The Gale readResearch

Steps per minute predicted death as well as walking speed (HR 0.83)

A UK cohort of 4,067 older adults followed for a median 11.8 years found that how briskly people walked in daily life, measured by a wrist accelerometer, tracked all-cause mortality about as well as a 2.44-metre clinic walking test.

By Gale Staff · October 25, 2026 · JAMA Network Open

Not reviewed by a clinician.

The short answer

Walking cadence measured in everyday life appears to carry about the same information about all-cause mortality as walking speed timed in a clinic. Among 4,067 Whitehall II participants (mean age 69.3 years), one standard deviation higher median cadence went with a hazard ratio of 0.83 (95% CI 0.77-0.90), against 0.77 (0.71-0.84) for clinic walking speed, and the three measures discriminated between those who died and those who survived almost identically (C index 0.744 to 0.747). This is an observational cohort that is about three-quarters male and drawn from a British civil-service population, so the findings describe an association and not an effect of walking faster, and the abstract does not report how many participants died.

Who to see

Geriatrician

Steps a minute in daily life said about as much about survival as walking speed timed in a clinic; a geriatrician can check your walking and mobility.

Find care →

The question after the headline: is a step count of any use here

A clinic can ask an older patient to walk a short, marked distance while someone times it, and the result, walking speed, has long been linked with how long people go on to live. A wristband that logs movement day and night is a different kind of measurement, with no stopwatch and no corridor. The natural question once both exist is whether the second says anything the first does not. A study published on 23 October 2026 in JAMA Network Open, by Rayane Haddadj and colleagues, set the two side by side in the same people.

The answer it gives is a measured one. The two approaches carried closely similar information about mortality, with one difference in how they added to a model that already held other risk factors. Neither the headline nor the abstract treats this as a reason to change what anyone does when they walk.

What the researchers did

The work draws on the Whitehall II accelerometer substudy in the United Kingdom, a population-based cohort. Baseline measurements were taken from January 2012 to December 2013, and participants were followed through 19 August 2024. To be included, a person needed valid accelerometer data, a clinical walking speed, the covariates the authors adjusted for, and electronic health record information. That left 4,067 participants, with a mean age of 69.3 years (standard deviation 5.7) and 3,024 men, or 74.4 percent. The median follow-up was 11.8 years, with an interquartile range of 11.5 to 12.1.

Clinical walking speed, in metres per second, came from a 2.44-metre walking test. The free-living measures came from nine days of wrist-worn accelerometer data, from which the team derived two cadence figures in steps per minute: the median walking cadence, and the 95th percentile walking cadence, which describes how fast a person walks during their quickest bouts rather than their typical ones.

All-cause mortality was taken from a British national mortality register. The authors used Cox proportional hazards models to estimate hazard ratios, each expressed per one standard deviation higher value of the measure, and judged how well each measure separated people who died from those who did not with Harrell's concordance index, a number where 0.5 is a coin toss and 1.0 is perfect. The analyses were run between September 2025 and February 2026.

What they found

In models containing mortality risk factors, covering sociodemographic, behavioural and health-related factors, all three measures were associated with lower all-cause mortality. Each standard deviation higher clinical walking speed went with a hazard ratio of 0.77 (95% confidence interval 0.71 to 0.84). Median walking cadence gave 0.83 (0.77 to 0.90), and 95th percentile cadence gave 0.82 (0.75 to 0.89). In plain terms, in this group, people who walked faster or at a higher cadence than average were less likely to die over roughly twelve years, after the authors' adjustments, and the cadence figures sit slightly closer to 1.0 than the clinic figure, with confidence intervals that overlap.

Discrimination was almost indistinguishable. The C index was 0.747 (0.730 to 0.764) for clinical walking speed, 0.744 (0.726 to 0.761) for median cadence and 0.745 (0.727 to 0.762) for 95th percentile cadence. A value near 0.75 is a useful but imperfect separation of those who died from those who did not, and the three differ by far less than the width of their own intervals.

The one place the measures parted is stated in a single sentence of the abstract: only clinical walking speed was associated with an improved discriminative ability of a model that already included mortality risk factors. The abstract gives no figure for that improvement, so the size of the gain cannot be quoted here. Cadence, in other words, gave similar discrimination in models that included the risk factors, but unlike the clinic test it was not associated with an improvement in discrimination over a model holding those risk factors.

