Inferential Statistics for Electrophysiological Analysis of Paretic Upper Limb-sensory Deficits and Muscle Imbalance in Patients with Acute Stroke

ABSTRACT

Objectives: The prevalence of somatosensory dysfunction after stroke exceeds 50%, and its severity is negatively correlated with upper limb function and the ability to perform self-care tasks. This study objectively assessed somatosensory deficits in the acute post-stroke phase and characterized muscle-output imbalance across severity levels.

Methods: This prospective observational pilot study was conducted at Saitama Medical University International Medical Center between June 2021 and July 2023. Forty-one patients (age 66 ± 10 years) with acute stroke were assessed for somatosensory deficits and muscle-output imbalance. The severity of somatosensory disturbance was grouped according to short-latency sensory evoked potentials (SSEP), and the change in the muscle co-contraction index (CCI) from surface electromyography (sEMG) was compared at two points within 1 week. The sEMG data for each SSEP group were replicated 1000 times using the bootstrap method.

Results: An interaction was observed across all movement directions. In the comparison analysis, significant changes were observed in the normal group except for elbow flexion on the non-paretic side. In the delayed group, only wrist flexion on the non-paretic side changed more than one standard deviation. The changes observed in the severe SSEP group were inferred as fluctuations in CCI, even in the early stages of stroke onset.

Conclusions: This study showed that changes in CCI of the upper extremity vary with the severity of somatosensory disturbance in patients with acute stroke and that muscle-output imbalance can change within 1 week of onset of acute stroke.

INTRODUCTION

Impairments in sensory modalities experienced by patients with stroke can lead to misidentification of objects and failure to avoid danger with the affected limb, interfering with daily activities and social engagement.¹⁾ More than 50% of stroke survivors experience somatosensory deficits,²⁾ and their severity is negatively correlated with upper limb function and self-care ability. Individuals with impaired proprioception are unable to control reaching movements of the upper limbs toward targets as close as 10 cm,³⁾ posing the risk of hindered upper limb activities of daily living (ADLs).

Despite somatosensory deficits being considered impediments to motor function and ADL recovery in patients with stroke, several challenges exist in evaluating these deficits during the acute post-stroke phase. First, efforts are focused on regaining motor function in the acute phase and attempts to recover sensory function are not explored as extensively as those for motor function. Because the evaluation of somatosensory deficits often relies on subjective patient reports, these deficits are less observable than motor impairments, and their severity is difficult to demonstrate objectively. Second, among the standardized assessments for somatosensory deficits are the Semmes–Weinstein Monofilaments test⁴⁾ and the visual analog scale, but the indicators of impairment severity in both tests rely on patient subjectivity, making quantitative evaluation difficult. Subjective reports in the acute post-stroke phase are prone to fluctuations because of impairments in consciousness and higher brain functions, compromising the reliability of the evaluation. Third, although the short-latency somatosensory evoked potential (SSEP) is a standardized, objective somatosensory assessment,⁵⁾ it is time-consuming, making its implementation challenging within the time constraints of the acute phase.

Past intervention studies on somatosensory deficits have reported that therapists do not consistently administer standardized assessments or interventions related to somatosensory function in clinical settings.^6,7) The majority of therapists employ non-standardized methods for assessing somatosensory deficits, and specific, standardized interventions targeting these deficits are rarely performed.⁸⁾ Past survey studies have shown that although rehabilitation staff consider the evaluation of somatosensory deficits important, the sensory modalities assessed are limited to proprioception and tactile stimuli.⁸⁾

The outcomes of human joint movements become sensory information at the sensory receptors and are transmitted to the cerebral hemispheres via ascending pathways. Proprioceptive stimuli from muscle spindles activate somatosensory and widespread motor areas.⁹⁾ In joint movements, the perception and control processes share a common neural substrate, enabling immediate motor control, and somatic sensation is perceived through motor error between motor commands and motor outcomes. Muscle synergies are coordinated patterns in which multiple muscles activate cooperatively at specific ratios, enabling the motor control system to reproduce complex muscle output patterns through their combination. Voluntary movements are adjusted through sensory feedback.¹⁰⁾ In motor–sensory integration, the perception of motor outcomes, proprioceptive sensations, and vestibular inputs are considered essential for updating and maintaining the stability of both the external and internal environments. In cases of somatosensory impairment, accurate perception of bodily states becomes difficult, disrupting both feedforward and feedback mechanisms. Consequently, the selection and adjustment of muscle synergies in the motor control process may become challenging. This impairment can disrupt the balance of coordinated output between antagonist and synergist muscles.

SSEP is used as a quantitative indicator of somatosensory deficits and is utilized for predicting prognosis as an objective electrophysiological examination.¹¹⁾ The SSEP waveform is used to infer the source of neural activity, and the N20 component occurring 20 ms after stimulation is considered to originate from the primary somatosensory cortex. N20 is also used as an objective indicator for examining sensory function in patients with stroke.⁵⁾ The method of differentiating the severity of somatosensory deficits using SSEP is expected to provide a highly accurate evaluation method that is unaffected by impairments in consciousness or higher brain functions.

The motor commands from the motor cortex induce skeletal muscle contraction. The objectivity and reproducibility of muscle output detected using ordinal scales, such as manual muscle testing, are inferior to those of electrophysiological methods. Electrophysiological muscle-strength evaluation includes surface electromyography (sEMG), which estimates the state of muscle contraction based on the integrated electromyographic values. sEMG is also utilized as an indicator of the coordinated action of agonist and antagonist muscles and can serve as an indicator of muscle output in cases where the feedback mechanism is impaired because of somatosensory deficits. The co-contraction index (CCI), calculated as the ratio of muscle output between the main active muscle to the antagonist muscle, has been used for the quantitative evaluation of muscle-output coordination. Previous studies have investigated the relationship between CCI and motor paralysis and upper extremity dysfunction and have shown a correlation with the assessment of motor paralysis.¹²⁾

In recent years, methods involving proprioceptive sensory–motor re-education through exercise therapy have been employed.¹³⁾ Robotic therapy has demonstrated effects through repeated, quantified stimulation. However, reports suggest that current sensory re-education interventions are not based on standardized assessments and are influenced by therapists’ empirical knowledge.^6,7) By selecting quantified assessment methods for somatosensory deficits, accurate tracking of symptom progression becomes possible. This is useful for estimating appropriate loading levels when providing program instructions for treatment interventions and determining the amount of motor and sensory stimulation, such as in robotic therapy.

