Home mirror therapy: a randomized controlled pilot study ...

Full Terms & Conditions of access and use can be found athttps://www.tandfonline.com/action/journalInformation?journalCode=idre20

Disability and Rehabilitation

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/idre20

Home mirror therapy: a randomized controlledpilot study comparing unimanual and bimanualmirror therapy for improved arm and handfunction post-stroke

Daniel Geller, Dawn M. Nilsen, Lori Quinn, Stephen Van Lew, ClaribellBayona & Glen Gillen

To cite this article: Daniel Geller, Dawn M. Nilsen, Lori Quinn, Stephen Van Lew, ClaribellBayona & Glen Gillen (2021): Home mirror therapy: a randomized controlled pilot study comparingunimanual and bimanual mirror therapy for improved arm and hand function post-stroke, Disabilityand Rehabilitation, DOI: 10.1080/09638288.2021.1973121

To link to this article: https://doi.org/10.1080/09638288.2021.1973121

Published online: 19 Sep 2021.

Submit your article to this journal

View related articles

View Crossmark data

ORIGINAL ARTICLE

Home mirror therapy: a randomized controlled pilot study comparing unimanualand bimanual mirror therapy for improved arm and hand function post-stroke

Daniel Gellera,b, Dawn M. Nilsenb, Lori Quinnc, Stephen Van Lewa, Claribell Bayonaa and Glen Gillenb

aDepartment of Occupational Therapy, New York University Langone Health, Rusk Rehabilitation, New York, NY, USA; bDepartment ofRehabilitation and Regenerative Medicine (Occupational Therapy), Vagelos College of Physicians and Surgeons, Columbia University, New York,NY, USA; cDepartment of Behavioral Sciences, Teachers College, Columbia University, New York, NY, USA

ABSTRACTPurpose: To compare home-based unimanual mirror therapy (UMT) and bimanual mirror therapy (BMT)for upper limb recovery in subacute/chronic stroke individuals with moderate-to-severe arm impairment.Method: Twenty-two participants were randomized into 1 of 3 groups: UMT, BMT or traditional occupa-tional therapy (TOT) home-based programs. The intervention was 6-weeks and consisted of OT 2days aweek, weekly sessions with the research OT, and 30-minutes of the home-based program 5days a week,according to group allocation. The Action Research Arm Test (ARAT), ABILHAND, Fugl-Meyer Assessment(FMA), grip strength, and Stroke Impact Scale (SIS) were used for outcome measures.Results: All groups significantly improved over time on all outcome measures and adhered to the pre-scribed dosage regardless of group (p<0.05). While there were no between-group differences, effect sizeand 95% confidence interval data suggest a clinical significance in favor of UMT as compared to the othergroups.Conclusions: All participants, regardless of home-based program, adhered to the prescribed dosage andsignificantly improved over time. Despite no between-group differences, effect size and 95% confidenceinterval data suggest that UMT may be more beneficial for individuals with moderate-to-severe armimpairment as compared to BMT or TOT.

ClinicalTrials.gov: #NCT02780440

� IMPLICATIONS FOR REHABILITATION� Home-based unimanual mirror therapy (UMT), bimanual mirror therapy (BMT), and traditional occupa-

tional therapy (TOT), when administered in conjunction with outpatient OT, are helpful for improvingupper limb recovery post-stroke.

� Home-based UMT may be more beneficial than BMT or TOT for improvement in upper limb motorfunction and activities of daily living of patients with moderate to severe arm impairmentpost-stroke.

ARTICLE HISTORYReceived 17 February 2021Revised 16 August 2021Accepted 17 August 2021

KEYWORDSStroke rehabilitation; stroke;upper limb; mirror therapy;occupational therapy

Introduction

Stroke is the leading cause of adult disability in the United Stateswith over 7 million survivors [1]. The majority of stroke survivorshave persistent hemiparesis, with over 85% experiencing upperlimb dysfunction, which is a significant barrier to function and lifeparticipation [2]. Thus, remediation of upper limb function is oftenthe target of standard care in stroke rehabilitation programs [3–7].However, despite various upper limb interventions, there contin-ues to be challenges such as cost, labor intensity, required arm/hand movements for participation [4,8] and low practice dosage[9,10], which limits patients’ upper limb recovery and functionpost-stroke.

Mirror therapy (MT) was first described by Ramachandran &Rogers-Ramachandran [11] to relieve phantom pain after amputa-tion and later introduced to treat hemiparesis post-stroke [12]. MTis an intervention in which a mirror box is placed in the mid-sagittal plane to the participant between the extremities. The

affected hand is placed in the mirror box and the unaffectedhand is placed outside of the box facing the mirror. During MT,the person moves the unaffected hand while watching the mirrorreflection, giving the visual illusion that the affected limb is mov-ing. MT not only has been shown to improve upper limb recoverypost-stroke, but requires minimal supervision, no arm/handrequirements for participation, and low cost, thus can be used asan adjunct to standard stroke care to increase practice dosage[13–15]. Furthermore, MT has been shown to facilitate neuroplas-tic changes in the brain through 3 possible underlying mechan-ism: (a) perceptual motor process; (b) direct facilitation of themotor network; and (c) activation of the mirror neuron system[16]. While these are three distinct hypotheses, it is possible thatit could be a combination of all three; however, there is still noclear understanding of the underlying mechanism of MT.

