Do various baseline characteristics of transversus abdominis and lumbar multifidus predict clinical...

PAIN�

xxx (2013) xxx–xxx

w w w . e l s e v i e r . c o m / l o c a t e / p a i n

Comprehensive review

Do various baseline characteristics of transversus abdominis and lumbarmultifidus predict clinical outcomes in nonspecific low back pain? Asystematic review

0304-3959/$36.00 � 2013 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.pain.2013.07.010

⇑ Corresponding author. Address: Department of Physical Therapy, University ofAlberta, 3-48 Corbett Hall, Edmonton, Alberta T6G 2G4, Canada. Tel.: +1 780 4926891; fax: +1 780 492 4429.

E-mail address: [email protected] (G.N. Kawchuk).

Please cite this article in press as: Wong AYL et al. Do various baseline characteristics of transversus abdominis and lumbar multifidus predict clinicomes in nonspecific low back pain? A systematic review. PAIN

�(2013), http://dx.doi.org/10.1016/j.pain.2013.07.010

Arnold Y.L. Wong a, Eric C. Parent a,b, Martha Funabashi a, Tasha R. Stanton c, Gregory N. Kawchuk a,⇑a Department of Physical Therapy, University of Alberta, Edmonton, Alberta, Canadab Glenrose Rehabilitation Hospital, Edmonton, Alberta, Canadac Sansom Institute for Health Research, School of Health Sciences, University of South Australia, Adelaide, South Australia, Australia

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

a r t i c l e i n f o a b s t r a c t

Article history:Received 26 March 2013Received in revised form 5 July 2013Accepted 10 July 2013Available online xxxx

Keywords:Effect modifierLow back painLumbar multifidusPrognostic factorSystematic reviewTransversus abdominisTreatment

Although individual reports suggest that baseline morphometry or activity of transversus abdominis orlumbar multifidus predict clinical outcome of low back pain (LBP), a related systematic review is unavail-able. Therefore, this review summarized evidence regarding the predictive value of these muscular char-acteristics. Candidate publications were identified from 6 electronic medical databases. After review, 5cohort studies were included. Although this review intended to encompass studies using different muscleassessment methods, all included studies coincidentally used ultrasound imaging. No research investi-gated the relation between static morphometry and clinical outcomes. Evidence synthesis showed lim-ited evidence supporting poor baseline transversus abdominis contraction thickness ratio as atreatment effect modifier favoring motor control exercise. Limited evidence supported that high baselinetransversus abdominis lateral slide was associated with higher pain intensity after various exercise inter-ventions at 1-year follow-up. However, there was limited evidence for the absence of relation betweenthe contraction thickness ratio of transversus abdominis or anticipatory onset of lateral abdominal mus-cles at baseline and the short- or long-term LBP intensity after exercise interventions. There was conflict-ing evidence for a relation between baseline percent thickness change of lumbar multifidus duringcontraction and the clinical outcomes of patients after various conservative treatments. Given study het-erogeneity, the small number of included studies and the inability of conventional greyscale B-modeultrasound imaging to measure muscle activity, our findings should be interpreted with caution. Furtherlarge-scale prospective studies that use appropriate technology (ie, electromyography to assess muscleactivity) should be conducted to investigate the predictive value of morphometry or activity of thesemuscles with respect to LBP-related outcomes measures.

� 2013 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction

Approximately 9.2% of the global population is affected by lowback pain (LBP), while LBP-related disability is the leading causeof disability in the world [91]. Although certain cases of LBP areascribed to specific pathology, 90% of those with LBP experienceLBP of unknown origin or pathology, known as nonspecific LBP [24].

Notwithstanding the lack of consensus regarding the causal rela-tion between deficits in spinal muscles and the onset of LBP, thetransversus abdominis (TrA) and lumbar multifidus (LM) play

important roles in intersegmental spinal control [32,46,53,73,75,100] and may affect the progression and recurrence of LBP. Ana-tomically, TrA connects to the lumbar vertebrae through the thora-columbar fascia, forming a corset-like structure encircling the trunk,which controls intra-abdominal pressure and vertebral stiffness[4,31,34,43]. In contrast, the LM muscles are deep paraspinal mus-cles with densely packed short muscle fibers that generate largeforces over short distances [20] for intersegmental control [9,100].

Various investigations have demonstrated associations betweenLBP and the characteristics of TrA/LM [15,18,57]. Research has re-vealed that patients with acute or chronic LBP have increased fatinfiltration and abnormal changes of type I and II fibers in LM[2,3,42,56,58,70], while patients with unilateral LBP displaylocalized asymmetrical LM atrophy at the painful vertebral level[15,33,70,92]. Functionally, patients with LBP demonstrate

cal out-

http://dx.doi.org/10.1016/j.pain.2013.07.010

mailto:[email protected]

http://www.elsevier.com/locate/pain


2 A.Y.L. Wong et al. / PAIN�


significantly less TrA/LM thickness changes than asymptomaticindividuals during voluntary tasks as measured by B-modeultrasound imaging (USI) [14,18,48,82] and a delayed anticipatoryonset of TrA or deep LM fibers as measured by intramuscular elec-tromyography during trunk loading or limb movement [37–39,57].Collectively, it is hypothesized that the aberrant morphometry,histology, and activation of these muscles in individuals with acute[16,35,47,75], chronic [61], and recurrent [75] LBP may cause pro-longed LBP and its recurrence [32,57].

Recent research suggests that baseline morphometry of TrA/LMmay be a treatment effect modifier (a characteristic that predictswho will or will not benefit specifically from a particular treatment)or a prognostic factor for LBP-related clinical outcomes [26,29,85].Specifically, decreased LM thickness change during contraction asmeasured by B-mode USI was correlated with the predictors for clin-ical success with a stabilization exercise program [29]. Poor baselineTrA lateral slide measured with B-mode USI may also predictpositive recovery of chronic LBP regardless of the intervention[85]. Taken together, a comprehensive review of evidence for thepredictive or treatment modification value of TrA/LM measurementsmay help clinicians predict LBP prognosis and determine the besttreatments for specific patient subgroups. However, to our knowl-edge, no systematic review on this topic has been conducted.

The primary objective of this review was to summarize theevidence with respect to the ability of morphometry, histology,or activation of TrA/LM to predict clinical outcomes and recurrenceof nonspecific LBP among untreated or conservatively treatedpatients. The predictive ability of TrA/LM includes the prognosticvalue and treatment effect modification. The secondary objectivewas to review whether baseline features of TrA/LM would havedifferential prognostic effects or modify treatment effects insubgroups of patients with different (1) chronicity, (2) age, and(3) inception or survival stage.

2. Methods

This review protocol was registered with PROSPERO(CRD42012002703). The methodology and reporting format of thisreview follows the recommendations and guidelines of the Pre-ferred Reporting Items of Systematic Reviews and Meta-analyses(PRISMA) [63] and the Meta-analysis of Observational Studies inEpidemiology (MOOSE) [80].

Relevant articles were systematically identified through Med-line, Embase, PEDro, SPORTDiscus, CINAHL, and the Cochrane Li-brary (from the beginning of each database up to December2012) using keywords, MeSH, and free-text words, which included:low back pain, LBP, backache, lumbago, lumbalgia, multifidus(truncated), LM, MF, transversus abdominis, TrA, lumbar muscle,erector spinae, stabilizing muscle, predict (truncated), follow-up,cohort, course, longitudinal, and randomized control (truncated).Appendix I shows the exact search strings utilized. In addition,ClinicalTrial.gov, NIH Clinical Center Clinical Research Studies,and Current Controlled Trials Register were searched to identifyrelevant ongoing research. The principal investigators of relevantongoing research, 11 prominent researchers who have publishedmore than 5 articles in this area, as well as the correspondingauthor of each included article, were contacted to identify anyadditional studies of potential relevance. The results of the searchstrategy are shown in Fig. 1.

2.1. Selection criteria of studies

Only full reports published in English, Chinese, French, or Portu-guese and meeting the following criteria were included for analysis(Appendix II).

Please cite this article in press as: Wong AYL et al. Do various baseline charactecomes in nonspecific low back pain? A systematic review. PAIN

�(2013), http:

2.1.1. DesignLongitudinal cohort research is the preferable design for identi-

fying a causal relation for prognostic factors [26,64], while ran-domized controlled trial design is desirable for evaluatingtreatment effect modifiers [78]. As such, the eligible study designincluded primary longitudinal (both prospective and retrospective)cohort studies and randomized controlled trials. Other eligiblestudy designs included observational studies, case series with 10or more subjects involving untreated or conservatively treated pa-tients with nonspecific LBP, systematic reviews, and meta-analy-ses. Studies from books were excluded as secondary sources ofinformation.

Studies that investigated ‘‘inception’’ or ‘‘survival’’ cohorts wereincluded where inception cohort refers to a group of patients earlyin the course of nonspecific LBP, while survival cohort refers to agroup of patients who were at various stages of the LBP develop-ment at the time of recruitment [13,41].

2.1.2. PopulationStudies that investigated self-ambulatory adults older than

18 years, with acute (<6 weeks), subacute (6 to 12 weeks), orchronic (>3 months) nonspecific LBP were included [28]. Nonspe-cific LBP was defined as pain or discomfort between the 12th ribcostal margin and above the gluteal folds, with or without leg pain,where pain is not attributed to specific physical cause or pathology[90,101]. Articles were excluded if they were (1) multiple reportsthat included duplicated results of identical patient cohorts or (2)studies involving less than 80% of participants with nonspecificLBP.

