How walking differs in people with diabetic neuropathy

A systematic review and meta-analysis published in Systematic Reviews on 6 November 2026 by Sacco and colleagues from the University of São Paulo, Brazil. It was registered in advance (PROSPERO CRD420251076593) and searched six databases up to 11 March 2026.

It included 43 observational studies with 4,464 participants, and 37 studies were pooled in the meta-analysis. Participants were adults with type 1 or type 2 diabetes. Anyone with a current ulcer or an amputation was excluded. Only barefoot walking was included, and studies in shoes, insoles or orthoses were excluded. For joint motion and step and timing measures, only data collected at the person's own chosen walking speed were used.

The review didn't directly compare people with neuropathy against people with diabetes alone. Any difference between those two groups below comes from reading the two comparisons side by side.

Neuropathy was diagnosed differently across studies, using the 10 g monofilament (with different thresholds), vibration perception threshold above 25 V, the Michigan Neuropathy Screening Instrument, tuning fork, nerve conduction studies, and in one study physician diagnosis. Severity wasn't separated out.

Forefoot peak pressure was 126 kPa higher in people with neuropathy (95% CI 39 to 213). Studies varied enormously (I² 99.3%), and the prediction interval ran from −180 to +432 kPa. When whole-forefoot values weren't available, the authors used metatarsal-region values, such as the first metatarsal, instead.

Forefoot peak pressure was 70 kPa higher in people with diabetes without neuropathy (31 to 110), with similar variation between studies.

Midfoot peak pressure was 58 kPa higher with neuropathy (2 to 115), only just significant and highly variable.

Rearfoot and hallux peak pressures didn't differ significantly in either group. With neuropathy, the pressure–time integral (pressure accumulated over the step) was higher at the rearfoot, by 5.5 kPa·s (1.1 to 9.8), and at the hallux, by 3.0 kPa·s (2.2 to 3.8), and the hallux result was consistent across studies (I² 0%).

People with neuropathy walked 0.09 m/s slower (0.02 to 0.17). People with diabetes without neuropathy walked 0.10 m/s slower (0.05 to 0.16).

Stride length was 10 cm shorter with neuropathy (5 to 15 cm). In diabetes without neuropathy, the difference wasn't significant.

Stance was longer with neuropathy, by 0.08 seconds (0.01 to 0.15) and by 1.1 percentage points of the gait cycle (0.3 to 2.0).

Knee range of motion during stance was 4.45° smaller with neuropathy (2.59 to 6.32), from 5 studies that agreed closely (I² 6%).

Maximum hip extension was 4.23° less with neuropathy (3.16 to 5.30), from only 3 studies. Overall hip range of motion with neuropathy didn't differ significantly.

Ankle range of motion didn't differ significantly in either group, and neither did maximum dorsiflexion or maximum plantarflexion with neuropathy.

In diabetes without neuropathy, hip range of motion was 1.63° smaller (1.33 to 1.92).

Ankle plantarflexion moment was lower in both groups: by 0.05 Nm/kg with neuropathy and 0.06 Nm/kg in diabetes without neuropathy.

Ankle power generation was 0.38 W/kg lower with neuropathy (0.06 to 0.70), and ankle power absorption was 0.11 W/kg higher (3 studies).

Hip flexion moment was lower in both groups, and knee extension moment was slightly lower with neuropathy (0.03 Nm/kg). Ground reaction force differences were very small.

The timing of peak tibialis anterior and lateral gastrocnemius activity didn't differ significantly with neuropathy. Vastus lateralis peaked 1.5% of the support phase later, a borderline result (p = 0.047). Lower tibialis anterior and gastrocnemius amplitude was described in individual studies but not pooled.

Study quality was rated with a modified Downs and Black checklist, and the authors describe most studies as good quality. The paper gives two different maximum scores for that checklist (17 and 18). There was no GRADE certainty rating. Very high variation between studies (I² above 90%) affected most of the plantar pressure comparisons, and a test for publication bias flagged 9 of 60 comparisons.

Funded by Brazilian public research agencies (FAPESP and CNPq). The authors declared no conflicts of interest. The lead author co-authored some of the included studies and, according to the paper, stayed out of quality rating and data extraction.

None of the gait differences were adjusted for walking speed. People with neuropathy walked slower, and slower walking on its own shortens stride and reduces joint ranges. The authors say the differences can't be put down to neuropathy alone.

The 126 kPa forefoot figure is an average across very different studies. The direction fits with offloading the forefoot in neuropathic feet. The size can't be applied to an individual patient, given a prediction interval from −180 to +432 kPa.

Changes also showed up in people classed as having diabetes without neuropathy, including 70 kPa higher forefoot pressure and slower walking. That could mean gait changes start before neuropathy is detectable, or that some had neuropathy their screening test missed. The review can't separate the two.

Every study was barefoot. Most patients spend their walking day in shoes, and footwear and orthosis studies were excluded.

All included studies were cross-sectional. They show differences at one point in time and can't show that neuropathy caused them or that they lead to ulcers.

The abstract's conclusion mentions “longer steps”. The pooled results show shorter strides with neuropathy.

Across 37 pooled studies, people with diabetic neuropathy walked more slowly than healthy controls, with shorter strides, higher forefoot pressure, less knee motion and less hip extension. Ankle motion and the timing of tibialis anterior and gastrocnemius activity didn't differ significantly. Most results varied widely between studies and none were adjusted for speed. It seems like the higher forefoot pressure is the most practice-relevant finding, even though its size is uncertain.