Influence of Support Surface Stability on Changes in Quadriceps Femoris Muscle Thickness During Progressive Squat Exercise in Patients with Stroke: A Randomized Controlled Trial — Thicker is not stronger

Clinical Question (PICO)

In selected inpatients 6 months to less than 1 year after stroke, with sufficient balance and cognition to follow a supervised squat program, does performing otherwise identical progressive squats on a balance pad rather than a stable floor produce greater affected-side quadriceps muscle-thickness increases after 4 weeks? The outcomes are four individual muscle measurements—not walking, independence, strength, or recurrent stroke. No subgroup treatment-effect analyses are reported.

Bottom Line

Unstable-surface squats produced greater increases in rectus femoris thickness: +0.78 mm between groups (95% CI 0.34–1.23; Holm-adjusted P = 0.006) and vastus medialis thickness: +0.44 mm (95% CI 0.09–0.79; adjusted P = 0.046). The abstract explicitly cautions that these changes do not demonstrate improved strength or functional performance. The other two quadriceps components did not show statistically significant between-group differences.

(Primary full text; Methods, Results, and Table 2.)

Design

Trial type: Single-center, open-label, parallel-group randomized controlled trial; outcome measurement was not blinded.
N: 32 randomized after 36 were screened.
- Unstable support-surface group: 16.
- Stable support-surface group: 16.
Randomization: Computer-generated random-number table, 1:1 allocation. The same principal investigator generated the sequence, enrolled participants, and assigned treatment; no allocation concealment was used. No stratification is described.
Setting: C Rehabilitation Hospital, Seoul, Republic of Korea.
Enrollment/study period: First enrollment June 26, 2023; study completed August 28, 2023. Published in September 2026; this is a look-back selection, not a newly conducted 2026 trial.
Follow-up: Baseline and completion of the 4-week intervention; no later durability assessment. A separate mean follow-up is not reported.
Analysis: All 32 randomized participants completed treatment and were analyzed; no post-randomization losses or imputation were reported. Independent-samples t tests compared change scores, with Welch correction when needed, 95% CIs, Hedges’ g, and Holm–Bonferroni adjustment across four comparisons. Baseline-adjusted ANCOVA was a post hoc sensitivity analysis.
Primary outcome: Change from baseline in ultrasound-measured thickness, in millimeters, of affected-side rectus femoris, vastus intermedius, vastus lateralis, and vastus medialis. No single primary muscle was designated in the published analysis. Prospective public registration is unavailable to confirm the outcome hierarchy.

Population

Inclusion Criteria

Stroke onset 6 months to less than 1 year previously.
Functional Ambulation Category ≥1.
Berg Balance Scale ≥41.
Korean Mini-Mental State Examination ≥24.
No significant visual, auditory, or vestibular impairment.
Ability to understand procedures and provide informed consent.

Exclusion Criteria

Orthostatic hypotension.
Robot-assisted therapy in addition to conventional therapy.
Scheduled discharge during the study period.
Previous surgery for a bone fracture.

Four patients were excluded before randomization: COVID-19 infection, peritonitis, guardian refusal, and stent-insertion surgery, one each. These screening events are not treatment-emergent adverse events.

Baseline Characteristics

Mean age: 65.31 ± 11.56 years with unstable support versus 62.81 ± 8.48 with stable support.
Male/female: 11/5 versus 8/8.
Right/left affected side: 8/8 in each group.
Mean body weight: 61.31 ± 12.09 versus 68.19 ± 13.35 kg.
Exact time since stroke, stroke subtype, lesion location, and actual baseline FAC/BBS/MMSE-K scores were not included in the reported muscle-thickness dataset; eligibility cutoffs must not be mistaken for observed baseline means.

Interventions

Unstable surface: Progressive squats on a balance pad, 30 minutes/session, five sessions/week for 4 weeks. Five sets of 10 repetitions during weeks 1–2; six sets of 10 during weeks 3–4; 60 seconds of rest between sets.
Stable surface: The identical schedule and repetitions on a flat support surface.
Concurrent care in both arms: Neurodevelopmental treatment, 30 minutes twice daily, 10 sessions/week; routine occupational therapy on the same frequency and duration, for 4 weeks.
Safety setup: Investigator on the affected side within arm’s reach, with a treatment table immediately behind the participant, positioned to permit sitting at 90° knee flexion. This was supervised inpatient training, not unsupervised home exercise.

