Modified Constraint-Induced Therapy in Patients With Chronic Stroke Exhibiting Minimal Movement Ability in the Affected Arm

Stephen J Page, Peter Levine


Background and Purpose The purpose of this study was to determine the efficacy of a reimbursable, outpatient modified constraint-induced therapy (mCIT) protocol administered to subjects with chronic stroke who initially exhibited minimal movement ability in their affected wrists and fingers.

Subjects The subjects were 4 individuals who had experienced a stroke more than 1 year prior to study entry (mean age [±SD]=60.25±1.98 years, mean time since stroke=37.5±23.2 months).

Method A multiple-baseline, preintervention-postintervention, single-blinded case series design was used. The intervention consisted of structured, ½-hour therapy sessions emphasizing affected arm use in valued activities, occurring 3 times per week for 10 weeks. Subjects’ less affected arms also were restrained 5 days per week for 5 hours per day during the same 10-week period. The main outcome measures were the Action Research Arm Test (ARAT), the Motor Activity Log (MAL), and the Fugl-Meyer Assessment of Motor Recovery (FM).

Results The subjects exhibited improvements in use of the more affected arm (+1.9, +1.8, +1.7, and +2.3 for subjects 1 through 4, respectively) and in quality of movement (+1.5, +2.1, +1.63, and +1.9 for subjects 1 through 4, respectively), as measured by the MAL. They also exhibited reduced impairment, as measured by the FM (+5.0, +6.5, +5.5, and +5.0 for subjects 1 through 4, respectively), and increased fine motor skill movement, as measured by the ARAT (+7.5, +7.0, +7.0, and +5.5 for subjects 1 through 4, respectively).

Discussion and Conclusion The findings demonstrated that mCIT participation was efficacious because it led to increased use of the affected arm and of function and to increased ability to perform valued activities. The subjects reported some new ability to perform some valued activities. These outcomes are significant because few therapies are believed to effectively increase use of the affected arm and function in this population.

Stroke is the leading cause of disability in the United States,1 and stroke-induced hemiparesis is the most frequent impairment managed by therapists.2 Yet, despite months of physical rehabilitation and billions of dollars spent annually,1,3 studies examining the efficacy of conventional stroke rehabilitation techniques have shown negligible efficacy.4,5

Increased affected arm use and function were reported after patients with chronic stroke (>1 year poststroke) participated in constraint-induced therapy (CIT),69 in which: (1) patients’ unaffected arms were restricted during 90% of waking hours over a 2-week period and (2) patients engaged in 6-hour activity sessions using their affected arms on 10 weekdays of the same 2-week period. Although these findings are promising, the authors of a recent CIT case study reported that the patient “grew tired of wearing the mitt and had difficulty with full adherence … ‘cheating’ with the uninvolved hand was a frequent temptation.”10(p851) Additionally, a recent survey11 administered to 208 patients with stroke and to 85 therapists showed that most patients would not want to participate in CIT, that more than 80% of the therapists felt that patients are extremely unlikely to adhere to CIT, and that many facilities do not have resources to administer CIT.

One of the only CIT randomized controlled trials9 reported that several subjects could not participate in their CIT trial because of home duties and that the effort that subjects had to put forth to participate was “strenuous.” Diminished physical deconditioning, comorbidities, or lack of appropriate support also may make it difficult for patients with stroke to participate in CIT. Thus, although CIT studies show that massed training increases movement ability, there remains a need for efficacious interventions that reduce hemiparesis within the confines of real-world clinics and payer restrictions.

Taub—one of the originators of CIT—and colleagues stated that “any technique that induces a patient to use an affected limb…should be considered therapeutically efficacious … (and is) … likely to produce the use-dependent cortical reorganization.”12(p243) Consistent with this claim, modified constraint-induced therapy (mCIT) has been shown to increase affected arm use and function in case studies,13,14 and randomized controlled pilot studies enrolling subjects with acute stroke (<14 days poststroke),15 subjects with subacute stroke (3–12 months poststroke),16,17 and subjects with chronic stroke (12 months poststroke).18 Besides its reimbursement within existing current procedural terminology codes, CIT and mCIT treatment effects appear to be comparable.19 Other data suggest that cortical reorganizations, brought about by increased affected arm use during mCIT, are responsible for the motor changes.20 The mCIT protocol combines structured, 1/2-hour, functional practice sessions 3 days per week with restriction of the unaffected arm 5 days per week for 5 hours per day. Both of these regimens occur during a 10-week period.

