Physiotherapy - Vol 94 - N2 2008

8/4/2019 Physiotherapy - Vol 94 - N2 2008

1/86

Physiotherapy V OLUME94, NUMBER 2, 2008

EditorMichele Harms PhD MSc GradDipPhys MCSPEditor

The Chartered Society of PhysiotherapyLondon, UK

Editorial OfcePhysiotherapy Editorial OfceElsevier Ltd, The Boulevard

Langford LaneKidlingtonOxford OX5 1GB, UKTel.: +44 (0)1865 843753Fax: +44 (0)1865 843992E-mail: [email protected]

Associate EditorsClaire Ballinger PhD MSc DipCOTGlasgow Caledonian UniversityGlasgow, UK

Richard Bohannon EdD, PT, NCS, FAHA, FAPTAUniversity of ConnecticutConnecticut, USA

Tracy Bury MSc GradDipPhys MCSPWCPT SecretariatLondon, UK

Vinette Cross PhD MMedEd MCSP DipTP CertEdUniversity of BirminghamBirmingham, UK

Mike Hurley PhD GradDipPhys MCSPKings College LondonLondon, UK

Sallie Lamb DPhil MSc GradDipPhys MCSPUniversity of Warwick

Warwick, UK

Gareth Noble PhD BSc(Hons)University of WalesSwansea, Wales

Sheila Payne PhD BA(Hons) RN DipN C. PhyscholLancaster UniversityLancaster, UK

Valerie Pomeroy PhD BA GradDipPhys FCSPUniversity of East Anglia

Norwich, UK

Gabrielle Rankin PhD MSc GradDipPhys MCSPChartered Society of PhysiotherapyLondon, UK

Patricia Roche PhD MSc BSc(Hons) MCSPQueen Margaret University CollegeEdinburgh, Scotland

Sally Singh PhD BSc MCSPGleneld HospitalLeicester, UK

Paul Watson PhD MSc GradDipPhys MCSPBSc(Hons) CertHSM PGCE DipPTUniversity of LeicesterLeicester, UK

International Advisory BoardD. Bader (London, UK)K. Bo (Oslo, Norway)M. OBrien (Ontario, Canada)C. Cott (Toronto, Canada)R. de Bie (Maastricht, The Netherlands)W. de Weerdt (Heverlee, Belgium)C. Eales (Johannesburg, South Africa)J. Hay Smith (Dunedin, New Zealand)

S. Jenkins (Perth, Australia)G. Jensen (Nebraska, USA)R. Ladyshewsky (Perth, Australia)I. Nara (Kobe, Japan)A. Nieuwboer (Heverlee, Belgium)P. Pothongsunun (Chiang Mai, Thailand)G. Wulf (Nevada, USA)

The journal of the Chartered Society of Physiotherapy Visit the journal website at www.intl.elsevierhealth.com/journals/phys

Amsterdam Boston Jena London New York Oxford Paris Philadelphia San Diego St. Louis


2/86

Physiotherapy 94 (2008) 8990

Editorial

Client-centred rehabilitation: what is it and how do we measure it?

Client centredness has emerged in recent years as animportant underlying principle for the delivery of health andrehabilitation services [1]. Increasingly, hospitals are imple-menting programmes to enhance client centredness of care,but there are few measures to evaluate hospital performancein these areas. In Ontario, the largest province in Canada,indicators of client centredness have been developed at bothclient and organisational level in order to facilitate the evalu-ation of programmes in the Hospital report: rehabilitation,a province-wide quality improvement and research initiativethat uses a balanced scorecard approach to evaluate the per-formance of all hospitals with designated rehabilitation beds.

A balanced scorecard approach [2] recognises that hos-pital care is a complicated activity requiring a multitude of skills, experiences and technologies [3]. Performance mea-surement must therefore include a balanced scorecard of indicators that, when taken together, provide insights intothe overall performance of a hospital. The balanced score-card approach describes performance across four quadrants

that are critical to the success of any healthcare organisa-tion: system integration and change (SIC) (the changes andinvestments that facilities need to make in order to moveahead in thenext 35years); clinical utilisation andoutcomes(elements of clinical success); client perspectives (dimen-sions of care that are most relevant to clients); and nancialperformance and condition.

The rst step in developing client-centredness indica-tors for the SIC and client perspectives quadrants requireddenition of the concept of interest, i.e. client-centred reha-bilitation. A review of the literature revealed that there wasno commonly used denition of client-centred rehabilitation[4], and that most available denitions focused on acute care,usually from the perspective of various health professionalsrather than clients.

In order to address these shortcomings, focus groupswereconductedwithadultrehabilitationclients withchronic phys-ically disabling conditions and rehabilitation professionalsin order to identify the important domains of client-centredrehabilitation [5,6] . From these studies, it was concluded thatclient-centred rehabilitation is a philosophy or approach tothe delivery of rehabilitation services that reects the needs

of individuals and groups of clients. Further, client-centredrehabilitation can be thought of at two levels: client (clientsareactively involved inmanaging their care andtheirrehabili-tation in partnership withservice providers); andorganisation(an approach to care that strives to incorporate clientperspec-tives into the provision of services). Table 1 shows the sevendomains of client-centred rehabilitation at client level andtheseven domains of client-centred rehabilitation at organisa-tional level identied in the research. These domains formedthebasisfor thedevelopment of indicators at twolevels: clientand organisation.

Creation of a tool to measure client centredness at clientlevel consistedof thedevelopmentof ve to sixquestionnaireitems for each of the seven domains of client-centred reha-bilitation, cognitive interviews with rehabilitation clients toevaluate the clarity and relevance of the items, and testing of the psychometricpropertiesof the toolvia a self-administeredsurvey mailed to over 1000 patients discharged from tworehabilitation facilities. The product of this pilot work was

a 30-item questionnaire the Client Centred RehabilitationQuestionnaire (CCRQ) which has demonstrated inter-itemand testretest reliability and concurrent validity [7].

Subsequently, the CCRQ was embedded within the clientperspectives of rehabilitation survey for the Hospital report:rehabilitation. For the 2005 report, it was sent to all clients(n = 11 464) discharged from designated rehabilitation bedswithin the province of Ontario in 2004. There was a 46%response rate ( n = 5274). Further construct validity testingwas undertaken by correlating each individual question andeach domain score in the CCRQ with a general measure of satisfaction (i.e. Please rate the overall quality of care andservices you received) that functioned as a gold standardof client satisfaction [8]. Clients perceived their quality of care to be higher when they felt that: they were involved indetermining their progress in rehabilitation; their physicalcomfort needs were being recognised and handled by staff;they were being treated with respect and dignity; and theyhad been given the information that they needed to managetheir condition in thecommunity. Most importantwas feelingthat they had been treated with respect and dignity, and thatthey were seen as a person and not just another case.

0031-9406/$ see front matter 2008 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.physio.2008.02.004
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://dx.doi.org/10.1016/j.physio.2008.02.004http://dx.doi.org/10.1016/j.physio.2008.02.004http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


3/86

90 Editorial / Physiotherapy 94 (2008) 8990

Table 1Domains of client-centred rehabilitation at client and organisational level

Domains at client level Domains at organisational level

Client participation in decision-making and goal-setting Client involvement in service planningClient-centred education (information is timely and appropriate) Evaluation of the programmes and services from the clients perspectiveEvaluation of outcomes from the clients perspective, not just from the

perspective of the rehabilitation professionalsOrganisational response to client feedback

Family involvement and support AccessibilityCoordination and continuity of care (clients often feel isolated and abandoned

after discharge from rehabilitation)Interdisciplinary approach

Emotional support (clients feel they are being listened to and treated withrespect and dignity)

Coordination and continuity with community services

Physical comfort (recognition and management of pain) Organisational support for client-centred care

Two indicators were used to measure client centrednessat organisational level. The rst, evidence of client-centredcare, measures the extent to which systems and processesare in place to ensure that the individual care that clientsreceive during their rehabilitation is client centred. It isderived from seven questions in the SIC questionnaire that

parallel the following four components of client-centredrehabilitation: client involvement in decision-making andgoal-setting; client-centred and family-centred education;family involvement; and emotional support mechanisms forclients/families and staff.

The second organisational level indicator, evidence of organisational client centredness, measures the extent towhich hospitals implement a client-centred approach toservice delivery at organisational level. The indicator isconstructed from four questions in the SIC questionnairethat address: the extent to which hospitals are collectingclient/family feedback, including evidence that client per-spectives are incorporated into service planning; strategiesthathospitalsutilise to disseminate client andfamily feedback results; and the existence of specic staff roles that facil-itate client-centred care (e.g. patient advocate/ombudsman,designated contact person assigned to each client and fam-ily, or designated staff who addresses equity issues). Moredetaileddenitionsof the twoindicators, includingweightingof each question, are available in the rehabilitation technicalsummary at http://www.hospitalreport.ca/ .

Preliminary examination of the inter-relationshipsbetween the client perspectives indicators and the twoclient-centredness SIC indicators found that hospitals whichactively involve clients and families within the organisation

are perceived by clients to provide care of higher quality, torecogniseand managephysicalcomfort needs,and to activelyinvolve clients in evaluating their rehabilitation outcomes.These results represent trends that could provide directionfor hospitals with respect to quality improvement efforts.