This study aimed to evaluate somatosensory deficits in the acute post-stroke phase using objective indicators and assess the associated changes in muscle-output imbalance across different severity levels. By examining changes in CCI by severity of somatosensory impairment rather than motor function, changes in CCI in the acute phase of stroke can be stratified from the degree of somatosensory impairment. Furthermore, the changes in upper limb motor function could be objectively assessed from the severity of somatosensory disturbance in the acute phase, which can be used to standardize the rehabilitation treatment load from the subacute to the chronic phase and to help in the selection of treatment methods. Therefore, this study hypothesized that muscle-output imbalance exhibits characteristic changes depending on the severity of somatosensory deficits.

MATERIALS AND METHODS

Study Design

This was a prospective observational pilot study. Given that somatosensory deficits are often assessed subjectively, making objective evaluation challenging, we adopted an approach using electroencephalography (EEG) and electromyography (EMG), which are less influenced by the patients’ altered consciousness or subjective sensory fluctuations, to analyze somatosensory deficits in patients with acute stroke.

Setting

The study was conducted at the Rehabilitation Center of our Medical Center. Data collection, approved by the ethics committee of the same center, was conducted between June 2021 and June 2023 (excluding the periods when the study was suspended because of the COVID-19 pandemic: July 2021–February 2022 and June–October 2022). This study enrolled patients with acute stroke who provided informed consent and were admitted during the survey period. Participants underwent SSEP measurements using EEG and sEMG recordings of the biceps and triceps brachii muscle activity, with two measurements recorded at least 2 days apart within 1 week after stroke onset.

Participants

Consecutive patients admitted to our center with acute stroke during the study period were screened for eligibility. The following inclusion criteria were used: 1) age 20 years or older with incidence of stroke, 2) prescribed occupational therapy, and 3) mild-to-moderate upper limb motor impairment, with a Fugl–Meyer Assessment of upper extremity (FMA-UE) score of 19 or higher for the upper extremity.¹⁴⁾ The following exclusion criteria were used: 1) previous history of stroke, 2) orthopedic disorders in the affected upper limb,¹⁵⁾ 3) diagnosed peripheral nerve disorder in the upper limb, 4) head trauma or surgical incision precluding SSEP electrode placement, 5) speech impairment, 6) cognitive impairment with a Mini-Mental State Examination-Japanese (MMSE-J) score less than 24,¹⁶⁾ and 7) missing data.

The reasons for ineligibility were reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.¹⁷⁾ Prior to the study, all participants received oral and written explanations regarding the research and provided written informed consent. This study was approved by the Ethics Committee of the Center for Appropriate Promotion of Clinical Research at Saitama Medical School International Medical Center (approval number 2021–008) and registered in the University Hospital Medical Information Network (UMIN000045608).

A priori power analysis using software (G*Power; Heinrich Heine University, Düsseldorf, Germany) indicated that the required sample size was 81 participants, with an estimated moderate effect size (δ) of 0.3, power of 0.80, and significance level (α) of 0.05, assuming three SSEP waveform groups similar to that in previous studies. The initial calculated sample size was 67, with an additional 20% (14 participants) to account for potential dropouts.

Measurements

SSEP measurements were performed using the MEB-9400 series Neuropack S1 (Nihon Kohden, Tokyo, Japan), with motor sub-threshold electrical stimulation applied to the median nerve territory of the affected upper limb. sEMG was recorded using the Delsys Trigno series (Inter Reha, Tokyo, Japan) for the biceps brachii, triceps brachii, flexor carpi radialis, and extensor carpi radialis muscles.

Upper limb function was assessed using the FMA-UE score to evaluate voluntary motor control in the paretic upper limb following stroke. Demographic and clinical data, including age, height, weight, sex, medical history, comorbidities, side of hemiparesis, diagnosis, lesion location, higher brain-function test results, and admission severity, were obtained from medical records.

EEG Measurement

SSEP peak latencies and amplitudes were measured following median nerve stimulation. According to the international 10–20 electrode-placement system, the active electrode was placed at CPc, and the reference electrode was placed at Fz. The median nerve was stimulated at the wrist with a 5-Hz rectangular wave pulse of 0.2 ms duration, above the motor threshold, and averaged over 200 responses. SSEP latency refers to the first positive peak, representing the average peak latency of the main waveform occurring around 20 ms. The normal upper limit for SSEP latency was defined as the mean plus two standard deviations (SDs) in healthy adults, which is 20.88 ms.¹⁸⁾ Participants were classified into three groups based on their SSEP waveforms: normal (peak latency within the normal upper limit), delayed (peak latency exceeding the normal upper limit), and unresponsive (no detectable peak).¹⁵⁾

sEMG Data Processing

sEMG data were recorded using the Delsys EMGworks Acquisition software and sampled at 2000 Hz. Prior to analysis, the EMG amplitude digital signal was processed as the root mean square (RMS) of the EMG signal using a 1-s moving window. To detect the maximum RMS amplitude, a 1-s window was moved in 100-ms increments across the entire maximum voluntary contraction period.

