The Time Is Brain Review

Square-Step Exercise After Ischemic Stroke: A Two-Ward Trial

· DOI: 10.1186/s12877-026-07990-x

stroke deep-dive rehabilitation falls

Editorial illustration of rehabilitation after stroke
AI-generated editorial illustration. Generated for editorial use.

Clinical Question (PICO)

In ambulatory older adults recovering from ischemic stroke and concerned about falling, does 12 weeks of square-step exercise added to usual care, compared with usual care alone, improve fear of falling, balance, daily function, or estimated fall risk? The paper reports no prespecified subgroup effects.

Bottom Line

In this 2026 single-center trial, the 12-week mean Short Falls Efficacy Scale–International (SFES-I) score was 8.64 with exercise versus 9.86 with usual care (P = 0.009); lower scores indicate less concern. Only two wards were randomized, and no cluster-adjusted analysis was reported, so the apparent precision and causal interpretation are fragile. A fall-risk scale is not a count of actual falls.

Design

  • Trial type: Single-center, two-ward cluster randomized controlled trial; assessors and analysts were reportedly blinded. Participants knew their own intervention, making meaningful participant masking uncertain.
  • N: 62 enrolled, 31 per ward; 57 completed (28 exercise, 29 control).
  • Randomization: One ward allocated to each arm by sealed-envelope draw; patients were consecutively enrolled within wards after ward assignment.
  • Setting: Two neurology units at a tertiary hospital in Mudanjiang, China.
  • Enrollment: January–September 2024; published 2026.
  • Follow-up: 12 weeks, spanning hospitalization and home rehabilitation.
  • Analysis: Reported independent and paired t-tests or nonparametric alternatives; results reflect completers. No adjustment for ward clustering or intention-to-treat analysis was described.
  • Primary outcome: SFES-I score at 12 weeks; seven activities, each scored 1–4, with higher scores indicating greater concern about falling.

Population

Inclusion criteria

  • Ischemic stroke confirmed by CT or MRI; age at least 60 years; SFES-I greater than 7.
  • Modified Barthel Index at least 60 and bilateral lower-limb strength at least grade 4.
  • Patient or cohabiting family member able to use a smartphone or comparable device; consent.

Exclusion criteria

  • Severe comorbidities, psychiatric disorders, or language, writing, visual or hearing impairments; participation in another exercise program.
  • Deterioration, death, readmission or withdrawal were listed as dropout criteria.

Baseline characteristics

  • Mean age 69.29 years in the exercise ward and 67.24 years in the control ward; male 19/28 and 19/29 among completers.
  • Mean NIHSS 1.68 and 1.62, respectively, among completers: predominantly mild deficits and preserved mobility.

Interventions

  • Square-step exercise plus routine care: Three 45–50 minute sessions weekly for 12 weeks, with warm-up, progressive stepping patterns and cool-down. Training started in hospital after clearance for ambulation and continued at home with family instruction, videos, logs and remote review.
  • Routine care: Neurology nursing, fall-prevention education and rehabilitation guidance, with telephone follow-up every four weeks.

Outcomes

Primary outcome — fear of falling: Among completers, 12-week SFES-I was 8.64 ± 1.73 versus 9.86 ± 1.66 (P = 0.009). Mean change was −6.68 versus −3.93 points; the between-group change comparison was P = 0.002. The table reports a 95% CI of −1.39 to −0.31 without clearly labeling its metric; it should not be presented as a points-based confidence interval. No subgroup estimates were reported.

Secondary outcomes:

  • Berg Balance Scale: 43.71 ± 4.82 versus 40.76 ± 4.66 at 12 weeks (P = 0.022); higher is better. Change comparison P = 0.001.
  • Modified Barthel Index: 94.89 ± 3.29 versus 93.72 ± 4.22 (P = 0.250); no statistically significant between-group post-treatment difference. The table's stated changes do not consistently match its baseline and final means.
  • Morse Fall Scale: 23.21 ± 13.49 versus 29.48 ± 9.10 (P = 0.046) at 12 weeks, but the between-group change comparison was P = 0.314. The table's reported changes do not consistently match the displayed baseline and follow-up means; these are risk scores, not observed fall rates.

Adverse events / safety: The discussion says no falls occurred during the intervention, but the paper does not provide a clear comparative adverse-event table or establish between-arm fall counts. Three exercise participants and two controls did not complete follow-up; exercise-related harms were not comprehensively quantified.

Figures

Example square-step pattern
Figure 1. Example of step pattern diagram. Xu et al.; CC BY-NC-ND 4.0. Click to expand.
Participant flow chart
Figure 2. Participant flow chart; IG, intervention group; CG, control group. Xu et al.; CC BY-NC-ND 4.0. Click to expand.