Two different MT protocols have been used in stroke rehabili-tation: unimanual mirror therapy (UMT) and bimanual mirror

CONTACT Daniel Geller [email protected] Department of Rehabilitation and Regenerative Medicine (Occupational Therapy), Vagelos College ofPhysician and Surgeons, Columbia University, 617 West 168th St. 3rd floor, New York, NY, 10032, USA� 2021 Informa UK Limited, trading as Taylor & Francis Group

DISABILITY AND REHABILITATIONhttps://doi.org/10.1080/09638288.2021.1973121

therapy (BMT). During both protocols, the affected hand is placedin the mirror box and the patient is instructed to view the mirrorreflection of the unaffected hand. During UMT the affected handis static, while during BMT, the affected hand moves in anattempt to duplicate the unaffected hand.

Both MT protocols have been shown to improve upper limbrecovery post-stroke [13–15]; however, results have been incon-sistent across the impairment and activity domains of theInternational Classification of Functioning, Disability and Health(ICF) [17]. In recent randomized controlled UMT and BMT studiesin subacute/chronic individuals, the results showed eitherimprovements only at the impairment level [18–23], or improve-ment at both domain levels [24–30]. Furthermore, while two stud-ies compared the MT protocols, neither were able to provide

definitive conclusion regarding clinical application of MT.Selles et al. [31], compared five different upper limb interventions,including UMT and BMT, during a 1-session reaching task withchronic stroke patients. Movement time results showed no signifi-cant differences between UMT and paretic limb training without amirror, while BMT showed significantly less improvement, sug-gesting UMT may be more beneficial than BMT. However, thefocus of the study was upper limb motor learning after 1 session,not clinical application. In the second study, a MT meta-analysisof 32 randomized controlled studies, showed UMT to have agreater effect on motor function and motor impairment as com-pared to BMT, as per SMD and 95% CI data. However, since therewere no statistically significant group differences, the researcherssuggested further research is needed regarding optimal

Figure 1. CONSORT flow diagram.

2 D. GELLER ET AL.

application of MT post-stroke [32]. Hence, there continues to belack of clarity regarding which MT protocol would most benefitupper limb recovery post-stroke.

The aim of this pilot study was to compare the efficacy ofhome-based UMT, BMT, and traditional OT (TOT), as an adjunct tooutpatient occupational therapy (OT) services, for upper limbrecovery post-stroke. To our knowledge this is the first random-ized controlled intervention study to compare UMT to BMT. It washypothesized that: (1) UMT would be more beneficial than BMT;(2) both MT groups would be more beneficial than TOT for upperlimb recovery in subacute/chronic stroke patients who have mod-erate-to-severe hemiparesis.

Material and methods

Participants

Participants were recruited from an outpatient OT department inthe New York metropolitan area. The inclusion criteria were: (1)aged 19 to 85 with a first-time stroke >3months; (2) Fugl-MeyerAssessment (FMA) score of 10–50 indicating moderate-to-severearm impairment [33]; (3) following directions; (4) ability to graspand release a washcloth with the affected hand. Exclusion criteriawere (1) complex medical problems and pre-existing neurologicalor psychiatric disease; (2) hearing and/or visual impairments; (3)perceptual deficits, such as apraxia, neglect, and visual agnosia;(4) botulinum toxin injection in the affected limb <3months priorto study inclusion; and (5) aphasia. Institutional review boardapproval was obtained in the academic hospital and all partici-pants provided written informed consent.

Study design

This was a single-blinded, randomized controlled study and wasregistered at ClinicalTrials.gov ID: NCT02780440. The participantswere pre-screened using the OTs initial assessment for age, timepost-stoke, multiple strokes, following directions, and hand func-tion (severe spasticity). This was followed by formal screening by1 of 2 trained OT assessors. Qualified participants were random-ized into 1 of 3 groups: UMT, BMT, or TOT (Figure 1). A non-research study OT performed randomization by a sealed envelopemethod. The same trained OT assessor, blind to group allocation,administered the pretest and posttest in the outpatient OT pro-gram in the rehabilitation department. In addition, non-researchOTs providing the conventional therapy twice a week were blindto group allocation.

Intervention

Following the pre-test, the research OT educated the participantsfor 1 h on the home program, which was modeled after 2 studies[15,34], according to group allocation. All participants were pro-vided with an instructional binder that included written instruc-tions, pictures of the exercises and tasks, a log to document timespent per exercise, and home program tools (e.g., cup, paper clipsetc.). Education included review of the binder though verbalinstruction, demonstration, and participant return demonstration.Mirror group participants were provided with a mirror box andeducated on the MT treatment protocol. UMT participants wereinstructed to keep the affected hand static in the mirror box(Figure 2(A)), while BMT participants were instructed to duplicatethe unaffected hand as best as possible (Figure 2(B)). The TOTparticipants were instructed to perform all tasks with the affectedlimb with no mirror. All participants were instructed to performthe home-based program 30-min a day and 5 days a week. Eachsession was divided into three 10-min categories: (1) moving thearm/hand, (2) functional task with objects, and (3) object manipu-lation [34] as per group allocation (Table 1). Furthermore, partici-pants were educated on the log, created by the primary researchOT, to track time (minutes) performing the home program and foradherence purposes. The 6-week intervention consisted of two45-min standard OT sessions in the clinic, one weekly 30-min ses-sion with the primary research OT, and the home-based program.The weekly meeting with the research OT included home pro-gram progression and review of the log. Weekly phone calls werealso made for adherence purposes.

Primary outcome measure

The Action Research Arm Test (ARAT) is a 19-item standardizedobjective assessment used to evaluate arm/hand functionpost-stroke at the activity level domain [35]. Scoring is based on a4-point ordinal scale with a total score of 57, whereby higherscores reflect greater recovery. It has excellent interrater and intra-rater reliability and construct validity with the FMA [36]. The min-imal clinical important difference (MCID) was established to be 5.7points for the ARAT in chronic stroke individuals [37].

