Dose-response association between walking speed and all-cause mortality: A systematic review and meta-analysis of cohort studies.
Level 3 - non-randomized controlled study
Systematic review and meta-analysis of observational cohort studies
PubMed 39133765 · doi:10.1080/02640414.2024.2390302
What was done
Authors searched PubMed, Web of Science, Embase, and Cochrane Library through September 2023 for cohort studies evaluating walking speed and all-cause mortality. They meta-analyzed hazard ratios (HR) comparing the fastest versus slowest walking-speed categories and conducted dose-response meta-analyses using multivariate models (mvmeta). Subgroup analyses evaluated sex, age (>65 years vs. younger), and measurement method (timed speed test vs. self-report). A total of 13 cohort studies with 530,841 participants were included, of which 11 provided data for dose-response analysis.
What was found
Individuals in the fastest walking-speed category had a 43% lower risk of all-cause mortality compared to those in the slowest category (HR = 0.57, 95% CI 0.48–0.66). An inverse linear dose-response relationship was observed overall, with each 0.1 m/s increase in walking speed associated with a 6% reduction in mortality risk (HR = 0.94, 95% CI 0.92–0.96). For participants older than 65 years, the dose-response relationship was non-linear, whereas linear dose-response relationships were observed for both timed objective tests and self-reported walking speed.
Why it matters
This meta-analysis quantifies the continuous prognostic value of walking speed on survival, reinforcing gait speed as a simple, informative biomarker of vitality and mortality risk.
Limits
The findings are derived entirely from observational cohort studies and cannot establish causality; underlying subclinical disease could drive slower walking speeds (reverse causality). The abstract does not report study-specific follow-up times, adjustment variables, or the exact shape of the non-linear curve in older adults, and self-reported walking speed carries subjective reporting bias.