In conclusion, client centredness is an important com-ponent of any rehabilitation programme. It is important for

organisations to utilise measures of client centredness thatcan be used for discriminative and evaluative purposes sothat quality of care canbe monitoredwithinandbetween pro-grammes. In Ontario, measures of client centredness at bothclient and organisational level have been developed and arebeing used as part of a province-wide quality improvement

initiative. Although developed in Canada, these measures arein the public domain and have the potential to be utilised inother rehabilitation service delivery systems.

Conict of interest: None.

References

[1] Picker Institute. Eyeon patient: a reportby PickerInstitute forthe Amer-ican Hospital Association. Boston, MA: Picker Institute Inc.; 2000. p.114.

[2] Kaplan RB, Norton DP. The balanced scorecard measures that driveperformance. Harvard Bus Rev 1992;70:719.

[3] Baker GR, Pink GH. A balanced scorecard for Canadian hospitals.

Healthcare Manage Forum 1995;8:713.[4] Cott CA, Boyle J, Fay J, Sutton D, Bowring J, Lineker S. Client-centredrehabilitation. Working Report 157. Toronto: Arthritis CommunityResearch and Evaluation Unit; 2001.

[5] Cott CA. Client-centred rehabilitation: client perspectives. Dis Rehabil2004;26:141122.

[6] Soever L, Cott CA, Boyle J. Client centred rehabilitation II: health careprofessionals perspectives. Working Report 160. Toronto: ArthritisCommunity Research and Evaluation Unit; 2003.

[7] Cott CA, Teare G, McGilton KS, Lineker S. Development and pilottesting of the client-centred rehabilitation questionnaire. Dis Rehabil2006;28:138797.

[8] Tyas J, Soever L, MacKay C, Cott CA. Client perspectives technicalsummary. Hospital report 2005: rehabilitation. Toronto: Hospital ReportResearch Collaborative, University of Toronto; 2005.

Cheryl A. Cott Department of Physical Therapy, Faculty of Medicine,

University of Toronto, 160500 University Avenue, Toronto,Ontario M5G 1V7, Canada

Tel.: +1 416 978 0301; fax: +1 416 946 8562. E-mail address: [email protected]

Available online at www.sciencedirect.com
http://www.hospitalreport.ca/http://-/?-mailto:[email protected]:[email protected]://-/?-http://www.hospitalreport.ca/


4/86

Physiotherapy94(2008)9196

Incidence and effect of leg length discrepancyfollowing total hip arthroplasty

D.J. Beard a, , J. Palan a, J.G. Andrew b, J. Nolan c, D.W. Murray a,

EPOS Study Groupa Nufeld Department of Orthopaedic Surgery, University of Oxford, Nufeld Orthopaedic Centre,

Oxford OX3 7LD, UK b Ysbyty Gwynedd District Hospital, Penrhosgarnedd, Bangor, UK

c Norfolk and Norwich University Hospital, Norwich, UK

Abstract

Objectives The clinical importance of a leg length discrepancy (LLD) following total hip arthroplasty (THA) remains controversial. Thisstudy was undertaken to determine the effects of LLD on clinical outcomes at up to 3 years follow-up.Design and setting Prospective, multicentre study.Participants Nine hundred and eighty-seven cases of primary THA, categorised into two main groups: the NoLLD group (LLD of less than10 mm) and the LLD group (LLD of 10 mm or more).Main outcome measures The primary outcome measure was the change in Oxford Hip Score ( OHS) at up to 3 years follow-up. Secondaryoutcome measures were length of operating time and hospital stay, and revision and dislocation rates. Potential predisposing factors for LLD,including body mass index, age and type of anaesthesia employed, were examined.Results At 3 years, the LLD group had a signicantly worse OHS [22.0; 95% condence interval (CI) 20.5 to 23.5] compared with theNoLLD group (23.8; 95% CI 23.1 to 24.5) ( P = 0.034). There were no signicant differences in revision ( P = 0.389) or dislocation ( P = 0.220)rates between the two groups. Use of an epidural was associated with a decreased incidence of developing an LLD of 10 mm ( P = 0.004).Conclusion A postoperative LLD of 10 mm or more leads to poorer functional outcomes. Further studies are needed to assess the impact of an LLD on clinical outcomes in the longer term. 2008 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved.

Keywords: Leg length discrepancy; Arthroplasty, hip; Dislocation; Revision; Oxford Hip Score

Introduction

Leg length discrepancy (LLD) has been well describedas a common complication following total hip arthroplasty(THA) [1,2]. The presence of LLD has been associated withback pain [3,4], increased risk of nerve injury [5] and dis-location [6], poor patient satisfaction [7] and the need forrevision surgery [8]. It has been cited as a major cause of liti-gation followingTHA [9]. In theliterature,thereis continuingdebate about the importance of LLD and its clinical effect.

Corresponding author. Tel.: +44 1865 227454; fax: +44 1865 227671. E-mail address: [email protected]

(D.J. Beard).

A recent paper by Konyves and Bannister [10] concludedthat patients with an LLD (longer leg length on operatedside) hada worse functional outcome compared with patientswho did not have an LLD. However, a study by White andDougall [11] showed no statistical association between LLDand patient satisfaction and outcome. However, in both of these studies, the cohort size was small and single centred.

Objectives

The aim of this study was to determine the clinical effectsof LLD following THA using the change in the Oxford HipScore ( OHS) as a primary outcome measure.

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-mailto:[email protected]://-/?-http://-/?-http://dx.doi.org/10.1016/j.physio.2008.01.005http://dx.doi.org/10.1016/j.physio.2008.01.005http://-/?-http://-/?-mailto:[email protected]://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


5/86

92 D.J. Beard et al. / Physiotherapy 94 (2008) 9196

Design and setting

This was a prospective non-randomised multicentre studyinvolving seven centres.

Participants

In total, 987 THAs were examined in this study, involvingconsultant andnon-consultant surgeons andutilising twodif-ferent surgical approaches (anterolateral and posterior). Thediagnosis in every case was that of primary osteoarthritis andall were unilateral hips. None of the patients had undergoneprevious surgery to the hip. In all cases, a cemented Exeterfemoral stem (Stryker Howmedica Osteonics, Mahwah, NewJersey, USA) was used with various acetabular components.

The true leg length was measured in millimetres (mm)using a direct tape measure method, with the patient in asupine position, from the anterior superior iliac spine to themedial malleolus of the operated side and non-operated side[1214]. The LLD was calculated as the difference betweenthe two measurements. The patients were categorised intotwo main groups: the NoLLD group, withan LLD of less than10mm (n = 794); and the LLD group, with an LLD of 10mmor more (n = 193). A subgroup analysis was also performedon a small number of patients ( n = 55) who had a severe LLDof 20 mm of more.

The mean LLD was 0.7 mm [standard deviation(SD) 1.8] and 14.9 mm (SD 7.4) in the NoLLD groupand the LLD group, respectively. In the subgroup of patientswith an LLD of 20 mm or more, the mean LLD was 24.7 mm

(SD 2.1). The overall mean LLD in the series was 3.5 mm.The LLD was assessed to determine whether the effect of thediscrepancy was to lengthen or shorten the affected leg. Inthe NoLLD group, 417 patients had equal leg lengths. Withinthis group, there were 110 patients with a minor LLD of lessthan 10 mm, of which 49 patients (45%) had a shorteningof the operated leg and 61 patients (56%) had a lengthen-ing of the operated leg. In the LLD group, 75 cases of LLDof 10 mm or more (39%) were shorter and 118 cases (61%)were longer ( P < 0.001, Chi-squared test). Within the LLD

group, the mean length of leg shortening was 13.9 mm andthe mean length of leg lengthening was +15.6 mm.

Patient demographics are shown in Table 1. There were293 men and 498 women in the NoLLD group, and 69men and 123 women in the LLD group. Patients in theNoLLD group were younger than those in the LLD group

(P = 0.014). There was no signicant difference in the meanbody mass index of the two groups ( P = 0.135). TheCharnleycategories for each group of patients were not signicantlydifferent ( P = 0.865). CategoryA describesunilateral hipdis-ease, Category B describesbilateral hipdisease andCategoryC describes multiple joint involvement which limits walkingability [15] (Table 1).

Main outcome measures

The OHS was used as a well-validated method [16,17] o f assessing clinical outcomes on a patient-centred basis. TheOHS is scored from 12 (best) to 60 (worst) and consists of 12 questions, each marked from one (best) to ve (worst).Pre- and postoperative OHSs were collected and the changein OHS ( OHS) was calculated at 3 months, 1 year and3 years. The OHS was also analysed as separate questioncomponents to identify if there were particular functionalactivities or symptoms which affected those patients with anLLD.Secondaryoutcome measures included the meanlengthof operating time and hospital stay, and revision and disloca-tion rates for each of the LLD groups. The type of anaestheticused for each patient was recorded to assess if there was anassociation between the type of anaesthetic used and LLD.

This was either a general anaesthetic, spinal anaesthesia oran epidural.

Statistics

For the outcome measures, analysis of variance andTukeys Post Hoc test were used to compare differences indatabetween the groups.Categorical and frequencydata wereanalysed using Chi-squared and Fishers Exact tests, and the

-levelofsignicancewasdenedaslessthan5%( P 0.05).