Criticisms

  • Only one cluster per arm: ward effects and treatment are inseparable. Patient-level tests cannot recover valid ward-level randomization uncertainty; the sample calculation did not report a cluster correction.
  • Allocation preceded consecutive enrollment, raising selection-bias concerns; the intervention included extra attention and caregiver contact.
  • Five participants were excluded from the reported completer analyses; missing-data assumptions and intention-to-treat estimates were not reported.
  • The 2026 trial registration was retrospective, after 2024 enrollment. Outcome reporting cannot be assumed prospectively locked.
  • Several displayed change values and confidence-interval labels are internally inconsistent; lack of multiplicity adjustment and the nominal P values warrant caution.
  • Mild strokes, preserved leg strength and smartphone access restrict applicability. Remote balance assessments may not be equivalent to in-person testing.

Funding

The paper lists the 2022 Basic Research Operating Expenses Project for Provincial Universities in Heilongjiang Province (2022-KYYWF-0703), Heilongjiang Provincial Health Commission Research Project (20241414050036), and Heilongjiang Provincial Social Sciences Planning Special Project (25EDH001). The funders' specific role was not stated. Registry: ChiCTR2600127548, retrospectively registered on July 2, 2026.

The paper

  • Authors. Xu X, Liu Y, Xu W, Lu J.
  • Title. Effects of square-step exercise on fear of falling in older patients with ischemic stroke: a cluster randomized controlled trial.
  • Journal. BMC Geriatrics.
  • Year. 2026.
  • DOI. 10.1186/s12877-026-07990-x
  • PMCID. PMC13673687. Figures distributed under CC BY-NC-ND 4.0.
Deep Dive — click to expand

What this is

A pragmatic rehabilitation hypothesis tested in 2024 and published in 2026: add supervised, progressively complex stepping to usual stroke care, then measure concern about falling after 12 weeks. The observed primary outcome favors exercise, but the trial randomized two wards, not 62 independent treatment units.

1. Shadow Audit

The control group also improved substantially: SFES-I fell 3.93 points versus 6.68 in the exercise group. Both final means remained above 7, the paper's threshold for concern. A better risk score and an assertion of no falls are not evidence that the intervention prevents falls; the between-group Morse change test was P = 0.314, and reported change figures contain discrepancies.

2. Inversion Engine

The reported 12-week SFES-I difference is −1.22 points in favor of exercise. A ward-level imbalance, selection effect or attention effect of at least 1.22 points in that direction would be sufficient to explain the observed final-score contrast without a true exercise-specific benefit. This is a sensitivity illustration, not an estimated bias or a cluster-adjusted confidence interval; two randomized wards cannot estimate ward-to-ward variation reliably.

3. Second-Order Catalyst

Rehabilitation therapists could first incorporate supervised stepping as an optional, individualized activity for medically stable, ambulatory patients who resemble this sample. A unit considering adoption should record staff time, caregiver support, adherence and actual falls, rather than interpreting a low Morse score as a demonstrated fall-prevention effect.

4. Asymmetric Leverage

A mat and progressive stepping may be inexpensive, but the tested package also involved demonstration, monitoring, family training, electronic check-ins and follow-up. The payoff is potentially meaningful for patients avoiding movement because of fear; the hidden cost is therapist and caregiver capacity. Neither cost-effectiveness nor a number needed to treat can be derived from these data.

5. Paradigm Destroyer

The useful challenge is to treat fear of falling as a rehabilitation target, not merely a label on a risk screen. A sensible next-morning protocol would assess fear and walking safety, then offer a therapist-supervised stepping option when appropriate. Track patient-centered outcomes and falls prospectively; do not replace established individualized fall-prevention measures on this trial alone.

MVP — Minimum Viable Proof

A larger prospective trial with multiple independently randomized wards or individual randomization, concealed enrollment, cluster-appropriate analysis and a prespecified meaningful SFES-I threshold should show a sustained benefit without increased falls or injury. An actual falls endpoint requires sufficient follow-up and event ascertainment; this study cannot supply that proof.

Best Combination

Combine this hypothesis-generating trial with the paper's cited exercise and balance literature, then assess the totality of randomized evidence and local rehabilitation resources before changing a pathway. The present study adds feasibility in a selected older ischemic-stroke population, not definitive comparative effectiveness.

Overvalue Warning

Do not mistake nominal patient-level P values for reliable cluster-randomized inference with one ward per group. Do not convert improvement on a fall-risk scale into a proven reduction in falls or a guarantee of safety for patients with severe deficits.

Educational use notice. The Time Is Brain Review is dedicated to medical education. Figures from published articles are embedded for teaching purposes, with attribution to the original publication. We do not claim copyright over any embedded figure; all rights remain with the original publisher and authors under the article's published license. If you are a rights holder and would like a figure removed, please contact the editorial team. See our editorial notice for the full policy.