Secondary outcome measures

The FMA upper-limb subscale is a 33-item standardized objectivemotor impairment measure for hemiplegia post-stroke patients[38]. Scoring is based on a 3-point ordinal scale with a maximumscore of 66, with higher scores indicating greater motor recovery.Its psychometric properties have been well established [39] with

Figure 2. Mirror therapy groups. (A) Unimanual mirror therapy: patient instructed to not move the affected hand in the mirror box. No objects are used in the mirrorbox. (B) Bimanual mirror therapy: patient instructed to move the affected hand in the mirror box as best as possible to duplicate the unaffected hand. The sameobjects are used for both hands (e.g., cup in bilateral hands)

UNIMANUAL AND BIMANUAL MIRROR THERAPY 3

an MCID of 5.25 points for the upper extremity FMA in chronicstroke individuals [40].

ABILHAND, an activity level measurement, is a valid and reli-able interview-based tool, measuring participants’ perceived diffi-culty with 23 bimanual hand activities. It has high reliability andmoderate correlation with grip strength, Box and Blocks Test, andthe Purdue Pegboard [41]. The MCID of the ABILHAND was estab-lished to be 0.26 to 0.35 logits in chronic stroke individuals [42].

Grip strength, an impairment level measurement, was eval-uated with a Jamar Dynamometer. Following standard clinicalprocedures, maximal grip strength was measured with the partici-pant seated, feet flat on the floor and shoulder distance apart,arm position with shoulder at 0 degrees, elbow at 90 degrees,forearm and wrist in neutral. The final score was taken from theaverage of three measurements with higher scores indicatinggreater grip strength. Dynamometer grip strength has good reli-ability and has been correlated with four upper extremity tests[43] and performance improvement [44].

The Stroke Impact Scale (SIS), version 3.0, is a subjective stand-ardized questionnaire assessing health related quality of life post-stroke. The global perception of stroke recovery, with moderatetest- retest reliability [45] is based on a 0–100 score with higherscores indicating greater recovery and quality of life.

Statistical analysis

Data were analyzed using IBM SPSS statistics software (Version 23,IBM Corp., Armonk, NY). Demographics, baseline measures, andhome-based program dosage data were compared using a univari-ate analysis of variance and Fisher’s exact test, as appropriate.Normal distribution was verified through Shapiro-Wilks test andabsolute z-score skewness and kurtosis less than 1.96 [46,47]. A 3(group) by 2 (test) mixed-model analysis of variance (ANOVA) wasperformed. Statistical significance was set at p<0.05. Effect sizeand 95% confidence intervals (CI) were calculated. Effect size ofCohen’s d can be interpreted as 0.2–0.5¼ small effect; 0.5–0.8¼moderate effect; and >0.8 ¼ large effect [48]. Individual data wasanalyzed to determine whether the participant met or exceededthe MCID with respect to the ARAT, ABILHAND, and FMA.

Results

Participants

Twenty-five participants were enrolled and 22 completed thestudy and were included in the final analysis. There were 3

withdrawals from the UMT group: (1) participant being dischargedfor medical issues, (2) after participant was consented and priorto starting the first treatment, the participant indicated the homeprogram would be too time consuming and decided not to par-ticipate in the trial (3) environmental factors (patient reported dif-ficulty setting up equipment) (Figure 1). Baseline demographics,outcome measures, and dosage results are presented in Table 2.There were no statistically significant differences between thegroups in all areas (p>0 .05). It should be noted that the strokeonset time was larger for the TOT as compared to the othergroups; however, this was due to one participant who was280months post-stroke.

Primary outcome measure

For the ARAT, there was a significant main effect of time, F (1,19)¼18.14, p< 0.001; however, there was no significant main effect ofgroup, F (1,19) ¼ 0.61, p¼ 0.69. There also was no significant inter-action between the groups and time, F (2,19) ¼ 0.61, p¼ 0.55.Thus, indicating ARAT scores improved pre to post interventionwithin each group with no significant differences between thegroups. There was a small (d ¼.38) and a moderate (d ¼0.55) effectsize in ARAT change scores, favoring UMT over both BMT and TOT,respectively. However, there was a trivial effect size (d ¼0.15) infavor of BMT over TOT (Table 3). In addition, 43% of UMT and BMTgroup participants exceeded the MCID of 5.7, compared to only25% of the TOT group participants.

Secondary outcome measures

As seen in Table 4, there was a significant main effect of time forthe ABILHAND, F (1,19) ¼12.16, p¼ 0.002; however, there was nosignificant main effect of group, F (1,19) ¼ 0.29, p¼ 0.60 Therealso was no significant interaction between the groups and time,F (2,19) ¼ 2.66, p¼ 0.096. Thus, indicating ABILHAND scoresimproved pre to post intervention within each group with no sig-nificant differences between the groups. The ABILHAND changescores showed a large effect size in favor of UMT over BMT(d¼ 0.87) and TOT (d¼ 1.22; MD 1.15; 95% CI 0.09 to 2.20), whileonly a small effect size of BMT over TOT (d¼ 0.38). In addition,100% of the UMT participants exceeded the MCID, while only43% and 63% of the BMT and TOT group participants, respect-ively, exceeded the MCID.

At the impairment level, there was a significant main effect oftime on the FMA F (1,19) ¼18.14, p< 0.001; however, there wasno significant main effect of group, F (1,19) ¼ .36, p¼ 0.70. There

Table 1. Home-based program divided into 3 categories.

