Skip to main content
Frontiers in Medicine logoLink to Frontiers in Medicine
. 2026 Nov 3;13:1979101. doi: 10.3389/fmed.2026.1979101

Dose-response relationship between isokinetic strength training intensity and lower limb spasticity improvement in stroke patients: a retrospective cohort study

Yang Zhang 1,†, Weigui Zhu 1,*,†
PMCID: PMC13774213  PMID: 43047527

Abstract

Background

Post-stroke spasticity (PSS) significantly impairs motor function recovery and quality of life in stroke survivors. While isokinetic strength training has demonstrated efficacy in neurological rehabilitation, the optimal training intensity for spasticity management remains unclear. This study aimed to investigate the dose-response relationship between different isokinetic training intensities and lower limb spasticity improvement in stroke patients.

Methods

This retrospective cohort study analyzed medical-record data from 392 stroke patients with lower limb spasticity who received rehabilitation therapy at our institution between January 2021 and December 2023. Patients were categorized into four groups based on isokinetic training intensity: control (standard rehabilitation only, n = 98), low-intensity (30% maximal voluntary contraction [MVC], n = 98), moderate-intensity (50% MVC, n = 98), and high-intensity (70% MVC, n = 98). The primary outcome was change in Modified Ashworth Scale (MAS) scores at 8 weeks. Secondary outcomes included Fugl-Meyer Assessment Lower Extremity (FMA-LE), Berg Balance Scale (BBS), 10-Meter Walk Test (10MWT), and isokinetic peak torque measurements. Covariate-adjusted and sensitivity analyses were performed to address confounding by indication.

Results

At 8 weeks, all groups demonstrated significant improvements from baseline (all P < 0.001). The high-intensity group achieved the greatest MAS reduction (1.07 ± 0.36 points), followed by moderate-intensity (1.02 ± 0.29), low-intensity (0.69 ± 0.48), and control (0.20 ± 0.45) groups (F = 95.789, P < 0.001, η2 = 0.425). These associations remained significant after adjustment for baseline covariates (all adjusted P < 0.001). The difference in MAS reduction between the moderate- and high-intensity groups was not statistically significant (1.02 versus 1.07 points, P = 0.813). Linear dose-response relationships were observed for all outcomes (all R2 > 0.40, P < 0.001). Between-group comparisons revealed significant differences between all training intensities for FMA-LE improvement (high: 7.23 ± 1.92; moderate: 5.64 ± 1.89; low: 3.48 ± 1.92; control: 1.27 ± 1.72 points; P < 0.001).

Conclusions

This study demonstrates a significant dose-response relationship between isokinetic strength training intensity and lower limb spasticity improvement in stroke patients. Higher training intensities (50%–70% MVC) were associated with superior outcomes across all measures, although no additional antispasticity benefit was demonstrated at 70% versus 50% MVC, suggesting that structured progressive resistance training may be incorporated into comprehensive post-stroke rehabilitation programs.

Keywords: dose-response, isokinetic training, lower extremity, motor function, muscle tone, post-stroke spasticity, stroke rehabilitation

Introduction

Stroke remains a leading cause of long-term disability worldwide, with approximately 80% of survivors experiencing motor impairment affecting their functional independence and quality of life (1). Among the various sequelae of stroke, post-stroke spasticity (PSS) affects 25%–43% of patients within the first year following the cerebrovascular event, presenting a significant barrier to motor recovery and rehabilitation outcomes (2, 3). Lower limb spasticity particularly impairs ambulation, balance, and activities of daily living, contributing to reduced mobility and increased fall risk (4, 5).

The pathophysiology of PSS involves complex alterations in the central nervous system, including loss of descending inhibitory control from the upper motor neurons, enhanced spinal reflexes, and secondary changes in muscle properties (6). These maladaptive changes result in velocity-dependent increases in muscle tone, which subsequently lead to muscle shortening, contractures, and further functional deterioration (7–9). Consequently, effective management of spasticity represents a critical component of comprehensive stroke rehabilitation (10, 11).

Current therapeutic approaches for PSS encompass pharmacological interventions, physical modalities, and various rehabilitation techniques (12). Among these, strength training has emerged as a promising strategy, challenging the traditional belief that resistance exercise might exacerbate spasticity (13). Recent evidence suggests that properly prescribed strength training not only improves muscle force generation but may also contribute to spasticity reduction through neuroplastic mechanisms. Notably, Engardt et al. demonstrated as early as 1995 that isokinetic quadriceps strength training did not exacerbate spasticity in patients with stroke (14–16).

Isokinetic exercise, characterized by constant angular velocity throughout the range of motion with accommodating resistance, offers several theoretical advantages for stroke rehabilitation (17, 18). This modality allows precise quantification of training parameters, enables safe maximal muscle activation, and provides objective assessment of muscle performance (19). Several randomized controlled trials have demonstrated the efficacy of isokinetic training in improving motor function following stroke (20, 21). However, the optimal training intensity for achieving spasticity reduction remains inadequately characterized.

The concept of dose-response relationships in rehabilitation medicine has gained considerable attention in recent years (22). Understanding how different training intensities affect clinical outcomes is essential for developing evidence-based rehabilitation protocols that maximize therapeutic benefits while ensuring patient safety (23). Previous studies examining extracorporeal shockwave therapy and botulinum toxin injections have demonstrated dose-dependent effects on spasticity management (24, 25). However, comparable evidence regarding isokinetic strength training intensity is notably limited.