Table 1Demographics and preoperative data for the two study groups

Total no. of total hiparthroplasties

n NoLLD group (LLD of less than 10 mm)

n LLD group (LLD of 10 mm or more)

P -value

Gender (total) 983 791 192 0.803Men 293 69Women 498 123

Mean age at operation (years) +1 SD 985 792 69.4 8.5 193 71.1 8.5 0.014Mean body mass index (kg/m 2) + 1 SD 936 752 27.6 5.0 184 27.0 4.3 0.135

Charnley category 939 754 185 0.865A 502 122B 141 33C 111 30

LLD, leg length discrepancy; SD, standard deviation.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


6/86


7/86


Table 3Mean change in Oxford Hip Score ( OHS) in the leg length discrepancy (LLD) group, at up to 3 years follow-up, comparing a shorter and longer LLD

Total no. of total hiparthroplasties

Shorter LLD Longer LLD P -value b

n Mean 95% CI n Mean 95% CI

Mean OHSa

3 months 170 68 18.6 16.3 to 20.9 102 17.8 16.0 to 19.6 0.568

1 year 188 73 22.1 19.9 to 24.3 115 22.8 21.3 to 24.5 0.5913 years 153 57 21.4 18.9 to 23.9 96 22.4 20.5 to 24.3 0.498

CI, condence intervals.a OHS: 0 (worst) to 48 (best).b P -value calculated using analysis of variance.

of 20 mm or more, scoring worse for Question 9 comparedwith the NoLLD group ( P = 0.001).

Secondary outcome measures

The percentage of patients in each of the three groupsneeding revision surgery was analysed. In the NoLLD group,six of 794 cases [relative frequency (RF) = 1%] needed revi-sion, and in the LLD group, three of 193 cases were revised(RF=2%; P = 0.389). There was no signicant difference indislocation rate between thegroups (NoLLD group:RF = 2%,16/794 cases; LLD group: RF = 1%, 1/193 cases; P = 0.220).Themean lengthof operatingtime for the two groupswas sig-nicantly different ( P = 0.038), as was the length of hospitalstay (P = 0.006). The NoLLD group had a longer mean oper-ating time (89.9 minutes) and a shorter length of hospital stay(9.0 days) compared with the LLD group (mean operatingtime 84.9 minutes, length of hospital stay 10.6 days).

Use of an epidural was associated with a decreased inci-

dence of having an LLD greater than 10 mm ( P = 0.004). Ahigher percentage of patients in the NoLLD group (29%,229/793 cases) received an epidural and had an LLD of less than 10 mm, whereas in the LLD group, the percent-age was 19% (36/193 cases). No signicant differences wereassociated with a general ( P = 0.307) or spinal ( P = 0.225)anaesthetic and an LLD of 10 mm or more.

Discussion

LLD has been reported in the literature as being a com-mon nding following primary THA, although there is alack of consensus about what constitutes a clinically signi-cant postoperative inequality [18]. The vast majority of thesepatients have an LLD of less than 10mm; in one series, 97%of patients undergoing THA had an LLD of less than 10 mm,with a mean LLD of 1 mm [19]. In another study, Turula et al. [14] f ound that LLD varied from 20 (shortened leg) to+15 mm (lengthened leg) with a mean of 2.8 mm. In a con-secutive series of 100 patients, Ranawat and Rodriguez [20]demonstrated that the mean LLD was 3.4 mm (range 10to 18 mm). In the present series, the relative frequency of patients with an LLD of less than 10 mm was 80%, with anoverall mean LLD of 3.5 mm. The incidence of an LLD of

20 mm or more was 6% (55/987 cases). Such a pronounceddiscrepancy has been described in the literature as beingpoorly tolerated by patients [21].

This study has demonstrated thathaving anLLD of 10mmor more is associated with having a signicantly poorer out-come in terms of the clinical benet of surgery ( OHS)compared with patients who either have equal leg lengthsor an LLD of less than 10 mm. Whether such a differencein OHS is clinically important is more difcult to assess.Murray et al. have described a two-point change in OHSas being the minimum clinical change perceived by patientsas meaningful and which may lead to changes in clinicalpractice [22].

Therewasno difference in OHSat3yearswhenashorterLLD was compared with a longer LLD. It is surprising tonote that those patients in the LLD group who had an LLD of 20 mmor more still gaineda considerablebenet from havinga THA and had a greater OHS at 3 years compared withpatientswho had anLLD of 10to


8/86


9/86


[6] Woo RY, Morrey BF. Dislocations after total hip arthroplasty. J BoneJoint Surg Am 1982;64:1295306.

[7] Maloney WJ, Keeney JA. Leg length discrepancy after total hip arthro-plasty. J Arthroplasty 2004;19(Suppl. 1):10810.

[8] Austin MS, Hozack WJ, Sharkey PF, Rothman RH. Stability and leglength equality in total hip arthroplasty. J Arthroplasty 2003;18(Suppl.1):8890.

[9] Attarian DE, Vail TP. Medicolegal aspects of hip and knee arthroplasty.Clin Orthop Relat Res 2005:726.

[10] Konyves A, Bannister GC. The importance of leg length discrepancyafter total hip arthroplasty. J Bone Joint Surg Br 2005;87:1557.

[11] White TO, Dougall TW. Arthroplasty of the hip. Leg length is notimportant. J Bone Joint Surg Br 2002;84:3358.

[12] Beattie P, Isaacson K, Riddle DL, Rothstein JM. Validity of derivedmeasurements of leg-length differences obtained by use of a tape mea-sure. Phys Ther 1990;70:1507.

[13] Parvizi J, Sharkey PF, Bissett GA, Rothman RH, Hozack WJ. Surgicaltreatment of limb-length discrepancy following total hip arthroplasty.J Bone Joint Surg Am 2003;85-A:23107.

[14] Turula KB, Friberg O, Lindholm TS, Tallroth K, Vankka E. Leglength inequality after total hip arthroplasty. Clin Orthop Relat Res1986:1638.

[15] Garellick G, Malchau H, Herberts P. Specic or general health out-comemeasures in theevaluationof totalhip replacement. A comparisonbetween theHarris hipscore andthe Nottingham HealthProle. J BoneJoint Surg Br 1998;80:6006.

[16] Dawson J, Fitzpatrick R, Carr A, Murray D. Questionnaire on the per-ceptions of patients about total hip replacement. J Bone Joint Surg Br1996;78:18590.

[17] Dawson J, Fitzpatrick R, Murray D, Carr A. Comparison of measuresto assess outcomes in total hip replacement surgery. Qual Health Care1996;5:818.

[18] Abraham WD,Dimon IIIJH. Leglengthdiscrepancy intotalhip arthro-plasty. Orthop Clin North Am 1992;23:2019.

[19] Woolson ST, Hartford JM, SawyerA. Results of a method of leg-lengthequalization for patients undergoing primary total hip replacement. JArthroplasty 1999;14:15964.

[20] Ranawat CS, Rodriguez JA. Functional leg-length inequality followingtotal hip arthroplasty. J Arthroplasty 1997;12:35964.

[21] Gross RH.Leg lengthdiscrepancy:how muchis toomuch? Orthopedics1978;1:30710.

[22] Murray DW, Fitzpatrick R, Rogers K, Pandit H, Beard DJ, Carr AJ, et al. The use of the Oxford hip and knee scores. J Bone Joint Surg Br2007;89:10104.

[23] Royall RM. The effect of sample size on the meaning of signicancetests. Am Statist 1986;40:3135.

[24] Holmstrom B, Laugaland K, Rawal N, Hallberg S. Combined spinalepiduralblock versus spinal andepiduralblock for orthopaedic surgery.Can J Anaesth 1993;40:6016.

[25] Jasty M, Webster W, Harris W. Management of limb length inequalityduring total hip replacement. Clin Orthop Relat Res 1996;333:16571.

[26] Brand RA, Yack HJ. Effects of leg length discrepancies on the forcesat the hip joint. Clin Orthop Relat Res 1996;333:17280.

[27] CallaghanJJ, HeithoffBE, Goetz DD,SullivanPM,PedersenDR, John-ston RC. Prevention of dislocation after hip arthroplasty: lessons fromlong-term followup. Clin Orthop Relat Res 2001;393:15762.

[28] Coventry MB. Late dislocations in patients with Charnley total hiparthroplasty. J Bone Joint Surg Am 1985;67:83241.

[29] Padgett DE, Warashina H. The unstable total hip replacement. ClinOrthop Relat Res 2004;420:729.

[30] Bhave A, Mont M, Tennis S, Nickey M, Starr R, Etienne G. Func-tional problems and treatment solutions after total hip and knee jointarthroplasty. J Bone Joint Surg Am 2005;87(Suppl. 2):921.

[31] Defrin R, Ben Benyamin S, Aldubi RD, Pick CG. Conservative cor-rection of leg-length discrepancies of 10 mm or less for the relief of chronic low back pain. Arch Phys Med Rehabil 2005;86:207580.

[32] Gurney B. Leg length discrepancy. Gait Posture 2002;15:195206.