For each movement direction (elbow flexion, extension, wrist palmar flexion, and dorsiflexion), the participants performed an 8-s maximum isometric contraction,¹⁹⁾ and the middle 5 s was analyzed, excluding the initial 3 s. The RMS values calculated during this 5-s period were used to determine the ratio of the antagonist-muscle activity to the agonist-muscle activity.^20,21) The resulting CCI was calculated as the ratio of antagonist-muscle output to agonist-muscle output for the paretic upper limb in each movement direction (Fig. 1). CCI provides an objective and simple assessment of motor control: CCI takes a value between 0 and 1, where 0 indicates that the primary and antagonist muscles in the movement did not overlap and 1 indicates that the two muscles completely overlapped and co-contracted. In this study, isometric maximal contraction was selected as the task, and because the task specified four movement directions for the direction of movement of the main active muscle, higher muscle output of the antagonist muscle contributed to a higher CCI value.^22,23)

Fig. 1.

Methodology for calculating CCI on the paretic and non-paretic sides using electromyographic (EMG) analysis. Left panel: EMG data from the paretic side. Right panel: EMG data from the non-paretic side. Each panel shows: x-axis, time (s), ranging from 0 to 7.5 s; y-axis, EMG amplitude (mV); blue waveform, EMG output of the main active muscle; green waveform, EMG output of the antagonist muscle; and pink shaded area (approximately 2.5–5 s), time frame for analysis. EMG waveforms are superimposed with 0 as the base axis. Root mean square (RMS) values, calculated from the EMG waveforms within the analyzed time frame (pink area), are displayed in corresponding colors. The CCI is calculated using the RMS values from the analyzed time frame. This method allows for quantitative comparison of muscle-activation patterns between paretic and non-paretic sides, providing insights into muscle-output imbalances in patients with stroke.

Upper Limb Function Assessment

The degree of upper limb motor impairment was evaluated using the upper extremity section of the FMA-UE score.

Tasks

The SSEP measurements were performed with the participants in a resting position with their eyes closed. Prior to the experiment, the median nerve territory was stimulated using the sEMG and SSEP electrode setups to ensure no pain was induced. For sEMG recordings, the participants were seated with their limbs secured, and they performed maximum isometric contractions. The movement directions included elbow flexion, extension, wrist palmar flexion, and dorsiflexion, with a total of eight trials for both the paretic and non-paretic sides.

Protocol

Figure 2 illustrates the study protocol. Eligible participants during the study period provided written informed consent within 4 days of stroke onset, followed by EEG and sEMG measurements. SSEP and sEMG recordings took approximately 60 min, with a second sEMG recording session lasting approximately 20 min. The sEMG measurements were recorded at two time points at least 2 days apart within 1 week after stroke onset. The follow-up period was set to a maximum of 1 week. For participants who could complete all measurements during hospitalization, the measurements were only recorded on the designated days, and they continued to receive inpatient rehabilitation that was similar to that in other patients for the remaining hospital stay.

Fig. 2.

Experimental design for somatosensory evoked potentials and electromyography assessment in upper limb muscle function. (A) Short-latency somatosensory evoked potentials (SSEP) test. The participants were categorized into normal, delayed, and abnormal groups based on SSEP latency. (B) Electromyography (EMG) measurement setup. EMG sensors were attached to the participant’s muscles, and data were transmitted via Wi-Fi link to a computer for analysis during maximal isometric contraction. (C) Timeline of measurements. SSEP test was conducted first, followed immediately by the initial EMG measurement. A second EMG measurement was performed within 1 week after symptom onset, allowing at least 2 days between the first and second EMG tests. The EMG data were used to calculate the CCI, comparing antagonist (root mean square [RMS]) to agonist (RMS) muscle activity.

Statistical Analysis

To investigate the impact of SSEP groups on muscle imbalance, repeated-measures analysis of variance was performed for each movement direction, with SSEP group and EMG measurement time as factors, followed by post-hoc multiple comparisons if an interaction was observed. Statistical analyses were performed using Jefferey’s Amazing Statistics Program Version 0.16.2 (accessed 30 September 2023, https://jasp-stats.org/). The statistical significance level was set at a 5% risk rate. In the absence of standardized CCI criteria, changes between the two measurement points were evaluated using the non-paretic side as a reference. Joint movement changes exceeding 1 SD from the changes in the non-paretic side were extracted.

The change in muscle imbalance was estimated for each severity level of sensory deficits (Fig. 3). To verify that this change was independent of the degree of upper limb motor impairment, the FMA-UE score was included as a covariate in the above analysis, and the results were examined for variability. To perform stratified analyses with a small sample size, a bootstrap resampling method was adopted, which is widely used in demographic studies.²⁴⁾ In this study, sEMG data sequences were randomly sampled from the actual sample to generate 1000 bootstrap data values. This exploratory and preliminary analysis aimed to assess feasibility and cannot be generalized.

Fig. 3.

Electromyographic (EMG) waveform patterns across somatosensory evoked potential (SSEP) classifications in patients with stroke: normal (left), delayed (center), and abnormal (right). Upper row: EMG data from the paretic side; lower row: EMG data from the non-paretic side. Each graph shows: x-axis, time; y-axis, EMG amplitude; blue area, activity of the main active muscle; green area, activity of the antagonist muscle. The graphs display RMS values of the EMG signals within the analyzed time frame.

RESULTS

Enrollment and Patient Characteristics

This study included 861 hospitalized patients with stroke. Of these, 49 provided consent, and 41 participants completed the study assessments. Among the 49 patients who consented, 5 voluntarily withdrew, 1 patient was found to have a prior stroke history not documented in their medical record, 1 patient had a peripheral intravenous line in the measurement of upper limb precluding sEMG recording, and 1 patient had a pacemaker implant, posing difficulty in electrical stimulation (Fig. 4). The participants’ mean age was 66 ± 10 years, height 163.3 ± 8.0 cm, and weight 64.1 ± 11.7 kg. The diagnoses were ischemic stroke in 35 (85.4%) patients and hemorrhagic stroke in 6 (14.6%; Table 1).

Fig. 4.

