More than step counts — slow walking speed, little running and few long walks predict cardiovascular mortality: a Walk Watch UK Biobank Study

Background Low daily step counts have traditionally been associated with cardiovascular death risk, suggesting other objective real-world gait measures may be complementary or better predictors. This study examined the relationship between real-world walking speed, quality, and walking bout distributions, measured using a wrist-worn device, and cardiovascular death in a large cohort of older people. Methods Participants aged 60to78 years from the UK Biobank who wore a wrist-worn device were included in this population-based observational cohort study. Gait data were analysed using Watch Walk methods. Cardiovascular death, defined as death within ten years of follow-up due to heart disease, stroke, or vascular conditions, was tracked using National Health Service databases. Minimally adjusted and multivariable Cox proportional-hazard models assessed the relationship between digital gait biomarkers and cardiovascular death. Results Among 38,766 participants, 485(1.3%) had cardiovascular deaths during follow-up. In minimally adjusted models, maximal walking speed, running duration, step count, longest walk duration, and the proportion of short walks were associated with cardiovascular death. In multivariable models adjusted for age, sex and smoking status, slower maximal walking speed, reduced daily running duration, and a higher proportion of short walks remained independent predictors. This model had a C-statistic of 0.75, comparable to traditional risk scores including SCORE2 and the Framingham Risk Score (both 0.74). Conclusions Walking speed, running duration, and the proportion of longer walks are key real-world walking characteristics to consider when assessing cardiovascular death risk. Predictive models with these measures demonstrate good accuracy, suggesting a non-invasive option for early risk assessment.