Modified constraint-induced therapy and CIT studies have typically required that subjects exhibit active extension in their affected wrists and fingers to be included. However, Bonifer et al21 reported appreciable affected arm use and motor function increases in patients who exhibited minimal affected hand ability. Most notably, Taub and colleagues22,23 described marked changes in subjects participating in CIT who initially were able “to lift a washrag off a tabletop using any type of prehension they could manage, and then release the rag.” Bonifer and Anderson24 administered CIT to patients who initially exhibited no isolated wrist movement, reporting some motor changes but no functional benefits.

Given CIT's limitations and the promise of mCIT, this study examined the efficacy of mCIT in patients with stroke who exhibited minimal initial affected hand movement. Specifically, we applied motor inclusion from above CIT studies,2123 enrolling individuals who could barely lift a washrag off a tabletop, as described above. Because CIT and mCIT studies have shown comparable treatment effects, we hypothesized that the subjects in this study would exhibit: (1) increased affected arm use, as measured by score increases on the Amount of Use (AOU) scale of the Motor Activity Log (MAL), (2) reductions in affected arm impairment, signified by score increases on the Fugl-Meyer Assessment of Motor Recovery (FM), and (3) increased distal movement ability, as measured by score increases on the Action Research Arm Test (ARAT).



Volunteers were recruited using advertisements placed in therapy clinics in hospitals in the midwestern United States. A research team member screened volunteers using the following inclusion criteria: (1) history of no more than one stroke; (2) ability to lift a washrag off a tabletop using any type of prehension they could manage and then release the rag; (3) stroke experienced more than 12 months prior to study enrollment; (4) a score of ≥69 on the Modified Mini Mental Status Examination25; (5) age greater than 18 years and less than 80 years; (6) and nonuse of the affected arm, defined as having an MAL score of less than 2.5 (described later).

We also applied the following exclusion criteria: (1) able to actively extend the wrist more than 10 degrees, extend 2 or more digits more than 10 degrees, and abduct the thumb more than 10 degrees; (2) excessive spasticity, defined as having a score of ≥3 on the Modified Ashworth Spasticity Scale26; (3) excessive pain in the affected upper limb, as measured by a score of ≥4 on a 10-point visual analog scale; (4) still enrolled in any form of physical rehabilitation; (5) participating in any experimental rehabilitation or drug studies; and (6) receipt of botulinum toxin A administered to the affected arm, wrist, or fingers within the past 4 months. Subjects were not excluded based on ambulation ability, or lack thereof. However, all subjects were taught to safely don and doff the restraint device during the first intervention session. Using these inclusion and exclusion criteria, we identified 4 volunteers (mean age [±SD]=60.25±1.98 years, mean time since stroke=37.5±23.2 months) who were eligible to participate in the study (Tab. 1).

Table 1.

Subject Characteristics

Outcome Measures

Instruments used for this study were applied in previous mCIT studies1318 and CIT studies12,21 and are responsive to changes following forced-use programs for people with chronic stroke.27 The ARAT,28 our primary outcome measure, is a 19-item test divided into 4 categories (grasp, grip, pinch, and gross movement), with each item graded on a 4-point ordinal scale (0=can perform no part of the test, 1=performs test partially, 2=completes test but takes abnormally long time or has great difficulty, and 3=performs test normally) for a total possible score of 57. The test is hierarchical in that, if the subject is able to perform the most difficult skill in each category, he or she will be able to perform the other items within the category and, thus, those items need not be tested. The ARAT has high intrarater (r=.99) and retest (r=.98) reliability and validity.28,29 The 66-point, upper-extremity section of the FM30 assesses several impairment dimensions using a 3-point ordinal scale (0=cannot perform, 1=can perform partially, and 2=can perform fully). The FM has been shown to have impressive test-retest reliability (total=.98–.99, subtests=.87–1.00),31 interrater reliability, and construct validity.32 The MAL is a semistructured interview measuring how subjects use their affected limbs for activities of daily living (ADL). Both subjects and caregivers use a 6-point AOU scale to rate how much the subject is using the affected arm and a 6-point Quality of Movement (QOM) scale to rate how well the subject is using the affected arm. Previous research33 indicates that the MAL is a valid and reliable scale of arm use and movement quality in real-world settings.

Testing and Intervention

A single-blinded, multiple-baseline, case series design was applied. After the subjects had signed consent forms approved by the local institutional review board and after screening, the FM, ARAT, and MAL were administered on 2 occasions 1 week apart. The measures were administered by an evaluator who was masked to the intervention in which the subjects were participating. The subjects also were mixed with other subjects who were being tested for other studies, thus further masking the rater. Following the second pretesting session, the subjects began the mCIT intervention.