2.1.3. Predictors or treatment effect modifiersIt is hypothesized that spinal stability is maintained by the

passive (such as disc, ligaments, and facets), active (spinal mus-cles), and neural control (sensory and neuromotor control) sub-systems. Dysfunctions in any of the 3 subsystems may increasethe risk of spinal injury by overloading the other subsystems[67–69]. As such, aberrant changes in morphometry, histology,or activation of TrA/LM may affect or predict future clinical out-comes or recurrence of LBP. To be included, studies had to inves-tigate the effect of static and dynamic morphometry, histology, oractivation of TrA/LM in predicting the clinical outcomes of pa-tients with nonspecific LBP. Static morphometry was defined asthe measurement of architectural characteristics (such as shape,cross-sectional area, volume, length, depth, diameter, and penna-tion angles) of a given muscle at rest [98], whereas dynamic mor-phometry referred to the measurement of change in architecturalfeatures of a muscle during contraction [98]. Histology was de-fined as the study of microscopic composition of a muscle (suchas muscle fiber types) [7]. Muscle activation indicated the changein myoelectric signals as measured by electromyography duringmuscle contraction [8]. Studies that investigated the onset of lat-eral abdominal muscles (TrA, obliquus internus, and obliquusexternus) as measured by M-mode USI or tissue Doppler imagingwere included because they are valid assessment of deep abdom-inal muscles function by taking account of intra- and interindivid-ual variability in the onset of the 3 muscles [59,60,85,88]. In orderto identify potential studies that investigated the morphometry,histology, or activation of TrA/LM at baseline, our search strategydid not place any restrictions on the type of muscle measurementtechniques.

2.1.4. Clinical outcomesClinical outcomes included type and duration of LBP symptoms,

LBP-related disability, return to work or sports, recurrence of LBP,LBP-related medications, and visits to health care professionals.

ristics of transversus abdominis and lumbar multifidus predict clinical out-//dx.doi.org/10.1016/j.pain.2013.07.010


Fig. 1. Flow diagram of the literature search.

A.Y.L. Wong et al. / PAIN�

xxx (2013) xxx–xxx 3

2.1.5. InterventionStudies with or without conservative treatments were eligible

for review as long as they met the population and study design cri-teria. Research that solely evaluated the effectiveness of treat-ments was excluded (ie, the research did not perform predictionor treatment effect modification analyses).

2.2. Selection of studies

The selection of studies was divided into 2 stages (Fig. 1). In thefirst stage, the identified citations from the databases were storedin RefWorks 2.0 (RefWorks-COS, USA), and duplicated citationswere excluded. A research assistant removed the identifiable infor-mation (except the title and abstract) of the citations [86]. Tworeviewers (AW and MF) screened the titles and abstracts indepen-dently and discarded the irrelevant citations using standardizedscreening forms according to the selection criteria. Each reviewerfilled in the forms and listed the reasons for the inclusion or theexclusion of articles. The screened lists were compared betweenthe 2 reviewers. Disagreement was resolved by consensus. To min-imize the risk of discarding studies incorrectly, articles that werechosen by either reviewer were included for the next stage of thereview. Piloting of the study selection process was conducted onthe first 100 potential citations. The rationales for inclusion andexclusion were discussed and clarified to ensure the consistencybetween reviewers. In the second stage, the full-text articles ofthe potentially eligible studies were retrieved. The research assis-tant then blacked out the information regarding authors, institu-tions, and journals. The same screening procedures used at thefirst stage were then repeated. If selection disagreement persistedafter the joint review process, a third reviewer (EP) was consultedfor a final decision. The reference lists of the included articles weresearched for relevant articles. Forward citation tracking with Sco-pus and Web of Science was conducted to identify relevant publi-cations that cited the included articles.


�(2013), http:

The interreviewer reliability at each selection stage was ana-lyzed by percent agreement and kappa coefficients. Kappa wasinterpreted as 0.00 to 0.20 for poor agreement, 0.21 to 0.40 for fairagreement, 0.41 to 0.60 for moderate agreement, 0.61 to 0.80 forgood agreement, and 0.81 to 1.00 for almost perfect agreement[54].

2.3. Risk of bias assessment of selected studies

There is no standardized assessment tool for appraising the riskof bias in prognostic studies [5]. On the basis of the recommenda-tions on evaluation of prognostic and treatment effect modificationstudies [1,5,25,27,81], a previously published risk of bias assess-ment tool was modified and adopted [10]. The modifications aimedto evaluate the risk of bias related to confounding variables andtreatment effect modifiers. The modified assessment tool covers7 potential bias areas, including patient population, follow-up ofpatients, prognostic factors, treatment, outcome measurement,confounding measurement, and statistical analysis (Table 1). Themaximum score of this checklist is 26. The cutoff point for distin-guishing a high-quality study is 50% of maximum score [10]. Astudy with higher scores indicates that the study has lower riskof bias. Because the cutoff point was arbitrary, sensitivity analysesusing 60% and 70% cutoff points were conducted to test the robust-ness of these cutoff points.

Assessment of Multiple Systematic Reviews (AMSTAR) wasused to evaluate the risk of bias in the included systematic reviews,if any. AMSTAR was chosen because it has shown high reliability aswell as face, construct, and external validity for assessing the riskof bias of systematic reviews [76,77].

The 2 reviewers conducted the risk of bias assessment indepen-dently on the included studies and were blinded to informationabout authors, institutions, and journals. The interreviewer reli-ability for risk of bias assessment was analyzed by percent agree-ment and intraclass correlation coefficient (ICC, model 3,1) of the



Table 1Risk of bias assessment tool for prospective cohort studies.a

Characteristic Criterion Value

1. Patient population Case definitionOperational definition of cases including exclusion criteria 2Operational definition of cases but no exclusion criteria 1No explicit definition of cases or cannot tell 0Source populationClear description of the source population 1Unclear or no description of the source population 0RepresentativenessPatients representative of clinical practice 2Patients unlikely to be representative of clinical practice 1Cannot determine 0Patient selectionInception cohort (defined in relationship to onset of symptoms) 2Survival cohort, including a subset of the sample with an acute episode/<3 mo (which is analyzed separately) 1Survival cohort, unable to define subsets within the cohort or cannot tell 0ParticipantsClinical and demographic characteristics described 2Insufficient description of participants characteristics 1No explicit description of participants characteristics 0

2. Study attrition/follow-up Follow-up (extent and length)Follow-up of >80% of total sample to at least 1 y as planned 3Follow-up of >80% of total sample for less than 1 y or patients followed for varying lengths of time within 1 y 2Follow-up <80% of total sample completed as planned 1Unclear 0Reasons for loss to follow-up and reportingReasons were provided with descriptive characteristics of this group/in case of no attrition 2Reasons were provided without descriptive characteristics of this group 1Unclear 0

3. Predictors (eg, prognostic factors/treatment effect modifiers)

A priori clear definition or description of and rationale for potential predictor(s) including information onstandardization or using reliable and validated measurements

2

A clear definition or description of and rationale for predictor(s) but insufficient detail on standardization orvalidation of measurements, or the predictor(s) was/were not specified in advance.

1

Inadequate description of/rationale for potential prognostic factors, or a post hoc factor 0

4. Treatment details Description and standardization and/or randomization of treatment (or description of control for confoundingactivities for study without treatment)b

2

Description of treatment but no standardization or randomization (or description of control for confoundingactivities for study without treatment)

1

No information on the treatment providedb 0

5. Outcome measures Blinded outcome criteria appropriate to the research questions with reports of standardized or validmeasurements

2

Unblinded outcome criteria appropriate to the research questions 1No explicit outcome criteria (eg, patient significantly improved) 0

6. Confounding variables Important confounding variables are clearly defined and measured with valid and reliable measurements 1Confounding variables are not defined or considered 0

7. Analysis Statistical analysisAdjusted proportions provided or appropriate multivariate techniques used to adjust for other prognostic factorsc 2Crude proportions but data stratified or presented in a manner which would allow for analysis of subsetsc 1Crude proportions for at least 1 response (eg, remission and/or recurrence) or inadequate sample size 0Selective reporting of resultsAll the results were presented 1Only some results were presented 0Effect sizeEffect sizes was given or sufficient information to calculate effect sizes (eg, odds ratios, relative risk, correlations,likelihood ratios, significance of the interaction tests)

1

Not reported 0Standard errors and/or confidence intervals for the estimatesStandard errors and/or confidence intervals for the estimates was given or sufficient information was given tocalculate them

1

Not reported 0

a Modified from Chorti et al. [10].b Score 0 for a treatment effect modifier if there is no randomization.c Ten or more events per independent variable per prognostic factor; 40 or more events per independent variable for treatment effect modifier; more participants are

needed as the number and the correlation of covariates increase.



final score. A consensus meeting was arranged to resolve the differ-ences between both reviewers.

2.4. Data extraction

Two reviewers independently extracted the following informa-tion: study design, participants’ characteristics (such as age,


�(2013), http:

gender, type and duration of LBP), sample size, intervention (suchas type, dose, duration and follow-up frequency), predictive fac-tors, clinical outcomes (such as functional outcome score, workstatus, and recurrence), statistical analysis (such as Cox regressionanalysis, log rank test, multivariate Kaplan-Meier survival analysis,and linear/multiple/logistic/Poisson regression models), and re-sults at multiple assessment time points (such as odds ratios, beta





coefficients, risk ratios, positive or negative likelihood ratios, andhazard ratios). If the studies assessed multiple outcomes with dif-ferent statistical methods, we solely extracted the information thatwas relevant to our research questions. To ensure accuracy of dataextraction, a consensus meeting was held between the reviewersto discuss the extracted data.

2.5. Synthesizing evidence for prognostic factors and treatment effectmodifiers

Depending on the study quality, homogeneity of study popula-tion (such as age, chronicity of LBP), type of predictive factors, out-come measures, and follow-up time, meta-analysis was consideredfor the studies with low risk of bias using a 50% cutoff point. Spe-cifically, meta-analyses for the prognostic/treatment modificationeffect of each muscle were planned on the basis of the follow-upduration (ie, immediately after treatment, less than 1 year, andmore than 1 year). Meta-analysis with reference to the secondaryobjectives (acute, subacute, and chronic LBP; age between 18 to40 years, 41 to 60 years, and above 60 years; and inception or sur-vival cohort) would be performed if the relevant data were homo-geneous. If meta-analysis was not conducted, qualitative analysiswas used to summarize the predictive value of each predictor.The level of evidence for each predictor was assessed using the cri-teria adopted in previous prognostic research [12]. The level of evi-dence (strong, moderate, limited, no, and inconclusive evidence)was classified on the basis of the risk of bias of the cohort studiesand the consistency of the research findings (Table 2). The resultsrepresent the extent of evidence that substantiate a given featureof TrA or LM in predicting the clinical outcomes of untreated orconservatively treated patients with nonspecific LBP. If P75% ofall the included studies reported a factor that showed a uniformassociation in the same direction, the evidence was consideredconsistent [12].