Flow chart of patient selection and classification for acute stroke study.

Table 1. Clinical characteristics of the patients

Characteristic	Classification	Attribute (n=41)
Age, years		66 ± 10
Height, cm		163.3 ± 8.0
Weight, kg		64.1 ± 11.7
Sex	Male	27 (65.9)
	Female	14 (34.2)
Stroke type	Ischemic	35 (85.4)
	Hemorrhagic	6 (14.6)
Lesion hemisphere	Right	15 (36.6)
	Left	26 (63.4)
Lesion location	A (cerebral cortex)	7 (17.1)
	B (cortex–thalamus)	29 (70.7)
	C (thalamus–brain stem)	4 (9.8)
	D (brain stem)	1 (2.4)
Thalamic lesion	Included	6 (14.6)
	Not included	35 (85.4)
Thumb Localizing Test	0 (normal)	33 (80.5)
	1 (slightly impaired)	4 (9.8)
	2 (moderately impaired)	1 (2.4)
	3 (severely impaired)	3 (7.3)
SSEP	Latency, ms	20.81 ± 0.83
	Amplitude, μV	0.48 ± 0.44
MMSE-J		27.6 ± 1.8
FMA-UE		58.3 ± 8.3

Data are presented as mean ± standard deviation for continuous variables and as number of patients (n) and percentage (%) for categorical variables.

Comparisons among SSEP Groups

Output Ratio in Paretic Upper Limb Muscle

Muscle imbalance showed a significant interaction effect (period × group) across all movement directions (P < 0.05; Fig. 5, Table 2). In post-hoc multiple comparisons, significant changes were observed in all joint movements on the paretic side (P < 0.05). Paretic wrist palmar flexion showed a change greater than 1 SD of the non-paretic side’s change (Table 3). Notably, wrist palmar flexion in the delayed group was the only joint movement that showed a change greater than 1 SD of the non-paretic side’s change across all movement directions (Fig. 5, Table 2). The detailed bootstrap analysis of joint motion before and after bootstrapping and the changes in the CCI with the direction of motion are shown in the Supplementary Material (S1, S2).

Fig. 5.

Comparisons of muscle output ratio during upper limb movements between the paretic and non-paretic sides across study groups. Output ratios of (A) brachioradialis muscle of the paretic side, (B) brachioradialis muscle of the non-paretic side, (C) triceps brachii muscle of the paretic side, (D) triceps brachii muscle of the non-paretic side, (E) flexor carpi radialis of the paretic side, (F) flexor carpi radialis of the non-paretic side, (G) extensor carpi radialis of the paretic side, and (H) extensor carpi radialis of the non-paretic side. The vertical axis represents the CCI, with higher values indicating increased antagonist muscle co-contraction. The horizontal axis shows two measurement time points within a week. Data points are color-coded by group: delayed (green), abnormal (orange), and normal (purple). Asterisk indicates significant difference (P < 0.001) between groups at each time point. An interaction between time and group was observed for both sides. Multiple comparisons revealed significant differences among all groups on the paretic side and between the normal and non-response groups on the non-paretic side.

Table 2. Results of repeated measures analysis of variance for output ratio in paretic and non-paretic upper limb muscles across movement directions

Side	Movement	Effect	F (df=[1/2, 2997])	P	η2
Paretic	Elbow flexion	Period	16.94	<0.001	0.002
		Group	254.61	<0.001	0.102
		Period × group	147.28	<0.001	0.026
	Elbow extension	Period	339.38	<0.001	0.018
		Group	161.95	<0.001	0.102
		Period × group	70.85	<0.001	0.007
	Wrist flexion	Period	1076.74	<0.001	0.009
		Group	397.28	<0.001	0.199
		Period × group	1041.22	<0.001	0.018
	Wrist extension	Period	219.94	<0.001	0.019
		Group	63.65	<0.001	0.029
		Period × group	3.20	0.041	0.001
Non-paretic	Elbow flexion	Period	35.78	<0.001	0.003
		Group	87.78	<0.001	0.040
		Period × group	328.72	<0.001	0.048
	Elbow extension	Period	0.71	0.400	0.000
		Group	308.34	<0.001	0.133
		Period × group	106.47	<0.001	0.015
	Wrist flexion	Period	3.37	0.067	0.000
		Group	115.65	<0.001	0.059
		Period × group	177.70	<0.001	0.019
	Wrist extension	Period	88.36	<0.001	0.007
		Group	22.37	<0.001	0.010
		Period × group	355.68	<0.001	0.054

Significant period × group interactions indicate differing time-course changes among SSEP groups. F-values are from repeated measures analysis of variance. Effects were considered statistically significan at P0.05

Table 3. Change in muscle-power output ratio per movement direction and standard error between groups classified by somatosensory evoked potential findings

Side	Movement direction	Normal (n=1000)	Delay (n=1000)	Abnormal (n=1000)	Overall SD (n=3000)
Paretic	Elbow flexion	0.18 ± 0.65	−0.24 ± 0.61	−0.06 ± 0.33	−0.04 ± 0.57
	Elbow extension	0.56 ± 1.03	0.18 ± 0.57	0.13 ± 0.93	0.29 ± 0.89
	Wrist flexion	−0.10 ± 0.55	1.06 ± 0.81 a	0.13 ± 0.36	0.36 ± 0.78
	Wrist dorsiflexion	0.17 ± 0.46	0.12 ± 0.56	0.16 ± 0.67	0.15 ± 0.56
Non-paretic	Elbow flexion	−0.01 ± 0.32	−0.12 ± 0.36	0.24 ± 0.28	0.03 ± 0.35
	Elbow extension	0.11 ± 0.34	−0.36 ± 0.53	0.30 ± 1.68	0.02 ± 1.07
	Wrist flexion	0.26 ± 0.55	−0.20 ± 0.48	−0.11 ± 0.70	−0.02 ± 0.61
	Wrist dorsiflexion	0.53 ± 0.84	−0.30 ± 0.28	0.14 ± 0.82	0.12 ± 0.77

Change in muscle-power output ratio for each movement direction (elbow flexion/extension, wrist flexion/dorsiflexion) on the paretic and non-paretic sides are shown across three groups classified by SSEP findings: Normal, Delay, and Abnormal. Values are presented in arbitrary units, multiplied by 10⁻¹. Data are presented as mean ± SD. The overall SD across all groups (n=3000) is included for reference.

