To compute statistical power, values of NeuroFlexor NC previously recorded in chronic stroke patients (33 ) (n = 20, mean 11.24, SD (11.96)) and in healthy subjects (28 (link)) (n = 13, mean 0.0 (SD 2.0)) were used. A 1-sided power calculation indicated a sample size of n=14 to achieve power = 0.8, p < 0.05. Descriptive statistics are presented as mean (standard deviation; SD) for normally distributed continuous data and as median (interquartile range; IQR) for ordinal and not normally distributed data (detected with the Shapiro–Wilk test). Spearman’s rank correlation (rs) was conducted to measure the correlation between NeuroFlexor components and age and anthropometric measurements. Sex differences were evaluated with a Mann–Whitney U test. In addition, the Mann–Whitney U test was used to evaluate differences between NC quantified in stroke patients and in healthy subjects at different stretch velocities.
After natural log transformation (applied to correct skewed distribution), a repeated measures analysis of variance (rm-ANOVA) investigated the difference in NC quantified at 120, 180 and 240°/s in the stroke patients. A further rm-ANOVA was performed in the sub-group of healthy subjects. In addition, a non-parametric Friedman test was conducted to confirm the differences in stroke patients’ NC depending on stretch velocities. Spearman’s rank correlation (rs) was used to investigate relationships between EMG signal and NC, and between the clinically scored muscle tone according to MAS and NC and the NeuroFlexor total resistance force.
To assess reliability, a 2-way random effects model, single measure, absolute-agreement, was used to generate an intraclass correlation coefficient model 2.1 (ICC2,1) with 95% CI. To rate the ICC coefficients, Currier’s suggestion (34 ) was used: 0.90 – 0.99 = high reliability, 0.80 – 0.89 = good reliability, 0.70 – 0.79 = fair reliability, and ≤ 0.69 = poor reliability. In addition, a test-retest repeatability coefficient, as an expression of the smallest real difference between measurements, was calculated by multiplying the standard error of measurement (SEM) by 2.77 (i.e. 1.96 × √2) (35 (link)). SEM represents the within-subjects standard deviation and was calculated as SD × √(1–ICC). A paired t-test was used to assess any systematic bias between the 2 sessions.
In healthy subjects, cut-off values for the NeuroFlexor components were established by adding 3 SD to the mean (32 (link)), after elimination of outliers defined with the interquartile method. This conservative approach ensured that almost all healthy subjects fall within the cut-off score and, therefore, that a measured value above the limit could be considered pathological. In addition, limits of normality of stretch-induced EMG amplitude were established for gastrocnemius and soleus muscles, by adding 3 SD to the mean. Receiver operating characteristic (ROC) curve analysis was then used to validate cut-off values for NC for both 30° and 40°, by comparing with pathological EMG amplitudes.
The level of statistical significance was set at p ≤ 0.05. All statistical analyses were performed using IBM SPSS Statistics for Windows, Version 27.0 (IBM Corp., Armonk, NY, USA).
After natural log transformation (applied to correct skewed distribution), a repeated measures analysis of variance (rm-ANOVA) investigated the difference in NC quantified at 120, 180 and 240°/s in the stroke patients. A further rm-ANOVA was performed in the sub-group of healthy subjects. In addition, a non-parametric Friedman test was conducted to confirm the differences in stroke patients’ NC depending on stretch velocities. Spearman’s rank correlation (rs) was used to investigate relationships between EMG signal and NC, and between the clinically scored muscle tone according to MAS and NC and the NeuroFlexor total resistance force.
To assess reliability, a 2-way random effects model, single measure, absolute-agreement, was used to generate an intraclass correlation coefficient model 2.1 (ICC2,1) with 95% CI. To rate the ICC coefficients, Currier’s suggestion (34 ) was used: 0.90 – 0.99 = high reliability, 0.80 – 0.89 = good reliability, 0.70 – 0.79 = fair reliability, and ≤ 0.69 = poor reliability. In addition, a test-retest repeatability coefficient, as an expression of the smallest real difference between measurements, was calculated by multiplying the standard error of measurement (SEM) by 2.77 (i.e. 1.96 × √2) (35 (link)). SEM represents the within-subjects standard deviation and was calculated as SD × √(1–ICC). A paired t-test was used to assess any systematic bias between the 2 sessions.
In healthy subjects, cut-off values for the NeuroFlexor components were established by adding 3 SD to the mean (32 (link)), after elimination of outliers defined with the interquartile method. This conservative approach ensured that almost all healthy subjects fall within the cut-off score and, therefore, that a measured value above the limit could be considered pathological. In addition, limits of normality of stretch-induced EMG amplitude were established for gastrocnemius and soleus muscles, by adding 3 SD to the mean. Receiver operating characteristic (ROC) curve analysis was then used to validate cut-off values for NC for both 30° and 40°, by comparing with pathological EMG amplitudes.
The level of statistical significance was set at p ≤ 0.05. All statistical analyses were performed using IBM SPSS Statistics for Windows, Version 27.0 (IBM Corp., Armonk, NY, USA).