As in previous studies,16,17 the mCIT intervention was begun 1 week after the second pretesting session. The mCIT intervention comprised 2 components. The first component consisted of 1/2-hour, one-on-one sessions of therapy, occurring 3 days per week over a 10-week period. This therapy included shaping techniques, which consisted of: (1) operant conditioning, applied in such a way that current motor capacity was extended using positive verbal encouragement and (2) challenging exercises that targeted deficient components of 2 or 3 activities chosen by the subjects with help from their therapist (eg, writing, using a fork). For example, some subjects worked on picking up and using a spoon or fork. These movements were broken into their smallest measurable elements (eg, reaching for the item, grasping it, bringing it to the mouth), and the deficient components for a particular subject were practiced. Progressively, the individual components were parsed together into the movement as each component was mastered, until the entire movement could be performed. Each element was usually timed so that the smallest performance changes could be observed. Verbal reinforcement was provided by the therapist when performance improvements were observed; when 3 or more unsuccessful attempts were made, reinforcement also was provided, but was never negative or discouraging. Use of modeling and prompts was frequent. In these ways, the positive motor behavior was “shaped” and encouraged using operant conditioning principles and frequent feedback.

During the 10-week period, subjects’ unaffected arms also were restrained every weekday for 5 hours identified as a time of frequent arm use. Their arms were restrained using a cotton hemi-sling, while their hands were placed in mesh, polystyrene-filled mitts with Velcro* straps around the wrist. “Homework” activities also were given to each subject, based on the subject's goals and movement deficits.

Although no difficulties with adherence have been documented in mCIT studies, prior to the beginning of the first therapy session, each subject who participated in the mCIT intervention and his caregiver met in a group meeting with the assigned physical therapist and the principal investigator (SJP), and a behavioral contract was discussed, reviewed, and signed. For additional sling and mitt use adherence tracking, subjects maintained a log in which they documented times of use and exercises performed during these times. The log acted as a reminder for subjects to don the slings and mitts, because therapists and the subject or caregiver had identified times when the subject would actively practice at home. However, therapists also noted the homework activities that subjects were to practice at home in the log. The log was brought to each therapy session for the therapist to review with the subject and his caregiver.

One week after therapy completion, each subject returned to the laboratory at which pretesting occurred, and the FM, ARAT, and MAL again were administered by the same examiner who administered the pretests. The examiner was masked in that he was unaware of the intervention in which the subjects had participated. Logs also were returned at this time.


Before intervention, the MAL AOU scores were 0.70, 0.73, 0.80, and 0.98 for subjects 1 through 4, respectively (Tab. 2). All subjects reported only occasionally using their affected limbs for ADL, and they reported mostly using their unaffected arms. All subjects’ preintervention MAL QOM scores suggested that the quality with which they performed the ADL tasks with the affected limb was low.

Table 2.

Subjects’ Scores on the Motor Activity Log Amount of Use (AOU) and Quality of Movement (QOM) Scales Before and After Intervention

After intervention, subjects uniformly exhibited increased use of the affected arm, with MAL AOU change scores of +1.9, +1.8, +1.7, and +2.3, for subjects 1 through 4, respectively (Tab. 2). Subjects reported using their affected limbs for ADL tasks such as eating finger foods and tying shoes.

The increases in use of the affected limb after the mCIT intervention were accompanied by increased FM and ARAT scores (Tab. 3). Specifically, after intervention, FM scores uniformly increased by +5.0, +6.5, +5.5, and +5.0 and ARAT scores increased by +7.5, +7.0, +7.0, and +5.5 for subjects 1 through 4, respectively. Positive motor changes on the FM and ARAT were accompanied by preintervention-postintervention MAL QOM change scores of +1.5, +2.1, +1.63, and +1.9 for subjects 1 through 4, respectively.

Table 3.

Subjects’ Scores on the Fugl-Meyer Assessment of Motor Recovery (FM) and the Action Research Arm Test (ARAT) Before and After Intervention


Modified constraint-induced therapy has been shown to have a comparable treatment effect to CIT,19 while requiring substantially less time to administer, and is reimbursed within most managed care plans. This study examined mCIT efficacy in subjects with stroke who exhibited less active affected hand and wrist movement compared with subjects in previous mCIT studies.1318,20

Before intervention, subjects exhibited stable motor deficits that had not changed since discharge, as suggested by their medical records and by reports from their physiatrists or physical therapists. Subjects’ scores on the FM and ARAT also did not change appreciably between pretesting sessions. All subjects also scored less than 2.5 on the MAL AOU scale, which was indicative of only occasionally using their affected arms for ADL tasks. After intervention, subjects displayed changes of +1.7 points or higher on the MAL AOU scale, and they reported attempting more ADL tasks with their affected limbs such as eating or attempting various dressing and self-care tasks. The magnitude of changes on the MAL AOU scale after the mCIT intervention during this very focused, 10-week period adds further credence to the finding that mCIT increases use of the affected limb. Recent work20 has confirmed that increased use of the affected arm during mCIT produces cortical reorganizations that are responsible for motor changes that are observed. We hypothesize that similar mechanisms were responsible for the motor changes that we observed, but this study did not examine this research question. Additional work is under way with this population to confirm this hypothesis.