Outcomes

Primary Outcomes — between-group change in muscle thickness at 4 weeks:

Values below are unstable minus stable support. The P values are Holm-adjusted unless otherwise stated. These are continuous differences, so HR/OR/RR are not applicable.

Muscle	Unstable-group change, mean ± SD	Stable-group change, mean ± SD	Difference (95% CI), mm	Hedges’ g	Unadjusted P	Adjusted P
Rectus femoris	1.27 ± 0.79 mm	0.49 ± 0.33 mm	0.78 (0.34 to 1.23)	1.27	0.002	0.006
Vastus intermedius	0.91 ± 1.53 mm	0.52 ± 0.63 mm	0.39 (−0.45 to 1.23)	0.32	0.353	0.353
Vastus lateralis	0.98 ± 0.66 mm	0.62 ± 0.50 mm	0.36 (−0.06 to 0.78)	0.60	0.090	0.181
Vastus medialis	1.03 ± 0.60 mm	0.58 ± 0.31 mm	0.44 (0.09 to 0.79)	0.90	0.015	0.046

The vastus medialis difference is the article’s reported estimate; subtracting the rounded group means gives a slightly different value. Do not silently replace the estimate with arithmetic on rounded inputs. The reported CIs are individual 95% intervals, not described as simultaneous multiplicity-adjusted intervals.

Secondary Outcomes / Sensitivity Analyses:

No gait-speed, falls-as-an-efficacy-endpoint, independence, strength, or quality-of-life outcomes are reported in this analysis.
Post hoc ANCOVA: Rectus femoris adjusted difference 0.78 mm (95% CI 0.33–1.23; Holm-adjusted P = 0.005); vastus medialis 0.45 mm (0.10–0.80; P = 0.042).
ANCOVA remained nonsignificant for vastus intermedius: 0.41 mm (−0.46 to 1.27; P = 0.346), and vastus lateralis: 0.35 mm (−0.07 to 0.77; P = 0.205).
Thickness increased within both groups for all four muscles, but without a non-exercise control this does not isolate squat exercise from recovery or concurrent rehabilitation.
Subgroups: None reported. The four muscles are distinct outcomes, not patient subgroups.

Adverse Events / Safety:

No falls or other intervention-related adverse events were observed in either arm during 4 weeks, and all participants completed the intervention. With only 16 participants exposed to unstable-surface training, zero observed events is not proof of safety in patients with poorer balance, cognitive impairment, or unsupervised conditions. Mortality, symptomatic intracranial hemorrhage, and recurrent vascular events were not reported as measured safety outcomes.

Figures

Participant flow — *Figure 1.* CONSORT flow diagram of participant enrollment, allocation, follow-up, and analysis.

Quadriceps ultrasound measurement sites — *Figure 2.* Representative RUSI images of quadriceps femoris muscle-thickness measurements. (A) Rectus femoris and vastus intermedius; (B) vastus lateralis; and (C) vastus medialis. Each panel represents a distinct ultrasound acquisition obtained at the corresponding anatomical measurement site. Muscle thickness was measured at the central vertical caliper line.

Rectus femoris thickness changes — *Figure 3.* Changes in rectus femoris muscle thickness from baseline to post-intervention in the unstable and stable support-surface groups. * p < 0.05 for within-group pre–post comparisons; † Holm-adjusted p < 0.05, when present, for between-group comparisons of change scores.

Vastus intermedius thickness changes — *Figure 4.* Changes in vastus intermedius muscle thickness from baseline to post-intervention in the unstable and stable support-surface groups. * p < 0.05 for within-group pre–post comparisons.

Vastus lateralis thickness changes — *Figure 5.* Changes in vastus lateralis muscle thickness from baseline to post-intervention in the unstable and stable support-surface groups. * p < 0.05 for within-group pre–post comparisons.

Vastus medialis thickness changes — *Figure 6.* Changes in vastus medialis muscle thickness from baseline to post-intervention in the unstable and stable support-surface groups. * p < 0.05 for within-group pre–post comparisons; † Holm-adjusted p < 0.05, when present, for between-group comparisons of change scores.

