REF-718 · Hollander, Heidt, Van Der Zwaard, Braumann & Zech, 2017 · Medicine & Science in Sports & Exercise, 2017 Apr;49 · Paper · PubMed
Populations that have been barefoot their whole lives have wider feet, smaller hallux angles and less ankle dorsiflexion at footstrike
And no difference in injury rates
Key Findings
15 studies, n=8399 participants, comparing habitually barefoot with habitually shod locomotion.
Foot shape differs: habitually barefoot populations have wider feet (pooled effect size 0.55, 95% CI 0.06-1.05) but not shorter feet (-0.22, CI -0.51 to 0.08), and a lower pedobarographically measured hallux angle (-1.16, CI -1.64 to -0.68; limited evidence).
Biomechanics differ: reduced ankle dorsiflexion at footstrike in barefoot populations (pooled effect size -3.47, CI -5.18 to -1.76; limited evidence).
Injury rates do not differ: no differences in relative injury rates between habitually barefoot and habitually shod groups — but limited evidence for a different body-part distribution of musculoskeletal injuries, and more foot pathologies but fewer foot deformities and defects in habitually barefoot runners.
Authors' conclusion: only limited or very limited evidence exists for long-term effects of habitual barefoot locomotion on biomechanics or health outcomes, and no evidence exists for effects on motor performance.
Practical Application
Across 15 studies (n=8399), habitually barefoot populations had wider feet, a lower hallux angle and less ankle dorsiflexion at footstrike than habitually shod ones, and no difference in relative injury rates — though injuries were distributed differently and foot pathologies were more common; the evidence was limited to very limited. Even a lifetime of barefoot locomotion does not show fewer injuries overall.
Inference, not tested: this weakens the "but they've adapted over years" reply to the short-trial evidence (REF-716).
What a lifetime barefoot changes is foot shape. Inference, not tested: a wide forefoot or a straight hallux in an athlete raised barefoot is morphology, not pathology, and argues for shoe fit by actual foot width rather than by category.
This is not a transition study and cannot be used as one. Growing up barefoot and switching as an adult are different exposures; the switching cost is in REF-715.
Treat "motor performance" claims for barefoot training as unsupported — the review found no evidence either way.
Limitations
Observational comparisons of populations, not randomised exposure. Habitually barefoot and habitually shod groups differ in geography, surface, occupation, body composition and healthcare access; footwear cannot be isolated.
The authors themselves rate the evidence limited to very limited — this entry describes the shape of the literature more than a settled effect.
Inference, not tested: injury ascertainment across such populations is likely heterogeneous, which makes a true difference harder to detect.
Mixed running and walking exposure; not specific to trained runners under high load.
The randomised footwear evidence (REF-716) came later and should be read alongside it. The authors call for prospective designs.
How this entry was checked and changed
Content correction 2026-09-27
"Injury does not differ" -> no difference in relative injury rates, but a different body-part distribution and more foot pathologies (fewer deformities) in habitually barefoot runners — now stated.
Added the pooled effect sizes and "no shorter feet"; type now says meta-analysis.
Removed the invented "switching at 35" example; morphology and ascertainment points labelled as inference.