3. Results

3.1. Literature search

The literature search of databases yielded 2321 potentially rel-evant articles (Fig. 1). A total of 2212 articles were excluded on thebasis of title, abstract, and duplication. In the first stage of screen-ing, the percent agreement between the 2 reviewers was 96.8%with a kappa coefficient of 0.69. One hundred fifteen full-text arti-cles were retrieved. Forty-eight articles were identified on the ba-sis of forward tracking, reference lists of selected articles, orsuggestions from experts. A total of 110 articles were excludedafter assessing the full text. The major reasons for exclusion werea cross-sectional or diagnostic study design, absence of assessmentof morphometry or activation of TrA/LM, and not assessing theability of TrA or LM morphometry/activation in predicting clinicaloutcomes. Five articles met the selection criteria [19,22,59,85,102].

Table 2Level of evidence.a

Level ofevidence

Description

Strong Consistent results (P80%) from at least 2 high-quality studiesModerate One high-quality study and consistent findings (P80%) in 1 or

more low-quality studiesLimited Findings in 1 high-quality cohort or consistent results (P80%)

among low-quality studiesNo No study identifiedConflicting Inconsistent results irrespective of study quality

a Adopted from Cornelius et al. [12].


�(2013), http:

The percent agreement between the 2 reviewers was 100% (kappacoefficient = 1.00) in the second stage of screening.

Three included articles investigated TrA [19,59,85] and 2 in-cluded studies examined LM (Table 3) [22,102]. Of these articles,one evaluated the treatment effect modification of TrA dynamicmorphometry [19] and the rest investigated the prognostic abilityof TrA/LM dynamic morphometry [22,59,85,102]. No includedstudies assessed the predictive ability of TrA or LM on LBPrecurrence.

3.2. Risk of bias assessment and synthesis of evidence

The risk of bias assessment scores is shown in Table 4. The rawagreement for these scores between 2 reviewers was 82%. TheICC3,1 was 0.76. Full agreement was reached after the consensusmeeting. When the cutoff point was set at 50% of the maximumrisk of bias score, all the included studies were classified as low riskof bias (Table 4). If the cutoff point was set at 60% of the maximumscore, 2 included articles were classified as high risk of bias (Ta-ble 4) [22,85]. When the cutoff point was set at 70% of the maxi-mum score, 4 included articles were classified as high risk of bias(Table 4) [22,59,85,102]. The major sources of high risk of biaswere due to (1) no inclusion of inception cohort, (2) less than1 year’s follow-up, (3) no a priori description of potential predic-tors, (4) no adjustment for confounders, (5) insufficient partici-pants for multiple statistical tests, and (6) selective reporting ofresults. Regardless of the cutoff points, the conclusions of thisreview remain unchanged. Because the data obtained from theincluded studies was clinically heterogeneous (such as differentphysical tests for TrA and LM, various clinical outcome measures,and inconsistent follow-up duration), meta-analysis was precluded[87].

3.3. Participants

The sample sizes of the included studies ranged from 25 to 87patients [19,22,59,85,102]. The average age of the participants inthese studies ranged from 33.3 to 54.9 years. Four of 5 includedarticles investigated patients with chronic LBP [19,59,85,102]. Forthe study that did not target patients with chronic LBP, a samplewith mixed duration of LBP was studied (ie, pain duration rangingfrom 2 days to 24 years; median duration, 186 days) [22]. In otherwords, all included studies recruited survival cohorts of patients.The participants of the included studies were recruited fromhospitals, physiotherapy and general practice clinics, and commu-nity-based advertisements [19,22,59,85,102]. None of the studiestargeted patients with acute/subacute LBP.

3.4. Physical tests for TrA and LM

Although the review aimed to include studies that assessedbaseline characteristics of TrA/LM using various measurementmethods, all included studies coincidentally measured TrA/LMusing various USI methods, which have published reliability andvalidity for measuring static and dynamic morphometry of TrA/LM [50,51,60,88,94,101]. Because the included studies did notuse electromyography to measure muscle activation, the findingsof this review were limited to the investigation of baseline TrA/LM morphometry. Two studies assessed voluntary TrA contractionratio during abdominal drawing-in maneuver task (ADIM) usingB-mode [85] and M-mode [59] USI, respectively. TrA contractionratio was the ratio between TrA muscle thickness at maximal con-traction and TrA muscle thickness at rest [85]. Similarly, automaticTrA contraction ratio during an isometric knee flexion/extensioncontraction task [19] and TrA lateral slide during ADIM [85] wasmeasured by B-mode USI. TrA lateral slide was quantified by the



Table 3Data extraction of included studies reporting the functions of transversus abdominis, lumbar multifidus, or lateral abdominal muscles of lumbar spine and the prediction of clinical outcomes in patients with low back pain.

Study Study population; average age andstandard deviation; follow-up

Intervention Muscle (prognosticfactors/treatmentmodifiers)

Outcomemeasure

Confounder adjustment Relation between baseline musclefunction and future clinical outcomes(95% confidence interval)

Ferreiraet al.(2010)[19]

Patients with chronic nonspecific lowback pain recruited from physicaltherapy departments at 3 teachinghospitals (n = 34); 45.4 ± 17.3 y formotor control group, 54.9 ± 11.3 y forgeneral exercise group, 45.4 ± 17.7 yfor spinal manipulation group; 8 wk

A randomized controlled trial. Eachgroup received 12 sessions over 8 wk:

Percentage thicknesschange of TrA at baselineas measured by B-modeUSI during an isometricknee flexion/extensiontask (treatment effectmodifiers)

(A) Globalimpressionof recovery

(B) RMDQ(C) Patient-

specificfunctionalscale

(D) Visual ana-logue scale

Corresponding baseline of clinicaloutcomes

Linear regression to compare theeffect of motor control exercise (vsgeneral exercise) on clinicaloutcomes using baseline percentagethickness change of TrA as anindependent variable:

(1) Motor control exercises: train-ing function of specific deepmuscles of lumbar region, coor-dination of deep trunk musclesand pelvic floor muscles withdiaphragmatic respiration pat-tern, incorporation of deeptrunk muscles contraction dur-ing functional tasks, homeexercises

(2) General exercises: strengthen-ing and stretching of the mainmuscles of the body, cardiovas-cular fitness, education, homeexercises

(3) Spinal manipulative therapy:joint mobilization but not thrustmanipulation techniques to thespine or pelvis

� Interaction effect a: �19.9 (�45.3to 5.5); P = .116� Interaction effect a: 16.1 (�34.0

to 66.2); P = .506� Interaction effect a: �13.7 (�67.2

to 39.9); P = .597� Interaction effect a: 18.1 (1.2 to

34.9); P = .037

Fritz et al.,2011[22]

Patients with low back pain with orwithout leg symptoms recruitedfrom physical therapy clinics andcommunity-based advertisements(n = 50); 33.3 ± 12.9 y; 1 wk

Uncontrolled study. Each patientreceived 2 sessions of spinalmanipulative therapy 3–4 d apart

Baseline LM percentagethickness change asmeasured by B-mode USIduring CALT in prone(prognostic factor)

Modified ODI No adjustment for any confounders Zero-order correlation betweenimmediate postspinal manipulationchange in LM percentage thicknesschange in the first session andpercentage ODI improvement in thefirst week, r = �0.013(nonsignificant)

Mannionet al.,2012[59]

Patients with chronic nonspecific lowback pain with or without referredpain recruited from rheumatology,orthopedics and neurology ofparticipating hospitals (n = 37);44 ± 12.3 y; 9 wk

Uncontrolled study. Each patientreceived 1 treatment session perweek for 9 wk.

Lateral abdominalmuscle function atbaseline:

RMDQ No adjustment for any confounders Correlation between baseline TrA-CRand RMDQ scores: r = 0.24,P = .20Correlation between baselineanticipatory onset of lateralabdominal muscles and RMDQ score:r = 0.04, P = .84Stepwise multipleregression model for identifyingunique predictors of RMDQ scoresdidn’t include baseline TrA-CR noranticipatory onset of lateralabdominal musclesNo significantcorrelation between baseline lateralabdominal muscle function andaverage pain or highest pain at 9 wk

Motor control exercise: trainingfunction of specific deep muscles oflumbar region, coordination of deeptrunk muscles and pelvic floormuscles with diaphragmaticrespiration pattern, incorporation ofdeep trunk muscles contractionduring functional tasks, education,home exercises

(i) Voluntary TrA-CRduring ADIM asmeasured M-modeUSI

(ii) Anticipatory onset oflateral abdominalmuscles (TrA, OI, OE)during rapid armmovement asmeasured by TDI(prognostic factors)

Pain graphicrating scale

Unsgaard-Tøndelet al.(2012)[85]

Participants with chronic nonspecificlow back pain recruited from generalpractitioner or physiotherapists andby advertising among staff at a localhospital (n = 87); 40.2 ± 11.2 y forpatients who experienced clinicallyimportant pain reduction and

Randomized controlled trial. Eachgroup received weekly sessions over 8wk:

Baseline abdominalmuscle function:

NPRS In linear regression model A, age,initial fear avoidance for physicalactivity, gender, body mass index,initial pain intensity and painduration were adjusted for.In linearregression model B, age, initial painlevel and pain duration were

The regression results were obtainedfrom the pooled sample of alltreatment groups.