This study aimed to investigate the dose-response relationship between different intensities of isokinetic strength training and lower limb spasticity improvement in stroke patients. We hypothesized that higher training intensities would be associated with greater reductions in spasticity and superior improvements in motor function, balance, and gait performance (26). The findings from this investigation may provide valuable guidance for clinical decision-making in stroke rehabilitation.

Methods

Study design and setting

This retrospective cohort study analyzed clinical data from stroke patients who received inpatient rehabilitation at the Department of Rehabilitation Medicine, Jiangyan TCM Hospital of Taizhou City, between January 2021 and December 2023. The study was conceived and initiated in 2024, after all treatments and outcome assessments had been completed, and data were abstracted retrospectively from hospital medical records; both the exposure (training intensity delivered during routine care) and the outcomes had therefore already occurred when the study began, consistent with the definition of a retrospective cohort study. The study protocol was approved by the Ethics Committee of Jiangyan TCM Hospital of Taizhou City (Approval No.: 2024-012-017), and the requirement for individual informed consent was waived due to the retrospective nature of the analysis. This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Participants

Patients were included if they met the following criteria (1): diagnosis of first-ever ischemic or hemorrhagic stroke confirmed by computed tomography or magnetic resonance imaging; (2) age between 35 and 80 years; (3) time since stroke onset between 1 and 24 months; (4) presence of lower limb spasticity with Modified Ashworth Scale (MAS) score ≥1 in the affected limb (27); (5) ability to maintain sitting balance independently; (6) completion of the 8-week rehabilitation program; and (7) complete medical records available for analysis.

Exclusion criteria included: (1) recurrent stroke or transient ischemic attack; (2) severe cognitive impairment (Mini-Mental State Examination score <20); (3) severe aphasia preventing comprehension of instructions; (4) unstable medical conditions (uncontrolled hypertension, cardiac arrhythmia, severe diabetes complications); (5) musculoskeletal conditions affecting the lower limbs (joint replacement, fractures, severe osteoarthritis); (6) concurrent use of oral antispasticity medications or botulinum toxin injections during the study period; and (7) inability to tolerate isokinetic testing procedures.

Group assignment

Patients were retrospectively categorized into four groups based on the isokinetic training intensity prescribed during their rehabilitation program: control group (standard rehabilitation without specific isokinetic training), low-intensity group (30% of maximal voluntary contraction [MVC]), moderate-intensity group (50% MVC), and high-intensity group (70% MVC). The training intensity was determined by the treating physician based on clinical assessment of individual patient tolerance and functional capacity at admission. This stratification reflects the routine practice of our department, in which isokinetic training is prescribed as a component of standard inpatient rehabilitation and the intensity is individualized: patients with lower baseline functional capacity, limited exercise tolerance, or relevant comorbidities were generally prescribed 30% MVC, whereas patients with better baseline strength and exercise tolerance were prescribed 50% or 70% MVC; the control group comprised patients whose routine program did not include isokinetic training. Because allocation was not randomized, potential confounding by indication was addressed through covariate-adjusted analyses and inverse probability weighting sensitivity analyses (see Statistical Analysis).

Intervention protocol

All participants received a standardized comprehensive rehabilitation program consisting of conventional physical therapy (60 min daily), occupational therapy (30 min daily), and neurodevelopmental treatment approaches (28). The program was administered five days per week for eight consecutive weeks.

In addition to the standard program, patients in the intervention groups received isokinetic strength training using the Biodex System 4 Pro dynamometer (Biodex Medical Systems, Shirley, NY, USA) (18, 19). Training sessions were conducted three times per week for eight weeks, with each session lasting 30 min. The isokinetic protocol targeted the knee extensors and flexors of the affected limb at angular velocities of 60°/s and 120°/s (20).

For the low-intensity group, training was performed at 30% of the baseline MVC determined during initial assessment. The moderate-intensity group trained at 50% MVC, while the high-intensity group trained at 70% MVC. Each session consisted of three sets of 10 repetitions at each angular velocity, with two-minute rest intervals between sets. Training intensity was progressively adjusted every two weeks based on reassessment of MVC values to maintain the prescribed percentage.

Outcome measures

All assessments were performed by trained physiotherapists who were not involved in the patients’ direct treatment. Evaluations were conducted at baseline and at the completion of the 8-week intervention period.

Primary outcome

The Modified Ashworth Scale (MAS) was used to assess spasticity of the ankle plantar flexors and knee flexors on the affected side (27). The MAS grades muscle tone on a 6-point ordinal scale (0, 1, 1+, 2, 3, 4), with higher scores indicating more severe spasticity. Spasticity was assessed at both the ankle plantar flexors and the knee flexors; for each participant, the higher of the two grades (representing the more severely affected muscle group) was recorded as the participant's MAS score and used in all analyses. For statistical analysis, the MAS 1 + grade was converted to 1.5. This numeric conversion is a widely used convention that facilitates parametric modelling of MAS change scores; its robustness was verified with non-parametric and ordinal sensitivity analyses (see Statistical Analysis).

