Individual and combined applied robotic hand rehabilitation and conventional rehabilitation for post-stroke hemiplegia: a prospective three-arm randomized study

Abstract

BACKGROUND

Conventional rehabilitation (CR) and robotic hand rehabilitation (RR) are effective rehabilitation options for post-stroke hemiplegia. Combining these two rehabilitation protocols may positively affect the results.

AIM

To compare the effectiveness of individual and combined applications of CR and RR on hand and upper extremity function, spasticity, grip and tip pinch strength in the treatment of post-stroke hemiplegia.

DESIGN

A three-arm, prospective randomized controlled trial.

SETTING

Multidisciplinary rehabilitation facility.

POPULATION

Sixty-six patients with post-stroke hemiplegia.

METHODS

Participants were randomized into three groups: RR, CR, and combined conventional and robotic hand rehabilitation (CR-RR). For the RR group, a passive and active finger motion program (40 minutes a day), along with a home-based rehabilitation program, was administered using the robotic device. For the CR group, traditional manual therapy techniques were used to improve activities of daily living and to achieve isometric contractions in weak muscles (60 minutes a day). For the CR-RR group, 60 minutes of conventional rehabilitation was followed by 40 minutes of hand rehabilitation using the robotic device. The patients in all groups were rehabilitated for 1 month, 5 days a week. The Fugl-Meyer Upper Extremity Assessment Scale was used as primary outcome measure for evaluating the upper extremity function. The secondary outcome was evaluated based on hand function, upper extremity and hand motor function, activities of daily living (ADLs), upper extremity spasticity, and hand strength.

RESULTS

CR-RR was significantly more effective than both CR and RR in improving upper extremity function, hand function, and ADLs. The only group that had no effect on upper extremity function was RR. All three rehabilitation protocols were ineffective in treating shoulder adductor and elbow flexor spasticity. While all three protocols effectively reduced spasticity in the wrist, fingers, and thumb flexors, CR was significantly less effective compared to the other groups. Although the CR-RR and RR groups had similar results in reducing wrist and finger flexor spasticity, the CR-RR group was significantly more effective than the RR group in reducing thumb spasticity. CR did not influence grip and tip pinch strength, whereas both CR-RR and RR led to significant improvements.

CONCLUSIONS

RR has no effect on upper extremity functional results. CR has no effect on grip and pinch strength and only a limited effect on wrist, finger, and thumb flexor spasticity. Combining conventional rehabilitation with robotic rehabilitation in the CR-RR protocol yields better results in daily living activities, upper extremity function, hand function, wrist, finger, and thumb flexor spasticity, as well as hand grip and tip pinch strength in post-stroke hemiplegia.

CLINICAL REHABILTATION IMPACT

Concomitant application of RR and CR achieves better outcomes in post-stroke hemiplegia.

Hemiplegia or hemiparesis is commonly seen after a stroke. Loss of motion, coordination, sensory and dexterity in the upper extremity causes difficulties in activities of daily living (ADLs). Upper extremity function, particularly hand function, plays a crucial role in determining the quality of life and independence after stroke. Numerous studies have been conducted on rehabilitation to improve upper extremity function. However, 65% of stroke survivors experience difficulties in ADLs due to upper extremity impairment despite various rehabilitation programs.¹ The inability to use the upper extremity in daily life leads a decline in both quality of life and social participation.² Studies report that loss of hand function is one of the most unfortunate long-term consequences of stroke.³

CR is most commonly preferred rehabilitation protocol after stroke. It involves a series of repetitive and intensive exercises. Patients perform partial or full-assisted motion in upper extremity and hand joints under the supervision of therapists. These exercises are effective in improving gross motor function and daily functional skills. However, deficits in upper extremity functions, especially hand functions, often remain.^{4, 5} Numerous studies have focused on rehabilitation, especially hand rehabilitation in order to improve outcomes after stroke. Due to the increasing number of stroke patients, there are therapist inadequacies in therapist-assisted CR. The number of therapist sessions is insufficient to meet increasing demand, necessitating the exploration of alternative rehabilitation methods that do not rely solely on therapists. Clinicians and engineers have collaborated to develop devices aimed at treating upper extremity dysfunction, with the main goal of restoring the patient’s intuitive control over their own limbs. Robotic rehabilitation provides cost-effective, repetitive and independent rehabilitation with less direct therapist supervision. While CR can be time-consuming for therapists, RR allows therapists to supervise the rehabilitation of more patients simultaneously.⁶ RR has been found to be more successful in improving hand function than in rehabilitating other upper extremity joints.⁷ Additionally, RR positively impacts both functional recovery of the hand and the reduction of motor impairments.⁸ Despite these advantages, there is no definitive evidence that RR is more effective than CR.⁹

There are numerous studies in the literature on the effects of CR and RR on hand and upper extremity functions in stroke patients. These studies have shown that both therapies positively influence outcomes. However, no comparative study has investigated the effects of CR and RR when applied individually and in combination on hand and upper extremity functions. It is possible that better hand function can be achieved by combining CR and RR in stroke rehabilitation. The aim of our study is to compare the early-term effects of individually and combined CR and RR on hand and upper extremity functions and hand strength in stroke patients. Our hypothesis is that the combined application of CR and RR will yield better functional outcomes of the hand and upper extremity.