Increases in MAL AOU scores were accompanied by marked motor changes in all subjects, as measured by the FM and ARAT. On the FM, subjects displayed positive changes in proximal affected arm areas, such as ability to abduct the shoulder and extend at the elbow. However, subjects also exhibited distal changes, such as increased mass flexion and extension and pincer grasp. On the ARAT, which measures fine motor function to a greater degree than the FM, changes were particularly seen on the grip and grasp scales. Positive motor changes on the FM and ARAT were corroborated by MAL QOM change scores of +1.5, +2.1, +1.63, and +1.9 for subjects 1 through 4, respectively. Functionally, the subjects were better able to perform some valued activities, including eating finger foods, tying shoes, reaching for and grabbing a cup, and grasping and eating utensil and bringing it to the mouth.

Although the motor changes observed in this study were clinically relevant, they were not as sizable as those seen in other subjects with less impairment who exhibited active wrist and finger movement before mCIT intervention. For example, one mCIT study of subjects in the acute stage of stroke (<14 days poststroke) who were minimally impaired (ie, able to actively extend at least 10° at the metacarpophalangeal and interphalangeal joints and 20° at the wrist) demonstrated mean improvements of +18.7 on the FM and +21.7 points on the ARAT.15 In another mCIT study,18 subjects who were in the chronic stage of stroke (1 year poststroke) and minimally impaired had FM and ARAT score changes of +10 or higher, although these changes were not as sizable as those observed in the acute stage. We speculate that greater movement ability, due to less impairment at baseline, is the primary reason why more sizable motor changes were observed in previous mCIT work with individuals who were less impaired than in the current study of individuals who were more impaired.

Although the intervention sessions were shorter than those required for CIT, impairment reductions in this study, as measured by the FM, compared favorably to those reported in a CIT study enrolling subjects with identical inclusion criteria.21 Indeed, using similar inclusion criteria, Bonifer and colleagues21 reported a mean change in FM scores of 6 points (from 29 to 35), whereas we found a mean change score of +5.5 in the current study. On the MAL AOU scale, Bonifer and colleagues21 reported that subjects’ scores changed from 0.77 to 2.51. Subjects in the current study exhibited MAL AOU change scores of +1.7 to +2.3; thus, very similar changes were found in this study for MAL AOU scores as well. A recent meta-analysis19 also found CIT and mCIT treatment effects to be highly comparable, and, in some cases, the mCIT treatment effect was higher. The results of the current study are consistent with this finding and provide useful preliminary data suggesting that mCIT and CIT treatment effects also are comparable in this population. More data with a greater number of subjects are needed to further confirm this finding.

When considered with previous mCIT study outcomes, the results from this study are encouraging for clinicians encountering patients who may not meet traditional mCIT motor inclusion criteria. Indeed, it was previously believed that only a minority of patients with stroke would be eligible for mCIT, given that only a small portion of this population exhibit the requisite active wrist and finger movement in the affected limb. However, the current study results offer the possibility of extending the number of patients who may benefit from mCIT. Furthermore, mCIT has been combined with electrical stimulation to produce clinically significant changes in patients who initially exhibited no movement,34 as well as being combined with botulinum toxin A in patients who initially exhibited high spasticity of the affected arm.35 Thus, an exciting milieu of therapeutic combinations is emerging for the affected arm in which singularly efficacious therapies are being combined effectively with mCIT to produce greater treatment effects. Additional research is warranted to determine whether mCIT can be efficacious in patients with no initial movement, as has been reported in CIT pilot work.24 Future researchers also should continue to identify therapies that may extend the mCIT treatment effect to individuals who may not meet the traditional motor inclusion criteria, as was attempted in our study.


The results of this study suggest that this reimbursable outpatient mCIT program increases use of the affected arm and function in patients with chronic stroke who do not meet traditional motor inclusion criteria. Before intervention, the subjects were barely able to lift a washrag off a tabletop using any type of prehension they could manage, and then release the rag; after intervention, the subjects were able to perform valued ADL tasks with their affected hands. These results add to a growing body of evidence suggesting that this reimbursable outpatient regimen increases limb use and function, even years after stroke.


  • Both authors provided concept/idea/research design. Dr Page provided writing, data analysis, project management, subjects, facilities/equipment, and institutional liaisons. Mr Levine provided data collection and clerical support.

  • * Velcro USA Inc, PO Box 5218, 406 Brown Ave, Manchester, NH 03103.

  • Received July 13, 2006.
  • Accepted March 5, 2007.


View Abstract