(1) Motor control exercises(40 min each): training func-tion of TrA, LM and pelvicfloor muscles, co-contraction

(i) TrA lateral slideduring ADIM asmeasured by B-modeUSI

(A) Linear regression of pain at 1-yfollow-up in relation to:� Baseline (i) b = 0.13 (0.00 to

0.26); R2 = 0.036

6A

.Y.L.Wong

etal./PA

IN�

xxx(2013)

xxx–xxx

Pleasecite

thisarticle

inpress

as:W

ongA

YL

etal.D

ovarious

baselinech

aracteristicsoftransversus

abdominis

andlum

barm

ultifidu

spredict

clinicalout-com

esin

nonspecific

lowback

pain?A

systematic

review.PA

IN�

(2013),http://dx.doi.org/10.1016/j.pain.2013.07.010


41.2 ± 11 y for patients who did nothave clinically important painreduction; 1 y

adjusted for.In logistic regressionmodel C, age, initial fear avoidancefor physical activity, gender, bodymass index, and pain duration wereadjusted for.In logistic regressionmodel D, only age and pain durationwere adjusted for

of deep trunk muscles andpelvic floor muscles, incorpo-ration of deep trunk musclescontraction during functionaltasks, home exercises

(2) General exercises (1 h): gen-eral trunk strengthening andstretching of main musclesof the body, cardiovascularfitness, education, homeexercises

(3) Sling exercises (40 min each):performing different progres-sive back exercises in slingsthat hung down from the ceil-ing. Participants needed tomaintain a neutral spine posi-tion during various sling exer-cises to activate both deepand superficial trunk muscles,home exercises

(ii) TrA-CR10 duringADIM as mea-sured by B-modeUSI

(iii) OI-CR10 duringADIM as mea-sured by B-modeUSI

(iv) Anticipatoryonset of abdomi-nal muscles dur-ing rapid armmovements asmeasured by M-mode USI (prog-nostic factors)

� Baseline (ii) b = �0.03 (�0.13 to0.09); R2 = 0.002� Baseline (iii) b = 0.00 (�0.22 to

0.22); R2 = 0.000� Baseline (iv) b = 0.00 (�0.03 to

0.03); R2 = 0.001

(B) Linear regression of pain at 1-yfollow-up in relation to both baselineand change in (i), (iii) and (iv), thebaseline variables (i, iii and iv)explained 35% of total variance ofpain at 1-y follow-up. P value for thismodel was not reported

(C) Logistic regression of clinicallyimportant reduction in pain (NPRSP2) at 1-y follow-up in relation to:� Baseline (i) OR = 0.76 (0.62 to

0.93)� Baseline (ii) OR = 1.05, (0.92 to

1.20)� Baseline (iii) OR = 0.86 (0.65 to

1.13)� Baseline (iv) OR = 0.99 (0.96 to

1.03)(D) Logistic regression of clinicallyimportant reduction in pain at 1-yfollow-up in relation to both baselineand change in (i), (iii) and (iv), onlybaseline (i) was statisticallysignificantly associated withimproved pain level, OR 0.75 (0.57 to0.98)

Zielinskiet al.(2012)[102]

Participants with chronic low backpain with or without recurrencesrecruited from local physical therapyand physician clinics and/or fromself-referral; for CPR-eligible patients(n = 11) 34.3 ± 10.8

One arm of a randomized controlledstudy. Each participants received 1treatment session per week over 6 wk

Percentage thicknesschange of LM thicknessas measured by B-modeUSI during a resisted ornonresisted CALT inprone (A prognosticfactor)

Modified ODI;NPRS

Participants’ initial ODI and NPRSscores.

Linear regression of posttreatmentchange of ODI in relation to baselinepercentage change of LM thicknessin:

CPR-ineligible patients (n = 14)42.1 ± 10 y for 6 wk

Motor control exercise: trainingfunction of specific deep muscles oflumbar region, strengthening trunkflexors, extensors, and obliquemuscles, education on proper bodybiomechanics and spinal protectionduring activities of daily living, homeexercises

CPR-eligible patients� r = 0.36, P = .06 (no resistance

task)� r = 0.14, P = .39 (resistance task)

CPR-ineligible patients� r = 0.014, P = .65 (no resistance


Linear regression of posttreatmentchange of NPRS in relation tobaseline percentage change of LMthickness in:CPR-eligible patients� r = 0.094, P = .89 (no resistance


CPR-ineligible patients� r = 0.065, P = .94 (no resistance


A.Y.L.W

onget

al./PAIN

�xxx

(2013)xxx–

xxx7

Pleasecite

thisarticle

inpress

as:W

ongA

YL

etal.D

ovarious

baselinech

aracteristicsoftransversus

abdominis

andlum

barm

ultifidu

spredict

clinicalout-com

esin

nonspecific

lowback

pain?A

systematic

review.PA

IN�

(2013),http://dx.doi.org/10.1016/j.pain.2013.07.010


AD

IMin

dica

tes

abdo

min

aldr

awin

g-in

man

euve

r;B

-mod

e,br

igh

tnes

sm

ode;

CA

LT,c

ontr

alat

eral

arm

lift

ing

task

;C

PR,c

lin

ical

pred

icti

onru

le;

CPR

-eli

gibl

epa

tien

ts,p

atie

nts

wh

oei

ther

had

lum

bar

spin

eh

ypom

obil

ity

and/

orh

adan

y3

of4

clin

ical

feat

ure

sid

enti

fied

byth

ecl

inic

alpr

edic

tion

rule

for

stab

iliz

atio

nex

erci

sesu

cces

s;C

PR-i

nel

igib

lepa

tien

ts,p

atie

nts

wh

on

eith

erh

adlu

mba

rsp

ine

hyp

omob

ilit

yn

orh

adan

y3

of4

clin

ical

feat

ure

sid

enti

fied

byth

ecl

inic

alpr

edic

tion

rule

for

stab

iliz

atio

nex

erci

sesu

cces

s;LM

,lu

mba

rm

ult

ifidu

s;M

-mod

e,m

otio

nm

ode;

NPR

S,n

um

eral

pain

rati

ng

scal

e;O

DI,

Osw

estr

ydi

sabi

lity

inde

x;O

E,ob

liqu

us

exte

rnu

s;O

I,ob

liqu

us

inte

rnu

s;O

I-C

R10

,ob

liqu

ein

tern

us

thic

knes

sco

ntr

acti

onra

tio

mea

sure

das

:(t

hic

knes

sof

obli

quu

sin

tern

us

thic

knes

sat

max

imu

mco

ntr

acti

on/o

bliq

uu

sin

tern

us

rest

ing

thic

knes

s)�

10;

OR

,od

dsra

tio;

RM

DQ

,R

olan

dM

orri

sdi

sabi

lity

ques

-ti

onn

aire

;TD

I,ti

ssu

eD

oppl

erim

agin

g;Tr

A,t

ran

sver

sus

abdo

min

is;

TrA

-CR

,tra

nsv

ersu

sab

dom

inis

thic

knes

sco

ntr

acti

onra

tio

mea

sure

das

:(t

hic

knes

sof

tran

sver

sus

abdo

min

isat

max

imu

mco

ntr

acti

on/t

ran

sver

sus

abdo

min

isre

stin

gth

ickn

ess)

;Tr

A-C

R10

,tra

nsv

ersu

sab

dom

inis

thic

knes

sco

ntr

acti

onra

tio

mea

sure

das

:(t

hic

knes

sof

tran

sver

sus

abdo

min

isat

max

imu

mco

ntr

acti

on/t

ran

sver

sus

abdo

min

isre

stin

gth

ickn

ess)�

10;

and

USI

,ult

raso

un

dim

agin

g.a

Inte

ract

ion

effe

ctin

dica

tes

wh

eth

erth

epo

sttr

eatm

ent

pain

inte

nsi

tyaf

ter

mot

orco

ntr

olex

erci

se(i

nco

ntr

ast

toge

ner

alex

erci

se)

isaf

fect

edby

the

part

icip

ant’

spe

rcen

tth

ickn

ess

chan

geof

TrA

atba

seli

ne.

Apo

siti

vein

tera

ctio

nef

fect

mea

ns

that

low

erba

seli

ne

perc

enta

geth

ickn

ess

chan

geof

TrA

has

abe

tter

pain

redu

ctio

nif

the

part

icip

ant

rece

ives

mot

orco

ntr

olex

erci

sera

ther

than

gen

eral

exer

cise

.




�(2013), http:

maximum lateral displacement of the V-shaped midline border ofTrA from its resting to contracted positions [85]. The anticipatoryonset of lateral abdominal muscles during a contralateral rapidarm movement task was measured by M-mode USI [85] and tissueDoppler imaging [59]. The anticipatory onset of lateral abdominalmuscles was expressed as the onset time of the earliest abdominalmuscle morphometric change with reference to the onset of thedeltoid muscle of the moving arm [85]. The automatic LM contrac-tion during 2 types of contralateral arm lifting task was measuredby B-mode USI [22,102]. The dynamic morphometry of LM duringcontraction was quantified by the percent change in LM musclethickness. Specifically, the change in LM thickness (contracted �rest) was expressed as a proportion of the thickness at rest (andmultiplied by 100) [102].

3.5. Adjustment of confounders

Three included studies adjusted for some confounders in therespective data analyses [19,85,102]. Ferreira and coworkers ad-justed for the baseline values of a numerical pain rating scale,the Roland Morris disability index, and a patient-specific functionalscale in the respective regression models [19]. Similarly, Zielinskiand colleagues performed regression analyses after adjusting forthe Oswestry disability index and numeric pain rating scores atbaseline [102]. Unsgaard-Tøndel and colleagues adjusted for age,gender, body mass index, pain duration, initial Fear-Avoidance Be-liefs Questionnaire physical score, and initial pain intensity in theirlinear and logistic regression models [85]. Two studies did not ad-just for any confounders [22,59].

3.6. Clinical outcome measures

All studies used both a self-reported pain rating scale and atleast one functional assessment scale as clinical outcome measures[19,22,59,85,102], but in contrast, Unsgaard-Tøndel and colleaguesinvestigated only the predictive ability of baseline dynamic mor-phometry of TrA/LM on pain intensity [85]. The Modified Oswestrydisability index and the Roland Morris Disability questionnairewere the commonly adopted functional assessment scales. Onestudy also included a patient-specific functional scale [19].

3.7. Intervention and follow-up

The included studies adopted different conservative treatmentsincluding sling exercise [85], motor control exercise[19,59,85,102], general exercise [19,85], spinal mobilization [19],and spinal manipulative therapy [22]. Although no limitation wasplaced on the type of conservative treatment evaluated, none ofthe included articles investigated pharmaceutical intervention orthe predictive ability of TrA/LM on untreated patients.

The follow-up time of the included studies ranged from 1 weekto 1 year. Four included studies showed less than 80% attritionwhen they reassessed the participants immediately after the treat-ment [19,22,59,102]. Although the attrition in the study involving1-year follow up was 79.8%, their linear and logistic regressionanalyses used data from 56.8% to 62.4% of the total number of par-ticipants recruited [85].