Categories of activity Exercises/Tasks

Moving the Arm/Hand � Reaching� Elbow flexion/extension� Forearm pronation/supination� Wrist flexion/extension� Hand grasp/release� Finger opposition� Finger tapping� Finger abduction/adduction

Functional tasks with objects � Simulate washing the table with rag� Placing cup in hand and move forward and back, side to side, and pouring motions� Reach to touch different areas of table� Hold mallet and play like drum� Hold egg shaker and shake in different directions

Object manipulation � Grasp and release objects of different sizes, shapes, and textures (e.g., paper clips, screws)� Flip cards� Transport objects from one side of tray to the other� Squeeze and manipulate a ball

4 D. GELLER ET AL.

also was no significant interaction between the groups and time,F (2,19) ¼ 1.58, p¼ 0.23. Thus, indicating FMA scores improvedpre to post intervention within each group with no significant dif-ferences between the groups. There was a large effect size for theFMA change scores in favor of UMT over both BMT and TOT(d ¼0.81 and d¼0.84, respectively), however, a trivial effect size(d¼ 0.09) favoring BMT over TOT. In addition, 71% of the UMTgroup participants exceeded the MCID of 5.25, while for BMT andTOT, 57% and 38%, respectively, exceeded this number. For gripstrength, there was a significant main effect of time, F (1,19)¼4.49, p¼ .048; however, there was no significant main effect ofgroup, F (1,19) ¼ 1.20, p¼ 0.32. There also was a no significantinteraction between the groups and time, F (2,19) ¼ 1.33,p¼ 0.29. Thus, indicating grip strength scores improved pre to

post intervention within each group with no significant differen-ces between the groups. For grip strength, there was a large andmoderate effect size in favor of BMT over UMT (d¼ �1.79; MD�6.19; 95% CI, �10.2 to �2.16) and BMT over TOT (d=-0.59),respectively (Table 4).

At the participation level, there was a significant main effectof time for total SIS, F (1,19) ¼6.17, p¼ 0.023; however, therewas no significant main effect of group, F (1,19) ¼ 0.13,p¼ 0.88. There also was no significant interaction between thegroups and time, F (2,19) ¼ 0.22, p¼ 0.81. Thus, indicating totalSIS scores improved pre to post intervention within each groupwith no significant differences between the groups. The effectsize data was either small or trivial when comparing the 3groups (Table 4).

Table 2. Participants characteristic, baseline data, and dosage.

Unimanual mirror therapy(N¼ 7)

Bimanual mirror therapy(N¼ 7)

Traditional occupational therapy(N¼ 8) p Value

Age (years) 57.6 ± 13.1 57.3 ± 15.0 65.3 ± 15.4 0.494a

Onset months 23.4 ± 32.9 17.4 ± 22.9 48.8 ± 94.5 0.584a

Affected side (right/left) 4/3 1/6 5/3 0.176b

Hand dominance (right/left) 7/0 7/0 5/3 0.082b

Female/Male 5/2 3/4 2/6 0.239b

Dosage (minutes) 878 ± 89.6 904 ± 82.0 943 ± 98.9 0.390a

ARAT 15.3 ± 13.1 22.7 ± 11.0 17.8 ± 14.9 0.570a

FMA 28.7 ± 11.0 33.6 ± 9.40 29.6 ± 9.61 0.591a

ABILHAND -.805 ± 1.26 .280 ± 1.14 -.112 ± 1.34 0.283a

Grip strength 10.9 ± 13.8 16.6 ± 16.7 18.6 ± 12.4 0.576a

SIS-Total 45.7 ± 18.4 42.1 ± 16.8 42.5 ± 13.6 0.901a

Values are mean and standard deviation. The p value resulted from a given comparison test between the three groups, and the test was (a) Univariate, (b) Fisher’sExact Test.ARAT: Action Research Arm Test; FMA: Fugl-Meyer Assessment; SIS: Stroke Impact Scale.

Table 3. Results of primary outcome measure: mean difference, 95% (CI), and effect size.

Outcome by group

MeanTime Diff Group Diff

ComparisonMD

95% (CI) Effect sizePre Post p-value p-value

ARAT (Activity)UMT 15.3 ± 13.1 22.3 ± 18.0 <0.001 0.55 UMT/TOT 3.13 (�3.22, 9.47) 0.55BMT 22.7 ± 11.0 27.3 ± 13.6 BMT/TOT 0.70 (�4.64, 6.03) 0.15TOT 17.8 ± 14.9 21.6 ± 17.2 UMT/BMT 2.43 (�5.10, 9.96) 0.38

Values are mean and standard devation. The p values from analysis of variance (ANOVA).ARAT: Action Research Arm Test; BMT: bimanual mirror therapy; CI: confidence interval; Diff: difference; MD: Mean difference; TOT: traditional occupational therapy;UMT: unimanual mirror therapy.

Table 4. Results of secondary outcome measures: mean difference, 95% (CI), and effect size.

