An Exploration of the Effectiveness of Different Intensity Protocols of Modified Constraint-Induced Therapy in Stroke: A Systematic Review

Purpose To examine the effectiveness of different modified Constraint-Inuced Therapy (mCIMT) protocol intensities on upper extremity motor function in adults with hemiplegia. Methods A search was conducted in PubMed, Scopus, EBSCO, and Cochrane Library for articles published between April 2010 and December 2021. Only randomized controlled trials (RCTs) were included. Studies were excluded if they used a sample of less than five, mCIMT in combination with other therapy, and/or if they were not written in English. Methodologic quality was assessed using the Cochrane collaboration risk of bias tool–2. Results Thirty-six RCTs with a total of 721 participants were included. Most researchers followed a moderate to low protocol intensity in terms of total treatment time and moderate to high intensity with regard to restriction time. Almost all of the upper limb motor function measures showed statistically significant improvements (p < .05) after mCIMT, irrespective of the protocol's intensity, but there was lack of high-quality studies. Statistically significant improvements did not always translate to clinical importance. Conclusions Low-intensity CIMT protocols may result in comparable improvements to more intensive ones but caution has to be taken when drawing conclusions due to high risk of bias studies.


Introduction
Stroke incidence has raised since 2000 and is now the second leading cause of death worldwide.Globally, low-and middle-income countries present the highest rates of stroke prevalence [1].In 2019, five leading risk factors for stroke were identified by the Global Burden of Diseases, Injuries, and Risk Factors Study, including high systolic blood pressure, high body mass index, high fasting plasma glucose, ambient particulate matter pollution, and smoking [2].
Manual dexterity is frequently impaired in stroke persons, resulting in long-term functional deficits [3].According to the International Classification of Functioning, Disability, and Health (ICF) [4], it is expected that the severity of upper limb impairment may correlate with high dependence levels in activities of daily living, lack of participation, and low quality of life.As Raghavan [5] described, the functional consequences of stroke on the upper limb may be connected with learned nonuse, learned bad use, and unlearning or forgetting after stroke.The learned nonuse phenomenon was originally described by Taub et al. [6], who argued that limited use of the hemiplegic upper limb is rarely the direct result of brain insult but rather the effect of a gradually developed behavioral neglect.Constraint-induced movement therapy (CIMT) was introduced as a treatment method aiming at reversing learned nonuse.The original protocol consisted of three basic elements: (1) 6 hours of task-oriented intensive therapy, for a total of 2-3 weeks, (2) a "transfer package" facilitating transfer gains to the real world, and (3) restriction of the unaffected upper limb for 90% of the patient's waking hours [7].
Despite the positive effects demonstrated for CIMT [8][9][10][11][12], this treatment method was initially considered difficult to implement clinically by Page et al. [13].Since then, researchers have applied different modified versions of the treatment (modified constraint-induced movement therapy or mCIMT), highly variable but typically characterized by reduced daily restraint and treatment time.As a matter of fact, reviewing of the literature reveals that there is more published research on mCIMT than the original CIMT, perhaps reflecting researchers' response to clinical demand.
Many studies have shown the effectiveness of mCIMT and even its superiority against other therapeutic methods, usually traditional care [14][15][16].However, mCIMT is not represented by a single protocol.Previous systematic reviews have explored either the effectiveness of CIMT and mCIMT together [8,11] or included all mCIMT protocols without differentiating between various intensities [17].In addition, most of the previous systematic reviews have examined the effectiveness of CIMT or mCIMT in comparison to other treatment rather than within the experimental group [8,17].Fleet et al. [18] focused their review on protocols that were offered in 3 sessions per week, for 10 weeks, but the daily session and restriction time were not controlled for.Nijland et al. [19] compared between two different intensities of mCIMT in acute and subacute stroke, but their conclusions have to be cautiously interpreted due to the small number of included studies.To our knowledge, there has been no published review up until now comparing the before-after effect of mCIMT.
This review is aimed at exploring for the first time the relative effectiveness of different intensities of mCIMT protocols by dividing studies into three intensity groups, corresponding to the overall treatment time and the splintwearing time set by each research group.Therefore, the results of clinical tests of upper limb motor function in persons with stroke were reviewed before and after application of mCIMT.

Methods
This systematic review was conducted according to the preferred reporting items for systematic reviews and metaanalysis (PRISMA) guidelines [20].
In order to explore the relative effectiveness of various intensity mCIMT protocols, studies were categorized by our research team, according to the overall treatment time as "low" (less than 20 hours), "moderate" (21-39 hours), or "high" (more than 40 hours) intensity, and according to the daily splint-wearing time as "low" (≤ 3 hours/day), "moderate" (4-9 hours/day), and "high" (≥ 10 hours/day).S: randomized controlled trials (RCTs).Therefore, articles were included if they were randomized controlled trials, satisfying the following criteria: having investigated at least one modified protocol of CIMT, providing data before and after the intervention; having applied treatment to adults (people 18 years and older), following stroke; having included outcome measures related to upper limb motor function; and having been written in English.The last published systematic review and meta-analysis that presented the comparison between mCIMT and traditional rehabilitation included studies from inception to April 10, 2010 [17].Therefore, this systematic review included studies published from April 10, 2010, to December 31, 2021.