Secondary outcomes

The Fugl-Meyer Assessment Lower Extremity (FMA-LE) subscale was administered to evaluate motor recovery of the lower limb (0–34 points) (29). The Berg Balance Scale (BBS) was used to assess functional balance (0–56 points) (30). The 10-Meter Walk Test (10MWT) was performed to measure comfortable walking speed (31). Isokinetic assessment of knee flexors and extensors was performed using the Biodex System 4 Pro at 60°/s, with peak torque (Nm) values recorded. Assessments focused on the affected limb; the contralateral (unaffected) limb was not systematically tested.

Statistical analysis

Sample size calculation was performed based on preliminary data suggesting a mean difference of 0.5 MAS points between groups with a pooled standard deviation of 0.8. Using one-way ANOVA with four groups, α=0.05, and power=0.80, a minimum of 64 patients per group was required.

Continuous variables were presented as mean ± standard deviation. Baseline characteristics were compared using one-way ANOVA for continuous variables and chi-square tests for categorical variables. Within-group changes were analyzed using paired t-tests. Between-group comparisons were performed using one-way ANOVA with post-hoc Tukey's HSD tests. Tukey's HSD was used as the sole post-hoc procedure for all pairwise comparisons. Effect sizes were calculated using eta-squared (η2). The dose-response relationship was examined using linear regression analysis. To address potential confounding by indication, analysis of covariance (ANCOVA) models were fitted for each outcome, with the change score as the dependent variable, group as the factor of interest, and age, sex, stroke type, stroke duration, and baseline MAS, FMA-LE, BBS, and 10MWT values as covariates; adjusted dose-response relationships were examined using multiple linear regression with training intensity (% MVC) as a continuous predictor. Because the MAS is an ordinal scale, between-group differences in MAS change were additionally examined using the Kruskal–Wallis test, and a proportional-odds ordinal logistic regression of MAS change on training intensity and the same covariates was fitted as a sensitivity analysis. Inverse probability weighting based on multinomial propensity scores estimated from the same baseline covariates (stabilized weights trimmed at the 1st and 99th percentiles) was applied as a further sensitivity analysis for confounding by indication. All statistical analyses were performed using Python 3.9 with SciPy 1.9.3 and statsmodels. A two-tailed P value <0.05 was considered statistically significant.

Results

Patient characteristics

A total of 486 stroke patients were screened for eligibility during the study period. After applying inclusion and exclusion criteria, 392 patients were included in the final analysis: 98 in the control group, 98 in the low-intensity group, 98 in the moderate-intensity group, and 98 in the high-intensity group (Figure 1). No patients were lost to follow-up, and all completed the 8-week intervention program. Review of medical records confirmed that no participant received oral antispasticity medications, botulinum toxin injections, intrathecal baclofen, or chemical neurolysis during the study period.

Figure 1.

Bar chart titled \“Modified Ashworth Scale Change by Treatment Intensity\” shows increasing MAS reduction scores from control (0.20), low (0.69), moderate (1.02), to high (1.07) intensities, with significant differences indicated by asterisks.

Dose-response relationships between isokinetic training intensity and outcome improvements. Small symbols represent individual participants (n = 392); large symbols represent group means with standard error bars; dashed lines indicate linear regression fitted to individual-level data, and the reported R2 and P values refer to these individual-level regressions. (A) MAS reduction; (B) FMA-LE improvement; (C) BBS improvement; (D) 10MWT reduction; (E) Knee extension peak torque gain; (F) Knee flexion peak torque gain.

Baseline demographic and clinical characteristics were well-balanced across the four groups (Table 1). The mean age ranged from 60.24 ± 10.65 years (control) to 62.40 ± 9.70 years (low-intensity), with no significant between-group differences (P = 0.456). Males constituted 50.0–64.3% of each group. Ischemic stroke was the predominant etiology (65.3–78.6%), and time since stroke onset averaged 11.52–13.67 months across groups. Baseline values for all outcome measures demonstrated no significant differences among groups (all P > 0.05).

Table 1.

Baseline demographic and clinical characteristics of study participants.

Characteristic Control (n = 98) Low (n = 98) Moderate (n = 98) High (n = 98) P
Age, years 60.24 ± 10.65 62.40 ± 9.70 61.80 ± 8.95 61.28 ± 10.07 0.456
Male sex, n (%) 62 (63.3) 49 (50.0) 63 (64.3) 58 (59.2) 0.128
Ischemic stroke, n (%) 70 (71.4) 64 (65.3) 69 (70.4) 77 (78.6) 0.197
Stroke duration, mo 13.64 ± 6.16 13.67 ± 6.76 11.52 ± 6.34 13.05 ± 6.41 0.067
Baseline MAS 1.92 ± 0.82 2.02 ± 0.81 2.14 ± 0.84 1.99 ± 0.81 0.280
Baseline FMA-LE 18.00 ± 4.30 17.77 ± 4.25 17.52 ± 4.64 18.29 ± 4.32 0.651
Baseline BBS 34.86 ± 7.41 36.50 ± 7.67 35.30 ± 8.24 34.70 ± 9.13 0.402
Baseline 10MWT, s 29.95 ± 12.07 27.82 ± 10.57 29.56 ± 11.87 28.00 ± 11.36 0.457

MAS, modified ashworth scale; FMA-LE, Fugl-Meyer assessment lower extremity; BBS, Berg balance scale; 10MWT, 10-meter walk test.