10/86


Systematic review

Assessing the effect of sample size, methodological quality and statisticalrigour on outcomes of randomised controlled trials on mobilisation,

manipulation and massage for low back pain of at least 6 weeks durationDries M. Hettinga a, Deirdre A. Hurley b, Anne Jackson a, ,

Stephen May c, Chris Mercer d, Lisa Roberts ea Research and Development, Chartered Society of Physiotherapy, 14 Bedford Row, London WC1R 4ED, UK b School of Physiotherapy and Performance Science, University College Dublin, Beleld, Dublin 4, Ireland

c Faculty for Health and Wellbeing, Shefeld Hallam University, Shefeld S10 2BP, UK d Physiotherapy Department, Worthing and Southlands Hospitals NHS Trust, Upper Shoreham, Shoreham-by-Sea BN43 6TQ, UK

e

School of Health Professions and Rehabilitation Sciences, University of Southampton, Higheld, Southampton SO17 1BJ, UK

Abstract

Objectives To assess the effect of sample size, methodological quality and statistical rigour on outcomes of randomised controlled trials(RCTs) on manual therapy (i.e. manipulation, mobilisation and/or massage) for non-specic low back pain (LBP) of at least 6 weeks duration,and to report results from RCTs with adequate sample size, methodological quality and statistical rigour.Data sources MedLine, EMBASE, CINAHL, AMED, Cochrane, PEDro and the library collection of the Chartered Society of Physiotherapy.Review methods RCTs were identied that compared manual therapy with a control or alternative intervention in adults with non-specicLBP of at least 6 weeks duration. The sample size, methodological quality (adapted 10-point van Tulder scale) and statistical rigour were thenassessed. RCTs were regarded as higher quality if they fullled the following three criteria: (a) >40 subjects in the manual therapy group; (b)scoring >5/10 on the Van Tulder scale; and (c) reporting statistical tests that compared the change in the intervention group with the changein the control group.Results Ten RCTs were included in the review but only two qualied as higher quality RCTs. Results from smaller trials and lower qualityRCTs showed more variation in differences between the intervention and control groups than larger or higher quality trials. Evidence fromlarge, high-quality RCTs with adequate statistical analyses showed that, for improvement in pain and function, a mobilisation/manipulationpackage is an effective intervention [compared with general practitioner (GP) care], whilst manipulation used in isolation showed no realbenets over sham manipulation or an alternative intervention. No higher quality evidence considering massage was identied.Conclusions Many RCTs in the area of manual therapy for LBP have shortcomings in sample size, methodological quality and/or statisticalrigour, but there remains evidence from higher quality RCTs to support the use of a manual therapy package, compared with GP care, fornon-specic LBP of at least 6 weeks duration. 2007 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved.

Keywords: Low back pain; Massage; Manipulation; Manual therapy; Systematic review

Introduction

Manual therapy, dened as massage, mobilisation and/ormanipulation in this review, is commonly used by physio-therapists to treat patients with low back pain (LBP) [1,2] .However, within manual therapy, different approaches to

Corresponding author. Tel.: +44 1903 212116; fax: +44 208 306 6653. E-mail address: [email protected] (A. Jackson).

treatment exist, ranging from massage and mobilisation tomanipulation. Unfortunately, the literature does not alwaysdistinguish between mobilisation and manipulation [3]. Con-sequently, results of randomised controlled trials (RCTs) inthis area are often difcult to interpret, thereby limiting theirvalue in academic and clinical practice. Moreover, RCTs inthe area of back pain often score low on methodological qual-ity scales and include small sample sizes [4,5] , which meansthat their ndings should be interpreted with caution.

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-mailto:[email protected]://-/?-http://-/?-http://-/?-http://-/?-http://dx.doi.org/10.1016/j.physio.2007.10.008http://dx.doi.org/10.1016/j.physio.2007.10.008http://-/?-http://-/?-mailto:[email protected]://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


11/86

98 D.M. Hettinga et al. / Physiotherapy 94 (2008) 97104

These shortcomingsin theevidence formanual therapy forLBP are problematic when developing guidance for clinicalpractice. The European clinical guidelines for the treatmentof chronic LBP do not recommend the use of massage, whilea short course of manipulation/mobilisation is recommended[6]. Some systematic reviews on massage for LBP, however,

whilst acknowledging the low quality of most RCTs, con-clude that massage might be benecial for patients with LBP[710] . In contrast, some reviews on spinal manipulation,including the latest Cochrane review, conclude that there isinsufcient evidence to recommend manipulation [11,12] .This shows that the evidence for manual therapy is still hotlydebated, and a denitive statement on its effectiveness is notyet available.

In order to facilitate this debate, it is desirable to focus onhigh-quality research. Assendelft et al. (1995) showed thatalthough the quality of most reviews on spinal manipulationfor LBP was low, those with the highest quality scores gener-ally supported the use of spinal manipulation [13]. This givessome insight into how methodological quality may impactupon outcome. A similar approach may be possible at RCTlevel, especially since the quality of RCTs in back pain isoften reported to be impaired [4,5] . A good understanding of how the quality, including sample size, methodological qual-ity and statistical rigour, of RCTs may affect the reportedoutcome is essential if research is to inform clinical prac-tice.

Inadequate sample sizes in RCTs can result in unreliableresults. Some systematic reviews, in particular those fromthe Cochrane Collaboration, visualise this by using forestplots [8]. Adequate statistical power (i.e. sample size) is

especially difcult to achieve in studies on LBP dueto itsnat-ural history of resolution. Furthermore, the heterogeneity of the LBP population may hinder the clinician when applyingthe research ndings to clinical practice. Only appropriatelypowered RCTswill detect statistically signicant differences[5]. Moreover, research in other areas has shown that conclu-sions from reviews on multiple smaller RCTs can conictwith the results from one larger RCT [14,15] , which illus-trates the need for large RCTs.

Secondly, low methodological quality can impair thevalidityof theresults fromRCTs. Systematic reviewsattemptto control for this using various quality checklists or scalesto assess the quality of the individual RCTs. Again, this is of particular importance to LBP trials as their methodologicalquality is often impaired [4]. Thirdly, the effectiveness of anintervention for LBP should be tested by analysing differ-ences in changes between two or more intervention groups.Since a large proportion of patients with LBP are reported torecoverspontaneously [16,17] , within-group analysiswillnotdistinguish the natural course of LBP from the intervention-specic effect. Moreover, excluding those with acute LBP(18 years with non-specic LBP of at least6 weeks duration. A description of subacute or chronicpatients was not sufcient; the actual duration of pain wasneeded.

Either massage, mobilisation and/or manipulation wasused as the single intervention for at least one group in thetrial. RCTs were not excluded if these manual therapieswere combined with other non-physiotherapy interven-tions, such as general practitioner (GP) care or medication.

Theeffectiveness of manual therapieswas tested in at leastone of the following areas: pain; function; psychologicalstatus; or return to work/sick leave.

The methodological quality of the included trials wasassessed by tworeviewers using an adapted version of thevanTulder criteria [18]. The original scale consists of 24 criteria,and 10 were identied as the most relevant for this review.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


12/86
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


13/86


Nine of thecriteriaaddressed internalvalidityof theRCT, andone criterion addressed similarities of main baseline charac-teristics/predictors (see Table 1 f or a list of all criteria). Theother criteria from the original 24-item van Tulder list thatwere not used to calculate the score in this review were eitherincluded in the scope of this review (e.g. eligibility criteria,

description of interventions) or qualitatively assessed (e.g.statistical criteria). Sample size was considered separately asthis is not included in the van Tulder criteria. Instead of usingthe total number of subjects in the RCTs, the number of subjects in the manual therapy group was used as an indicationof sample size. This controls for RCTs with a relatively largesamplesize but multiple subgroups. Thesmallest RCTs ( 15subjects in the manual therapy group) that also scored verylow on the methodological quality scale ( 2 out of 10) wereexcluded from this review.

Quantitative analysis

Most of the RCTs monitored pain and function before andafter intervention, and therefore these two outcomes wereanalysed quantitatively. Any outcome measure for pain andfunction was included in this analysis. For every RCT, thechange in pain and function after any follow-up period wascalculated as a percentage of the baseline score of that group[(post-score baseline score)/baseline score 100%]. Animprovement in outcome was expressed as a positive value,while any worsening in symptoms was expressed as a neg-ative value. These percentage changes were plotted so thatthe change in the manual therapy group was displayed on the x-axis, and the control or alternative intervention group was

displayed on the y-axis (Figs. 1 and 2). Some RCTs reportedP -values for this difference in change, while the statistical

Fig. 1. The change in pain as a percentage of the baseline value for theintervention group compared with the same parameter of the comparatorgroup. Black points representresultsfrom larger randomised controlled trials( 40 subjects in manual therapy group) with a quality score of 5 or more.Open points represent results from smaller and/or lower quality trials. Theline y = x represents the line of no difference between the change in themanual therapy group and the control group.

Fig. 2. The change in function as a percentage of the baseline value for theintervention group compared with the same parameter of the comparatorgroup. All points are taken from larger randomised controlled trials ( 40

subjectsin manualtherapy group)with a quality score of 5 or more. Theline y= x represents the line of no difference between the change in the manualtherapy group and the control group.

signicance of this value was unknown in other cases. TheseLAbbe plots give a clear picture of all the data in the RCTsand facilitate identication of possible outliers [19]. Pointstowards the bottom-right corner of the plots indicate that themanual therapy intervention was better, while points towardsthe top-left corner favour the control interventions. Pointsalong the line y = x indicate that the effectiveness of the man-ual therapy intervention was similar to that of the controlintervention.