Materials and methods

Study design

This prospective randomized single blind comparison study was performed in the rehabilitation department of Gaziosmanpaşa Research and Training Hospital.

Ethical considerations

This study was approved by the Ethics Committee of Gaziosmanpaşa Training and Research Hospital (2022-143) and was performed in accordance with the tenets of the Declaration of Helsinki.¹⁰ All methods were performed according to relevant guidelines and regulations. After informing the patients about the possible side effects of the treatments, written informed consent was obtained from each patient.

Randomization and blinding

The original study was a prospective randomized single blind comparison study. No changes were made to the design or protocol of the study. Patients were randomly assigned to one of three groups:

• group 1 (RR): the patients underwent robotic hand rehabilitation;

• group 2 (CR): the patients underwent conventional rehabilitation;

• group 3 (combined rehabilitation: CR-RR): the patients underwent combined conventional and robotic hand rehabilitation at the same period.

Randomization was performed using the closed envelope method by an investigator who was not involved in patient care. Group allocation was concealed in sealed opaque envelopes, which were handed an investigator who was not involved in patient care or assessment. The therapist opened the envelope 6 hours prior to the start of the patient’s treatment.

The same physiotherapist, who was blinded to the group assignments, conducted both the pre-treatment and 1^st month follow-up evaluations of the patients.

Participants

The primary outcome of our study was functional score at the first months. We calculated the sample size based on the Fugl-Meyer upper extremity assessment scale of a previous study.¹¹

Assuming a 7-point improvement in Fugl-Meyer upper extremity assessment scale in the concomitant rehabilitation group with a α of 0.05, and a β error of 0.8, the study required 20 patients in each group. Allowing a 10% dropout rate, 22 patients were recruited in each group.

Sixty-six patients with post-stroke hemiplegia who would undergo rehabilitation at Gaziosmanpaşa Training and Research Hospital were included in our study between November 2022 and May 2023.

Inclusion criteria

• Diagnosed with cortical or subcortical stroke confirmed by radiological imaging;

• over 18 years of age;

• Mini-Mental Scale (MMS) = 24-30;

• the rehabilitation of the patient was performed at the rehabilitation department of Gaziosmanpaşa Research and Training Hospital.

Exclusion criteria

• Having a stroke lesion in other areas of the brain such as the cerebellum and brainstem;

• presence of joint or bone pathologies in the affected upper extremity;

• excessive spasticity in the elbow or wrist joints of the affected upper extremity (Modified Ashworth Scale score >3);

• having previously surgical treatment history on the affected extremity;

• having an additional neurological disease;

• not complying with treatment recommendations or inadequately applying;

• severe cognitive deficits (Levels of Cognitive Functioning-Revised, LCF-R <6).

Rehabilitation protocols

Robotic hand rehabilitation

RR group received hand rehabilitation through the robotic device Gloreha (Figure 1).¹¹ A home-based rehabilitation program was used to these patients for hand and upper extremity rehabilitation in addition to RR.

Figure 1.

—Hand exercises were performed with each finger mobilized individually using the robotic device Gloreha, with the presence of visual feedback.

Each rehabilitation session consisted of six parts:¹²

• a sequence of 17 cycles of digital flexion-extension joint motions, from the thumb to the fifth finger (7 min);

• a sequence of 23 cycles of motion to counting from one to five (7 min);

• a sequence of 70 cycles of motions including thumb-finger opposition motions from the second to the fifth finger (7 min);

• a sequence of 28 cycles of motions including wave-like finger motions (7 min);

• a sequence of 42 cycles of motions including fist opening/closing (7 min);

• a sequence of 20 cycles of motions including flexion-extension of the fingers alternated with flexion-extension of the thumb (5 min).

The patients underwent RR in the hospital 5 days per week for 1 month.

In Group 1, patients underwent a home-based intervention based on the Bobath concept in addition to RR.¹² An experienced team of physicians and physiotherapists designed a personalized home-based rehabilitation program for each patient. This program included facilitation of movements such as scapular protraction, shoulder flexion, elbow extension, wrist extension, and finger extension and opposition. The physiotherapist provided instructions for the program both verbally and through written brochures. Patients participated in the home-based rehabilitation program for 1-hour sessions every day.

Conventional rehabilitation

CR group received 60 minutes of consecutive occupational therapy sessions at the hospital, 5 days per week for 1 month. The rehabilitation program included strength, balance, manual dexterity exercises, and stretching/weight-bearing exercises for the affected arm. Treatments focused on practice of specific functional tasks when possible. These tasks included reaching and grasping various objects, isolated hand motions (writing, playing instruments, molding putty, cooking), and whole-body activities (swinging a racquet or basketball handling skills). The rehabilitation program also included training in ADLs. Each patient followed individualized programs based on their assessment and personal goals. Manual therapy techniques were used to obtain isometric contractions in weak muscles. Additionally, patients received mobilization and stretching exercises to restricted joints, as needed, to increase their range of motion.¹³

Combined rehabilitation

The patients underwent 60 minutes of CR followed by 40 minutes of hand rehabilitation through the robotic device Gloreha. A home-based rehabilitation program was not included for this group.