3.8. Statistical analysis

Linear regression [59,85,102], logistic regression [19,85], andcorrelation [22,59] were commonly used in the included studiesto quantify the predictive value of TrA and LM dynamic morpho-metric variables. However, the report of statistical results variedamong studies. Although some studies reported odds ratios orinteraction effects together with the corresponding confidence



Table 4Risk of bias assessment scores of the included studies.

Risk of bias item Study

Ferreira et al., 2010[19]

Fritz et al., 2011[22]

Mannion et al., 2012[59]

Unsgaard-Tøndel et al., 2012[85]

Zielinski et al., 2012[102]

1a. Case definition 2 2 2 2 21b. Source population 1 1 1 1 11c. Representativeness 2 2 2 2 11d. Patient selection 0 0 0 0 01e. Participants 2 2 2 2 22a. Follow-up 2 2 2 1 22b. Reasons for attrition 2 0 1 0 23. Prognostic factors/modifiers 1 1 1 1 14. Treatment details 2 2 2 2 15. Outcome measures 2 1 2 1 26. Confounders 1 0 0 1 17a. Analysis 0 0 0 0 07b. Selective reporting of results 0 1 0 0 17c. Effect size 1 1 1 1 17d. Standard errors/confidence

intervals1 0 0 1 0

Total scores 19 15 16 15 17Risk of bias (cutoff at 50%)a Low Low Low Low LowRisk of bias (cutoff at 60%)a Low High Low High LowRisk of bias (cutoff at 70%)a Low High High High High

a Cutoff score at 50%, n = 13; at 60%, n = 16; at 70%, n = 18.



intervals [19,85], others simply reported a P value or ‘‘nonsignifi-cant’’ [22,59,102]. Most studies estimated the relation betweenpredictors and clinical outcomes based on significance testing ofthe corresponding absolute values in outcome scales [19,22,59,102], but one study used clinically important pain reductionfor analysis [85].

3.9. Static morphometry and histology of TrA or LM in predictingclinical outcomes of LBP

The literature search did not identify any study that investi-gated the static morphometry or histology of TrA or LM in predict-ing clinical outcomes or recurrence of LBP in patients withnonspecific LBP. As such, there was no evidence to support the pre-dictive value of TrA/LM static morphometry or histology on clinicaloutcomes of nonspecific LBP.

3.10. Baseline dynamic morphometry of TrA or LM as a prognosticfactor for clinical outcomes of LBP

Table 3 summarizes the data extracted from the included stud-ies that investigated the dynamic morphometry of TrA/LM at base-line in predicting clinical outcomes of LBP. Depending on the typesof physical tests, the predictive value of TrA/LM varies. Unsgaard-Tøndel and coworkers found that better baseline TrA lateral slideduring ADIM in patients with chronic LBP was associated withhigher pain intensity at 1-year after various types of exercise ther-apy (beta coefficient = 0.13) [85]. They also noted that better base-line TrA lateral slide was more likely to have less clinicallyimportant pain improvement (odds ratio = 0.76) at 1 year’s fol-low-up, regardless of the type of exercise treatment [85]. However,the baseline TrA contraction ratio (contracted TrA thickness/re-laxed TrA thickness) during ADIM was not correlated withimprovement in clinical outcomes (pain intensity or LBP-relateddisability) of patients irrespective of the type of exercise interven-tion [59,85].

Conflicting predictive results were also noted in baseline per-cent LM thickness change. Fritz and colleagues found that baselinepercent thickness change of LM (during a contralateral arm liftingtask) was not associated with LBP-related disability after 1-week


�(2013), http:

spinal manipulative therapy [22]. A 6-week motor control exercisestudy also revealed that baseline percent LM thickness changecould not predict posttreatment pain intensity [102]. However,the same research found that a larger percent LM thickness changeduring a resisted contralateral arm lifting task at baseline was aweak but significant predictor of improved posttreatment LBP-re-lated disability in a subgroup of patients who were unlikely to im-prove with stabilization exercise (r = 0.38, P = .013) [102].

Overall, depending on the types of physical tests, there was lim-ited evidence about whether baseline TrA dynamic morphometricvariables were prognostic factors for LBP-related disability or painintensity in patients with chronic LBP after various exercise inter-ventions (Table 5). Similarly, there was limited evidence in supportof no relation between baseline LM dynamic morphometry andposttreatment LBP intensity, but there was conflicting evidencefor a relation between baseline LM dynamic morphometry andposttreatment LBP-related disability (Table 5).

3.11. Baseline dynamic morphometry of TrA or LM as a treatmenteffect modifier for clinical outcomes of LBP

One study reported that baseline dynamic morphometry of TrAwas a treatment effect modifier [19]. Ferreira and coworkers foundthat patients with poorer baseline TrA thickness contraction ratioreceiving a course of motor control exercise yielded better painreduction than those receiving a general exercise regimen at 8-week follow-up (Table 3) [19]. No study has investigated the treat-ment effect modification ability of LM at baseline.

Collectively, as a result of limited publication on this topic, lim-ited evidence corroborated that poorer automatic TrA contractionat baseline was a treatment effect modifier favoring motor controlexercise over general exercise (Table 5). There was no evidenceregarding the treatment effect modification of baseline LM dy-namic morphometry.

3.12. Baseline anticipatory onset of lateral abdominal muscles or LMmorphometric change in predicting clinical outcomes of LBP

Two studies reported that baseline anticipatory (feed-forward)onset timing of lateral abdominal muscles morphometric change



Table 5Level of evidence for conclusions on the neuromuscular function of transversus abdominis or lumbar multifidus as a treatment effect modifier and prognostic factor with studyquality appraisal cutoff point at 50%.

Prognostic factor or treatment moderator No. of cohort studies withpositive or no results andrisk of bias of the studies(high/low)

Level of evidence

Muscle Factor identified Clinical outcome Positiveresult

No result a Summarizedresult

Prognosticfactor

Treatment effectmodifier

High Low High Low

TrA Percentage thickness changeof TrAduring an isometric kneeflexion/extension taskb

RMDQ [18] No Limited

Patient-specificfunctional scale

[18] No Limited

Global perceivedrecovery

[18] No Limited

VAS [18] Positive LimitedTrA-CR during ADIM RMDQ [61] No LimitedTrA-CR10 during ADIM PGRS/NPRS (short term) [61] No Limited

PGRS/NPRS (long term) [85] No LimitedTrA lateral slide NPRS [85] Positive Limited

Lateral abdominalmuscles

Anticipatory onset of lateralabdominalmuscles before arm elevation

RMDQ [61] No Limited

PGRS/NPRS (short term) [61] No LimitedPGRS/NPRS (long term) [85] No Limited

LM LM percentage thicknesschange duringCALT c

Modified ODI [98] [21] Inconclusive Conflicting

NPRS [98] No Limited

ADIM, abdominal drawing-in maneuver; CALT, contralateral arm lifting task; conflicting, conflicting level of evidence; high, cohort study with high risk of bias; limited,limited level of evidence; low, cohort study with low risk of bias; LM, lumbar multifidus; inconclusive, contradictive findings regarding the factor as a treatment modifier orprognostic factor; NPRS, numeric pain rating scale; ODI, Oswestry disability index; PGRS, Pain graphic rating scale; positive, results supported the factor as a treatmentmodifier or a prognostic factor; RMDQ, Roland Morris disability questionnaire; strong, strong level of evidence; TrA-CR, transversus abdominis thickness contraction ratiomeasured as: (thickness of transversus abdominis at maximum contraction/transversus abdominis resting thickness); TrA-CR10, transversus abdominis thickness contractionratio measured as: (thickness of transversus abdominis at maximum contraction/transversus abdominis resting thickness) � 10.

a Result did not support the factor as a treatment modifier or a prognostic factor.b Percentage thickness change of TrA measured as change in muscle thickness as percentage of resting thickness.c Lumbar multifidus percent thickness change expressed as change in muscle thickness as percentage of resting thickness.



could not predict short- or long-term clinical outcomes (LBP-re-lated disability or pain intensity) of individuals with chronic LBPafter various types of exercise interventions (Table 3) [59,85]. Nostudy has investigated the predictive value of the anticipatory on-set of LM morphometric change on the clinical outcomes of pa-tients with nonspecific LBP.

There was limited evidence against the role of anticipatory on-set of lateral abdominal muscles morphometric change in predict-ing the short- or long-term clinical outcomes of patients withchronic LBP after various exercise regimens (Table 5) [59,85]. Therewas no evidence concerning the value of feed-forward onset of LMmorphometric change in predicting the clinical outcomes of pa-tients with nonspecific LBP.

3.13. Effect of chronicity, age, and inception/survival cohorts onpredictive ability of TrA/LM

Because the majority of the participants in the included studiesshould be classified as patients with chronic LBP, the value of mor-phometry of TrA/LM in predicting clinical outcomes of patientswith acute/subacute LBP remains unknown. Similarly, becausethe mean ages of the participants in the included studies werefairly homogenous, subgroup analysis based on different agegroups was precluded. No included study recruited inception pa-tient cohorts. As such, no subgroup analysis of inception and sur-vival cohorts was performed.


�(2013), http:

4. Discussion

The primary objective of this review was to summarize theevidence regarding the ability of morphometry, histology, oractivation of TrA/LM to predict clinical outcomes and LBP recur-rence among untreated or conservatively treated patients. Therewas limited evidence that baseline TrA contraction ratio andanticipatory onset of lateral abdominal muscles morphometricchange were not related to the short- or long-term clinical out-comes of patients with chronic nonspecific LBP after variousexercise interventions. Depending on the type of physical test,there was conflicting evidence regarding the dynamic morphom-etry of TrA/LM in predicting LBP-related disability or pain reduc-tion in patients with chronic nonspecific LBP after variousconservative treatments. No identified research investigated thebaseline static morphometry of TrA/LM in predicting clinical out-comes of patients with nonspecific LBP. Similarly, there has yetto be a study that examines the value of baseline TrA/LM char-acteristics in predicting LBP recurrence. Interestingly, limited evi-dence showed that poorer baseline TrA contraction ratio was atreatment effect modifier favoring motor control exercise overgeneral exercise.