(Primary full text; Methods, Results, and Table 2.)

Criticisms

The surrogate gap is explicit: A thicker muscle is not necessarily stronger, and neither measurement establishes better walking or independence. No study-specific minimal detectable change or clinically important thickness threshold was established.
Selection and assessment bias: Allocation was not concealed; one unblinded investigator enrolled, allocated, treated, and measured participants. Averaging three ultrasound measurements does not remove systematic assessor bias.
Registration was late: CRIS KCT0012481 was registered August 19, 2026—after the 2023 study had finished. The authors attribute the delay to administrative misunderstanding and report prospective ethics approval, but that does not substitute for prospective public outcome registration.
Small, selected population: One center, 16 participants per arm, relatively preserved balance and cognition, and only 4 weeks of follow-up. Actual stroke characteristics were unavailable for the analysis.
Multiplicity handled, measurement reliability unresolved: Holm adjustment is a strength, and post hoc ANCOVA is concordant. Nevertheless, trial-specific intra-rater reliability and minimum detectable changes were not measured. Nonsignificant results for two muscles do not establish equivalence.
No exercise-versus-no-exercise comparison: Both arms received substantial conventional therapy. The randomized contrast is the support surface, not the overall value of squats or of rehabilitation.

Funding

The article states that there was no external funding; no external sponsor or funder role is reported. The authors declare no conflicts of interest. Registration: KCT0012481, CRIS, retrospectively registered August 19, 2026; no NCT identifier is reported. Ethics approval: Sahmyook University, SYU 2023-05-013-002, June 20, 2023.

The paper

Authors. Hui Ju Nam and Ga-Yeon Kim.
Title. Influence of Support Surface Stability on Changes in Quadriceps Femoris Muscle Thickness During Progressive Squat Exercise in Patients with Stroke: A Randomized Controlled Trial.
Journal. Medicina (Kaunas), 62(9):1749.
Year. 2026; published September 11, 2026.
DOI. 10.3390/medicina62091749
PMCID. PMC13609271
PMID. 42796355.
License. Creative Commons Attribution 4.0. Figures reproduced without substantive modification.

Deep Dive — click to expand

What this is

This is a rehabilitation mechanism study, not a disability-outcome trial. Although published in 2026, it reports a supervised inpatient exercise program conducted in 2023. Thirty-two selected stroke survivors completed the same squat dose on different surfaces; the balance-pad arm gained more thickness in two of four quadriceps components. The useful question is whether a low-cost change in exercise conditions deserves further testing—not whether foam pads have already improved stroke recovery. Numerical results below come from the paper; implementation suggestions are editorial interpretation rather than guideline recommendations.

(Primary full text; Methods, Results, and Table 2.)

1. Shadow Audit

The headline can make a selective imaging response sound like a whole-patient recovery effect. Yet the advantage was 0.78 mm in rectus femoris and 0.44 mm in vastus medialis, with no measured strength or walking benefit. Meanwhile, vastus lateralis had an estimated effect of 0.36 mm with a CI spanning −0.06 to 0.78: neither demonstrated superiority nor established absence of an effect. The two positive muscles do not validate a global “stronger quadriceps” claim.

The paper is commendably direct about the larger shadow: the same person knew assignment and controlled enrollment, treatment, and the ultrasound probe. When effects are fractions of a millimeter and trial-specific reliability is unknown, an assessor’s expectations can matter. Post hoc ANCOVA addresses baseline thickness, not unblinded measurement or unconcealed allocation. The three-year gap between conduct and registration also leaves prospective outcome selection unverifiable from the public record.

2. Inversion Engine

To erase the point-estimate advantage entirely, an intervention-correlated measurement bias would have to account for 0.78 mm in rectus femoris or 0.44 mm in vastus medialis. To reach the lower edge of the reported individual 95% CIs requires only 0.34 mm and 0.09 mm, respectively. These are sensitivity thought experiments, not estimates of actual bias and not reanalyses of the Holm-adjusted tests.