MAS scores represent the higher of the ankle plantar flexor and knee flexor grades for each participant.

Primary outcome: modified ashworth scale

At baseline, MAS scores were similar across all groups (control: 1.92 ± 0.82; low: 2.02 ± 0.81; moderate: 2.14 ± 0.84; high: 1.99 ± 0.81; P = 0.280). As specified in the Methods, each participant's MAS score was the higher of the ankle plantar flexor and knee flexor grades. Following the 8-week intervention, all groups demonstrated significant within-group improvements (all P < 0.001). The high-intensity group achieved a mean MAS reduction of 1.07 ± 0.36 points, followed by the moderate-intensity group (1.02 ± 0.29 points), low-intensity group (0.69 ± 0.48 points), and control group (0.20 ± 0.45 points).

One-way ANOVA revealed a significant main effect of group on MAS change scores (F = 95.789, P < 0.001), with a large effect size (η2 = 0.425). Post-hoc analysis demonstrated significant differences between the control group and all intervention groups (all P < 0.001), between low-intensity and moderate-intensity groups (P < 0.001), and between low-intensity and high-intensity groups (P < 0.001). The difference between moderate and high-intensity groups did not reach statistical significance (P = 0.813, Tukey's HSD) (Figure 2).

Figure 2.

Flowchart diagram of a clinical study showing 486 assessed for eligibility, 94 excluded for specific reasons, and 392 enrolled, allocated equally into four groups: control, low-intensity, moderate-intensity, and high-intensity isokinetic training. All groups had 98 participants, no losses to follow-up, and all 392 participants analyzed with no exclusions or missing data.

Study flow diagram showing participant enrollment, allocation, follow-up, and analysis.

Secondary outcomes

Motor function (FMA-LE)

Significant improvements were observed in all groups following intervention (Table 2). The high-intensity group demonstrated the greatest improvement (7.23 ± 1.92 points), followed by moderate-intensity (5.64 ± 1.89 points), low-intensity (3.48 ± 1.92 points), and control (1.27 ± 1.72 points) groups. Between-group comparisons revealed significant differences across all pairwise comparisons (all P < 0.001). The ANOVA yielded F = 190.483 (P < 0.001) with η2 = 0.596.

Table 2.

Outcome changes following 8-week intervention.

Outcome Control Low Moderate High F P η 2
MAS Reduction 0.20 ± 0.45 0.69 ± 0.48 1.02 ± 0.29 1.07 ± 0.36 95.789 <0.001 0.425
FMA-LE Gain 1.27 ± 1.72 3.48 ± 1.92 5.64 ± 1.89 7.23 ± 1.92 190.483 <0.001 0.596
BBS Gain 2.00 ± 2.43 5.00 ± 2.37 7.90 ± 2.89 9.84 ± 2.85 164.067 <0.001 0.559
10MWT Reduction, s 1.16 ± 2.77 3.60 ± 2.36 6.35 ± 2.86 7.96 ± 2.92 118.515 <0.001 0.478
Knee Ext PT Gain, Nm 2.56 ± 4.25 7.67 ± 4.65 11.88 ± 4.70 16.04 ± 4.17 165.174 <0.001 0.561
Knee Flex PT Gain, Nm 1.58 ± 3.37 5.49 ± 3.25 8.55 ± 3.41 10.13 ± 3.02 130.315 <0.001 0.502

All within-group changes were statistically significant (P < 0.001) for all groups. MAS scores represent the higher of the ankle plantar flexor and knee flexor grades for each participant.

Balance (BBS)

The pattern of improvement followed the dose-response gradient. BBS scores increased by 9.84 ± 2.85 points in the high-intensity group, 7.90 ± 2.89 points in the moderate-intensity group, 5.00 ± 2.37 points in the low-intensity group, and 2.00 ± 2.43 points in the control group. All between-group comparisons were statistically significant (F = 164.067, P < 0.001, η2 = 0.559) (Figure 3).

Figure 3.

Four grouped bar charts labeled A to D display pre- and post-intervention data for four treatment groups: Control, Low, Moderate, and High, across different outcomes. Chart A shows Fugl-Meyer Assessment Lower Extremity points increasing post-intervention. Chart B shows Berg Balance Scale scores improving post-intervention. Chart C shows decreased times post-intervention on the 10-Meter Walk Test. Chart D displays increased knee extension peak torque post-intervention. All post-intervention results indicate statistically significant differences (denoted by triple asterisks) for all groups.

Pre- and post-intervention comparisons for key outcomes by treatment group. Gray bars, baseline; colored bars, post-intervention. *** P < 0.001 for within-group paired comparisons. (A) Fugl-Meyer Assessment Lower Extremity; (B) Berg Balance Scale; (C) 10-Meter Walk Test; (D) Knee extension peak torque.

Gait performance (10MWT)

Walking time decreased significantly in all intervention groups. The high-intensity group showed the greatest improvement (reduction of 7.96 ± 2.92 s), followed by moderate-intensity (6.35 ± 2.86 s), low-intensity (3.60 ± 2.36 s), and control (1.16 ± 2.77 s) groups. Between-group ANOVA was significant (F = 118.515, P < 0.001, η2 = 0.478).

