curve of at least 0.70 and found only six qualifying articles, across six different populations. [4] The wider 54-to-80-metre estimate is a narrative review rather than a pooled analysis. [3] Neither is wrong; they answer slightly different questions, and the honest position is to report both with their provenance rather than to pick the convenient one.
The reference equations come from 290 healthy adults in a single study and explain about 40% of variance, with an erratum subsequently issued. [2] The stroke reliability estimate is strong but rests on five pooled studies with little supporting measurement-error data. [5]
What would change the grade: MDC values for the neurological populations in which this test is routinely used, and MCID replication within populations rather than across them.
Common questions
How much change counts as real?
Ask first which population. Across six ROC-derived studies in adults with pathology, MCIDs ranged from 14.0 to 30.5 metres. [4] In COPD specifically, an older review estimated 54 to 80 metres and suggested about 86 metres for confidence in an individual patient. [3] Report the value you are using and its source alongside the result.
What distance should a healthy adult manage?
In 117 healthy men and 173 healthy women aged 40 to 80, median distances were 576 metres for men and 494 for women. Gender-specific regression equations were derived for computing percent predicted, and they explain about 40% of the variance. [2] Use the corrected source, since an erratum was later published.
Does it matter if I encourage the patient?
Yes, measurably. In the multiple sclerosis meta-analysis, provision of encouragement moderated the effect of the disease on performance, as did course layout. [6] The American Thoracic Society protocol standardises the phrasing and timing of encouragement for exactly this reason. [1] Whatever you do, do the same thing every time.
Can I use a shorter corridor?
Only if you accept that the result is not comparable with anything else. Course layout was a significant moderator of performance. [6] More turns means more deceleration and acceleration, and a shorter distance for the same physiological effort.
Is it reliable after stroke?
Test-retest reliability is high, with a pooled ICC of 0.98 (95% CI 0.98 to 0.99) from five studies. The gap is that only one study examined inter-rater and intra-rater reliability, and SEM and MDC were rarely reported — so you can trust the repeat measurement more than you can interpret an individual change score. [5]
What does it tell me in multiple sclerosis?
It captures a large disease effect: people with multiple sclerosis walked 177.2 metres less than healthy controls on average, and those with mild disability walked 185.19 metres further than those with moderate-to-severe disability, both with large effect sizes. [6] See also multiple sclerosis for the intervention evidence.
References
- . ATS statement: guidelines for the six-minute walk test. American Journal of Respiratory and Critical Care Medicine. 2002 Jul 1;166(1):111–7. doi:10.1164/ajrccm.166.1.at1102 PMID 12091180 Society guideline
- Enright PL, Sherrill DL. Reference equations for the six-minute walk in healthy adults. American Journal of Respiratory and Critical Care Medicine. 1998 Nov;158(5 Pt 1):1384–7. doi:10.1164/ajrccm.158.5.9710086 PMID 9817683 Reference equation study
- Wise RA, Brown CD. Minimal clinically important differences in the six-minute walk test and the incremental shuttle walking test. COPD: Journal of Chronic Obstructive Pulmonary Disease. 2005 Mar;2(1):125–9. doi:10.1081/copd-200050527 PMID 17136972 Narrative review
- Bohannon RW, Crouch R. Minimal clinically important difference for change in 6-minute walk test distance of adults with pathology: a systematic review. Journal of Evaluation in Clinical Practice. 2017 Apr;23(2):377–381. doi:10.1111/jep.12629 PMID 27592691 Systematic review
- Macchiavelli A, Giffone A, Ferrarello F, et al. Reliability of the six-minute walk test in individuals with stroke: systematic review and meta-analysis. Neurological Sciences. 2021 Jan;42(1):81–87. doi:10.1007/s10072-020-04829-0 PMID 33064231 Systematic review and meta-analysis
- Cederberg KLJ, Sikes EM, Bartolucci AA, et al. Walking endurance in multiple sclerosis: Meta-analysis of six-minute walk test performance. Gait & Posture. 2019 Sep;73:147–153. doi:10.1016/j.gaitpost.2019.07.125 PMID 31326830 Meta-analysis
About this resource
Using this in clinic
Every figure here is traceable to its source.
Every distance, MCID and reliability coefficient on this page carries the population it came from, because in this test the population is what makes the number mean anything. Where a value could not be verified against the paper it came from, it is not on this page, and the omission is stated rather than filled with a number from a secondary source.