The durations of the conventional and robotic rehabilitation protocols were determined based on an evaluation of similar studies with proven success in the literature.^12-14

Evaluation

The primary outcome of our study was evaluated using the Fugl-Meyer Upper Extremity Assessment Scale to assess upper extremity function.

The secondary outcome of our study was assessed based on hand function, upper extremity and hand motor function, ADLs, upper extremity spasticity, and hand strength. Evaluations were conducted using Brunnstrom’s Hemiplegia Recovery Staging for Upper Extremity Motor Function, Brunnstrom’s Hemiplegia Recovery Staging for Hand Motor Function, the Fugl-Meyer Upper Extremity Assessment Scale Hand Evaluation, and the ABILHAND Stroke Hand Ability Questionnaire. The Barthel Index for Activities of Daily Living was used to evaluate daily living activities. Spasticity was assessed using the Modified Ashworth Scale. Grip strength was measured with a dynamometer (Jamar^®, Preston^™, Jackson, MI, USA), and pinch strength was measured with a specially designed instrument (Pinch Gauge, B&R Engineering^™, Santa Fe Springs, CA, USA). Measurements were taken three times, and the average value was recorded.

The patients were evaluated before rehabilitation and at the 1^st month follow-up. The same therapist who was blind to the type of rehabilitation evaluated the patients before treatment and at the 1^st month follow-up.

Statistical analysis

Statistical analysis was performed using the Statistical Package for the Social Sciences for Windows (Version 22.0; SPSS Inc., Chicago, IL, USA). The conformity of the parameters to a normal distribution was assessed using the Kolmogorov-Smirnov and Shapiro-Wilk Tests. In addition to descriptive statistical methods, a one-way ANOVA test was used to compare quantitative data and parameters with normal distribution between groups. The Kruskal-Wallis test was used for comparing parameters that did not follow a normal distribution between groups, and Dunn’s Test was applied to determine the group responsible for the difference. The Wilcoxon signed-rank test was used to compare parameters that did not show a normal distribution within groups. The Chi-square Test and Fisher Freeman Halton Exact Chi-Square Test were used for comparing qualitative data. Differences were considered statistically significant at P values <0.05.

Responsiveness

In addition to the statistical evaluation, the responsiveness of the evaluation of upper extremity function, hand function, ADLs, spasticity, grip and tip pinch strength in patients with stroke was measured by the SRM. The change scores in different assessments could be measured with a standardized unit by the SRM. The SRM was classified using Cohen’s criteria. The magnitude of SRM indicates the strength of the change. The value below 0.2 were considered unresponsive, while those between 0.2 and 0.5 were classified as mildly responsive. The value ranging from 0.5 to 0.8 were regarded as moderately responsive, and those exceeding 0.8 were considered markedly responsive. A high SRM suggests a strong change and indicates that the treatment may be effective.¹⁵

MCID estimation

When a patient’s condition improves, the MCID is scaled positively, whereas it is scaled negatively when the patient’s condition worsens. A distribution-based method was adopted to measure MCIDs. There are various methods for estimating MCID, one of which is effect size. Bias can be reduced by standardizing MCID to effect size. The proportion of patients whose score changes exceeded the distribution-based MCID thresholds was examined. In different studies, these thresholds varied based on effect sizes of 0.2 SD, 0.5 SD, or 0.8 SD. In this study, effect sizes of 0.5 SD and 0.8 SD were used.

Data availability

The data associated with the paper are not publicly available but are available from the corresponding author on reasonable request.

Results

Sixty-six patients with post-stroke hemiplegia were included in our study. Three patients were excluded during the follow-up period: one had a new cerebral hemorrhage, one had experienced a myocardial infarction, 1one discontinued rehabilitation during the second week. Therefore, the pre-treatment and 1^st month follow-up results of the remaining 63 patients were evaluated. A total of 21 patients could be evaluated at the 1^st month follow-up in each group (Figure 2). No side effects were observed in any of the patients after the rehabilitation.

Figure 2.

—Flow chart for patient recruitment and follow-up.

A total of 63 patients, 30 females (47.6%) and 33 males (52.4%), aged between 53 and 89 years, were included in our study. The mean age was 67.57±8.50 years. There were no significant differences in age, sex, body mass index, stroke type, time between stroke and start of rehabilitation, affected body side, and dominant/non-dominant hand and intensive care period between the groups (P>0.05) (Table I).

Table I. —Demographic characteristics of the groups.