The stronger opposite clinical conclusion—stable surfaces produce better patient outcomes—would not require reversing these thickness results at all. It would require demonstrating better strength, walking, or safety despite smaller imaging changes. None of those comparative functional measurements exists here. That is the central inversion: the clinical question remains open even if every reported ultrasound result is exactly correct.

3. Second-Order Catalyst

The first plausible adopter is a supervised inpatient rehabilitation service resembling the Seoul study site, not an acute stroke call team or an unsupervised home program. Editorial inference: therapists could consider a carefully monitored local feasibility evaluation for patients meeting the trial’s balance and cognition thresholds, while preserving therapist guarding and the nearby treatment table. Its first deliverable should be feasibility, reproducible measurement, and a patient-centered outcome—not a hospital-wide order that everyone squat on foam.

Four weeks is the observed response horizon, not an implementation deadline or proof of durability. Adding a balance demand may also increase supervision requirements. The study’s fixed exercise time cannot tell us whether the surface change saves staff time or improves throughput.

4. Asymmetric Leverage

The intervention changes a surface rather than purchasing a complex device, so the financial barrier may be modest; this paper does not provide a cost-effectiveness analysis. Calculated from the protocol, each participant was prescribed 20 squat sessions, totaling 600 minutes, or 10 hours, and 1,100 repetitions over 4 weeks. Those are prescribed totals, not independently measured adherence. Both arms had the same dose, so the putative leverage is a surface modification within existing therapy—not extra treatment exposure.

The upside could be a scalable improvement if the imaging advantage proves to translate into meaningful function. The downside is mistaking an inexpensive tool for a free decision: guarding, selection, and falls surveillance still consume attention. The study is far too small to quantify that trade-off reliably. Low acquisition cost does not reduce the evidence required for a clinical-benefit claim.

5. Paradigm Destroyer

The habit this paper should kill is using muscle size as a shortcut for functional recovery. It does not kill stable-surface squats, conventional rehabilitation, or the need for safety assessment.

Proposed two-sentence protocol clarification: “Record muscle-thickness changes as an imaging outcome, not as evidence of improved walking or strength; any evaluation of unstable-surface squats must also measure patient-centered function. Restrict initial evaluation to appropriately selected, supervised patients and retain guarding and adverse-event monitoring rather than extrapolating this small trial to all stroke survivors.”

MVP — Minimum Viable Proof

The minimum existing statement is narrow: among 32 selected patients, the unstable surface produced an additional 0.78 mm rectus femoris and 0.44 mm vastus medialis thickness increase at 4 weeks, surviving the authors’ multiplicity correction. That establishes a reported structural signal, not a reason to change bedside functional claims.

The minimum practice-changing proof would be an adequately powered, prospectively registered randomized comparison with concealed allocation, blinded assessment, a prespecified clinically meaningful functional outcome, and sufficient safety follow-up. Ultrasound changes should exceed a trial-specific measurement-error threshold. No numerical functional-benefit threshold can be derived from this paper because those outcomes were not measured.

Best Combination

Read this as one step in a causal chain: muscle loss after stroke; muscle structure associated with strength; a surface modification changing structure; and finally better function. The article cites English et al.’s systematic review of post-stroke muscle loss and Akazawa et al.’s cross-sectional study relating quadriceps characteristics to strength. Those references support the biological rationale as discussed by the authors, not proof that this intervention completes the chain. Their findings are contextual here, not independently reappraised evidence.

A rigorous next study would combine the present surface-controlled design with independent assessors, validated ultrasound reproducibility, actual stroke-severity characterization, and prespecified functional testing. Retain the article’s useful elements—equal exercise dose, all randomized participants accounted for, and multiplicity control—but do not combine observational structure–strength associations with this surrogate RCT as though the combination proves improved mobility.

Overvalue Warning

Do not translate large standardized effects into large clinical benefits. Hedges’ g of 1.27 and 0.90 describe muscle-thickness change relative to variability; they are not disability effect sizes or percentages of walking recovery. The absolute effects remain submillimeter, and clinical importance is unestablished.
Do not translate zero observed falls into permission for unsupervised use. Only 16 patients received unstable-surface training, for 4 weeks, under guarding and restrictive eligibility. Neither safety in more impaired patients nor persistence of benefit is established.

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.