^a Only delayed wrist-flexion showed changes that exceeded the SD of the non-paretic side (overall).

Output Ratio in Non-paretic Upper Limb Muscle

Similar to that on the paretic side, muscle imbalance (CCI) showed a significant interaction effect (period × group) in all movement directions (P < 0.05; Fig. 5, Table 2) on the non-paretic side. In contrast, no main effect of period was observed for elbow extension and wrist flexion (Table 2). In post-hoc multiple comparisons, only the normal group showed no significant change in elbow flexion, whereas significant changes were observed in other movement directions (P < 0.05). On the non-paretic side, only the delayed group showed increased muscle imbalance at the second measurement, exhibiting a characteristic pattern compared with the other SSEP groups (Fig. 5).

Comparisons among SSEP Groups

The SSEP latency was 20.26 ± 0.52 ms and 21.64 ± 0.41 ms in the normal and delayed groups, respectively. The amplitude was 0.488 ± 0.524 µV and 0.461 ± 0.292 µV in the normal and delayed groups, respectively (Tables 4 and 5). The MMSE-J and FMA-UE scores showed no statistically significant differences among the groups.

Table 4. Comparison of basic attributes

Variable	df	Chi-square	P value
Sex	2	1.995	0.369
Stroke type	2	2.811	0.245
Lesion location	6	2.403	0.879
Thalamic lesion	2	0.409	0.815

This table shows the patient background characteristics of the three groups (normal, delayed, and abnormal), classified according to the short somatosensory evoked potential (SSEP) findings. Intergroup comparisons were performed using the chi-square test. P values are the results of the chi-squared test. *P < 0.05.

Table 5. Comparison of basic attributes, short latency sensory evoked potentials (SSEP), cognitive function (MMSE-J), and motor function (FMA-UE) among groups classified by SSEP findings

Item	Normal (n=15)	Delay (n=10)	Abnormal (n=16)	P value
Age, years	63 ± 9	63 ± 12	70 ± 8	0.13
Height, cm	164.3 ± 9.2	164.0 ± 4.8	162.0 ± 8.5	0.69
Weight, kg	62.8 ± 11.7	70.8 ± 8.0	61.0 ± 12.0	0.09
SSEP latency, ms	20.3 ± 0.5	21.64 ± 0.41	–	<0.001*
SSEP amplitude, μV	0.49 ± 0.52	0.46 ± 0.29	–	0.88
MMSE-J	27.7 ± 1.5	28.2 ± 2.0	27.2 ± 1.9	0.38
FMA-UE	59.1 ± 8.9	56.3 ± 8.2	58.9 ± 8.2	0.68
Thumb Localizing Test	0.2 ± 0.6	0.7 ± 1.3	0.4 ± 0.9	0.40

One-way analysis of variance was used for group comparisons. A statistically significant difference was observed in SSEP latency. P values are the results of one-way analysis of variance.

*P <0.01

DISCUSSION

This preliminary inferential analysis demonstrated that in patients with acute stroke, upper limb CCI changes differed according to the severity of somatosensory deficits, as classified by the SSEP results. For wrist palmar flexion, the delayed-SSEP group exhibited a decrease in paretic upper limb CCI at the second measurement. On the non-paretic side, the delayed-SSEP group showed an increase in upper limb CCI at the second measurement. These findings suggest that the degree of somatosensory deficits influences changes in upper limb CCI, supporting the hypothesis that muscle imbalance exhibits distinct characteristics depending on the severity of somatosensory deficits.

In this study, patients with delayed SSEP showed decreased paretic side CCI. Although previous studies have compared CCI during movement tasks between healthy participants and patients with stroke, the measurement period covered the subacute to chronic phases, and the CCI for each muscle was different.²¹⁾ In the present study, to evaluate changes in CCI on the paretic side during early onset, values on the non-paretic side were used as the reference. SSEP is considered to reflect the activity of the somatosensory cortex during median nerve stimulation,²⁵⁾ and somatosensory deficits caused by decreased activity in the postcentral gyrus are associated with amplitude reduction. After stroke, SSEP latencies are delayed and amplitudes are decreased in the paretic side compared to the non-paretic side, indicating demyelination and nerve damage.^26,27) SSEP involves the medial lemniscus pathway and reflects proprioceptive and fine-touch sensations. A decrease in proprioceptive sensation because of somatosensory deficits can reduce sensory feedback during movement, leading to impaired muscle-output adjustment and movement-control imbalance. We suggest that this phenomenon may contribute to the impaired co-contraction control of antagonist muscles relative to agonist muscles, as observed in the CCI. In the present study, changes in CCI early in the onset period were compared at two sites on the non-paretic side, but future studies could monitor changes in CCI over time based on sEMG as a new clinically feasible assessment method for improving upper limb function in post-stroke rehabilitation. In addition, the ability to monitor changes in CCI over time is expected to contribute to the development of treatment strategies based on changes in CCI.