Demographics	Group RR	Group CR	Group CR-RR	P value
Age, years	67.19±7.83	66.05±9.14	69.48±8.55	0.420 a
BMI, kg/m2	28.14±4.74	27.28±3.65	26.80±3.78	0.560 a
Sex
Female	10 (47.6%)	11 (52.4%)	9 (42.9%)	0.826 b
Male	11 (52.4%)	10 (47.6%)	12 (57.1%)
Stroke type
Ischemic	15 (71.4%)	18 (85.7%)	19 (90.5%)	0.343 c
Hemorrhagic	6 (28.6%)	3 (14.3%)	2 (9.5%)
Period stroke to rehabilitation (months)	20.05±26.91	16.57±19.68	12.57±19.76	0.349 d
Extremity
Right	11 (52.4%)	10 (47.6%)	5 (23.8%)	0.131 b
Left	10 (47.6%)	11 (52.4%)	16 (76.2%)
Affected party dominance
Dominant	11 (52.4%)	11 (52.4%)	6 (28.6%)	0.200 b
Non-dominant	10 (47.6%)	10 (47.6%)	15 (71.4%)
Intensive care period (day)	11.76±15.83	14.86±19.85	8.28±14.0	0.557 d

BMI: Body Mass Index. ^a One-way ANOVA test; ^b Chi-square Test; ^c Fisher Freeman Halton Exact Test; ^d Kruskal Wallis Test.

Evaluation of upper extremity function

There were no statistically significant differences between the groups in terms of Brunnstrom’s Hemiplegia Recovery Staging for Upper Extremity Motor Staging and Fugl-Meyer upper extremity assessment scale scores before treatment (P>0.05) (Table II).

Table II. —The evaluation of the upper extremity function.

Scale	Timepoint	Group RR	Group CR	Group CR-RR	P value a
Fugl-Meyer upper extremity assessment scale	Pre-rehabilitation	10.1±10.68	12.29±11.93	11.33±11.86	0.859
	1st month	10.38±10.89	21.67±14.42	23.86±14.13	0.001*
	P value b	0.125	0.001*	0.001*
Brunnstrom’s Hemiplegia Recovery Staging Upper Extremity Motor Staging	Pre-rehabilitation	2.38±1.5	3±2.19	2.19±1.83	0.263
	1st month	2.52±1.66	3.95±2.13	4.52±2.5	0.008*
	P value b	0.180	0.001*	0.001*

^a Kruskal Wallis Test; ^b Wilcoxon Sign Test; *statistically significant difference.

In RR group, there were no statistically significant differences in Brunnstrom’s Hemiplegia Recovery Staging for Upper Extremity Motor Staging and Fugl-Meyer upper extremity assessment scale scores at 1^st month follow-up compared to pre-treatment (P>0.05). However, in CR and CR-RR groups, a statistically significant improvement was observed (P=0.001) (Table II).

Markedly responsive (SRM>0.8) results were achieved in the CR and CR-RR groups in terms of Brunnstrom’s Hemiplegia Recovery Staging for Upper Extremity Motor Staging and Fugl-Meyer upper extremity assessment scale scores (Table III).

Table III. —The responsiveness and MCID estimates for the evaluation of upper extremity function using various assessment scales.

Assessment scale	Group RR			Group CR			Group CR-RR
		MCID estimates			MCID estimates			MCID estimates
	SRM	0.5 SD Score	0.8 SD Score	SRM	0.5 SD Score	0.8 SD Score	SRM	0.5 SD Score	0.8 SD Score
Fugl-Meyer upper extremity assessment scale	0.44	5.34	8.54	1.00*	5.96	9.54	0.82*	5.93	9.48
Brunnstrom’s Hemiplegia Recovery Staging Upper Extremity Motor Staging	0.22	0.75	1.20	1.18*	1.09	1.75	1.64*	0.91	1.46
Brunnstrom’s Hemiplegia Recovery Staging Hand Motor Staging	1.00*	0.65	1.04	1.08*	0.53	0.84	1.55*	0.75	1.20
Fugl-Meyer upper extremity assessment scale Hand Evaluation	1.19*	1.40	2.24	1.23*	1.6	2.56	1.16*	1.91	3.06
ABILHAND Stroke Hand Ability Questionnaire	1.58*	1.41	2.25	1.99*	1.64	2.62	1.52*	1.98	3.16
Barthel Index for Activities of Daily Living	1.05*	15.63	25.00	1.31*	13.83	22.12	1.60*	13.07	20.91

*Statistically significant difference.

A statistically significant difference was found between the groups in terms of Brunnstrom’s Hemiplegia Recovery Staging for Upper Extremity Motor Staging score (p=0.008) and Fugl-Meyer upper extremity assessment scale score (P=0.001) at 1^st month follow-up (Table II). Post-hoc tests showed that the 1^st month Brunnstrom’s Hemiplegia Recovery Staging for Upper Extremity Motor Staging score in the CR group (P=0.041) and CR-RR group (P=0.013) was significantly higher compared to RR group. Similarly, the Fugl-Meyer upper extremity assessment scale score were significantly higher in the CR group (P=0.005) and CR-RR group (P=0.001) than in RR group. There were no statistically significant differences between CR and CR-RR groups (P>0.05).

Evaluation of hand function

There were no statistically significant differences between the groups in terms of Brunnstrom’s Hemiplegia Recovery Staging Hand Motor Staging, Fugl-Meyer upper extremity assessment scale Hand Evaluation, and ABILHAND Stroke Hand Ability Questionnaire scores before treatment (P>0.05) (Table IV).