To assess the effect of sample size and methodologicalquality, the difference in change (percentage change in themanual therapy group minus percentage change in the con-trol/alternative intervention) was plotted against the qualityscore of the trial ( Fig. 3) and the number of subjects in themanual therapy group ( Fig. 4).

Fig.3. Thedifference inchange (percentage change in manual therapygroupminus percentage change in the control group) presented in the randomisedcontrolled trials and the methodological quality score of the correspondingtrial.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


14/86

D.M. Hettinga et al. / Physiotherapy 94 (2008) 97104 101

Fig. 4. The difference in change (percentage change in the manual therapygroup minus the percentage change in the control group) presented in therandomisedcontrolled trials andthe number ofsubjectsin themanualtherapygroup of the corresponding trial.

Qualitative analysis

Since small sample size, low methodological qualityand inadequate statistical testing can distort the results of RCTs, only results from large, high-quality RCTs with ade-quate statistical tests comparing the difference in changebetween intervention groups were considered for the qualita-tive analysis.To distinguish larger from smaller trialsandlowquality fromhigher quality trials, arbitrarycut-off points wereagreed by expert panel consensus. Methodological qualitywas assessed on a 10-point scale and RCTs were consideredto be of lower methodological quality if they scored less than

themedian score on the10-pointquality score.Consequently,RCTs scoring the median score or higher were considered tobe of higher methodological quality.

Although a power analysis should determine adequatesample size, a pragmatic approach was used in this reviewdue to limitations in the available data. An RCT with 40subjects in the manual therapy group was dened as large.This cut-off point was adapted from work by Moore et al.on the effect of chance on variability in patients response topain relief interventions [20].

The third criterion for best available evidence in thisreview was adequate statistical testing to determine the levelof statistical signicance for the difference in change betweenthe manual therapy group and the control/alternative inter-ventiongroup.Given that LBPis reported to improve in manypatients over time [16,17] , comparing pre- and post-scoreswithin one group is not sufcient to test the effectiveness of an intervention. This review only considered the differencein change between the manual therapy group and the com-parator group, and the level of statistical signicance for thatdifference in change.

Although all RCTs that fullled the three inclusion crite-ria were included in the quantitative analysis, except the verysmall RCTs that also scored low on the quality scale, onlyRCTs that were large, high-quality and employed adequate

statistical testswere used to recommendspecic manual ther-apy interventions for chronic non-specic LBP.

Results

Ten RCTs fullled all inclusion criteria for this review(Table 1); eight of these compared manual therapy witha control or alternative intervention [2128] , and twocompared various forms of manual therapy [29,30] . Themethodological quality of the RCTs ranged from 1/10 to7/10 and the median score was 5/10. The sample size of the RCTs ranged widely from 24 to 1334, and the numberof subjects in the manual therapy group ranged from nineto 353. Five of the 10 RCTs were classied as larger ( 40subjects in manual therapy group) [21,23,2527] . Only threeRCTs were both large and of higher methodological quality[21,26,27] . However, only two of these three trials reportedsufcient statistical analyses to meet the inclusion criteria

for the qualitative analysis [26,27] .

Quantitative analysis

For the quantitative analysis, eight RCTs were consideredsince two trials compared different manual therapy inter-ventions with one another. Of these eight RCTs, six trialsreported pain values that could be used to calculate a percent-age change score [21,23,24,2628] , and three trials reportedfunction scores that could be used to calculate a percentagechange score [21,26,27] . Fig. 1 gives the percentage changein pain for the manual therapy group and the correspondingcontrol group; 16 of the 27 points came from larger, higherquality RCTs, and two of these points were statistically sig-nicant. Fig. 2 gives the same information for function; allthese points came from larger higher quality trials and threepoints were statistically signicant at the 5% level.

Based on the percentage change in the manual therapygroup and the control/alternative intervention group (shownin Figs. 1 and 2), the difference in change between manualtherapy and control/alternative intervention was calculated.This difference in change was plotted against the method-ological quality score of the RCT ( Fig. 3) and the numberof subjects in the manual therapy group ( Fig. 4). From thelimited number of observations shown in Figs. 3 and 4, it can

be seen that the smaller RCTs and those of lower quality tendto show more variation in the difference in change than thelarger and higher quality RCTs.

Qualitative analysis

Based on the criteria of sample size, methodological qual-ityandstatistical rigour, there is evidencefrom higher qualityRCTs that states:

manipulation in combination with mobilisation is moreeffective than GP care for pain reliefandfunction improve-ment [27];
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


15/86


manipulation in combination with mobilisation is as effec-tive as GP care for reduction in fear-avoidance behaviour[27]; and

manipulation is as effective as sham manipulation or aneducation programme for pain relief and improvement infunction and psychological status (relief of depression)

[26].

Discussion

The aim of this review was to assess the extent to whichmethodological quality, sample size and statistical rigouraffected the conclusions from RCTs on mobilisation, manip-ulation and/or massage for persistent LBP. The authorsproposed that only evidence from RCTs that scored high onthese three criteria should be used to guide clinical practice.The results showed that the smaller and lower quality stud-ies displayed more variation in outcome, and therefore theresults of these studies should be interpreted with caution.It should be noted that this conclusion is based on a limitednumber of observations (see Table 1 ), although the ndingson impaired sample size and methodological quality in LBPtrials have been reported previously [4,5] . Moreover, simi-lar analyses on exercise interventions for LBP suggest thatsample size and methodological quality can indeed result inmisinterpretation of RCTs [31].

Evidence from RCTs that fullled the study criteria onmethodologicalquality, sample sizeandstatistical rigour sug-gested that mobilisation, in combination with manipulation,was a useful and effective additional therapy to standard GP

care forpainandfunction [27]. Manipulation usedin isolationappeared to be less effective,sincetherewashigh-quality evi-dence that such an intervention was of equal benet to shammanipulation or an education programme [26]. No clear evi-dence statement could be made for massage interventions asno large, high-quality RCT with adequate statistical analysiswas identied.

The justication for only considering RCTs that fullledall three criteria came from the quantitative analyses. Fig. 3shows that there is a tendency for more variance in differencein change in the lower quality RCTs. A similar tendency wasevident for the smaller RCTs ( Fig. 4). Moreover, the smallerRCTs tended to result more often in positive results for themanual therapy group; in Fig. 4, more points are locatedabove the line y = 0. This may have been caused by publica-tion bias as smaller trials with negative results are less likelyto be published. Differences in subject characteristics couldalso have played a role since larger trials have to recruit froma larger pool of subjects, which may be more generalisablethan more selective recruitment from smaller trials.

It is widely reported that LBP improves over time formany [16,17] , and this is evident from Figs. 1 and 2. Ittherefore seems reasonableto expect additional improvementfrom effective treatments, and this requires comparison witha second group that either receives a control or alternative

intervention. Analysing change scores within one group orcomparing post-scores of two groups seems inadequate todetect the true effectiveness of an intervention.

Most other systematic reviews on manual therapy for LBPreport effect sizes when comparing and summarising RCTs.However, this was not done in the present review because:

(a) a measure of variation and the mean are needed to calcu-late effect size and it has been reported previously that thisis not always reported in RCTs on manual therapy for LBP[12]; (b) an effect size may be more difcult for cliniciansto interpret, especially towards their patients, whereas a per-centage is a more intuitive parameter; and (c) pooling effectsizes in systematic reviews can be useful but does not reectthe diversity in the treatments studied, especially in manualtherapy where the same terms have been used to describedifferent interventions [3].

The use of percentages in the current reviewalsohas somelimitations. It does not take into consideration any variancepresent in the subject groups, nor does it indicate a clini-cally and statistically important difference. Nevertheless, itis of interest to analyse original data in various ways, and themethodology used here mayreveal new insightsor strengthenexisting views on trials in the area of manual therapy forLBP.