Muscle strength

Knee extension peak torque increased substantially across intervention groups. The high-intensity group achieved a gain of 16.04 ± 4.17 Nm, compared with 11.88 ± 4.70 Nm in the moderate-intensity group, 7.67 ± 4.65 Nm in the low-intensity group, and 2.56 ± 4.25 Nm in the control group (F = 165.174, P < 0.001, η2 = 0.561). Similar dose-response patterns were observed for knee flexion peak torque (Table 2).

Dose-response analysis

Linear regression analysis revealed significant dose-response relationships for all outcome measures (Figure 4). All R2 and P values reported in this subsection and in Figure 4 were derived from regressions fitted to individual-level data (n = 392). The strongest linear relationship was observed for FMA-LE improvement (R2 = 0.592, slope = 0.0868, P < 0.001), indicating that each 10% increment in training intensity was associated with approximately 0.87 additional points of motor function improvement. For the primary outcome (MAS reduction), the linear model explained 40.1% of the variance (R2 = 0.401, slope=0.0130, P < 0.001). Substantial dose-response relationships were also demonstrated for BBS improvement (R2 = 0.556), 10MWT reduction (R2 = 0.473), knee extension peak torque gain (R2 = 0.559), and knee flexion peak torque gain (R2 = 0.496). All dose-response trends were statistically significant (all P < 0.001). Because the control group received less total treatment time than the intervention groups, the dose-response analysis was repeated within the three isokinetic training groups only: the gradient remained positive and statistically significant but was attenuated (MAS reduction: slope 0.09 points per 10% MVC, R2 = 0.136, P < 0.001; FMA-LE improvement: slope 0.94 points per 10% MVC, R2 = 0.393, P < 0.001), reflecting the large low-to-moderate intensity contrast and the plateau between 50% and 70% MVC.

Figure 4.

Six-panel figure showing line graphs of training intensity versus outcome measures: higher training intensity correlates with greater improvements or reductions in MAS, FMA-LE, BBS, 10MWT, knee extension PT, and knee flexion PT. Each panel includes a trendline, error bars, and statistical significance (all p < 0.001, high R squared values).

Modified ashworth scale (MAS) reduction by treatment intensity group. Bar heights represent mean change scores; error bars indicate standard error of the mean. *** P < 0.001 for between-group comparisons.

Covariate-adjusted and sensitivity analyses

After adjustment for age, sex, stroke type, stroke duration, and baseline MAS, FMA-LE, BBS, and 10MWT values, the group effect remained significant for all outcomes (MAS reduction: adjusted F = 93.731; FMA-LE gain: F = 189.354; BBS gain: F = 167.359; 10MWT reduction: F = 114.755; knee extension peak torque gain: F = 167.825; knee flexion peak torque gain: F = 128.438; all P < 0.001; Table 3). Adjusted pairwise contrasts for MAS reduction confirmed significantly greater improvements in all intervention groups versus control and in the moderate- and high-intensity groups versus the low-intensity group (all adjusted P < 0.001), whereas the difference between the moderate- and high-intensity groups remained non-significant (adjusted mean difference 0.05 points, 95% CI −0.07 to 0.16, P = 0.431). Adjusted dose-response slopes were essentially unchanged (0.13 MAS points and 0.87 FMA-LE points per 10% MVC increment; all P < 0.001). Sensitivity analyses treating the MAS as an ordinal variable yielded consistent results: the Kruskal–Wallis test confirmed significant between-group differences in MAS change (H = 173.701, P < 0.001; median reductions: control 0, low 1, moderate 1, and high 1 point), and proportional-odds ordinal logistic regression showed that each 10% MVC increment in training intensity was associated with 2.11-fold higher odds of a greater MAS reduction (P < 0.001). Inverse probability weighting likewise preserved the dose-response gradient (weighted mean MAS reductions versus control: low 0.47, moderate 0.83, and high 0.87 points; high versus moderate 0.03 points, P = 0.511), indicating that the findings are robust to confounding by indication.

Table 3.

Covariate-adjusted analyses of outcome changes.

Outcome Adjusted F Adjusted P Adjusted slope per 10% MVC P
MAS reduction 93.731 <0.001 0.13 points <0.001
FMA-LE gain 189.354 <0.001 0.87 points <0.001
BBS gain 167.359 <0.001 1.13 points <0.001
10MWT reduction, s 114.755 <0.001 1.00 s <0.001
Knee extension peak torque gain, Nm 167.825 <0.001 1.95 Nm <0.001
Knee flexion peak torque gain, Nm 128.438 <0.001 1.25 Nm <0.001

ANCOVA models adjusted for age, sex, stroke type, stroke duration, and baseline MAS, FMA-LE, BBS, and 10MWT values. Slope = adjusted change in outcome per 10% increment in training intensity (multiple linear regression). Sensitivity analyses: Kruskal–Wallis test for MAS change, H = 173.701, P < 0.001; proportional-odds ordinal logistic regression, OR 2.11 per 10% MVC increment, P < 0.001; inverse probability weighting preserved all between-group gradients.

Adverse events

No serious adverse events were reported during the study period. Minor transient symptoms included muscle soreness (reported in 12.2% of high-intensity, 8.2% of moderate-intensity, and 3.1% of low-intensity participants) and fatigue (15.3%, 11.2%, and 6.1%, respectively). All symptoms resolved within 24–48 h without intervention and did not necessitate modification of the training protocol.