Table IV. —The evaluation of the hand function.

Parameter	Timepoint	Group RR	Group CR	Group CR-RR	P value a
Brunnstrom’s Hemiplegia Recovery Staging Hand Motor Staging	Pre-rehabilitation	2.24±1.3	2.33±1.06	2.1±1.51	0.472
	1st month	3.1±1	3.19±1.08	4.14±0.65	0.001*
	P value b	0.002*	0.001*	0.001*
Fugl-Meyer upper extremity assessment scale Hand Evaluation	Pre-rehabilitation	2.43±2.8	2.33±3.2	2.81±3.83	0.754
	1st month	4.86±4.34	4.81±3.41	7.86±3.69	0.018*
	P value b	0.001*	0.001*	0.001*
ABILHAND Stroke Hand Ability Questionnaire	Pre-rehabilitation	5.33±2.82	4.48±3.28	6.19±3.96	0.281
	1st month	8.52±3.23	8.95±4.57	12.33±4.49	0.009*
	P value b	0.001*	0.001*	0.001*

^a Kruskal Wallis Test; ^b Wilcoxon Sign Test; *statistically significant difference.

In all groups, there was statistically significant improvement in Brunnstrom’s Hemiplegia Recovery Staging Hand Motor Staging, Fugl-Meyer upper extremity assessment scale Hand Evaluation, and ABILHAND Stroke Hand Ability Questionnaire scores at 1^st month follow-up compared to pre-treatment (P<0.05) (Table IV).

Markedly responsive (SRM>0.8) results were achieved in all groups in terms of the Brunnstrom’s Hemiplegia Recovery Staging Hand Motor Staging, Fugl-Meyer upper extremity assessment scale Hand Evaluation, and ABILHAND Stroke Hand Ability Questionnaire scores (Table III).

A statistically significant difference was found between the groups at the 1^st month follow-up in terms of Brunnstrom’s Hemiplegia Recovery Staging for Hand Motor Staging (P=0.001), Fugl-Meyer upper extremity assessment scale Hand Evaluation (P=0.018), and ABILHAND Stroke Hand Ability Questionnaire scores (P=0.009) (Table IV). Post-hoc tests showed that the 1^st month Brunnstrom’s Hemiplegia Recovery Staging for Hand Motor Staging score in CR-RR group was significantly higher compared to RR group (P=0.002) and CR group (P=0.006). Similarly, the Fugl-Meyer upper extremity assessment scale Hand Evaluation score was significantly higher in CR-RR group than in RR group (P=0.041) and CR group (P=0.045). ABILHAND Stroke Hand Ability Questionnaire scores were also significantly higher in the CR-RR group than in RR group (P=0.020) and CR group (P=0.028).

There were no statistically significant differences between RR and CR groups (P>0.05).

Evaluation of ADLs

There were no statistically significant differences between the groups in terms of pre-treatment Barthel Index for Activities of Daily Living scores (P=0.942). In all groups, the increase in the 1^st month score compared to the pre-treatment Barthel Index for Activities of Daily Living score is statistically significant (P=0.001) (Table V).

Table V. —The evaluation of daily living activity.

Barthel Index for Activities of Daily Living	Group RR	Group CR	Group CR-RR	P value a
Pre-rehabilitation	41.19±31.26	43.57±27.66	40.71±26.14	0.942
1st month	44.28±31.08	62.38±19.85	66.19±20.12	0.037*
P value b	0.001*	0.001*	0.001*

^a Kruskal Wallis Test; ^b Wilcoxon Sign Test; *statistically significant difference.

Markedly responsive (SRM > 0.8) results were achieved in all groups for the Barthel Index for Activities of Daily Living score (Table III).

A statistically significant difference was found between the groups regarding the 1^st month Barthel Index for Activities of Daily Living scores after treatment (P=0.037) (Table V). Post hoc tests revealed that the 1^st month Barthel Index for Activities of Daily Living scores in the CR-RR group were significantly higher than in the RR group (P=0.045), with no statistically significant difference between the other groups (P>0.05).

The evaluation of spasticity

Modified Ashworth Scale scores of shoulder adduction, elbow flexor

Nineteen patients with shoulder adduction spasticity were evaluated in the RR group, 16 patients in the CR group, and 19 patients in the CR-RR group. Additionally, 17 patients with elbow flexor spasticity were assessed in the RR group, 18 in the CR group, and 16 in the CR-RR group. There were no statistically significant differences between the groups before treatment and at the 1^st month follow-up (P>0.05). Similarly, no statistically significant difference was observed within any of the groups between pre-treatment and the 1^st month follow-up (P>0.05) (Table VI). It was determined that there was no improvement in Shoulder Adduction and Elbow Flexor spasticity after treatment in the groups.

Table VI. —The evaluation of Modified Ashworth Scale.