Outcome by group

MeanTime Diff Group Diff

ComparisonMD

95% (CI) Effect sizePre Post p-value p-value

FMA (Impairment)UMT 28.7 ± 11.0 37.3 ± 12.0 <0.001 0.23 UMT/TOT 4.20 (�1.37, 9.76) 0.84BMT 33.6 ± 9.40 38.7 ± 9.96 BMT/TOT 0.49 (�5.24, 6.21) 0.09TOT 29.6 ± 9.61 34.0 ± 5.40 UMT/BMT 3.71 (�1.66, 9.08) 0.81

Grip strength (Impairment)UMT 10.9 ± 13.8 10.6 ± 13.7 0.048 0.29 UMT/TOT �5.19 (�15.1, 4.70) �0.59BMT 16.6 ± 16.7 22.4 ± 18.0 BMT/TOT 1.00 (�9.12, 11.1) 0.11TOT 18.6 ± 12.4 23.4 ± 9.91 UMT/BMT �6.19 (�10.2,�2.16) �1.79

ABILHAND (Activity)UMI �.805 ± 1.26 .546 ± 1.24 0.002 0.096 UMT/TOT 1.15 (0.09, 2.20) 1.22BMT .280 ± 1.14 .897 ± 1.87 BMT/TOT 0.41 (�0.81, 1.63) 0.38TOT �.112 ± 1.34 .091 ± 1.36 UMT/BMT 0.73 (�0.25, 1.72) 0.87

SIS-Total (Participation)UMT 45.7 ± 18.4 54.3 ± 18.0 0.023 0.81 UMT/TOT �0.18 (�17.4, 17.0) �0.01BMT 42.1 ± 16.8 56.9 ± 19.8 BMT/TOT 5.95 (�16.8, 28.7) 0.29TOT 42.5 ± 13.6 51.3 ± 10.3 UMT/BMT �6.13 (�34.2, 21.9) �0.25

Values are mean and standard deviation. The p values from analysis of variance (ANOVA).BMT: bimanual mirror therapy; CI: confidence interval; Diff: difference; FMA: Fugl-Meyer Assessment; MD: Mean Difference; SIS: Stroke Impact Scale; TOT: traditionaloccupational therapy; UMT: unimanual mirror therapy.

UNIMANUAL AND BIMANUAL MIRROR THERAPY 5

Although the groups did not differ significantly on the ARATand FMA baseline data, the BMT group was clearly higher func-tioning. For this reason, an analysis of covariance (ANCOVA) onthe posttest scores using the pretest scores as a covariate wasperformed. The ANCOVA results for the ARAT and FMA showedno statistically significant between group differences (p¼ 0.33 and0.23, respectively).

Discussion

To our knowledge, this is the first intervention study to comparehome-based UMT and BMT in subacute/chronic stroke individualswith moderate to severe hemiparesis. This 6-week home-basedprogram demonstrated excellent adherence regardless of groupdesignation. Although Yao et al. [49], reported adherence instroke rehabilitation to be low, this study was modeled after pre-vious home-based MT studies that showed adherence success[15,34]. Possible explanations for the participant adherence couldbe due to the structured clinic sessions, weekly sessions with theresearch OT, log tracking and weekly reminders by phone callor email.

UMT versus BMT

It was hypothesized that UMT would be more beneficial for upperlimb recovery as compared to BMT. While there were no betweengroup differences, there were indications of clinical significance infavor of UMT at the activity level. The results showed a small tolarge effect size in favor of UMT, as per the ARAT and ABILHAND,respectively. Furthermore, the ABILHAND had an MD¼ 0.73; 95%CI, �0.25 to 1.72, suggesting that UMT may be more beneficial,but larger sample sizes are needed [48]. In addition, all UMT par-ticipants exceeded the MCID, while less than half met this in theBMT group. These results are consistent with Morkish et al. [32]research, which showed that while both UMT and BMT groupsimproved, the UMT groups had considerably large changes onmotor function and impairment as compared to BMT. In addition,previous randomized controlled MT studies showed significantimprovement on the ARAT for the UMT studies [24–26], but notfor the BMT studies [15,50].

At the impairment level, the FMA data suggested that UMTmay be more clinically beneficial than BMT for motor recovery.This is consistent with data presented by Selles et al. [31] wherebymovement time (impairment level) was not statistically differentbetween UMT and arm training without a mirror, while BMT wasstatistically less effective. In addition, while data presented byMorkisch et al. [32] showed that both MT protocols reducedmotor impairments, the UMT analysis showed more robustchanges. On the other hand, our grip strength data suggestedBMT as more clinically relevant as compared to UMT. This mayhave occurred due to the implicit difference between the mirrorprotocols, whereby in BMT the affected hand moves, while inUMT the hand remains static. Another explanation could be thatsince the sample size was small, the difference in the baselinedata for grip strength for BMT and UMT may have partially led tothe large effect size in favor of BMT. Although grip strength iscorrelated with improved UE function [43], our results suggestthat UMT may be more clinically relevant for the activity leveldomain. Possibly, the visual feedback from the mirror, which hasbeen shown to improve internal representation of limb dynamics[51], may positively influenced the activity level domains. Sarlegnaet al. [51] examined the effect of visual feedback on a novelreaching task with a deafferented patient as compared to healthy

matched controls. The results showed that the deafferentedpatient was able to adapt to the novel reaching condition, withthe use of the visual feedback, similarly to the controls. In otherwords, the peak velocities (limb dynamics) were similar for thepatient and the healthy controls, thus the similar reaching results.The researchers concluded that visual feedback may be able toupdate limb dynamics, thus improve reaching capability.

Previous studies have examined the effect of BMT on participa-tion, but this is the first study to examine the effects of UMT onparticipation. Participation results showed a significant effect oftime for all groups, but no between group differences. Similar toour results regarding BMT and participation, Thieme et al. [52]showed a main effect of time for participation as measured bythe SIS total score. In contrast to our findings, Michielsen et al.[15] showed no effect of time on participation. This inconsistencymay have occurred because the researchers used the EQ-5Dinstead of the SIS to measure participation. However, participationis a complex construct consisting of multiple layers, not only mak-ing it challenging to measure [53], but also doubtful that oneintervention type could effectively change life participation [54].

