Everyday Walking Cadence From Wearables Predicted Mortality as Well as Gait-Speed Tests

That is the finding of a study in JAMA Network Open. It is an observational cohort study, so it shows links, not cause and effect. The researchers drew on the Whitehall II accelerometer substudy in the UK and analysed 4,067 participants. Their mean age was 69.3 years, and 3,024 of them (74.4%) were male.

Clinics often measure walking speed with a short test. Here, it was a 2.44-metre walk, recorded in meters per second. The team compared it with walking cadence, which is steps per minute. Cadence came from 9 days of data from a wrist-worn accelerometer, a motion sensor, as people went about daily life.

Two cadence measures were used: the median cadence and the 95th percentile cadence. The median is the typical pace. The 95th percentile reflects the faster end of a person's walking.

Baseline measures were taken between January 2012 and December 2013. Deaths were tracked through a British national mortality register until August 19, 2024. The median follow-up was 11.8 years.

All three measures were linked to lower risk of death from any cause. The figures are per 1 standard deviation higher, a standard step up in each measure. These results came from models that accounted for sociodemographic, behavioral and health-related risk factors. For clinical walking speed, the hazard ratio was 0.77 (95% CI, 0.71-0.84). For median cadence it was 0.83 (0.77-0.90). For 95th percentile cadence it was 0.82 (0.75-0.89). A hazard ratio below 1 means lower risk.

The researchers also checked how well each measure separated people who died from those who did not. They used the Harrell concordance index, where 1.0 is perfect and 0.5 is chance. The scores were very close: 0.747 for clinical walking speed, 0.744 for median cadence and 0.745 for 95th percentile cadence. The confidence intervals overlapped almost entirely.

There was one difference. Only clinical walking speed was associated with improved discriminative ability when added to a model that already included mortality risk factors.

Cadence did not show that added gain.

The authors conclude that both measures can be used in population-based studies. They say the choice may depend on research objectives, implementation constraints and the strengths of each method.

There are limits. The study describes an older cohort, mostly male, so the results may not apply to everyone. Because the study is observational, it cannot show that walking faster extends life. Slower walking could simply reflect underlying illness.

Still, the result matters for how healthspan might be tracked. A sensor worn during normal life gave prediction similar to a clinic test. That suggests everyday pace could be measured at scale without a visit. Whether cadence-based measures hold up in more diverse groups, and whether they add useful information in clinical settings, where the study found only the clinic test improved the models.