Discussion

This retrospective cohort study provides evidence of a significant dose-response association between isokinetic strength training intensity and improvement in lower limb spasticity among stroke patients. Our findings demonstrate that higher training intensities (50%–70% MVC) were associated with superior outcomes across all measured parameters, including spasticity reduction, motor function recovery, balance improvement, and gait performance enhancement. Importantly, MAS reduction did not differ significantly between the moderate- and high-intensity groups (1.02 versus 1.07 points, P = 0.813); thus, although our data support training intensities of at least 50% MVC, they do not demonstrate additional antispasticity benefit at 70% MVC over 50% MVC. These results have important implications for the design and implementation of evidence-based rehabilitation programs for post-stroke spasticity.

The observed dose-dependent improvement in MAS scores challenges the historical perspective that vigorous exercise might worsen spasticity in neurological conditions (13, 15). Traditional rehabilitation approaches often advocated for low-intensity interventions based on concerns that forceful muscle contractions could trigger or aggravate hypertonicity. However, accumulating evidence from recent studies supports the safety and efficacy of progressive resistance training in stroke rehabilitation (14, 21). Our data extend this body of knowledge by demonstrating that not only is higher-intensity training safe, but higher intensities are also associated with greater spasticity reduction than lower intensities or standard care alone.

Several mechanisms may explain the beneficial effects of high-intensity isokinetic training on spasticity (6, 9). First, strength training promotes neuroplasticity through enhanced corticospinal excitability and reorganization of motor representations, which may restore more normal patterns of muscle activation and reciprocal inhibition (7, 12). Second, improved muscle strength allows for better control of limb movement, potentially reducing the compensatory co-contraction patterns that often accompany weakness (15). Third, repetitive active movement through full range of motion may prevent or reverse the secondary viscoelastic changes in muscle tissue that contribute to the clinical manifestation of spasticity (8, 10).

The magnitude of improvement observed in the moderate- and high-intensity groups warrants particular attention. The mean MAS reductions of 1.02 and 1.07 points, respectively, exceed the commonly cited minimal clinically important difference of 0.5–1.0 points for this outcome measure (24, 25). Furthermore, the FMA-LE improvement of 7.23 points substantially surpasses the minimal detectable change of 3–4 points established for this scale (32). These findings suggest that moderate- to high-intensity isokinetic training is associated with clinically meaningful benefits beyond what would be achieved with standard rehabilitation alone.

The role of the unaffected limb merits comment. In the present study, isokinetic training and all assessments targeted the affected limb, and the contralateral limb was not systematically evaluated. Because compensatory use and strength changes of the unaffected side can contribute to gait performance after stroke, part of the observed improvement in walking speed may reflect bilateral functional gains rather than paretic-limb effects alone. Future studies should incorporate bilateral strength and spasticity assessments to clarify the contribution of the unaffected limb to gait improvement.

Several limitations of this study should be acknowledged. First, the retrospective design precludes establishing definitive causal relationships and introduces potential selection bias in group assignment. Second, the single-center nature of the study may limit generalizability. Third, the 8-week follow-up period does not allow assessment of long-term durability. Fourth, we did not include neurophysiological assessments that could provide mechanistic insights. Fifth, although patients receiving antispastic therapies during the study period were excluded and no such therapies were identified on record review, unmeasured co-interventions cannot be fully excluded. Sixth, upper limb spasticity was not assessed, and the present findings therefore apply only to the lower limb. Seventh, the unaffected limb was not systematically evaluated, as discussed above. Eighth, the control group received less total treatment time than the intervention groups; contrasts involving the control group may therefore partly reflect differences in overall treatment dose rather than isokinetic intensity per se, which we addressed by repeating the dose-response analysis within the three training groups. Finally, the use of the MAS as the primary spasticity measure has recognized limitations in sensitivity and reliability (27).

Conclusion

In conclusion, this study demonstrates a significant dose-response relationship between isokinetic strength training intensity and improvement in lower limb spasticity following stroke. Higher training intensities (50%–70% MVC) were associated with superior outcomes in spasticity reduction, motor function recovery, balance, and gait performance compared to lower intensities or standard rehabilitation alone. However, spasticity reduction did not differ significantly between 50% and 70% MVC, and the present data therefore do not support the superiority of the highest intensity over moderate intensity for spasticity management. These findings support the incorporation of structured progressive resistance training at appropriate intensities into comprehensive post-stroke rehabilitation programs. Clinicians should consider individual patient factors when prescribing training intensity, but the evidence suggests that adequate stimulus intensity is necessary to optimize rehabilitation outcomes.

Acknowledgments

The authors thank the rehabilitation team at the Department of Rehabilitation Medicine for their dedication to patient care and assistance with data collection.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Footnotes

Edited by: Salvatore Facciorusso, University of Foggia, Italy

Reviewed by: Luciano Bissolotti, Casa di Cura Domus Salutis, Fondazione Teresa Camplani, Italy

Donlaya Phisitsettakorn, School of Integrative Medicine, Mae Fah Luang University, Thailand

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

The studies involving humans were approved by the Ethics Committee of Jiangyan TCM Hospital of Taizhou City. The studies were conducted in accordance with the local legislation and institutional requirements. The ethics committee/institutional review board waived the requirement of written informed consent for participation from the participants or the participants’ legal guardians/next of kin because the retrospective nature of the analysis.