Parameter	Timepoint	Group RR	Group CR	Group CR-RR	P value a
Shoulder adductor	Pre-rehabilitation	1.42±0.69	1.37±0.71	1.36±0.76	0.861
	1st month	1.26±0.73	1.18±0.65	1.15±0.37	0.840
	P value b	0.250	0.250	0.125
Elbow flexor	Pre-rehabilitation	1.18±0.53	1.44±0.7	1.38±0.5	0.389
	1st month	1.06±0.6	1.17±0.7	1.13±0.5	0.737
	P value b	0.500	0.156	0.125
Wrist flexor	Pre-rehabilitation	1.35±0.61	1.41±0.62	1.33±0.49	0.940
	1st month	0.53±0.51	1.12±0.6	0.33±0.49	0.001*
	P value b	0.001*	0.025*	0.001*
Finger flexor	Pre-rehabilitation	1.5±0.71	1.4±0.6	1.25±0.45	0.588
	1st month	0.61±0.7	1.15±0.59	0.31±0.48	0.001*
	P value b	0.001*	0.025*	0.001*
Thumb flexor	Pre-rehabilitation	1.59±0.8	1.3±0.66	1.29±0.47	0.489
	1st month	1.18±0.53	1.05±0.69	0.41±0.51	0.001*
	P value b	0.020*	0.025*	0.001*

^a Kruskal Wallis Test; ^b Wilcoxon Sign Test; *statistically significant difference.

Markedly responsive (SRM>0.8) results were not achieved in any group for Modified Ashworth Scale scores of shoulder adduction and elbow flexors (Table VII).

Table VII. —The responsiveness and MCID estimates of the Modified Ashworth Scale in patients with stroke.

Parameter	Group RR			Group CR			Group CR-RR
		MCID estimates			MCID estimates			MCID estimates
	SRM	0.5 SD Score	0.8 SD Score	SRM	0.5 SD Score	0.8 SD Score	SRM	0.5 SD Score	0.8 SD Score
Shoulder adductor	0.42	0.35	0.55	0.43	0.36	0.57	0.33	0.38	0.61
Elbow flexor	0.24	0.27	1.20	0.42	0.27	0.56	0.65	0.25	0.40
Wrist flexor	1.56*	0.31	0.49	0.63	0.31	0.50	2.65*	0.25	0.39
Finger flexor	1.52*	0.36	0.57	0.56	0.30	0.48	1.63*	0.23	0.36
Thumb flexor	0.45	0.40	0.64	0.58	0.33	0.53	1.47*	0.24	0.38

*Statistically significant difference.

Modified Ashworth Scale scores of wrist flexor, finger flexor

Seventeen patients with wrist flexor spasticity were evaluated in the RR group, 17 patients in the CR group, and 15 patients in the CR-RR group. Additionally, 18 patients with finger flexor spasticity were assessed in the RR group, 20 in the CR group, and 16 in the CR-RR group. There were no statistically significant differences between the groups before treatment (P>0.05) (Table VI). In all groups, there was statistically significant improvement at 1^st month follow-up compared to pre-treatment (P<0.05) (Table VI). There was a statistically significant improvement between the groups in terms of the Modified Ashworth Scale scores for wrist flexor and finger flexor at the 1^st month follow-up (P=0.001) (Table VI). Post hoc tests showed that, for the Modified Ashworth Scale score of the wrist flexor, the 1^st month follow-up scores of the RR group (P=0.038) and CR-RR group (P=0.001) were significantly better than those of the CR group.

Similarly, for the Modified Ashworth Scale score of the finger flexor, the 1^st month scores of the RR group (P=0.029) and CR-RR group (P=0.001) were significantly better than CR group. No statistically significant difference was found between CR-RR and RR groups in terms of the Modified Ashworth Scale scores for wrist flexor and finger flexor at the 1^st month follow-up (P>0.05).

Markedly responsive (SRM>0.8) results were achieved in RR group and CR-RR group for Modified Ashworth Scale scores of wrist flexor, finger flexor (Table VII).

Modified Ashworth Scale scores of thumb flexor

Seventeen patients with thumb flexor spasticity were evaluated in the RR group, 19 patients in the CR group, and 17 patients in the CR-RR group. There were no statistically significant differences between the groups before treatment (P=0.489) (Table VI). In all groups, there was statistically significant improvement at 1^st month follow-up compared to pre-treatment (P<0.05) (Table VI). There was a statistically significant improvement between the groups in terms of the Modified Ashworth Scale scores for thumb flexor at the 1^st month follow-up (P=0.001) (Table VI). Post-hoc tests showed that, for the Modified Ashworth Scale score of thumb flexor, the 1^st month follow-up scores of the CR-RR group were significantly better than those of RR group (P=0.002) and CR group (P=0.010). No statistically significant difference was found between RR and CR groups in terms of the Modified Ashworth Scale scores for thumb flexor at the 1^st month follow-up (P>0.05).

Markedly responsive (SRM>0.8) results were achieved only in CR-RR group for Modified Ashworth Scale Scores of Thumb flexor (Table VII).

Evaluation of grip and tip pinch strength

There were no statistically significant differences between the groups in terms of pre-treatment grip strength and tip pinch strength (P>0.05) (Table VIII).

Table VIII. —The evaluation of hand grip and pinch strength.