Study Selection
2.1.2.Exclusion Criteria.Exclusion criteria ensured that the systematic review consisted only of research studies pertaining to the study population and mCIMT.Excluded studies comprised those which provided mCIMT in addition to any other intervention, with sample size less than 5 in the mCIMT group, unfinished studies, or studies that did not contain specific, numeric results (including grey literature).Additionally, kinematic analysis or brain physiology measures were excluded.

2.
2. Data Sources.Two investigators (I.M. and C.A.) conducted a structured search of the following databases: PubMed, Scopus, EBSCO, and Cochrane Library having used the MeSH terms included in Table 1.There were no additional advanced search methods used for each of the databases listed.

Data Extraction.
Two investigators (C.A. and I.M.) screened the abstracts identified in the database searches.C.A. applied the inclusion and exclusion criteria to the abstracts; if the information in the abstract did not meet the selection criteria, the study was excluded at this stage, and the reason for exclusion was recorded.Duplicate articles were removed.The remaining full-length articles were then retrieved and reviewed by C.A. to further determine that the study was in accordance with the selection criteria.The final decision was reached after the senior investigator (P.P.) had performed another review.

2.4.
Classifying the Evidence and/or Determining the Level of Confidence and Recommendations.In order to assess the risk of bias, the Cochrane risk-of-bias tool for randomized trials (RοB-2) was used independently by two researchers (I.M. and C.A). RoB-2 consists of five areas for RCT bias risk assessment.These five domains assess the risk of bias arising from the randomization process, deviations from intended interventions, missing outcome data, outcome measurements, and selection of reported results.During the assessment, each domain is rated as low, high, or having some concerns.Conflicting assessment ratings were resolved by consensus.At the end, each domain's results were calculated to reach an overall risk-of-bias rating of poor, fair, or good.
2.5.Data Synthesis and Analysis.Specific variables of interest were extracted from the studies and added to summary tables.Summary tables were then used to classify studies according to the following categories: treatment hours, splint wearing time, and outcome measures related to upper limb motor function.Out of these outcome measures, the

Statistical Analysis.
To satisfy the study's aims, pre and postintervention means were collected for the same group.The fact that there was lack of two independent group differences to compare at the same time prevented us from conducting a meta-analysis.Pre-post standardised mean differences (SMDs) are influenced by natural processes, personal characteristics, and settings, and these cannot be discerned from the effects of the intervention.In addition, pre-post SMDs should be avoided if the scores between pre and postintervention are not independent from each other [21].
2.7.Sensitivity Analysis.The influence of individual studies was examined by omitting low-quality studies to see the extent to which inferences depend on a particular study or group of studies (sensitivity analysis).

Study Selection.
The search identified 2,769 papers, of which 875 full-text articles were assessed for eligibility.Out of these, 700 studies were excluded through title/abstract screening because they followed study designs other than RCTs or the population and interventions did not match the study's inclusion criteria.The second screening through reading of the whole manuscript resulted in exclusion of 53 studies due to intervention characteristics (i.e. they did not use mCIMT or used combination of treatments), outcome measures, and methodological issues (i.e.sample size).Seventy studies were excluded because they had not published any results until December 2021.One [22] additional study and 16 more that were listed as grey literature were excluded from the study because there were no numerical data, leading to the 36 studies of this systematic review (Figure 1).
Out of the 721 participants, 57 withdrew or did not comply with the study protocol.In more detail, 14 dropped      Sign.

Yes a
Moderate/low Bhardwaj et al. [33] Box and block test 1 Sign. - Rehabilitation Research and Practice Sign.
Yes a MAL-QOM Sign.
N/R Sign.

Outcome Measures.
The outcome measures reviewed were the ones related to upper limb motor function, while graphs were plotted for the most commonly used ones.The latter included the FMA-UE, MAL (QOM and AOU), WMFT (PT and FA), and ARAT.
After sensitivity analysis to include only the low risk of bias studies, 4 out of the 36 studies were separately examined [23,32,45,49].Out of these studies, three focused on chronic [23,32,45] and one on acute stroke [49].Statistical significance was shown in all outcome measures between pre and postintervention measurements and also between follow-up and preintervention.Statistical significance however does not coincide with clinical importance across all measures and all studies.