Author contributions

YZ: Conceptualization, Data curation, Formal analysis, Writing – original draft, Writing – review & editing. WZ: Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

  • 1.Feigin VL, Stark BA, Johnson CO, Roth GA, Bisignano C, Abady GG, et al. and, Collaborators GBDS. Global, regional, and national burden of stroke and its risk factors, 1990–2019: a systematic analysis for the global burden of disease study 2019. Lancet Neurol. (2021) 20(10):795–820. 10.1016/S1474-4422(21)00252-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Urban PP, Wolf T, Uebele M, Marx J, Vogt T, Stoeter P, et al. Occurence and clinical predictors of spasticity after ischemic stroke. Stroke. (2010) 41(9):2016–20. 10.1161/STROKEAHA.110.581991 [DOI] [PubMed] [Google Scholar]
  • 3.Wissel J, Schelosky LD, Scott J, Christe W, Faiss JH, Mueller J. Early development of spasticity following stroke: a prospective, observational trial. J Neurol. (2010) 257(7):1067–72. 10.1007/s00415-010-5463-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Afzal B, Noor R, Alghwiri AA, Umar A. Effects of extracorporeal shockwave therapy combined with neuroplasticity-based training protocol on lower limb spasticity in stroke patients: a randomized controlled trial with ultrasonography evaluation. NeuroRehabilitation. (2026) 58(1):154–72. 10.1177/10538135251403026 [DOI] [PubMed] [Google Scholar]
  • 5.Huang Y, Tang Y, Zhang Z, Shen L, Guo Z, Chen Z, et al. Novel lower limb robot-assisted gait phase stimulation enhances walking and balance in stroke patients: a pilot trial. J Neuroeng Rehabil. (2026) 23(1):63. 10.1186/s12984-025-01871-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Li S, Francisco GE. New insights into the pathophysiology of post-stroke spasticity. Front Hum Neurosci. (2015) 9:192. 10.3389/fnhum.2015.00192 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Zhang Q-Y, Wang J-X, Ma L-X, Zhong J-S, Li F, Qian X, et al. Motion-style scalp acupuncture ameliorates post-stroke muscle spasticity through cerebral blood flow augmentation and 5-HT(2A)R-mediated spinal KCC2 reactivation. Brain Behav. (2025) 15(12):e71064. 10.1002/brb3.71064 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Picelli A, Filippetti M, Pontillo A, Dimitrova E, Valè N, Di Censo R, et al. Changes in tibialis anterior muscle activity following tibial nerve block in adults with spastic equinovarus foot: an observational pilot study. J Rehabil Med. (2025) 57:jrm44405. 10.2340/jrm.v57.44405 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Gracies JM. Pathophysiology of spastic paresis. I: paresis and soft tissue changes. Muscle Nerve. (2005) 31(5):535–51. 10.1002/mus.20284 [DOI] [PubMed] [Google Scholar]
  • 10.Shabbir MS, Saeed A, Tariq S, Shabbir N, Nasir R, Riaz H. Effects of segmental muscle vibration on flexor and extensor groups of the upper limb in enhancing functional recovery after stroke: a randomized trial. J Bodyw Mov Ther. (2025) 45:1–7. 10.1016/j.jbmt.2025.07.019 [DOI] [PubMed] [Google Scholar]
  • 11.Savevska CG, Dimitrova EN, Koevska V, Mitrevska B, Gocevska M, Kalcovska B, et al. Radial extracorporeal shock wave therapy decreases hand spasticity after stroke. Pril (Makedon Akad Nauk Umet Odd Med Nauki). (2025) 46(3):73–82. 10.2478/prilozi-2025-0024 [DOI] [PubMed] [Google Scholar]
  • 12.Lin X, Du K, Ding R, Chen Q, Niu X, Yang Q, et al. Effects of theta burst peripheral magnetic stimulation on cortical excitability and upper limb function after stroke: protocol for a randomized controlled trial. Front Neurol. (2025) 16:1702806. 10.3389/fneur.2025.1702806 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Zhu T, Liu K, Ni BY, Li L, Jin HP, Wu W. Effects of extracorporeal shock wave therapy combined with isokinetic strength training on spastic calf triceps in patients after a stroke: a double-blinded randomised controlled trial. Neurol Res. (2023) 45(11):1019–25. 10.1080/01616412.2023.2255413 [DOI] [PubMed] [Google Scholar]
  • 14.Chu F, Gao L, Zhou J, Jia F, Chen J, Tang W, et al. Transcranial direct current stimulation combined with isokinetic strength training to improve lower limb motor function in chronic stroke survivors: a randomized controlled study. PM R. (2025) 17(12):1411–20. 10.1002/pmrj.13429 [DOI] [PubMed] [Google Scholar]
  • 15.Patten C, Lexell J, Brown HE. Weakness and strength training in persons with poststroke hemiplegia: rationale, method, and efficacy. J Rehabil Res Dev. (2004) 41(3A):293–312. 10.1682/JRRD.2004.03.0293 [DOI] [PubMed] [Google Scholar]
  • 16.Engardt M, Knutsson E, Jonsson M, Sternhag M. Dynamic muscle strength training in stroke patients: effects on knee extension torque, electromyographic activity, and motor function. Arch Phys Med Rehabil. (1995) 76(5):419–25. 10.1016/S0003-9993(95)80570-2 [DOI] [PubMed] [Google Scholar]