In addition to the statistical limitations, this review alsoraises somemethodological issues. Firstly, the strict inclusioncriteria used to identify relevant RCTs facilitated compar-ison, and also resulted in the inclusion of only 10 RCTs.Secondly, systematic reviews are only as good as the compo-nent studies andthis was reected in thequantitative analysis.Thirdly, the validity and reliability of the outcome measures

for pain and function were not formally assessed, althoughmost RCTs used standard outcome measures such as visualanalogue scales for pain and the Roland-Morris DisabilityQuestionnaire for function. Furthermore, the study denitionof higher quality trials, i.e. 40 subjects in the manual ther-apy group, 5 out of 10 on the adapted van Tulder scale, andstatistical analyses testing the difference in change betweeninterventions, is based on consensus from the authors andthe wider Guideline Development Group involved in thisproject (see Acknowledgements). Ideally, a power calcula-tionshould reveal the appropriate sample size,butconductinga power analysis for all RCTs retrospectively wasbeyond thecapacity of this study. An arbitrary cut-off point was there-fore agreed, although some evidence has been identied tosupport this cut-off point [20], and this value has been usedin a similar review on exercise [31]. Whilst the use of thecut-off point in methodological quality score (i.e. the medianvalue) was arbitrary, it does distinguish the higher from thelower quality studies. The use of a total score implies thatshortcomings in certain methodological aspects can be com-pensated for by fullling other criteria. However, this is arestriction of all methodologicalquality scores. Nevertheless,such scores, due to their simplicity, are now commonly usedin reviews. All scores from the RCTs included in this revieware displayed in Table 1 . Finally, grouping changes in pain or
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


16/86

D.M. Hettinga et al. / Physiotherapy 94 (2008) 97104 103

function at any follow-up time was justied by the fact thatno correlation was found between follow-up time and dif-ference in change (not displayed in gures). This does not,however, suggest that there were no differences in effective-ness between follow-up pointswithin trials; it merely justiespooling all follow-up points in the gures presented in this

review. Similar results were found in a review on exercise forLBP [31].To ensure that future RCTs can be used to inform clinical

practice, the following points should be kept in mind in thedesign and analysis of RCTs:

descriptionof the interventions andprecise useof terms forthe various elements of the intervention should be explicitand consistent;

an adequate sample size must be determined in advance toensure detection of statistically signicant and clinicallyrelevant effects, and a power calculation is essential;

highest methodological quality must be aimed for in thedesign and reporting of the trial;

relevant and valid outcome measures should be used, forexample the suggestions made by the Cochrane Collabo-ration to use at least outcome measures for pain, function,global measure of improvement and return to work [18];

the change achieved with the primary intervention shouldbe compared with the change achieved with a control oralternative intervention, since such an analysis can revealthe true value of an intervention; and

identication of the characteristicsof subgroups for whomspecic treatments, such as manual therapy, may be mosteffective.

In conclusion, this systematic review shows that limita-tions in sample size, methodological quality and statisticalrigour could have impaired the validity of the results fromRCTs in the eld of manual therapy for LBP of at least 6weeks duration. Nevertheless, the ndings from the mostrobust RCTs supported the use of a comprehensive manualtherapy package consisting of mobilisation and manipula-tion over GP care. At this point in time, manipulation usedin isolation cannot be recommended. Likewise, no recom-mendations canbe made for massage interventions due to thelack of RCTs fulllingall criteria for best available evidence.There isa need forfurther high-qualityresearchinto theuse of manual therapy in the management of persistent LBP. More-over, it would be benecial to identify the characteristics of subgroups in non-specic LBP whowould benet most frommanual therapy. For manipulation, clinical prediction ruleshave been suggested [32]. However, a reliable set of criteriafor subclassication of LBP has yet to be established; move-ment pattern, sensory, neuromuscular and/or psychologicalparameters have all been suggested [33]. There is a needfor more thorough research into the effectiveness of variousphysiotherapy interventions to ensure that physiotherapistscan continue to play a key role in evidence-based medicineand practice.

Acknowledgements

The authors would like to thank the people who assistedwith conducting this review (Dr. Katherine Deane, Jo Jor-dan, Jenni Hall and Sharlene Ting), the library staff at theCSP (Samantha Molloy, Linda Grifths,Andrea Peace, Anna

Sewarniak and Alison Jinks), other staff at the CSP (AlexWarne, Helen Whittaker and Susan Williams), and all mem-bers of the Guideline Development Group (in addition tothe authors, Dr. Panos Barlos, Sarah Ferguson, Susan Green-halgh, Dr. Vicki Harding, Prof. Jennifer Klaber-Moffett, Dr.Denis Martin, Jude Monteath, Nia Taylor and Dr. SteveWoby).

Funding: CSP and the CSPs Charitable Trust, UK.

Conict of interest: None.

References

[1] Foster NE, Thompson KA, Baxter GD, Allen JM. Management of nonspecic low back pain by physiotherapists in Britain and Ire-land. A descriptive questionnaire of current clinical practice. Spine1999;24:133242.

[2] Gracey JH, McDonough SM, Baxter GD. Physiotherapy managementof low back pain:a survey of current practice inNorthernIreland. Spine2002;27:40611.

[3] Kotoulas M. The use and misuse of the terms manipulation andmobilization in the literature establishing their efcacy in thetreatment of lumbar spine disorders. Physiother Can 2002;54:5361.

[4] Koes BW, Malmivaara A, van Tulder MW. Trends in methodologicalquality of randomised clinical trials in low back pain. Best Pract Res

Clin Rheumatol 2005;19:52939.[5] Bouter LM, van Tulder MW, Koes BW. Methodologic issues inlow back pain research in primary care. Spine 1998;23:201420.

[6] Airaksinen O, Brox JI, Cedraschi C, Hildebrandt J, Klaber-MoffettJ, Kovacs F, et al. Chapter 4. European guidelines for the manage-ment of chronic nonspecic low back pain. EurSpine J 2006;15(Suppl.2):s192300.

[7] Cherkin DC, Sherman KJ, Deyo RA, Shekelle PG. A review of theevidence for the effectiveness, safety, and cost of acupuncture, mas-sage therapy, and spinal manipulation for back pain. Ann Intern Med2003;138:898906.

[8] Furlan AD, Brosseau L, Imamura M, Irvin E. Massage for low back pain. Cochrane Database Syst Rev 2002;2:CD001929.

[9] Ernst E. Massage therapy for low back pain: a systematic review. J Pain

Symptom Manage 1999;17:659.[10] Ernst E. The use, efcacy and costs of complimentary/alternative ther-apies for low back pain. Eur J Phys Med Rehabil 1998;8:537.

[11] Koes BW, Assendelft WJ, van der Heijden GJ, Bouter LM. Spinalmanipulation for low back pain. An updated systematic review of ran-domized clinical trials. Spine 1996;21:286071.

[12] Assendelft WJ, Morton SC, Yu EI, Suttorp MJ, Shekelle PG. Spinalmanipulative therapy for low back pain. Cochrane Database Syst Rev2004;1:CD000447.

[13] AssendelftWJ, Koes BW, Knipschild PG, Bouter LM. The relationshipbetween methodological quality and conclusions in reviews of spinalmanipulation. JAMA 1995;274:19428.

[14] Kjaergard LL, Villumsen J, Gluud C. Reported methodologic qualityand discrepancies between large and small randomized trials in meta-analyses. Ann Intern Med 2001;135:9829.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


17/86


[15] LeLorier J, Gregoire G, Benhaddad A, Lapierre J, Derderian F. Dis-crepancies between meta-analyses and subsequent large randomized,controlled trials. N Engl J Med 1997;337:53642.

[16] Waddell G. Volvo award in clinical sciences. A new clinical model forthe treatment of low-back pain. Spine 1987;12:63244.

[17] Spitzer WO, LeBalnc FE, Dupuis M. Scientic approach to the assess-mentand managementof activity-relatedspinal disorders: a monographfor clinicians: report of the Quebec Task Force on Spinal Disorders.Spine 1987;12:S154.

[18] van Tulder MW, Assendelft WJ, Koes BW, Bouter LM. Methodguidelines for systematic reviews in the Cochrane Collaboration Back Review Group for Spinal Disorders. Spine 1997;22:232330.

[19] Song F. Exploring heterogeneity in meta-analysis: is the LAbbe plotuseful? J Clin Epidemiol 1999;52:72530.

[20] Moore RA, Gavaghan D, Tramer MR, Collins SL, McQuay HJ. Size iseverything large amounts of information areneeded to overcomeran-dom effects in estimating direction and magnitude of treatment effects.Pain 1998;78:20916.

[21] Cherkin DC, Eisenberg D, Sherman KJ, Barlow W, Kaptchuk TJ,Street J, et al. Randomized trial comparing traditional Chinese medicalacupuncture, therapeutic massage, and self-care education for chroniclow back pain. Arch Intern Med 2001;161:10818.

[22] Evans DP, Burke MS, Lloyd KN, Roberts EE, Roberts GM. Lumbarspinal manipulation on trial. Part I clinical assessment. RheumatolRehabil 1978;17:4653.

[23] Gibson T, Grahame R, Harkness J, Woo P, Blagrave P, Hills R. Con-trolled comparison of short-wave diathermytreatmentwith osteopathictreatment in non-specic low back pain. Lancet 1985;1:125861.

[24] Hernandez-Reif M, Field T, Krasnegor J, Theakston H. Lower back pain is reduced and range of motion increased after massage therapy.Int J Neurosci 2001;106:13145.

[25] Postacchini F, Facchini M, Palieri P. Efcacy of various forms of conservative treatment in low back pain. A comparative study. Neuro-Orthopedics 1988;6:2835.

[26] Triano JJ, McGregor M, Hondras MA, Brennan PC. Manipulativetherapy versus education programs in chronic low back pain. Spine1995;20:94855.

[27] UK BEAM trial team. United Kingdom back pain exercise andmanipulation (UK BEAM) randomised trial: effectiveness of physi-cal treatments for back pain in primary care. BMJ 2004;329:137781.

[28] Waagen GN, Haldeman S, Cook G, Lopez D, DeBoer KF. Short termtrial of chiropractic adjustments for the relief of chronic low back pain.Man Med 1986;2:637.

[29] Descarreaux M, Blouin JS, Drolet M, Papadimitriou S, Teasdale N.Efcacy of preventive spinal manipulation for chronic low-back painand related disabilities: a preliminary study. J Manip Physiol Ther2004;27:50914.

