Energy Cost of Walking in Relation to Impairment and Functional Independence After Stroke: The Role of Walking Speed.

Thirty stroke survivors in inpatient or outpatient rehabilitation at a single French hospital (Limoges) were assessed in two sessions within 14 days. In the indoor session they walked a 60-metre corridor and climbed/descended five floors (110 steps); in the outdoor session they walked a 120-metre grassy loop, a 4% uphill slope of about 200 metres, and the same slope downhill. All tasks were performed at each person's own comfortable pace. Oxygen consumption was measured with a portable breath-by-breath system, and the energy cost of walking was calculated as oxygen consumption divided by walking speed. The researchers then correlated energy cost with two clinical scores — the NIHSS (neurological deficit, 0–42) and the Barthel Index (daily independence, 0–100) — using rank correlations, both with and without adjusting for walking speed. No correction for multiple comparisons was applied; the authors label all results exploratory. Missing values were excluded (per-protocol analysis).

Walking speed ranged from 0.3 m/s (stairs) to 0.9 m/s (flat and outdoor tasks). Oxygen consumption was highest during stair ascent (mean 16.8 mL/kg/min) and lowest during stair descent (10 mL/kg/min). Energy cost was highest during stair ascent (1.24 mL/kg/m) and lowest during overground walking (0.24 mL/kg/m). Unadjusted correlations between energy cost and the NIHSS were significant for overground walking (ρ = 0.44), stair descent (ρ = 0.44), downhill walking (ρ = 0.38), and irregular-terrain walking (ρ = 0.59); stair ascent was borderline (ρ = 0.37, p = 0.05). Unadjusted correlations between energy cost and the Barthel Index were significant for overground (ρ = −0.41), stair descent (ρ = −0.38), uphill (ρ = −0.48), irregular terrain (ρ = −0.59), and downhill (ρ = −0.58). After adjusting for walking speed, none of these correlations remained significant for any task. Two participants had data excluded from specific tasks because they could not sustain the required three-minute duration.

If you are measuring the energy cost of walking in a stroke patient, the number you get is largely determined by how fast they walk. It does not independently tell you about their neurological deficit or their daily independence beyond what their gait speed already tells you. The authors do note that more demanding tasks (irregular terrain, slopes) showed stronger unadjusted correlations than flat-ground walking, so a simple corridor walk may understate a patient's real-world energetic burden. However, because the correlations all disappear once speed is accounted for, a high energy-cost number should not be read as evidence of a specific neurological problem or of poor ADL function that speed does not already capture.

Only 30 participants, all from one hospital, which limits generalisability and statistical power. The cohort had very mild deficits (NIHSS median 2) and near-complete independence (Barthel median 100), so the study says little about patients with moderate-to-severe impairment; the authors note a likely ceiling effect on the Barthel Index. Cross-sectional design: no causal inferences can be drawn. No correction for multiple comparisons across the six tasks and two clinical scales; the authors themselves flag the results as exploratory. Nine of the 30 participants used walking aids, which may have influenced speed, oxygen consumption, and energy cost. The NIHSS is a global neurological scale that includes non-motor domains; it may be less sensitive than a motor-specific measure (e.g., Fugl-Meyer) for capturing the deficits most relevant to walking energetics. Energy cost is mathematically defined as oxygen consumption divided by speed, so adjusting for speed in the same analysis is somewhat circular and complicates interpretation of any 'independent' effect. Two participants could not complete the required duration for certain tasks and had those data points excluded.

Funded by the Limoges University Hospital and the University of Limoges. The authors declare no conflicts of interest. No industry sponsor is named.

Do not read the significant unadjusted correlations (e.g., ρ = 0.59 for irregular-terrain walking) as evidence that the energy cost of walking independently reflects a patient's neurological deficit or ADL independence. Every one of those correlations became non-significant once walking speed was accounted for, and the authors themselves state that the associations are largely explained by gait velocity. In addition, the sample was small (n = 30), the deficits were mild, no multiple-comparison correction was applied, and the design is cross-sectional, so none of the individual correlations should be treated as a reliable clinical marker.