Parameter	Timepoint	Group RR	Group CR	Group CR-RR	P value a
Grip strength (kg)	Pre-rehabilitation	1.19±3.12	0.98±1.7	1.38±1.71	0.268
	1st month	4.28±3.4	1.52±2.11	5.71±3.48	0.002*
	P value b	0.001*	0.062	<0.001*
Tip pinch strength (kg)	Pre-rehabilitation	0.64±0.92	0.54±0.92	0.90±1.31	0.691
	1st month	2.85±2.67	1.09±2.01	3.16±2.73	0.012*
	P value b	0.001*	0.125	0.002*

^a Kruskal Wallis Test; ^b Wilcoxon Sign Test; *statistically significant difference.

In CR group, there were no statistically significant differences in grip strength and tip pinch strength scores at 1^st month follow-up compared to pre-treatment (P>0.05). However, in RR and CR-RR groups, a statistically significant improvement was observed (P<0.05) (Table VIII).

Markedly responsive (SRM>0.8) results were achieved in RR group and CR-RR group in terms of grip and tip pinch strength (Table IX).

Table IX. —The responsiveness and MCID estimates of hand grip and pinch strength in patients with stroke.

Parameter	RR Group			CR Group			CR-RR Group
		MCID estimates			MCID estimates			MCID estimates
	SRM	0.5 SD Score	0.8 SD Score	SRM	0.5 SD Score	0.8 SD Score	SRM	0.5 SD Score	0.8 SD Score
Grip strength	1.30	1.56	2.50	0.68	0.85	1.36	1.70	0.86	1.37
Tip pinch strength	0.94	0.46	0.74	0.43	0.46	0.74	0.89	0.66	1.05

A statistically significant difference was found between the groups in terms of grip strength and tip pinch strength score (P<0.05) at 1^st month follow-up (Table VIII). Post hoc tests showed that the 1^st month grip strength score in the RR group (P=0.006) and CR-RR group (P<0.001) was significantly higher compared to the CR group. Similarly, tip pinch strength score were significantly higher in RR group (P=0.038) and CR-RR group (P=0.017) than in CR group. There were no statistically significant differences between RR and CR-RR groups (P>0.05).

Discussion

CR-RR is more effective than either CR or RR alone in improving hand and upper extremity function, ADLs, spasticity in the wrist, finger, and thumb flexors, as well as grip and tip pinch strength in stroke patients. While CR significantly improves ADLs, upper extremity and hand function, and spasticity in the wrist, finger, and thumb flexors, it is not as successful as CR-RR. CR also fails in improving grip and tip pinch strength. RR, combined with home-based rehabilitation, is effective for ADLs and hand function but is no effective for upper extremity function. Although RR reduces spasticity in the wrist, finger, and thumb flexors, it is not as effective as CR-RR in addressing thumb flexor spasticity. However, it does improve handgrip and tip pinch strength similarly to CR-RR.

Upper extremity robotic rehabilitation has recognized as a safe and feasible treatment to complement post-stroke rehabilitation. Various research groups have developed robotic devices that provide passive and/or active motions for upper extremity rehabilitation.⁷ Many studies have conducted on the effect of the robotic rehabilitation devices on treatment. CR, a frequently used rehabilitation method in stroke rehabilitation, was generally preferred as the control group. Numerous studies have compared CR with robotic rehabilitation.

A large multicenter, randomized, controlled trial investigated the effect of equal doses of robotic and conventional therapy on the results. There were no significant functional differences between both rehabilitation protocols at 12^th week. The results at three months showed improvement in both groups compared to pre-treatment.¹⁶ Similar results were obtained in our study. In RR group, no significant improvement was found in upper extremity function after treatment, likely due to the fact that the robotic device used in our study only targeted finger rehabilitation. A significant improvement in upper extremity function might be achieved with a robotic device that also rehabilitates the elbow and shoulder joints.

In a similar randomized controlled study comparing robotic rehabilitation with CR, 44 subacute and chronic stroke patients were evaluated. No significant difference were observed between these rehabilitation programs in terms of outcomes at 3, 6, 12 and 24 week.¹⁷ The robotic devices used in these studies were capable of rehabilitating all joints of the upper extremity. Unlike these robots, the robot in our study only has hand rehabilitation capacity. Since the device did not impact the rehabilitation of the other joints of the upper extremity, additional home-based rehabilitation was included for the robotic group. In contrast to these studies, similar results were not obtained between RR and CR in terms of upper extremity function in our study. RR was ineffective for improving upper extremity function in our study, likely due to the inadequacy of home-based rehabilitation in addressing the needs of other joints.

Hand function is crucial for overall upper extremity function. It is essential for performing ADLs, is often severely impaired after a stroke, and full recovery is unlikely. Although successful outcomes have been reported for hand functions with CR, hand rehabilitation robots have been developed to achieve even better results. Many studies have explored the efficacy of hand rehabilitation robots.