[30] Haas M, Groupp E, Kraemer DF. Dose-response for chiropractic careof chronic low back pain. Spine J 2004;4:57483.

[31] Hettinga DM, Jackson A, Klaber-Moffett J, May S, Mercer C, WobyS. A systematic review and synthesis of higher quality evidenceof the effectiveness of exercise interventions for non-specic lowback pain of at least six weeks duration. Phys Ther Rev 2007;12(3):22132.

[32] Childs JD, Fritz JM, Flynn TW, Irrgang JJ, Johnson KK, MajkowskiGR, et al. A clinical prediction rule to identify patients with low back painmost likely to benet fromspinal manipulation: a validation study.Ann Intern Med 2004;141:9208.

[33] Jull G. Classication systems in clinical practice. Phys Res Int2004;9:iiiiv.



18/86


Narrative review

Constraint-induced movement therapy: a narrative reviewAnna Tuke

School of Psychology, University of Nottingham, Nottingham NG7 2RD, UK

Abstract

Objectives To identify factors relevant to implementing constraint-induced movement therapy (CIMT) within the clinical setting.Data sources AMED (1995January 2007), CINHAL (1982January 2007), Medline (1996January 2007) and EMBASE (1996January2007) were searched to identify relevant studies.Review methods Criteria for inclusion of trials in this study were that trial participants were over 18 years of age, had had a stroke, andCIMT or modied CIMT was compared with either no intervention, modied CIMT or alternative treatment. Modied CIMT had to includeboth constraint and training components. The study design was either a quasi-randomised controlled trial or a randomised controlled trial.Trials had to be published in English and score 4 or more on the PEDro scale.Results Twelve eligible studies were identied. The quality of the studies varied, although there was evidence that this improved with morerecent studies achieving higher PEDro scores. Patient selection criteria and the components related to the delivery of CIMT were identied asrelevant factors. Thirteen different patient selection criteria were identied: age; length of time post stroke; specied side of hemiplegia; handdominance; spasticity; pain; balance and mobility; hand function; range of active and passive movement; cognitive impairment; perception;sensation; and communication. Ten components were identied as being relevant to the actual delivery of CIMT: type of CIMT; type of constraint; constraint wear time; excluded activities; shaping; shaping dosage; group versus individual treatment; environment; potentialharms of CIMT; and compliance. A third relevant consideration was the selection of outcome measures. Signicant variability was identiedin many aspects of CIMT, although there was evidence of greater standardisation in more recent studies.Conclusion The development of CIMT for stroke patients has provided clinicians with a treatment technique for a dened patient groupthat is now supported by a considerable evidence base. CIMT is a complex intervention and the optimum intensity and length of treatment

remains unknown. Transferring CIMT into the clinical environment has been hampered by the lack of standardisation in many aspects of theintervention. However, there is evidence that this is improving. Implementation and evaluation in the clinical environment would strengthenthe evidence base. 2007 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved.

Keywords: Constraint-induced movement therapy; Stroke; Clinical application

Introduction

Constraint-induced movement therapy (CIMT), a newtreatment for the upper limb following stroke, has emergedfrom the concept of learnt misuse and has been cited as oneof the few neurorehabilitation techniques developed from ascientic and theoretical foundation [1]. Early investigationson the effect of somatosensory deafferentation in monkeyssuggested that during a period of diaschisis, repeated unsuc-

Present address: Inpatient Physiotherapy Department, Babington Hospi-tal(DerbyshireCountyPrimaryCare Trust),Belper, DerbyRoad,DerbyshireDE56 1WX, UK. Fax: +44 1773 525086.

E-mail address: anna [email protected] .

cessful attempts to use the paretic upper limb resulted inthe monkey learning not to use that limb. This phenomenon,termed learned non-use, continued after the diaschisis hadpassed and the limb was potentially functional. However, theapplication of training techniques and/or constraint increasedthe spontaneous use of the paretic limb. It was later hypothe-sised that learned non-use could occur in humansafter stroke,and that the application of appropriate training regimes couldreduce learned non-use and increase spontaneous use of theparetic limb [2]. The technique consists of two components:constraint of the non-paretic upper limb; and shaping incre-mental repetitive task practise [3]. The combination of theseis thought to cause extinction of the learned non-use, result-ing in signicant improvementsin speedof task performance,

http://-/?-http://-/?-http://-/?-mailto:[email protected]://-/?-http://-/?-http://-/?-http://-/?-http://dx.doi.org/10.1016/j.physio.2007.07.007http://dx.doi.org/10.1016/j.physio.2007.07.007http://-/?-http://-/?-mailto:[email protected]://-/?-http://-/?-


19/86

106 A. Tuke / Physiotherapy 94 (2008) 105114

Table 1Criteria for inclusion of trials in this study

Trial participants were over 18 years of age and had had a stroke CIMT or modied CIMT was compared with either no intervention, mod-

ied CIMT or alternative treatment Modied CIMT had to comprise both constraint and training components Randomised controlled trial or a quasi-randomised controlled trial

Published in the English language in a journalCIMT, constraint-induced movement therapy.

functional ability and increased use of the paretic upper limb[35] .

A systematic review of randomised controlled trials of CIMT in stroke concluded that CIMT may improve upperlimb function following stroke for some patients when com-pared to alternative or no treatment [6]. This conclusionwas subsequently supported following a meta-analysis of the effectiveness of CIMT [7]. The completion of a largerandomised controlled trial conrmed that CIMT producesstatistically signicant changes in upperlimbfunction,whichwere maintained at 1 year [8]. The National Clinical Guide-lines for Stroke recommended the use of CIMT in patientswith at least 10 degrees of active wrist and nger extension,who are more than 1 year post stroke and who can walk inde-pendently without an aid [9]. There is some evidence thattherapists are sceptical about the use of CIMT within a clin-ical environment. Page et al. (2002) investigated therapistopinions about CIMT through a self-reported questionnaire,and found that the principal concerns were patient adherenceand safety [10]. At present, there is no reference to the practi-calapplicationof CIMT withinanyof thepublished literature.Therefore, the aim of this paper is to identify and review the

issues relevant to clinicians considering the development of a CIMT programme.

Method

A computerised search of relevant bibliographic databaseswas undertaken to identify all randomised control trials. Theinclusion criteria are listed in Table 1 and the search historyis given in Table 2 . The reference lists of all included trialswere searched manually.

The methodological quality of the studies was assessed

using the PEDro scale [11] . As blinding of therapists andsubjects is not possible with CIMT, the maximum score was8/10. Trials scoring 3 or less were not included.

Table 2Search history

Database Years Search terms No. of hits No. selected

AMED 1985January 2007 Constraint 159 8CINHAL 1982January 2007 Constraint 328 3Medline 1996January 2007 Constraint 2108 0Medline 1996January 2007 CIMT 108 5EMBASE 1996January 2007 Constraint 106 0

CIMT, constraint-induced movement therapy.

Results

In total, 16 studies were identied. Twelve studies scored4 or more on the PEDro scale ( Table 3 ). Eight of the studieswere small scale,with less than 20participants [3,1218] , andthree studies recruited 66, 23 and 69 participants [4,19,20] .

The largest trial recruited 222 stroke patients [8].

Selection criteria

In total, 24 selection criteria were identied ( Table 4 );these have been condensedinto 13 categories relevant to clin-icians. The number included for each study ranged from 2[18] to 15 [8,15] . The most frequently used criteria were therequirement to have more than 20 degrees of active extensionat thewrist anda denedlevel of cognitive impairment. Handdominance and its relationship to the hemiplegic side was theleast used criterion.

Length of time post stroke

CIMT has been applied to chronic, subacute andacute stroke patients. The studies recruited participants atthe following times post stroke: more than 12 months[3,4,13,14,20] ; between 3 and 9 months [8,16] ; more than6 months [17]; between 1 and 6 months [10]; less than 4months [15]; and less than 14 days [18,19] .

Age

Six out of the 12 studies did not specify an upper age limit[8,13,1619] . In theremainingsixstudies, theupper agelimitranged from 75 to 95 years ( Table 4 ). One study reported thatage did not inuence the results and therefore should not beused to exclude participants [8].

Specied side of hemiplegia and hand dominance

Two studies excluded participants on the basis of handdominance and its relationship to side of stroke [3,4] . Theformer reported that this was for ease of test administrationandthe latter didnot specify a reason. Another study reportedthat there was a small, but not clinically relevant, differencein treatment gains between participants with paresis of theirpre-stroke dominant side compared with those with paresisof their pre-stroke non-dominant side [8].