The caveat that matters most

A hazard ratio of 0.83 is a relative figure, and the abstract does not give the number of deaths or the absolute risk in any group, so what it means in percentage points over twelve years cannot be worked out from what is published there. It is also a statement about a group of people who happen to walk differently, not about what happens if one person changes how they walk. Slower walking can be an early sign of illness that has not yet been diagnosed, so some of the association may run from poor health to slow walking rather than the other way round, and adjustment for measured risk factors cannot fully remove that.

The population adds a second limit. Whitehall II participants were recruited from the British civil service, and in this analysis about three in four were men. Those who also had valid accelerometer data, a clinic walking measurement and linked health records are a further selected subset of that group.

Why it still matters

Clinic walking tests need a person, a measured distance and a visit. Wrist accelerometers can collect data in large research cohorts, as they did in this one. If a cadence measure captured over nine ordinary days gives about the same ranking of mortality risk as a stopwatch test, large population studies may use whichever one suits their design.

The authors' own conclusion is measured in the same way: both measures can be used in population-based studies, and the choice may depend on research objectives, implementation constraints and the strengths of each method. That is a statement about measurement in research, and not a claim that a wristband replaces a clinical assessment for any individual.

What this study can't tell you

  • Whether walking faster or at a higher cadence lowers the chance of dying: this is an observational cohort, and slower walking may reflect illness that was already developing.
  • How many people died, and what the absolute risk was in each group: the abstract reports hazard ratios and concordance indices but not the number of deaths or event rates.
  • How large the added discrimination from clinical walking speed was: the abstract states that only clinical walking speed improved a model containing mortality risk factors, with no figure given.
  • Whether the findings hold in women, in younger people or in other populations: participants had a mean age of 69.3 years, 74.4% were male, and all came from the Whitehall II cohort in the UK.
  • What any one person's cadence or walking speed means for that person's own outlook: the estimates describe groups, per standard deviation, in models adjusted for the authors' chosen covariates.
  • Whether people who lacked valid accelerometer data, a walking test or linked records differ from those analysed, since the analysis required all of them.

The Gale read

This paper makes a modest, well-bounded claim and the abstract supports it. A clinic timing and a wristband-derived cadence landed within a few thousandths of each other on discrimination (0.744 to 0.747), with hazard ratios per standard deviation of 0.77, 0.83 and 0.82 whose confidence intervals overlap. That is a reasonable basis for saying that free-living cadence is a plausible measurement tool in population research. It is not a basis for saying more. The one sentence that separates the two methods, that only clinical walking speed improved discrimination beyond other risk factors, arrives without a number. The abstract also omits deaths and absolute risks, so a reader cannot translate 0.83 into a probability, and the cohort is older, mostly male and drawn from one British employment group. What the study gives, then, is a comparison of two measurement methods in one cohort, not guidance about how any individual should move.

Common questions

is walking speed linked to how long you live

In this UK cohort of 4,067 adults with a mean age of 69.3, yes: each standard deviation higher clinic walking speed went with a hazard ratio of 0.77 (95% CI 0.71-0.84) for all-cause mortality over a median 11.8 years, after adjustment for risk factors. It is an association in an observational study, so it does not show that walking faster extends life, and slower walking may partly reflect existing illness.

what is a good walking cadence in steps per minute

The abstract does not give a threshold or a target. It reports hazard ratios per one standard deviation higher cadence, 0.83 for median walking cadence and 0.82 for 95th percentile cadence, without publishing the standard deviation in steps per minute, so it cannot name a cadence that counts as good. The finding is that cadence tracked mortality about as well as clinic walking speed in this group.

can a fitness tracker predict mortality as well as a walking test

In this study a wrist accelerometer worn for nine days gave about the same discrimination as a 2.44-metre clinic walking test, with C index values of 0.744 for median cadence and 0.747 for walking speed. The accelerometer here was a wrist-worn device worn for nine days in a mostly male, mostly older cohort, and only the clinic test improved a model that already included mortality risk factors, so the result does not show that a consumer tracker can stand in for a clinic assessment.

Sources

  1. 1.Haddadj R, Danilevicz IM, Rowlands AV, van Hees VT, Landre B, Le Faucheur A, Singh-Manoux A, Mork PJ, Sabia S. Clinical Walking Speed, Free-Living Walking Cadence, and All-Cause Mortality. JAMA Network Open. 2026;9(10):e2640122. doi:10.1001/jamanetworkopen.2026.40122 link

Read the study →

1 sources, numbered by first appearance. General health information, not medical advice. AI-assisted editorial content — every citation independently verified. Editorial policy

More from Gale Staff

The Latest Health Research →

Published October 25, 2026 · Gale Staff

All Gale Staff analyses →