While no definitive conclusions can be drawn, the data sug-gested that UMT may be preferential to BMT for impairment andactivity gains. However, it should be noted that although therewere no statistically significant baseline differences on the ARATand FMA, it was clear that the BMT group had higher motor func-tion, which may have impacted the results. One possible explan-ation for UMT being more clinically relevant than BMT may bethat during BMT a visual/proprioceptive conflict occurred (mirrorvisual image conflicting with proprioception of the movingaffected hand), which caused a decline in the positive effects ofthe visual feedback [31]. This conflict has been shown to decreasemotor performance in healthy adults [55–56]. Another explanationcould be the concept of cross-limb transfer, whereby bilateralgains occur from unilateral motor training [57]. One stroke studyshowed significant gains in the affected lower limb, in bothstrength and gait after strength training of the unaffected lowerlimb [58]. Another study, in healthy adults, showed positivebehavioral and physiological outcomes due to cross-limb transferwith and without MT [59].

Mirror groups versus TOT

It was hypothesized that both mirror groups would be morebeneficial for upper limb recovery as compared to TOT. Althoughthere were no between group differences, there were indicationsof clinical significance in favor of both mirror groups, especiallyfor UMT. Regarding the activity level measures, both the ARATand ABILHAND showed moderate to large effect sizes for UMTover TOT as compared to small effect size for BMT over TOT.Furthermore, the ABILHAND data (MD¼ 1.15; 95% CI, 0.09- 2.2)suggested a strong clinical significance in favor of UMT over TOT.Thus, the activity level data showed a clinical significance in favorof UMT over TOT, however, minimal difference between BMTand TOT.

At the impairment level, the FMA data suggested that UMTmay be more clinically beneficial than TOT. This is consistent withseveral randomized controlled studies that reported improvementon the FMA with UMT [18,20,25,26] compared to control groups.The grip strength data, however, showed an increase in both BMTand TOT groups and a decline in the UMT group. Possibly, BMTand TOT are clinically more relevant for improving grip strengthdue to the repetitive movement of the affected hand.

6 D. GELLER ET AL.

While no definitive conclusions can be drawn, UMT may bemore beneficial and clinically relevant than TOT for improvementsat the activity and impairment level, except for grip strength. Inaddition, BMT appeared to have minimal to no clinical advantageas compared to TOT. One possible explanation could be thatsince the participants had moderate/severe upper limb impair-ments (poor intrinsic feedback), augmented feedback (visual feed-back) would be a beneficial strategy for motor recovery. Thus, MTparticipants (visual feedback) would benefit to a greater extent ascomparted to TOT participants who relied solely on intrinsic feed-back [57]. While BMT also used augmented feedback, it could beargued that the visual/proprioceptive conflict may have degradedthe positive effects of the visual feedback as compared to UMT[31]. Overall, data from this pilot study suggest a full scalerandomized controlled trial is warranted. Thus, to determine sam-ple size, a power analysis was performed using the ARAT effectsize data comparing UMT to BMT. The sample calculation wascomputed with G�power, an online tool (available at https://www.macupdate.com/app/mac/24037/g-power), with the statistical testANOVA: with repeated measures, between factor, a set at 0.05,power (1-b) set at 0.80, and effect size set at 0.38. The sample cal-culation was N¼ 54.

There were several limitations in this study. First, the samplesize was small, which makes it difficult to generalize to the popu-lation and may have created a type II error, which reflects a fail-ure to detect group differences if present. Furthermore, thedifferences in the baseline characteristic of the 3 groups mayhave affected the results given the small sample size. Second,because there was no long-term follow-up, there was no under-standing if improvements made post-intervention were main-tained, improved, or declined over time. Third, while all of theevaluators were trained on the assessment through lecture, dem-onstration, and videos, there was no formal interrater reliabilitytesting. Fourth, all of the participants received traditional occupa-tional therapy; however, the treating OTs differed per participantdue to logistics of the study and the facility. Due to variability ofthe OTs’ experience and treating expertise, the participants maynot have received the same care. However, all the OTs are man-aged under one clinical supervisor, and thus receive similar assist-ance and feedback, attend the same department in-services, andattend similar continuing education classes. Fifth, the log datashowed that all participants adhered to the home program, how-ever, this was all subjective data. In a future randomized con-trolled study, video calls would increase the ability to monitorparticipants, corroborate adherence and help with therapyprogression.

Disclosure of statement

No potential conflict of interest was reported by authors.

References

[1] Virani SS, Alonso A, Benjamin EJ, et al. Heart disease andstroke statistics—2020 update: a report from the. Am HeartAssoc Circ. 2020;141:e139–e596.

[2] Nakayama H, Jørgensen HS, Raaschou HO, et al. Recoveryof upper extremity function in stroke patients: the copen-hagen stroke study. Arch Phys Med Rehabil. 1994;75(4):394–398.

[3] Kim JH. The effects of training using EMG biofeedback onstroke patients upper extremity functions. J Phys Ther Sci.2017;29(6):1085–1088.

[4] Kwakkel G, Veerbeek JM, van Weegen EH, et al. Constraint-induced movement therapy after stroke. Lancet Neurol.2015;14(2):224–234.

[5] Lee MJ, Lee JH, Koo HM, et al. Effectiveness of bilateral armtraining for improving extremity function and activities ofdaily living performance in hemiplegic patients. J Strokeand Cerebrovasc Dis. 2017;26(5):1020–1025.

[6] Nilsen DM, Gillen G, Geller D, et al. Effectiveness of interven-tions to improve occupational performance of people withmotor impairments after stroke: an evidence-based review.Am J Occup Ther. 2015;69(1):6901180030p1–6901180030p9.DOI:10.5014/ajot.2015.011965.