On the non-paretic side, patients with delayed SSEPs exhibited different changes compared to that on the paretic side, with the delayed group showing a tendency toward no decrease in CCI across all movement directions. A possible explanation for these phenomena is the ipsilesional deficit, a characteristic symptom observed in the non-paretic side after stroke. Ipsilesional deficits involve impairments in movement control and coordination on the side ipsilateral to the cerebral hemisphere lesion.^28,29) These deficits are known to persist from the acute to chronic stroke stages³⁰⁾ and are predicted to be caused by impaired inhibitory pathways involving the corpus callosum and vulnerable gamma-aminobutyric acid neurons, in addition to the severity of motor impairment.^31,32) The observed differences in changes in upper limb CCI stratified by somatosensory deficits on the non-paretic side in patients with acute stroke may be attributed to ipsilesional deficits. Additionally, the somatosensory deficits on the paretic side, observed in patients with delayed SSEPs, could also be a contributing factor. Somatosensory deficits can reduce the amount of ascending sensory information available for body monitoring, thereby affecting the body schema.

The delayed-SSEP group exhibited a distinct pattern of CCI changes, particularly in wrist-flexion movements, which differed from those in both normal and abnormal groups. Although the underlying mechanisms require further investigation, several factors may contribute to this finding. First, the delayed-SSEP group represents patients with intermediate somatosensory pathway damage,^25,26) which may result in different motor control adaptations compared to those with completely preserved or completely absent somatosensory function. Previous studies have suggested that partial sensory loss can trigger distinct compensatory mechanisms,^9,10) although these studies primarily focused on patients with chronic stroke. Second, our finding that wrist movements showed the most pronounced changes is consistent with those of previous reports that distal upper limb movements are more sensitive to somatosensory deficits than proximal movements.³⁾ However, whether this sensitivity translates to specific adaptation patterns in the acute phase remains to be established. Third, the statistical robustness of our bootstrap approach suggests that this finding reflects meaningful biological differences rather than methodological artifacts. Nevertheless, we cannot definitively rule out the possibility that unmeasured factors contribute to the observed pattern. If confirmed in larger studies, the identification of distinct CCI patterns in delayed-SSEP patients could have important clinical implications for stratifying rehabilitation approaches based on the severity of somatosensory deficit. However, additional research is needed to establish whether these early patterns predict functional outcomes and whether they can guide treatment selection.

This study has several limitations. First, we did not perform detailed lesion analysis using imaging diagnostics for the patients with stroke. To clarify the interlimb coordination of upper limb movements, an assessment of the extent of damage to the neural fibers traversing the corpus callosum would be necessary. The severity of ipsilesional deficits in the non-paretic upper limb has been shown to correlate with the axonal density of fibers projecting from the corpus callosum to the premotor cortex in the non-dominant hemisphere.³²⁾ An analysis of brain magnetic resonance imaging (MRI) scans would be required to determine if the patients in this study had damage to the fibers connecting the corpus callosum and premotor cortex. Fractional anisotropy values from brain MRI scans reflect the integrity of neural fibers, with higher values indicating increased axonal density, myelination, and pathway geometry, as well as the presence of fiber-crossing pathways.³³⁾ Second, we did not assess the degree of damage to the somatosensory pathways, such as the thalamus and medial lemniscus. By clarifying the location of nerve fiber damage caused by stroke and changes in muscle-output patterns, treatment strategies could be optimized based on imaging diagnosis. Specifically, this enables the selection of appropriate target conditions for sensory impairment and motor control training, providing a basis for avoiding excessive training loads. Although we obtained diagnostic information from medical records, we did not collect data to estimate the severity of brain damage, such as infarct regions for ischemic stroke or hemorrhage volume for hemorrhagic stroke. Image analysis would be required to obtain such data. By stratifying the severity of brain damage through image analysis, changes in recovery levels according to severity could be determined. This enables prognosis prediction based on recovery levels, contributes to the selection of treatment methods, and allows for the selection of appropriate exercise content without imposing excessive burden on patients. Third, we only performed a single SSEP measurement, precluding the assessment of temporal changes in somatosensory deficits. In acute stroke, neurological symptoms can change significantly, which is clinically well known. A single assessment during an unstable pathological state may not accurately reflect the recovery stage of somatosensory deficits or the extent of damage to neural pathways. In this study, the SSEP measurements were limited to a single time point to reduce patient burden and improve study feasibility. Furthermore, measurement noise may not have been completely eliminated. One method to reduce noise is to set the averaging count to 200 repetitions,¹⁵⁾ which can effectively improve amplitude analysis by obtaining sufficient averaging. Multiple data collections enable highly sensitive detection of changes in somatosensory impairment during the acute phase, making it possible to determine the extent of recovery from sensory impairment. The extent of recovery from sensory impairment enables prognosis prediction and serves as an exploratory starting point for intervention in sensory impairment. Fourth, the inclusion criteria based on an MMSE-J score of at least 24 and the high FMA-UE scores suggest that our study sample consisted mostly of patients with mild stroke. Therefore, our results may not be applicable to more severe cases. Fifth, because our results were based on bootstrap resampling and inferential statistics, their validity for real-time measured data needs to be verified by analyzing the contribution of the estimated values to the observed data. Given the limited number of cases—resulting from a temporary suspension of data collection during the COVID-19 pandemic—the statistical power and generalizability of the findings may be reduced. Future research needs to involve verification using the bootstrap method, which will require reinforcement with actual measurement values. Reinforcement with actual measurement values will clarify the accuracy of data obtained through inferential analysis and enable estimation analysis of CCI with small sample sizes, which will lead to enhanced clinical applicability. In addition, clarification of changes in CCI caused by somatosensory disorders will raise expectations for its applicability as training data in machine learning. Sixth, this was a single-center study, and the generalizability of our findings to other facilities remains uncertain. A multicenter validation of the change in CCI depending on the severity of somatosensory impairment may provide insights on the value of clinical interventions for somatosensory impairment in physical therapy and occupational therapy.