  • 17.Luo Y-Y, Huang C, Song Z, Nazari V, Wong AY-L, Yang L, et al. Wearable ultrasound-imaging-based visual feedback (UVF) training for ankle rehabilitation of chronic stroke survivors: a proof-of-concept randomized crossover study. Biosensors (Basel). (2025) 15(6):365. 10.3390/bios15060365 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Lu X, Qiao C, Wang H, Li Y, Wang J, Wang C, et al. Application of isokinetic dynamometry data in predicting gait deviation index using machine learning in stroke patients: a cross-sectional study. Sensors (Basel). (2024) 24(22):7258. 10.3390/s24227258 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.CHEN L, YANG Y, CHEN S, CHEN Y, LIN J, ZHAO Y, et al. Development and psychometric evaluation of an integrative manual muscle test for patients with post-stroke hemiplegia. Eur J Phys Rehabil Med. (2025) 61(6):809–20. 10.23736/S1973-9087.25.09153-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Kim CM, Eng JJ, MacIntyre DL, Dawson AS, Im CM, Eng JJ, et al. Effects of isokinetic strength training on walking in persons with stroke: a double-blind controlled pilot study. J Stroke Cerebrovasc Dis. (2001) 10(6):265–73. 10.1053/jscd.2001.123775 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Xie R, Sun J, Yu J, Zhang Y, Jin H, Zhu X, et al. Efficacy of bilateral soft exoskeleton training for people with subacute stroke: a randomized controlled trial. J Neuroeng Rehabil. (2025) 23(1):48. 10.1186/s12984-025-01847-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Ng SSM, Chen P, Li S, Chan PWT, Tsui AKY, Lai CYY, et al. Psychometric properties of functional gait assessment in people with stroke. BMC Neurol. (2026) 26(1):83. 10.1186/s12883-025-04591-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Lee JM, Kim EJ, Yang SP, Choi SJ, An SH. Relationships among the performances of the 5mWT, 10mWT, 5'10mWT, and 6MWT in individuals with stroke. Physiother Theory Pract. (2026) 42(1):127–34. 10.1080/09593985.2025.2562904 [DOI] [PubMed] [Google Scholar]
  • 24.Yang S-M, Lin H-H, Chen Y-H, Lu Y-L, Wu C-H, Chen W-S, et al. Dose-response effectiveness of focused shockwave therapy on ultrasonographic muscular properties in patients with stroke exhibiting ankle spasticity. J Neuroeng Rehabil. (2025) 22(1):184. 10.1186/s12984-025-01724-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Yang SM, Chen YH, Lu YL, Wu CH, Chen WS, Lin MT. The dose effectiveness of extracorporeal shockwave on plantar flexor spasticity of ankle in stroke patients: a randomized controlled trial. J Neuroeng Rehabil. (2024) 21(1):176. 10.1186/s12984-024-01473-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Wang X, Sun S, Chen Y, Zhang X, Deng L, Zhang J, et al. Efficacy of dual-target iTBS on gait function and brain activation in stroke patients: a randomized, single-blinded, sham-controlled study. Front Neurol. (2025) 16:1678850. 10.3389/fneur.2025.1678850 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Bohannon RW, Smith MB. Interrater reliability of a modified ashworth scale of muscle spasticity. Phys Ther. (1987) 67(2):206–7. 10.1093/ptj/67.2.206 [DOI] [PubMed] [Google Scholar]
  • 28.Leng H, Meng Q, Bao C, Zhang N, Deng Y, Zhou L, et al. Visual and visual-proprioceptive training improve gait via sensory reweighting after chronic stroke: a randomized controlled trial. J Neuroeng Rehabil. (2026) 23(1):88. 10.1186/s12984-025-01855-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Fugl-Meyer AR, Jääskö L, Leyman I, Olsson S, Steglind S. The post-stroke hemiplegic patient. 1. A method for evaluation of physical performance. Scand J Rehabil Med. (1975) 7(1):13–31. 10.2340/1650197771331 [DOI] [PubMed] [Google Scholar]
  • 30.Lapanan K, Chaplin C, Lord SR, Menant JC. The effect of interactive cognitive-motor training on balance, mobility and cognition in chronic stroke survivors: a systematic review and meta-analysis. Am J Phys Med Rehabil. (2026) 105:495–503. 10.1097/PHM.0000000000002896 [DOI] [PubMed] [Google Scholar]
  • 31.Wang X, Zhang K, Lin X, Liu Z, Wang L, Zhang F, et al. Efficacy of acupuncture-based adjunctive therapies in alleviating spasticity and motor function in patients with post-stroke spasticity: a systematic review and network meta-analysis. Complement Ther Med. (2025) 95:103282. 10.1016/j.ctim.2025.103282 [DOI] [PubMed] [Google Scholar]
  • 32.Hiengkaew V, Jitaree K, Chaikeeree N. Minimal detectable changes of the Berg balance scale, Fugl-Meyer assessment scale, timed “up & go” test, gait speeds, and 2-minute walk test in individuals with chronic stroke with different degrees of ankle plantarflexor tone. Arch Phys Med Rehabil. (2012) 93(7):1201–8. 10.1016/j.apmr.2012.01.014 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.


Articles from Frontiers in Medicine are provided here courtesy of Frontiers Media SA

RESOURCES