How you walk may flag lung disease years ahead

Chronic obstructive pulmonary disease is a leading cause of disability and death, and it is frequently diagnosed late because it develops slowly and quietly. Existing prediction tools depend largely on clinic-based assessments and detailed self-report questionnaires, which are resource-intensive, hard to scale and vulnerable to recall bias. Meanwhile, many older adults already wear devices that record how they move all day. The question this study set out to answer is whether everyday mobility captured passively at the wrist carries an early signal of impending COPD, and whether it could match the performance of conventional models.

The researchers analysed 28,251 UK Biobank participants aged 60-78 who had worn wrist-worn accelerometers. Digital gait biomarkers were extracted from the raw signals using signal-processing and machine-learning algorithms, describing features such as walking speed, running duration and the duration and distribution of walking bouts. Incident COPD was identified through linked electronic health records. Associations were tested with Cox proportional hazards models adjusted for age, sex, body mass index, smoking pack-years, air pollution exposure and asthma history, with internal validation, and discrimination summarised using Harrell's concordance index.

Over a mean follow-up of 8.5 years (SD 1.3), 639 participants — 2.26% of the cohort — developed COPD. Four digital gait biomarkers were independently associated with incident disease: lower running duration, slower maximal walking speed, shorter walking bout duration and a lower proportion of walks longer than 8 seconds. Combining those four movement measures with four easily collected items — age, sex, smoking pack-years and asthma history — produced a model with a Harrell's concordance index of 0.80, which the authors describe as comparable to existing models that require clinic-based tests and long questionnaires. The associations persisted after adjustment for smoking, pollution and asthma.

This extends a growing body of work showing that passively recorded movement carries prognostic information well beyond step counts. It fits the pattern in which slower and more fragmented walking precedes clinical diagnosis across several conditions, and it suggests the signal is not merely a proxy for smoking or existing asthma, since those were adjusted for. What it cannot establish is direction: reduced mobility may be an early consequence of undetected lung impairment rather than a cause of it. Only internal validation was performed, so external testing in other cohorts and in more diverse populations is the obvious next step.

This is a reason to view how you move day to day — how fast you can walk at your quickest, whether you still run at all, how long your continuous walking bouts last — as information about your health, not just your activity level. It does not mean walking faster prevents lung disease; the study only shows the pattern precedes diagnosis. For older adults, a drift towards slower, more broken-up walking is worth noticing rather than dismissing as ordinary ageing. Any actual diagnosis of lung disease still requires clinical testing.