This study investigated changes in CCI during the acute phase of stroke and examined their relationship with sensory impairment. Previous studies on CCI have mainly focused on its association with motor paralysis and muscle tone, with limited attention on sensory dysfunction. Moreover, most reports have been restricted to patients in the subacute or chronic phases.³⁴⁾ In contrast, our study emphasizes the early post-onset period, contributing new insights into the role of sensory impairment in motor coordination. This is considered a notable strength of our research. As a future research direction, prospective observational studies with appropriate sample sizes in multicenter collaborative studies that stratify the severity of brain injury are necessary to improve the generalizability of the results of this study. Longitudinal studies that identify changes in sensory impairment over time and stratified assessments of disease severity and sensory impairment severity will help establish their clinical applicability in numerous clinical settings.

CONCLUSION

The results of this study suggest that in patients with acute stroke, the changes in muscle imbalance differ depending on the severity of somatosensory deficits, not only on the paretic side, but also on the non-paretic side.

ACKNOWLEDGMENTS

We extend our deepest gratitude to all the patients who cooperated in this study. We thank Professor Shinichi Takahashi of the Department of Stroke Medicine, Professor Hiroki Kurita of the Department of Stroke Surgery, Professor Shinya Kouyama of the Department of Endovascular Medicine, Professor Hidetoshi Takahashi of the Department of Rehabilitation Medicine, and occupational therapists Mayumi Fujita, Tetsuya Okihara, Tomonori Takeda, Yoshiyuki Sawada, and Tsuyoshi Kouda of the International Medical Center, Saitama Medical University, for their cooperation in conducting this clinical study. We thank Professor Toyohiro Hamaguchi, Associate Professor Takashi Yasojima, and Assistant Professor Kohei Koizumi of the Department of Rehabilitation Science, Graduate School of Health and Welfare, Saitama Prefectural University, for their guidance in the planning, implementation, and analysis of the study results. This work was supported by the Hidaka Research Project of Saitama Medical University International Medical Center’s “Grant-in-Aid for Young Physician Training Research Project” under Grant number 5-D-1–04.

CONFLICTS OF INTEREST

The authors declare no conflict of interest.