Spasticity

Of the ve studies that reported excessive spasticity as anexclusion criterion, three reported using the Modied Ash-worth Scale [10,14,15] and one reported using the MotorAssessment Scale [19]. One study listed excessive spastic-ity as an exclusion criterion but did not report how it was
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


20/86

Table 3PEDro scores

Author Randomallocation

Concealedallocation

Baselinecomparability

Blindsubjects

Blindtherapist

Blindassessors

One keyoutcome from85% of subjects

Intention-to-treatanalysis

Betcom

Taub 1993 [3] Y N Y N N Y Y N YVan der Lee 1999 [4] N Y N N N Y Y Y Y

Dromerick 2000 [19] Y N Y N N Y Y N YPage 2002 [12] Y N N N N Y Y N YWittenberg 2003 [13] Y N Y N N Y Y N YPage 2004 [14] Y Y Y N N Y Y N YPage 2005 [15] Y N Y N N Y Y Y YAlberts 2004 [16] Y N Y N N N Y Y YSuputtitada 2004 [20] Y N Y N N Y Y N YBrogardh 2006 [17] Y Y N N N Y N Y YRo 2006 [18] Y N Y N N Y Y Y YWolf 2006 [8] Y Y Y N N Y Y Y Y

Table 4Inclusion criteria

Taub 1993[3]

Van der Lee 1999[4]

Dromerick 2000[19]

Page 2002[12]

Wittenberg 2003[13]

Page 2004[14]

Alberts 2004[16]

Suputtitada 2004[20]

P[

Time post stroke >12 months >12 months 12 months >12 months 39 months >12 months 20 >20 >20 >20 >20 >20 >Extension at the ngers (degrees) >10 >10 >10 Extension at the metacarpophalangeal

joints (degrees)>10 >10 >10 >10 >

Extension at the interphalangeal joints(degrees)

>10 >10 >10 >10 >

Passive range of movement + Movement at the thumb (degrees) >10 Minimum/maximum hand function + + + + + + +Balance + + + + Mobility + + + + Cognition + + + + + + + +Perception + + + Sensation + + + Communication + + + + Spasticity + + + +Pain + + + + , not listed as a criterion; +, listed as a criterion.
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


21/86

108 A. Tuke / Physiotherapy 94 (2008) 105114

assessed [3]. None of these studies documented which jointswere assessed.

Pain

Five studies reported using excessive pain as an exclusion

criterion [8,10,1416] . All of these studies reported using avisual analogue scale. Two studies did not report a maximumscore, but excluded patients if they had excessive pain inany joint of the paretic extremity [8,16] . Page et al., (2002),(2004) and (2005) excluded participants scoring 4 or moreon a 10 point visual analog pain scale.

Balance and mobility

In the six studies that stipulated balance and mobility ascriteria, participants were at the following times post stroke:morethan12months [3,4,20] ; between3and9months [8,16] ;and more than 6 months [17]. None of these studies reportedusing a standardisedmeasure of balance and/or mobility. Theinclusion criteria for the six studies were: able to mobilisewithout an assistive device at all times [3,20] ; able to walk inside without a stick [4]; independent toilet transfers, inde-pendent 2-minute standandindependent sitstand [8,16] ; andable to mobilise 20 metres in 40 seconds [17].

Out of the six studies that did not stipulate balance andmobility as criteria, participants were at the following timespost stroke: more than 12 months [13,14] ; between 1 and 6months [15]; less than 4 months [10]; and less than 14 days[18,19] .

Hand function

Only one study did not specify hand function as an inclu-sion criterion [3]. Of the 11 studies that did specify handfunction as an inclusion criterion, six used the Motor Activ-ity Log (MAL), excluding participants if they scored morethan 2.5 [8,10,1416] or 2.7 [13]. Two studies excluded par-ticipants if they scored more than 51 on the Action ResearchArm (ARA) Test (maximum score 57) [4,20] . Two stud-ies reported using the National Institute of Health StrokeScale (NIHSS), although the exclusion scores were differ-ent [18,19] . Another study recruited participants scoring lessthan 65 on the Sollermans Hand Function Test (maximumscore 80) [17].

Range of active and passive movement

Most studies specied 20 degrees of active wrist extensionand 10 degrees of active nger extension ( Table 4 ). However,the operational details of testing these movements were onlyreported by two studies, both of which specied the rangeof movement and the number of repetitions required at each joint [8,16] . One of these studies included lower functioningparticipants who had only 10 degrees of wrist extension, 10degrees of thumb extension/abduction and at least 10 degrees

of extension in two additional digits [8]. There is evidencethat these lower function patients also beneted from CIMT[8].

Twostudies specied thepassiverange of movement at theshoulder (90 degrees of abduction and exion, 45 degrees of external rotation), elbow (no less than 30 degrees of elbow

extension, 45 degrees of forearm supination and pronationfrom neutral), wrist (extension to neutral) andmetacarpopha-langeal joint (no contracture greater than 30 degrees) [8,16] .

Cognitive impairment

A minimum level of cognitive function was reported as anexclusion criterion in all papers except two [13,18] . TheMini-Mental State Examination (MMSE) was reported in fourstudies [3,4,8,16] . The minimum scores used were 26, 22, 24and 24 respectively out of a total score of 30. One study usedthe NIHSSto assess cognitive function, requiringparticipantsto score 0 or 1 on the consciousness, communication andneglect items [19]. Three studies stipulated that participantshad to score more than70 on the modied MMSE [12,14,15] .One study stated cognitive impairment as an exclusion crite-rion but did not report how it was measured [20].

Perception

Three studies assessed perception [3,4,19] using the fol-lowing range of tests: test of visual neglect [3]; lettercancellation test [3,4]; line bisection test [4]; and the NIHSS[19]. The effect of CIMT on perceptual impairments is cur-rently unknown.

Sensation

One study reported impaired sensation as an exclusioncriterion [20]; however, its measurement was not dened.Two studies rated sensation using the Abbreviated Sensory-Perceptual Examination [3] and a dichotomous scale [4],although neither specied impaired sensation as an exclu-sion criterion. The latter reported that patients with sensorydisorders receiving CIMT exceeded the mean improvementin the ARA Test compared with patients with sensory disor-ders who received bimanual training [4]. This exceeded the

minimal clinically important difference and was maintainedat 1-year follow-up.

Communication

Six studies reported impaired communication as an inclu-sion criterion. Of these, one study did not mention themeasureused [20]. Another study reported that patientswith alevelof aphasia that prevented the completion of the outcomemeasures were excluded [18]. The remaining four studiesused the following outcome measures: Token test [3,17] ;Stichting Afasie Netherlands [3]; and the NIHSS [19].
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


22/86

A. Tuke / Physiotherapy 94 (2008) 105114 109

Outcome measures

Table 5 illustrates the outcome measures used to assesschange in upper limb function following CIMT. In total, 12differentmeasureswere identied, andthe MAL, WolfMotorFunction Test (WMFT), Fugl Meyer Test and ARA Test were

used most commonly. A number of different versions of theMAL exist and a clear description of the scoring system isnot reported.

Intervention

Ten relevant factors were identied in the delivery of CIMT: type of CIMT; type of constraint; length of timethe constraint was worn (wear time); excluded activities;shaping; shaping dosage; group versus individual treatment;environment; potential harms of CIMT; and compliance.

Type of CIMT

CIMT can be classied into the traditional format and amodied format. Seven studies reported using the traditionalformat [3,4,8,13,16,17,20] . Studies investigating the tradi-tional format recruited participants who were 3 months [16],6 months [17], between 3 and 9 months [8], or more than 12months post stroke [3,4,13,20]

Five studies reported using a modied format[10,14,15,18,19] , although a number of different varia-tions exist. The modied format was applied to participantswho were less than 14 days [18,19] , between 1 and 4 months

[10], more than 12 months [14], or less than 4 months [15]post stroke. None of the 12 studies reviewed demonstratedthat any one of these formats was superior.

Type of constraint

The type of constraint varied between studies ( Table 6 ).Three studies reported using a resting splint and sling[3,4,13] . Three other studies reported using mesh,polystyrene-lled mitts and a sling [12,14,15] . The remain-ing studies reported using either a mitt [8,1619] or a glove[20]. There does not appear to be a relationship between thetype of constraint, the type of CIMT or the length of timepost stroke. None of the 12 studies reviewed compared onetype of constraint with another.

Wear time

In the studies investigating traditional CIMT, four stud-ies requested participants to wear the constraint for 90%of their waking hours [3,8,16,17] , and two studies did notreport the length of time that participants were requestedto wear the constraint [4,20] . One study reported a weartime of 6 hours during week days and 4 hours at weekends[13]. Extending the mitt-wearing regime following a tradi-

tional CIMT programme didnot enhance the treatment effect[17].

Thestudies investigatingmodied CIMT reported that themitt should be worn for more than 6 hours per day [19], orfor 5 hours on week days that were identied as a periodof frequent hand use [10,14,15] . One study requested that

participants should aim to wear the mitt for 90% of wakinghours [18].

Activities excluded

Four studies reported activities for which the constraintshould not be worn ( Table 6 ). These ranged from activi-ties where balance or safety would have been compromised[3,18] to travelling, sleeping,dressing [4] and toileting [3,17] .The other studies did not report any activities for which theconstraint should not be worn.

Compliance

Six studies assessed compliance with the constraint-wearing regime ( Table 6). Five studies reported issuingparticipants with a log book [4,10,14,15,17] , and two studiesreported using informal patient/therapist interviews [12,15] .One study included a behavioural contract, a caregivers con-tract, mitt compliance device and daily schedule to increaselevels of compliance [8]. Another study provided oral andwritten instructions to carers for those discharged prior to theendof treatment [18]. Four studies provided comments aboutthe level of compliance [3,10,15,17] ; these were predomi-nantly positive. One study reported that the mitt compliancedevice consistedof a physical sensorand timer [8]. Onthefewoccasions that the self-reports did not match the output fromthe device, participantswere info

Documents

Physiotherapy - Vol 94 - N2 2008