No subgroup analysis (based on age, chronicity, and inceptionor survival cohorts) was performed in this review, given the ab-sence of relevant subgroup classification and analysis in the pri-mary studies.





4.1. Characteristics of the included studies

4.1.1. Sample sizeThe sample sizes of the included studies were relatively small.

Theoretically, each candidate predictor in a multivariable regressionmodel requires at least 10 participants to ensure the precision ofestimation and to prevent erroneous associations [11,55]. Four timesmore participants are required if a study aims to estimate the effectof a treatment effect modifier [6]. However, 4 included studies didnot meet these criteria [22,59,85,102]. Conducting regression analy-sis with a suboptimal sample size might reduce the statistical powerto identify statistically significant associations [71].

4.1.2. Physical testsThe dynamic morphometry of a muscle measured with B-mode

USI is affected by many factors. Although muscle activation maychange muscle dimensions, many factors affect the magnitude ofthe dimensional change. Specifically, dynamic morphometry of amuscle in 2-dimensional ultrasound images is influenced by theresting state of the muscle, the extensibility of the musculotendi-nous unit, the type of contraction, the competing force from nearbymuscles, the out-of-plane muscle motion, the orientation of theultrasound transducer, and the operator’s experience [97]. As such,the reported correlation coefficient between the magnitude ofchange in abdominal muscle thickness and the corresponding elec-tromyography ranged from 0.14 to 0.93 [8,36,45,62,96]. Given that4 included studies used B-mode USI to measure dynamic mor-phometry of TrA/LM without reporting their operators’ qualifica-tion [19,22,85,102], their findings should be interpreted withcaution [95]. Future studies should use electromyography or novelmeasurement methods [40,66] to measure the activation of trunkmuscles [99].

Although ADIM was commonly used to test TrA activity [59,85],the performance of ADIM relies on the examiner’s instruction andthe participant’s capability to perform the desirable action [52].Inconsistent ADIM performance might explain the absence of cor-relation between baseline TrA contraction ratio and LBP intensityat 9 weeks’ or 1 year’s follow-up [59,85]. Instead of using ADIM,assessment of automatic TrA activity using an active straight legraise test [52] or an isometric knee flexion/extension task [18] ispreferable because these tests eliminate the voluntary componentof ADIM.

The great variability in onset timing of lateral abdominal mus-cles in individuals with [23,89] and without [60,88] LBP may ac-count for the lack of relation between baseline anticipatory onsetof these muscles and future clinical outcomes of LBP. Future re-search should investigate whether the predictive ability of feed-forward activation of TrA exists in certain patient subgroups usingintramuscular electromyography.

4.1.3. Intervention and follow-upThe treatment intensity of the included studies varied from 2

sessions over 1 week to 9 weeks of once-weekly treatment[19,22,59,85,102]. These results might not be generalizable totreatments of different frequencies and durations. Additionally,some treatments (such as general exercise) in the included studiesmight improve clinical outcomes through the alternations of self-belief about disability or exercise [44,49,74,93]. Therefore, baselineTrA/LM characteristics might be less likely to predict the clinicaloutcomes of patients receiving these treatments.

Four included studies investigated the association between pre-dictors and the immediate posttreatment clinical outcomes[19,22,59,102]. Although this might be the best stage to estimatethe predictive value of various baseline predictors, it would beimperative to investigate their predictive ability at a longer-termfollow-up.


�(2013), http:

4.1.4. Adjustment of confoundersWe determined that if our evidence synthesis discarded the

studies that did not adjust for confounders [22,59], our conclusionswould remain. Psychological factors, which have been identified asindependent predictors of LBP outcomes [21,59,72], were not con-sidered in most of the included studies [19,22,59,102]. Analyzingdata without adjusting for psychological factors may affect esti-mates of the predictive ability of TrA/LM [83,84].

4.1.5. Statistical analysisFour included studies [19,22,59,102] analyzed the relation be-

tween predictors and clinical outcomes using the absolute valuesin outcome scales. One study used clinically important pain reduc-tion for analysis [85]—a measure argued to have greater clinicalsignificance, although it requires a larger sample size to obtain asignificant result.

Two studies used multiple statistical tests [85,102], which in-crease the risk of false-positive findings. Zielinski and coworkersused 8 separate linear regression models to investigate whetherbaseline LM percent thickness change could predict the postexer-cise clinical outcomes in 25 patients [102]. Similarly, Unsgaard-Tøndel and colleagues used multiple statistical tests and observedonly baseline TrA lateral slide was related to the LBP intensity ofpatients with chronic LBP at 1 year’s follow up [85]. Future studiesshould replicate these experiments with larger sample sizes and apriori hypotheses to minimize the risk of accepting false positiveresults.

4.2. Strengths of this review

This review has multiple strengths. To improve the coverage ofrelevant articles, we identified candidate publications in 4 lan-guages through a systematic search and contacted the correspond-ing authors of the included articles and prominent researchers inthe area. To enhance the accuracy and consistency of data extrac-tion, we used standardized forms for screening and data extraction,and conducted a pilot exercise. Further, this review protocol wasregistered in PROSPERO to improve the credibility.

Our risk of bias assessment tool has incorporated the recom-mended criteria for evaluating prognostic studies and treatmenteffect modification studies [5,10,25,27]. To provide a comprehen-sive overview of the assessment results, we reported specific itemsin detail. We also conducted 2 sensitivity analyses using 60% and70% cutoff points to test the robustness of the assessment. The re-sults showed that different cutoff points would not change theconclusions of this review.

4.3. Limitations

Like other systematic reviews, the present review may be sub-ject to publication bias where positive studies are selectively pub-lished [1,17]. To optimize the chance of including publications withpositive and negative results, a comprehensive search strategy wasused. Although a funnel plot was not used in this review, given theheterogeneity and small number of included studies [79], the re-sults of our included studies only marginally support or even rejectthe value of baseline features of TrA/LM in predicting clinical out-comes of patients with LBP. Therefore, our review is unlikely to beaffected by publication bias.

4.4. Implications

This review was conducted on the basis of the hypothesis thataberrant changes in morphometry or activation of TrA/LM may af-fect the progression or recurrence of LBP [32,65]. However, differ-ent factors may affect our findings. Firstly, the deficits in TrA/LM





may be the consequence of LBP rather than the cause of LBP [30].This premise is supported by the fact that experimental paincauses changes in anticipatory onset of TrA [35] and dynamic mor-phometry of TrA/LM [47]. Hence, baseline deficits of TrA/LM maynot predict LBP prognosis. Secondly, all included studies recruitedparticipants with chronic nonspecific LBP whose prognosis may beaffected by multiple factors (such as duration of pain and copingstrategy) [93]. Therefore, our results might have been different ifthe included studies had involved patients with acute/subacuteLBP. Further, given the heterogeneity of nonspecific LBP, baselinecharacteristics of TrA/LM may only predict clinical outcomes ofcertain patient subgroups. Thirdly, 4 [19,59,85,102] of 5 includedstudies [19,22,59,85,102] reported the posttreatment changes inTrA/LM characteristics. However, regardless of the presence or ab-sence of posttreatment TrA/LM characteristic changes, there wereonly weak associations between baseline TrA/LM characteristicsand future clinical outcomes. The weak relations might be attrib-uted to the small sample size or the heterogeneity of survivalpatient cohorts. Finally, because B-mode USI cannot measuremuscle activation, our findings might have been different if theincluded studies had used electromyography. Overall, despite theconflicting evidence regarding baseline TrA/LM deficits in predict-ing clinical outcomes of patients with LBP, some longitudinal stud-ies substantiate the deficits of TrA/LM as risk factors for thedevelopment of LBP or lower limb injury in healthy individuals(Appendix III).

Given the above, high-quality large-scale longitudinal studieswith long-term follow-up and adjustment for confounders areneeded to clarify the potential role of TrA/LM in predicting theprognosis or recurrence of LBP. Future studies should also investi-gate whether the posttreatment changes in TrA/LM characteristicsaccompany the improvement of clinical outcomes.

Conflict of interest

The authors report no conflict of interest.

Acknowledgments

The authors would like to thank librarian Jeanette Buckinghamfor the design of the literature search. We also thank Lydia Dani forassisting the procedures of removing the duplicated citations andthe identifiable information of the potential citations. A.Y.L.W. issupported by the Alberta Innovates-Health Solutions Graduate Stu-dentship and the Golden Key Graduate Scholar Award. G.N.K. issupported by the Canadian Research Chair Program.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.pain.2013.07.010.

References

[1] Altman DG. Systematic reviews of evaluations of prognostic variables. BMJ2001;323:224–8.

[2] Bajek S, Bobinac D, Bajek G, Vranic TS, Lah B, Dragojevic DM. Muscle fiber typedistribution in multifidus muscle in cases of lumbar disc herniation. Acta MedOkayama 2000;54:235–41.

[3] Barker KL, Shamley DR, Jackson D. Changes in the cross-sectional area ofmultifidus and psoas in patients with unilateral back pain: the relationship topain and disability. Spine 2004;29:E515–9.

[4] Barker PJ, Guggenheimer KT, Grkovic I, Briggs CA, Jones DC, Thomas CDL,Hodges PW. Effects of tensioning the lumbar fasciae on segmental stiffnessduring flexion and extension: Young Investigator Award winner. Spine2006;31:397–405.

[5] Beattie PF, Nelson RM. Evaluating research studies that address prognosis forpatients receiving physical therapy care: a clinical update. Phys Ther2007;87:1527–35.


�(2013), http:

[6] Brookes ST, Whitely E, Egger M, Smith GD, Mulheran PA, Peters TJ. Subgroupanalyses in randomized trials: risks of subgroup-specific analyses; power andsample size for interaction test. J Clin Epidemiol 2004;57:229–36.

[7] Brown S, Gregory D, Carr J, Ward S, Masuda K, Lieber R. Adaptation to themultifidus muscle in response to experimentally induced intervertebral discdegeneration. Spine 2011;36:1728–36.

[8] Brown SHM, McGill SM. A comparison of ultrasound and electromyographymeasures of force and activation to examine the mechanics of abdominal wallcontraction. Clin Biomech (Bristol, Avon) 2010;25:115–23.