A pilot feasibility study evaluated robotic rehabilitation in 21 hemiplegic patients. As in our study’s RR group, home-based rehabilitation was also applied in this study. There were no significant differences in VAS, the Ashworth spasticity index, or hand edema at 2-month compared to pre-treatment. However, significant improvements were noted in hand motor function and grip strength. This study demonstrated that home-based rehabilitation combined with the Glore Lite hand rehabilitation robot is suitable and reliable for post-stroke upper extremity rehabilitation at home. Our study revealed that the hand rehabilitation robot and home-based rehabilitation together were successful in hand functions and ADLs, but not in upper extremity function. While it improved wrist, finger and thumb flexor spasticity, it was not as effective as CR-RR. Improvements in hand grip and tip pinch strength were observed, similar to this study.¹²

Another randomized controlled pilot study compared isolated hand rehabilitation robots with conventional hand rehabilitation. The robot device provided computer-controlled, repetitive, passive mobilization of the fingers with multisensory feedback, as in our study. Significantly better results were achieved with RR compared to conventional hand rehabilitation in terms of hand motor function, grip and pinch strength. It was concluded that hand robotic rehabilitation is feasible and effective in recovering fine manual dexterity and strength and in reducing arm disability in hemiplegic patients. It was obtained that although home-based rehabilitation in addition to RR, upper extremity functional results did not improve, but there was a significant improvement in hand functions in our study as this study. While this study indicated a reduction in arm disability, it raises doubts about the sufficiency of hand rehabilitation alone.¹⁴

While hand functional recovery is important for overall upper extremity recovery, rehabilitation of other joints is also crucial. In CR-RR group, both conventional rehabilitation and robotic hand rehabilitation were applied. CR provided rehabilitation for both the upper extremity and hand, and additional robotic hand rehabilitation was included. This combined CR-RR program resulted in more successful outcomes for both upper extremity and hand function.

In another randomized control trial, no significant differences were found between RR and CR regarding upper extremity and hand functional results at the 8th and 32nd weeks.¹⁸ In our study, no significant differences in hand functional results between RR and CR groups; however, CR achieved significantly better results in upper extremity function. It was determined that CR was effective for upper extremity function, while RR was ineffective. The CR-RR yielded significantly better results in terms of both upper extremity and hand functions compared to RR and CR alone.

A study evaluating 33 patients compared CR and RR over the same period as our study. Both rehabilitation methods improved motor and functional outcomes. In RR, only the fingers and wrist joints were targeted. Despite this limitation, a significant improvement was found in the Fugl-Meyer upper extremity assessment scale and Fugl-Meyer shoulder/elbow extremity assessment scale in RR compared to the pre-treatment.¹⁹ Our study included a home-based upper extremity rehabilitation program alongside RR; however, no improvements were detected in upper extremity functional scales. In contrast to this study, our study revealed that RR alone was insufficient to improve upper extremity functions.

Our study revealed that although RR is effective for ADLs, hand function, and addressing wrist, finger and thumb flexor spasticity, it is not effective for improving overall upper extremity function, nor for hand grip and tip pinch strength, contrary to this study.

There is only one case report in the literature documenting the application of both RR and CR together. This report described CR-RR treatment for a ten-year-old girl with right hemiparesis following ischemic stroke due to probable primary central nervous system angiitis. The rehabilitation consisted of 10 sessions, resulting in improvements in activities of daily living, upper extremity functions, and force control, consistent with our findings.²⁰

Limitations of the study

All patients were chronic stroke survivors, with more than six months having passed since the onset. The effectiveness of rehabilitation protocols could also be evaluated in subacute stroke cases. Subacute stroke cases were not included in this study, as they would disrupt the homogeneous distribution of the participants. However, subacute stroke cases could be evaluated in a separate study. Short-term results were assessed, and a longer follow-up evaluation would be necessary to determine whether the significant improvements could be maintained after rehabilitation. The physiotherapist performing both sets of rehabilitation could not be blinded to the group allocation, while CR was administered to the patients by different physiotherapists. However, the physiotherapists who applied CR were selected from among experienced physiotherapists who had been working in this unit for a long time. The CR for each patient was consistently administered by the same physiotherapist until the end of the treatment. The duration of rehabilitation is also crucial for the success of treatment. The rehabilitation durations among the groups in the study are not equal. In this study, RR and CR durations were determined based on successful studies in the literature. To eliminate differences in rehabilitation durations between groups, reducing the duration in the CR-RR rehabilitation group could have negatively impacted the effectiveness of both CR and RR. However, in a separate study, the effectiveness of the CR-RR rehabilitation protocol could be evaluated by shortening the durations of conventional rehabilitation and robotic rehabilitation.

Conclusions

Although RR provides successful results in terms of ADLs, hand function, hand motor strength and hand spasticity, it is insufficient for treating upper extremity function and spasticity of upper extremity joints other than the hand. The combined application of RR and CR achieved better results for ADLs, upper extremity function, hand functional results, and thumb flexor spasticity compared than RR alone.

In post-stroke hemiplegic patients, combining CR and RR protocols is essential to achieve both better functional outcomes and greater improvement in hand spasticity, rather than applying either protocol alone.

The study was registered in the ClinicalTrials.gov protocol registration and results system under the title ‘Combined Robotic Hand Rehabilitation and Conventional Rehabilitation for Post-stroke Rehabilitation’. ClinicalTrials.gov ID is NCT06184191.