[7] Thomas LH, French B, Coupe J, et al. Repetitive task train-ing for improved functional ability after stroke activity amajor update of a cochrane review and Meta-analysis.Stroke. 2017;48(4):e102-3.

[8] Maciejasz P, Eschweiler J, Gerlach-Hahn K, et al. A survey ofrobotic devices for upper limb rehabilitation. J NeuroengRehab. 2014;11:1–29.

[9] Lang CE, MacDonald J, Gnip C. Counting repetitions: anobservational study of outpatient therapy for people withhemiparesis post-stroke. J Neurol Phys Ther. 2007;31(1):3–10.

[10] Lang CE, MacDonald J, Reisman D, et al. Observation ofamounts of movement practice provided during strokerehabilitation. Arch Phys Med Rehabil. 2009;90(10):1692–1698.

[11] Ramachandran VS, Rogers-Ramachandran D, Cobb C.Touching the phantom limb. Nature. 1995;377(6549):489–490.

[12] Altschuler EL, Wisdom SB, Stone L, et al. Rehabilitation ofhemiparesis after stroke with a mirror. Lancet. 1999;353(9169):2035–2036.1.

[13] Thieme H, Morkisch N, Mehrholz J, et al. Mirror therapy forimproving function after stroke update of a cochranereview. Stroke. 2019;50(2):e26-27.

[14] Amasyali SY, Yaliman A. Comparison of the effects of mirrortherapy and electromyography-triggered neuromuscularstimulation on hand function in stroke patients: a pilotstudy. Int J Rehabil Res. 2016;39(4):302–307.

[15] Michielsen ME, Selles RW, van der Geest JN, et al. Motorrecovery and cortical reorganization after mirror therapy inchronic stroke patients: a phase II randomized controlledtrial. Neurorehabil Neural Repair. 2011a;25(3):223–233.

[16] Deconinck FJ, Smorenburg AR, Benham A, et al. Reflectionson mirror therapy: a systematic review of the effect of mir-ror visual feedback on the brain. Neurorehabil NeuralRepair. 2015;29(4):313–349.

[17] World Health Organization (WHO). 2001. International clas-sification of functioning, disability, and health (ICF).Geneva: World Health Organization.

[18] Arya KN, Pandian S, Kumar D, et al. Task-based mirror ther-apy augmenting motor recovery in poststroke hemiparesis:a randomized controlled trial. J Stroke Cerebrovasc Dis.2015;24(8):1738–1748.

[19] Colomer C, No�e E, Llorens R. Mirror therapy in chronicstroke survivors with severely impaired upper limb func-tion: a randomized controlled trial. Eur J Phys Rehabil Med.2016;52(3):271–278.

[20] Gurbuz N, Afsar SI, Ayas S, et al. Effect of mirror therapy onupper extremity motor function in stroke patients: arandomized controlled trial. J Phys Ther Sci. 2016;28(9):2501–2506.

UNIMANUAL AND BIMANUAL MIRROR THERAPY 7

[21] Mirela Cristina L, Matei D, Ignat B, et al. Mirror therapyenhances upper extremity motor recovery in strokepatients. Acta Neurol Belg. 2015;115(4):597–603. 2015

[22] Lee MM, Cho HY, Song CH. The mirror therapy programenhances upper-limb motor recovery and motor functionin acute stroke patients. Am J Phys Med Rehabil. 2012;91(8):689–700.

[23] Wu CY, Huang PC, Chen YT, et al. Effects of mirror therapyon motor and sensory recovery in chronic stroke: arandomized controlled trial. Arch Phys Med Rehabil. 2013;94(6):1023–1030. 2013

[24] Invernizzi M, Negrini S, Carda S, et al. The value of addingmirror therapy for upper limb motor recovery of subacutestroke patients: a randomized controlled trial. Eur J PhysRehabil Med. 2013;49:311–317.

[25] Khandare SS, Singaravelan RM, Khatri SM. Comparison oftask specific exercises and mirror therapy to improve upperlimb function in subacute stroke patients. J Dental Med Sci.2013;7:5–14.

[26] Kim K, Lee S, Kim D, et al. Effects of mirror therapy com-bined with motor tasks on upper extremity function andactivities daily living of stroke patients. J Phys Ther Sci.2016;28(2):483–487.

[27] Park Y, Chang M, Kim KM, et al. The effects of mirror ther-apy with tasks on upper extremity function and self-care instroke patients. J Phys Ther Sci. 2015;27(5):1499–1501.

[28] Lim KB, Lee HJ, Yoo J, et al. Efficacy of mirror therapy con-taining functional tasks in Poststroke Patients. Ann RehabilMed. 2016;40(4):629–636.

[29] Rajappan R, Abudaheer S, Selvaganapathy K, et al. Effect ofmirror therapy on hemiparetic upper extremity in subacutestroke patients. Int J Physiother. 2015;2:1041–1046.

[30] Samuelkamaleshkumar S, Reethajanetsureka S, PauljebarajP, et al. Mirror therapy enhances motor performance in theparetic upper limb after stroke: a randomized controlledtrial. Arch Phys Med Rehabil. 2014;95(11):2000–2005.

[31] Selles RW, Michielsen ME, Bussmann J, et al. Effects of amirror-induced visual illusion on a reaching task in strokepatients: Implications for mirror therapy training.Neurorehabil Neural Repair. 2014;28(7):652–659.

[32] Morkisch N, Thieme H, Dohl C. How to perform mirror ther-apy after stroke? Evidence from a Meta-analysis. RestorNeurol Neurosci. 2019;37(5):421–435.