[9] Cholewicki J, Panjabi MM, Khachatryan A. Stabilizing function of trunk flexor–extensor muscles around a neutral spine posture. Spine 1997;22:2207–12.

[10] Chorti A, Chortis A, Strimpakos N, McCarthy C. The prognostic value ofsymptom responses in the conservative management of spinal pain: asystematic review. Spine 2009;34:2686–99.

[11] Concato J, Feinstein A, Holford T. The risk of determining risk withmultivariable models. Ann Intern Med 1993;118:201–10.

[12] Cornelius LR, van der Klink JJL, Groothoff JW, Brouwer S. Prognostic factors oflong term disability due to mental disorders: a systematic review. J OccupRehabil 2011;21:259–74.

[13] Costa L, Henschke N, Maher CG, Refshauge KM, Herbert RD, McAuley JH, DasA, Costa LO. Prognosis of chronic low back pain: design of an inception cohortstudy. BMC Musculoskelet Disord 2007;8:11.

[14] Critchley D, Coutts F. Abdominal muscle function in chronic low back painpatients measurement with real-time ultrasound scanning. Physiotherapy2002;88:322–32.

[15] Danneels LA, Vanderstraeten GG, Cambier DC, Witvrouw EE, De Cuyper HJ. CTimaging of trunk muscles in chronic low back pain patients and healthycontrol subjects. Eur Spine J 2000;9:266–72.

[16] Dickx N, Cagnie B, Parlevliet T, Lavens A, Danneels L. The effect of unilateralmuscle pain on recruitment of the lumbar multifidus during automaticcontraction. An experimental pain study. Man Ther 2010;15:364–9.

[17] Easterbrook PJ, Gopalan R, Berlin JA, Matthews DR. Publication bias in clinicalresearch. Lancet 1991;337:867–72.

[18] Ferreira PH, Ferreira ML, Hodges PW. Changes in recruitment of theabdominal muscles in people with low back pain: ultrasound measurementof muscle activity. Spine 2004;29:2560–6.

[19] Ferreira PH, Ferreira ML, Maher CG, Refshauge K, Herbert RD, Hodges PW.Changes in recruitment of transversus abdominis correlate with disability inpeople with chronic low back pain. Br J Sports Med 2010;44:1166–72.

[20] Freeman MD, Woodham MA, Woodham AW. The role of the lumbarmultifidus in chronic low back pain: a review. PMR 2010;2:124–46.

[21] Friedrich M, Hahne J, Wepner F. A controlled examination of medical andpsychosocial factors associated with low back pain in combination withwidespread musculoskeletal pain. Phys Ther 2009;89:786–803.

[22] Fritz JM, Koppenhaver SL, Kawchuk GN, Teyhen DS, Hebert JJ, Childs JD.Preliminary investigation of the mechanisms underlying the effects ofmanipulation: exploration of a multi-variate model including spinalstiffness, multifidus recruitment, and clinical findings. Spine2011;36:1772–81.

[23] Gubler D, Mannion AF, Schenk P, Gorelick M, Helbling D, Gerber H, Toma V,Sprott H. Ultrasound tissue Doppler imaging reveals no delay in abdominalmuscle feed-forward activity during rapid arm movements in patients withchronic low back pain. Spine 2010;35:1506–13.

[24] Haldeman S, Kopansky-Giles D, Hurwitz EL, Hoy D, Mark Erwin W,Dagenais S, Kawchuk G, Stromqvist B, Walsh N. Advancements in themanagement of spine disorders. Best Pract Res Clin Rheumatol 2012;26:263–80.

[25] Hamberg-van Reenen HH, Ariëns GAM, Blatter BM, van Mechelen W, BongersPM. A systematic review of the relation between physical capacity and futurelow back and neck/shoulder pain. PAIN� 2007;130:93–107.

[26] Hancock M, Herbert RD, Maher CG. A guide to interpretation of studiesinvestigating subgroups of responders to physical therapy interventions. PhysTher 2009;89:698–704.

[27] Hayden JA, Côté P, Bombardier C. Evaluation of the quality of prognosisstudies in systematic reviews. Ann Intern Med 2006;144:427–37.

[28] Hayden JA, van Tulder MW, Malmivaara AV, Koes BW. Meta-analysis:exercise therapy for nonspecific low back pain. Ann Intern Med2005;142:765–75.

[29] Hebert JJ, Koppenhaver SL, Magel JS, Fritz JM. The relationship of transversusabdominis and lumbar multifidus activation and prognostic factors forclinical success with a stabilization exercise program: a cross-sectionalstudy. Arch Phys Med Rehabil 2010;91:78–85.

[30] Hides J, Stanton W, Mendis MD, Sexton M. The relationship of transversusabdominis and lumbar multifidus clinical muscle tests in patients withchronic low back pain. Man Ther 2011;16:573–7.

[31] Hides J, Wilson S, Stanton W, McMahon S, Keto H, McMahon K, Bryant M,Richardson C. An MRI investigation into the function of the transversusabdominis muscle during ‘‘drawing-in’’ of the abdominal wall. Spine2006;31:E175–8.

[32] Hides JA, Jull GA, Richardson CA. Long-term effects of specific stabilizingexercises for first-episode low back pain. Spine 2001;26:E243–8.

[33] Hides JA, Stokes MJ, Saide M, Jull GA, Cooper DH. Evidence of lumbarmultifidus muscle wasting ipsilateral to symptoms in patients with acute/subacute low back pain. Spine 1994;19:165–72.

[34] Hodges PW, Eriksson AEM, Shirley D, Gandevia SC. Intra-abdominal pressureincreases stiffness of the lumbar spine. J Biomech 2005;38:1873–80.



http://refhub.elsevier.com/S0304-3959(13)00378-3/h0005






































































































[35] Hodges PW, Moseley GL, Gabrielsson A, Gandevia SC. Experimental musclepain changes feedforward postural responses of the trunk muscles. Exp BrainRes 2003;151:262–71.

[36] Hodges PW, Pengel LHM, Herbert RD, Gandevia SC. Measurement of musclecontraction with ultrasound imaging. Muscle Nerve 2003;27:682–92.

[37] Hodges PW, Richardson CA. Altered trunk muscle recruitment in people withlow back pain with upper limb movement at different speeds. Arch Phys MedRehabil 1999;80:1005–12.

[38] Hodges PW, Richardson CA. Delayed postural contraction of transversusabdominis in low back pain associated with movement of the lower limb. JSpinal Disord 1998;11:46–56.

[39] Hodges PW, Richardson CA. Inefficient muscular stabilization of the lumbarspine associated with low back pain: a motor control evaluation oftransversus abdominis. Spine 1996;21:2640–50.

[40] Hu Y, Wong YL, Lu WW, Kawchuk GN. Creation of an asymmetrical gradientof back muscle activity and spinal stiffness during asymmetrical hipextension. Clin Biomech (Bristol, Avon) 2009;24:799–806.

[41] Hudak PL, Cole DC, Haines AT. Understanding prognosis to improverehabilitation: the example of lateral elbow pain. Arch Phys Med Rehabil1996;77:586–93.

[42] Hyun JK, Lee JY, Lee SJ, Jeon JY. Asymmetric atrophy of multifidus muscle inpatients with unilateral lumbosacral radiculopathy. Spine2007;32:E598–602.

[43] Jemmett RS, Macdonald DA, Agur AMR. Anatomical relationships betweenselected segmental muscles of the lumbar spine in the context of multi-planar segmental motion: a preliminary investigation. Man Ther2004;9:203–10.

[44] Jensen MP, Turner JA, Romano JM. Correlates of improvement in multidisciplinarytreatment of chronic pain. J Consult Clin Psychol 1994;62:172–9.

[45] John EK, Beith ID. Can activity within the external abdominal oblique bemeasured using real-time ultrasound imaging? Clin Biomech (Bristol, Avon)2007;22:972–9.

[46] Kay AG. An extensive literature review of the lumbar multifidus: anatomy. JMan Manip Ther 2000;8:102–14.

[47] Kiesel KB, Uhl T, Underwood FB, Nitz AJ. Rehabilitative ultrasoundmeasurement of select trunk muscle activation during induced pain. ManTher 2008;13:132–8.

[48] Kiesel KB, Underwood FB, Mattacola CG, Matacolla C, Nitz AJ, Malone TR. Acomparison of select trunk muscle thickness change between subjects withlow back pain classified in the treatment-based classification system andasymptomatic controls. J Orthop Sports Phys Ther 2007;37:596–607.

[49] Klaber Moffett JA, Carr J, Howarth E. High fear-avoiders of physical activitybenefit from an exercise program for patients with back pain. Spine2004;29:1167–72.

[50] Koppenhaver S, Hebert J, Fritz J, Parent E, Teyhen D, Magel J. Reliability ofrehabilitative ultrasound imaging of the transversus abdominis and lumbarmultifidus muscles. Arch Phys Med Rehabil 2009;90:87–94.

[51] Koppenhaver S, Hebert J, Parent E, Fritz J. Rehabilitative ultrasound imaging isa valid measure of trunk muscle size and activation during most isometricsub-maximal contractions: a systematic review. Aust J Physiother2009;55:153–69.

[52] Koppenhaver SL, Fritz JM, Hebert JJ, Kawchuk GN, Childs JD, Parent EC, GillNW, Teyhen DS. Association between changes in abdominal and lumbarmultifidus muscle thickness and clinical improvement after spinalmanipulation. J Orthop Sports Phys Ther 2011;41:389–99.

[53] Koumantakis G, Watson P. Trunk muscle stabilization training plus generalexercise versus general exercise only: randomized controlled trial of patientswith recurrent low back pain. Phys Ther 2005;85:209–25.

[54] Landis JR, Koch GG. The measurement of observer agreement for categoricaldata. Biometrics 1977;33:159–74.

[55] Lewis S. Regression analysis. Pract Neurol 2007;7:259–64.[56] Luoto S, Heliovaara M, Hurri H, Alaranta H. Static back endurance and the risk

of low-back pain. Clin Biomech (Bristol, Avon) 1995;10:323–24.[57] MacDonald D, Moseley GL, Hodges PW. Why do some patients keep hurting

their back? Evidence of ongoing back muscle dysfunction during remissionfrom recurrent back pain. Pain� 2009;142:183–8.