REFERENCES

• 1.  Motomura N, Yamadori A, Asaba H, Sakai T, Sawada T: Failure to manipulate objects secondary to active touch disturbance. Cortex 1990;26:473–477. PMID:2249449, DOI:10.1016/S0010-9452(13)80098-6
• 2.  Carey LM: Somatosensory loss after stroke. Crit Rev 1995;7:51–91.
• 3.  Ghez C, Sainburg R: Proprioceptive control of interjoint coordination. Can J Physiol Pharmacol 1995;73:273–284. PMID:7621366, DOI:10.1139/y95-038
• 4.  Suda M, Kawakami M, Okuyama K, Ishii R, Oshima O, Hijikata N, Nakamura T, Oka A, Kondo K, Liu M: Validity and reliability of the Semmes-Weinstein Monofilament test and the thumb localizing test in patients with stroke. Front Neurol 2021;11:625917. PMID:33584520, DOI:10.3389/fneur.2020.625917
• 5.  Macerollo A, Brown MJ, Kilner JM, Chen R: Neurophysiological changes measured using somatosensory evoked potentials. Trends Neurosci 2018;41:294–310. PMID:29549962, DOI:10.1016/j.tins.2018.02.007
• 6.  Pumpa LU, Cahill LS, Carey LM: Somatosensory assessment and treatment after stroke: an evidence‐practice gap. Aust Occup Ther J 2015;62:93–104. PMID:25615889, DOI:10.1111/1440-1630.12170
• 7.  Doyle S, Bennett S, Gustafsson L: Occupational therapy for upper limb post-stroke sensory impairments: a survey. Br J Occup Ther 2013;76:434–442. DOI:10.4276/030802213X13807217284143
• 8.  Sidarta A, Lim YC, Wong RA, Tan IO, Kuah CW, Ang WT: Current clinical practice in managing somatosensory impairments and the use of technology in stroke rehabilitation. PLoS One 2022;17:e0270693. PMID:35951544, DOI:10.1371/journal.pone.0270693
• 9.  Naito E, Nakashima T, Kito T, Aramaki Y, Okada T, Sadato N: Human limb‐specific and non‐limb‐specific brain representations during kinesthetic illusory movements of the upper and lower extremities. Eur J Neurosci 2007;25:3476–3487. PMID:17553017, DOI:10.1111/j.1460-9568.2007.05587.x
• 10.  Pruszynski JA, Kurtzer I, Nashed JY, Omrani M, Brouwer B, Scott SH: Primary motor cortex underlies multi-joint integration for fast feedback control. Nature 2011;478:387–390. PMID:21964335, DOI:10.1038/nature10436
• 11.  Chae J, Yang G, Park BK, Labatia I: Muscle weakness and cocontraction in upper limb hemiparesis: relationship to motor impairment and physical disability. Neurorehabil Neural Repair 2002;16:241–248. PMID:12234087, DOI:10.1177/154596830201600303
• 12.  Domenick Sridharan N, Chaer RA, Thirumala PD, Balzer J, Long B, Tzeng E, Makaroun MS, Avgerinos ED: Somatosensory evoked potentials and electroencephalography during carotid endarterectomy predict late stroke but not death. Ann Vasc Surg 2017;38:105–112. PMID:27521822, DOI:10.1016/j.avsg.2016.07.064
• 13.  De Bruyn N, Saenen L, Thijs L, Van Gils A, Ceulemans E, Essers B, Lafosse C, Michielsen M, Beyens H, Schillebeeckx F, Alaerts K, Verheyden G: Sensorimotor vs. motor upper limb therapy for patients with motor and somatosensory deficits: a randomized controlled trial in the early rehabilitation phase after stroke. Front Neurol 2020;11:597666. PMID:33343498, DOI:10.3389/fneur.2020.597666
• 14.  Woodbury ML, Velozo CA, Richards LG, Duncan PW: Rasch analysis staging methodology to classify upper extremity movement impairment after stroke. Arch Phys Med Rehabil 2013;94:1527–1533. PMID:23529144, DOI:10.1016/j.apmr.2013.03.007
• 15.  Kim M, Lee HH, Lee J: Does isolated somatosensory impairment affect the balance and ambulation of patients with supratentorial stroke after the acute phase? J Clin Neurosci 2020;74:109–114. PMID:32051098, DOI:10.1016/j.jocn.2020.01.084
• 16.  Mori E, Mitani Y, Yamadori A: Usefulness of a Japanese version of the Mini-Mental State Test in neurological patients [in Japanese]. Neuropsychology 1985;1:2–10.
• 17.  von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, STROBE Initiative: The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Epidemiology 2007;18:800–804. PMID:18049194, DOI:10.1097/EDE.0b013e3181577654
• 18.  Tobimatsu S, Fukui R, Kato M, Kobayashi T, Kuroiwa Y: Multimodality evoked potentials in patients and carriers with adrenoleukodystrophy and adrenomyeloneuropathy. Evok Pot 1985;62:18–24. PMID:2578374, DOI:10.1016/0168-5597(85)90031-0
• 19.  Barker RN, Brauer S, Carson R: Training-induced changes in the pattern of triceps to biceps activation during reaching tasks after chronic and severe stroke. Exp Brain Res 2009;196:483–496. PMID:19504088, DOI:10.1007/s00221-009-1872-8
• 20.  Frisoli A, Procopio C, Chisari C, Creatini I, Bonfiglio L, Bergamasco M, Rossi B, Carboncini M: Positive effects of robotic exoskeleton training of upper limb reaching movements after stroke. J Neuroeng Rehabil 2012;9:36. PMID:22681653, DOI:10.1186/1743-0003-9-36
• 21.  Bandini V, Carpinella I, Marzegan A, Jonsdottir J, Frigo CA, Avanzino L, Pelosin E, Ferrarin M, Lencioni T: Surface-electromyography-based co-contraction index for monitoring upper limb improvements in post-stroke rehabilitation: a pilot randomized controlled trial secondary analysis. Sensors (Basel) 2023;23:7320. PMID:37687775, DOI:10.3390/s23177320
• 22.  Rosa MC, Marques A, Demain S, Metcalf CD: Lower limb co-contraction during walking in subjects with stroke: a systematic review. J Electromyogr Kinesiol 2014;24:1–10. PMID:24246405, DOI:10.1016/j.jelekin.2013.10.016
• 23.  Ervilha UF, Graven-Nielsen T, Duarte M: A simple test of muscle coactivation estimation using electromyography. Braz J Med Biol Res 2012;45:977–981. PMID:22641413, DOI:10.1590/S0100-879X2012007500092
• 24.  Morillas F, Valero J: Random resampling numerical simulations applied to a SEIR compartmental model. Eur Phys J Plus 2021;136:1067. PMID:34722097, DOI:10.1140/epjp/s13360-021-02003-9
• 25.  Tamura Y, Ueki Y, Lin P, Vorbach S, Mima T, Kakigi R, Hallett M: Disordered plasticity in the primary somatosensory cortex in focal hand dystonia. Brain 2009;132:749–755. PMID:19151081, DOI:10.1093/brain/awn348
• 26.  Keren O, Ring H, Solzi P, Pratt H, Groswasser Z: Upper limb somatosensory evoked potentials as a predictor of rehabilitation progress in dominant hemisphere stroke patients. Stroke 1993;24:1789–1793. PMID:8248956, DOI:10.1161/01.STR.24.12.1789
• 27.  Prakash R, Song Y, Reddy J: Prognostic value of somatosensory-evoked potentials in neurology: a critical review in hypoxic encephalopathy. Neurol India 2016;64:396–404. PMID:27147145, DOI:10.4103/0028-3886.181555
• 28.  Jones RD, Donaldson IM, Parkin PJ: Impairment and recovery of ipsilateral sensory-motor function following unilateral cerebral infarction. Brain 1989;112:113–132. PMID:2917274, DOI:10.1093/brain/112.1.113
• 29.  Brodal A: Self-observations and neuro-anatomical considerations after a stroke. Brain 1973;96:675–694. PMID:4773860, DOI:10.1093/brain/96.4.675
• 30.  Noskin O, Krakauer JW, Lazar RM, Festa JR, Handy C, O’Brien KA, Marshall RS: Ipsilateral motor dysfunction from unilateral stroke: implications for the functional neuroanatomy of hemiparesis. J Neurol Neurosurg Psychiatry 2008;79:401–406. PMID:17635970, DOI:10.1136/jnnp.2007.118463
• 31.  Delvaux V, Alagona G, Gérard P, De Pasqua V, Pennisi G, de Noordhout AM: Post-stroke reorganization of hand motor area: a 1-year prospective follow-up with focal transcranial magnetic stimulation. Clin Neurophysiol 2003;114:1217–1225. PMID:12842718, DOI:10.1016/S1388-2457(03)00070-1
• 32.  Kwon BM, Lee Y, Lee HH, Ko N, Kim H, Kim BR, Moon WJ, Lee J: Relationship between ipsilesional upper extremity motor function and corpus callosum integrity in patients with unilateral stroke: a diffusion tensor imaging study. Brain Neurorehabil 2022;15:e7. PMID:36743845, DOI:10.12786/bn.2022.15.e7
• 33.  Johansen-Berg H, Della-Maggiore V, Behrens TE, Smith SM, Paus T: Integrity of white matter in the corpus callosum correlates with bimanual co-ordination skills. Neuroimage 2007;36:T16–T21. PMID:17499163, DOI:10.1016/j.neuroimage.2007.03.041
• 34.  Wang Y, Zhong L, Jin M, Liao D, Privitera AJ, Wong AY, Fong GC, Bao SC, Sun R: Assessing stroke-induced abnormal muscle coactivation in the upper limb using the surface EMG co-contraction Index: a systematic review. J Electromyogr Kinesiol 2025;81:102985. PMID:39847816, DOI:10.1016/j.jelekin.2025.102985