[33] Woodbury CA, Velozo CA, Richards LG, et al. Rasch analysisstaging methodology to classify upper extremity move-ment impairment after stroke. Arch Phys Med Rehabil.2013;94(8):1527–1533.

[34] Nilsen DM, DiRusso T. Using mirror therapy in the homeenvironment: a case report. Am J Occup Ther. 2014;68(3):e84–e89.

[35] Bushnell C, Bettger J, Cockroft K, et al. Chronic stroke out-come measures for motor function intervention Trials:Expert Panel Recommendations. Circ Cardiovasc QualOutcomes. 2015;8(6 Suppl 3):S163–S9.

[36] Yozbatiran N, Der-Yeghiaian L, Cramer SC. A standardizedapproach to performing the action research arm test.Neurorehabil Neural Repair. 2008;22(1):78–90.

[37] Van der Lee J, De Groot V, Beckerman H, et al. The intra-and interrater reliability of the action research arm test: apractical test of upper extremity function in patients withstroke. Arch Phys Med Rehabil. 2001;82(1):14–19.

[38] Fugl-Meyer AR, J€a€ask€o L, Leyman I, et al. The post-strokehemiplegic patient. Scand J Rehabil Med. 1975;7(1):13–31.

[39] Platz T, Pinkowski C, van Wijck F, et al. Reliability and valid-ity of arm function assessment with standardized guide-lines for the Fugl-Meyer Test, Action Research Arm Testand Box and Block Test: a multicentre study. Clin Rehabil.2005;19(4):404–411.

[40] Page S, Fulk G, Boyne P. Clinically important differences forthe upper-extremity Fugl-Meyer Scale in people with min-imal to moderate impairment due to chronic stroke. PhysTher. 2012;92(6):791–798.

[41] Simone A, Rota V, Tesio L, et al. Generic ABILHANDquestionnaire can measure manual ability across a varietyof motor impairments. Int J Rehabil Res. 2011;34(2):131–140.

[42] Wang TN, Lin KC, Wu CY, et al. Validity, responsiveness,and clinically important difference of the ABILHAND ques-tionnaire in patients with stroke. Arch Phys Med Rehabil.2011;92(7):1086–1091.

[43] Boissy P, Bourbonnais D, Carlotti MM, et al. Maximal gripforce in chronic stroke subjects and its relationship to glo-bal upper extremity function. Clin Rehabil. 1999;13(4):354–362.

[44] Sunderland A, Tinson D, Bradley L, et al. Arm function afterstroke. An evaluation of grip strength as a measure ofrecovery and a prognostic indicator. J Neurol NeurosurgPsychiatry. 1989;52(11):1267–1272.

[45] Vellone E, Savini S, Fida R, et al. Psychometric evaluation ofthe stroke impact scale 3.0. J Cardiovasc Nurs. 2015;30(3):229–241.

[46] Ghasemi A, Zahediasl S. Normality tests for statistical ana-lysis: a guide for non-statisticians. Int J Endocrinol Metab.2012;10(2):486–489.

[47] Kim HY. Statistical notes for clinical researchers: assessingnormal distribution (2) using skewness and kurtosis. RestorDent Endod. 2013;38(1):52–54.

[48] Page P. Beyond statistical significance: clinical interpret-ation of rehabilitation research and literature. Int J SportsPhys Ther. 2014;9(5):726–736.

[49] Yao M, Chen J, Jing J, et al. Defining the rehabilitationadherence curve and adherence phases of stroke patients:an observational study. Patient Prefer Adherence. 2017;11:1435–1441.

[50] Dohle C, P€ullen J, Nakaten A, et al. Mirror therapy pro-motes recovery from severe hemiparesis: a randomizedcontrolled trial. Neurorehabil Neural Repair. 2009;23(3):209–217.

[51] Sarlegna FR, Malfait N, Bringoux L, et al. Force-field adapta-tion without proprioception: can vision be used to modellimb dynamics? Neuropsychol. 2010;48(1):60–67.

[52] Thieme H, Bayn M, Wurg M, et al. Mirror therapy forpatients with severe arm paresis after stroke: a randomizedcontrolled study. Clin Rehabil. 2013;27(4):314–324.

[53] Chang F, Coster W. Conceptualizing the construct of par-ticipation in adults with disabilities. Arch Phys Med Rehabil.2014;95(9):1791–1798.

[54] Salter K, Jutai JW, Teasell R, et al. Issues for selection ofoutcome measures in stroke rehabilitation: ICF participa-tion. Disabil Rehabil. 2005;27(9):507–528.

[55] Holmes NP, Crozier G, Spence C. When mirrors lie: "visualcapture" of arm position impairs reaching performance .Cogn Affect Behav Neurosci. 2004;4(2):193–200.

[56] Lajoie Y, Paillard J, Teasdale N, et al. Mirror drawing in adeafferented patient and normal subjects: Visuoproprioceptiveconflict. Neurology. 1992;42(5):1104–1106.

8 D. GELLER ET AL.

[57] Magill RA, Anderson D, eds. Motor learning and control:concepts and applications (10th ed.) New York (NY):McGraw-Hill; 2014.

[58] Dragert K, Zehr E. High-intensity unilateral dorsiflexoresistance training results in bilateral neuromuscular

plasticity after stroke. Exp Brain Res. 2013;225(1):93–104.

[59] Nojima I, Mima T, Koganemaru S, et al. Human motor plas-ticity induced by mirror visual feedback. J Neurosci. 2012;32(4):1293–1300.

UNIMANUAL AND BIMANUAL MIRROR THERAPY 9