Discussion
This review identified 36 RCTs, corresponding to the inclusion criteria, and 721 participants in total.Almost all studies showed statistically significant improvements after mCIMT in the perceived arm motor function, as measured by MAL and arm motor impairment measures, in agreement to previous research [8,17].Some of the improvements were  17 Rehabilitation Research and Practice clinically important, and some were not, especially when follow-up measurements were compared to postintervention values.Nineteen studies [23, 24, 41-43, 46-48, 50, 51, 53, 25, 26, 29, 30, 32, 38-40] included at least one follow-up measurement and showed statistically significant retainment of improvements for most outcome measures.However, there were many studies in which there was misreporting of results or statistical significance.
The results of this study show no major differences in the effectiveness of mCIMT protocols of different treatment intensities, and there is no support to the "more is better" notion.This comes in agreement with other studies [8,11,59], who also found that the dose of intervention did not influence the results.Fleet et al. [18] reviewed 15 studies of mCIMT that included 10-week intervention with frequency of 3 times per week; out of these studies, 10 used 30minute sessions, and 5 used 60-minute sessions.Although the authors did not provide separate analysis between the two intensity protocols, graphical displays do not seem to support any substantial intensity effect.Similarly, Sterr et al. [59] found that using or not a constraint and having more or less shaping training did not significantly change the treatment outcome in low-functioning stroke persons.In fact, Nijland et al. [19] concluded that low-intensity protocols seem to be more beneficial when it comes to acute and subacute stroke.
However, suggesting that any protocol may be effective is an oversimplification, since it is widely acceptable and evidence based [60,61] that intensity relates to neuroplasticity and directly translates to functional improvements.Reviewing the literature reveals the large variability of protocols, participants with different characteristics, and the methodological limitations evident in most studies.Some of them were conducted with less than 10 participants or with a lack of blinding procedures, while significant drop-out rates during postintervention or follow-up may have been an additional reason for the observed high risk of bias.
The results after sensitivity analysis concluded on four high-quality studies.Two of these studies used a highintensity treatment protocol, and the other two used a low and moderate one.The results showed statistical significance (that, however, was not always coinciding with clinical importance) postintervention and at follow-up when follow-up was compared with preintervention values.On the contrary, when   follow-up measurements were compared with postintervention, the results did not show clinical significance in almost all cases, indicating that further improvements were not achieved after the end of intervention.
Studies have used a variability of outcome measures, but almost all of them have used the MAL, as this measure forms part of the behavioral transfer package of CIMT.This means that MAL improvements are expected within any CIMT program, as it is the main goal of this intervention to encourage patients to try and use their upper limb as much as possible in order to reverse learned nonuse.Therefore, an increase in MAL is expected to occur during intervention and does not necessarily indicate long-lasting improvements.However, what may be an indicator of real-world improvements is when MAL values remain increased (or increase even more) at follow-up, and this is one reason why follow-up should always be included in CIMT studies.Among the 4 high-quality studies included in this review, none shows clinically significant retention of improvements in either MAL-AOU or MAL-QOM.
The heterogeneity of protocols, outcome measures, and population characteristics, which has been noted by other researchers [8,17], makes it very difficult to reach more definite conclusions.Future studies should focus on comparing different intensities of mCIMT while also considering the severity of impairment and time after stroke and how these may relate to the overall effectiveness.It may also be worth considering for future reviews to include only specific categories of studies in terms of methodology, i.e., high-quality studies only, and/or in terms of patient characteristics, i.e., acute/chronic and functional level.
Most of the studies included in this review did not provide any information on the training background and experience of therapists providing CIT, which might be a factor influencing the outcomes.Moreover, future studies should provide clear reports on the home-based part of the intervention, which has been hardly described by most researchers in addition to information related to the third element of CIMT, i.e., the transfer package and how this has been incorporated within the protocol.

Study Limitations.
A meta-analysis may have provided clearer results, but this was not possible with the methodology employed by the present study.To allow comparisons,  this study analyzed only the first follow-up measurement, even though some studies included more than one; thus, some data on long-term effectiveness have been lost in this review.Furthermore, follow-up measurements took place at different times in each study.Reviewing all studies of mCIMT has provided a significant amount of information but also confusion due to the variability of methodologies compared.
4.2.Conclusions.Modified CIMT may be an effective intervention for persons with stroke but long-lasting results need to be further researched.This review does not provide support to the theory that the higher the intensity of treatment, the better the outcome, pointing to the fact that apart from the intensity, the actual content and structure of therapy matter.The ideal protocol intensity is yet to be confirmed, and it is very likely that this shall be different according to patient's characteristics, i.e., time since stroke and functional level.More sound methodologies in future studies may provide more reliable results.

Table 1 :
Search keywords.OR Cerebral stroke OR Cerebrovascular disorder OR Cerebrovascular disease OR Cerebrovascular accident OR CVA OR Brain vascular accident OR Cerebral infarction OR Brain injury OR Brain hemorrhage OR Brain ischemia OR Brain infarction *

Table 2 :
Characteristics of included studies.
following were identified as the most commonly used by researchers: The Fugl-Meyer assessment for upper extremity (FMA-UE), Motor Activity Log quality of movement (MAL-QOM) and amount of use (MAL-AOU), Wolf Motor Function Test performance time (WMFT-PT) and functional ability (WMFT-FA), and Action Research Arm Test (ARAT).Microsoft 365 Excel 2021 was used for data analysis and graph plotting.

Table 3 :
Treatment intensity and outcome measurements.

Table 4 :
Methodologic quality of included studies.
presents a graphical display of WMFT-FA data in relation to various intensity mCIMT protocols used in each study.