[58] Mannion A, Käser L, Weber E, Rhyner A, Dvorak J, Muntener M. Influence ofage and duration of symptoms on fibre type distribution and size of the backmuscles in chronic low back pain patients. Eur Spine J 2000;9:273–81.

[59] Mannion AF, Caporaso F, Pulkovski N, Sprott H. Spine stabilisation exercisesin the treatment of chronic low back pain: a good clinical outcome is notassociated with improved abdominal muscle function. Eur Spine J2012;21:1301–10.

[60] Mannion AF, Pulkovski N, Schenk P, Hodges PW, Gerber H, Loupas T, GorelickM, Sprott H. A new method for the noninvasive determination of abdominalmuscle feedforward activity based on tissue velocity information from tissueDoppler imaging. J Appl Physiol 2008;104:1192–201.

[61] Marshall P, Murphy B. Delayed abdominal muscle onsets and self-reportmeasures of pain and disability in chronic low back pain. J ElectromyogrKinesiol 2010;20:833–9.

[62] McMeeken JM, Beith ID, Newham DJ, Milligan P, Critchley DJ. The relationshipbetween EMG and change in thickness of transversus abdominis. ClinBiomech (Bristol, Avon) 2004;19:337–42.

[63] Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group. Preferred reportingitems for systematic reviews and meta-analyses: the PRISMA statement. Int JSurg 2009;151:264–70.


�(2013), http:

[64] Moons K, Royston P, Vergouwe Y, Grobbee D, Altman D. Prognosis andprognostic research: what, why, and how? BMJ 2009;338:b375.

[65] Moseley GL. Impaired trunk muscle function in sub-acute neck pain: etiologicin the subsequent development of low back pain? Man Ther 2004;9:157–63.

[66] Paalanne N, Niinimäki J, Karppinen J, Taimela S, Mutanen P, Takatalo J,Korpelainen R, Tervonen O. Assessment of association between low back painand paraspinal muscle atrophy using opposed-phase MR imaging. apopulation-based study among young adults. Spine 2011;36:1961–8.

[67] Panjabi MM. Clinical spinal instability and low back pain. J ElectromyogrKinesiol 2003;13:371–9.

[68] Panjabi MM. The stabilizing system of the spine. Part I. Function, dysfunction,adaptation, and enhancement. J Spinal Disord 1992;5:383–9.

[69] Panjabi MM. The stabilizing system of the spine. Part II. Neutral zone andinstability hypothesis. J Spinal Disord 1992;5:390–6.

[70] Parkkola R, Rytökoski U, Kormano M. Magnetic resonance imaging of thediscs and trunk muscles in patients with chronic low back pain and healthycontrol subjects. Spine 1993;18:830–6.

[71] Portney L, Watkins M. Foundations of clinical research: applications topractice. 3rd ed. Upper Saddle River, NJ: Prentice-Hall; 2009.

[72] Prkachin KM, Schultz I, Berkowitz J, Hughes E, Hunt D. Assessing painbehaviour of low-back pain patients in real time: concurrent validity andexaminer sensitivity. Behav Res Ther 2002;40:595–607.

[73] Quint U, Wilke HJ, Shirazi-Adl A, Parnianpour M, Loer F, Claes LE. Importanceof the intersegmental trunk muscles for the stability of the lumbar spine: abiomechanical study in vitro. Spine 1998;23:1937–45.

[74] Rainville J, Ahern DK, Phalen L, Childs LA, Sutherland R. The association ofpain with physical activities in chronic low back pain. Spine 1992;17:1060–4.

[75] Richardson C, Hodges P, Hides J. Therapeutic exercise for lumbopelvicstabilization: a motor control approach for the treatment and prevention oflow back pain. 2nd ed. New York: Churchill Livingstone; 2004.

[76] Shea BJ, Grimshaw JM, Wells GA, Boers M, Andersson N, Hamel C, Porter AC,Tugwell P, Moher D, Bouter LM. Development of AMSTAR: a measurementtool to assess the methodological quality of systematic reviews. BMC Med ResMethodol 2007;7:10.

[77] Shea BJ, Hamel C, Wells GA, Bouter LM, Kristjansson E, Grimshaw J, Henry DA,Boers M. AMSTAR is a reliable and valid measurement tool to assess themethodological quality of systematic reviews. J Clin Epidemiol 2009;62:1013–20.

[78] Stanton TR, Hancock MJ, Maher CG, Koes BW. Critical appraisal of clinicalprediction rules that aim to optimize treatment selection for musculoskeletalconditions. Phys Ther 2010;90:843–54.

[79] Sterne JAC, Sutton AJ, Ioannidis JPA, Terrin N, Jones DR, Lau J, Carpenter J,Rucker G, Harbord RM, Schmid CH, Tetzlaff J, Deeks JJ, Peters J, Macaskill P,Schwarzer G, Duval S, Altman DG, Moher D, Higgins JPT. Recommendationsfor examining and interpreting funnel plot asymmetry in meta-analyses ofrandomised controlled trials. BMJ 2011;343:d4002.

[80] Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D,Becker BJ, Sipe TA, Thacker SB, Meta-analysis Of Observational Studies inEpidemiology (MOOSE) Group. Meta-analysis of observational studies inepidemiology: a proposal for reporting. JAMA 2000;283:2008–12.

[81] Sun X, Briel M, Busse JW, You JJ, Akl EA, Mejza F, Bala MM, Mertz D, Diaz-Granados N, Vandvik PO, Malaga G, Srinathan SK, Dahm P, Johnston BC,Alonso-Coello P, Hassouneh B, Walter SD, Heels-Ansdell D, Bhatnagar N,Altman DG, Guyatt GH. Credibility of claims of subgroup effects inrandomised controlled trials: systematic review. BMJ 2012;344:e1553.

[82] Teyhen DS, Williamson JN, Carlson NH, Suttles ST, O’Laughlin SJ, Whittaker JL,Goffar SL, Childs JD. Ultrasound characteristics of the deep abdominalmuscles during the active straight leg raise test. Arch Phys Med Rehabil2009;90:761–7.

[83] Tosteson TD, Buzas JS, Demidenko E, Karagas M. Power and sample sizecalculations for generalized regression models with covariate measurementerror. Stat Med 2003;22:1069–82.

[84] Turner EL, Perel P, Clayton T, Edwards P, Hernández AV, Roberts I, Shakur H,Steyerberg EW, CRASH trial collaborators. Covariate adjustment increasedpower in randomized controlled trials: an example in traumatic brain injury.J Clin Epidemiol 2012;65:474–81.

[85] Unsgaard-Tøndel M, Nilsen TIL, Magnussen J, Vasseljen O. Is activation oftransversus abdominis and obliquus internus abdominis associated withlong-term changes in chronic low back pain? A prospective study with 1-yearfollow-up. BMJ Sports Med 2012;46:729–34.

[86] van Tulder M, Furlan A, Bombardier C, Bouter L, Editorial Board of theCochrane Collaboration Back Review Group. Updated method guidelines forsystematic reviews in the Cochrane Collaboration Back Review Group. Spine2003;28:1290–9.

[87] van Tulder MW, Assendelft WJJ, Koes BW, Bouter LM. Method guidelines forsystematic reviews in the Cochrane Collaboration Back Review Group forSpinal Disorders. Spine 1997;22:2323–30.

[88] Vasseljen O, Fladmark AM, Westad C, Torp HG. Onset in abdominal musclesrecorded simultaneously by ultrasound imaging and intramuscularelectromyography. J Electromyogr Kinesiol 2009;19:e23–31.

[89] Vasseljen O, Unsgaard-Tøndel M, Westad C, Mork PJ. Effect of core stabilityexercises on feed-forward activation of deep abdominal muscles in chroniclow back pain: a randomized controlled trial. Spine 2012;37:1101–8.

[90] Verkerk K, Luijsterburg PA, Ronchetti I, Miedema HS, Pool-Goudzwaard A, vanWingerden JP, Koes BW. Course and prognosis of recovery for chronic non-specific low back pain: design, therapy program and baseline data of aprospective cohort study. BMC Musculoskelet Disord 2011;12:252.




































































































































































[91] Vos T, Flaxman AD, Naghavi M, Lozano R, Michaud C, Ezzati M, Shibuya K,Salomon JA, Abadalla S, Aboyans V, Abraham J, Ackeman I, Aggarwal R, AhnSY, Ali MK, Alvarado M, Anderson HR, Anderson LM, Andrews KG, Atkinson C,Baddour LM, Bahalim AN, Barker-Collo S, Barrero LH, Bartels DH, Basanez MG,Baxter A, Bell ML, Benjamin EJ, Bennett D, Bernabe E, Bhalla K, Bhandari B,Bikbov B, Abdulhak AB, Birbeck G, Black JA, Blencowe H, Blore JD, Blyth F,Bolliger I, Bonaventure A, Boufous S, Bourne R, Bossussinesq M, Braithwaite T,Brayne C, Bridgett L, Brooker S, Brooks P, Brugha TS, Bryan-Hancock C, BucelloC, Buchbinder R, Buckle G, Budke CM, Burch M, Burney P, Burstein R, CalabriaB, Campbell B, Canter CE, Carabin H, Carapetis J, Camona L, Cella C, Charlson F,Chen H, Cheung AT, Chou D, Chugh SS, Coffeng LE, Colan SD, Colquhoun S,Colson KE, Candon J, Connor MD, Cooper LT, Corriere M, Cortinovis M,Couville de Vaccaro K, Couser W, Cowie BC, Criqui MH, Cross M, DabhadkarKC, Dahiya M, Dahodwala N, Damsere-Derry J, Danaei G, Davis A, De Leo D,Degenhardt L, Dellavalle R, Delossantos A, Denenberg J, Derrett S, Jarlais DesDC, Dharmaratne SD, Dherani M, Diaz-Tome C, Dolk H, Dorsey ER, Driscoll T,Duber H, Ebel B, Edmond K, Elbaz A, Ali SE, Erskine H, Erwin PJ, Espindola P,Ewoigbokhan SE, Farazadfar F, Feigin V, Felson DT, Ferrari A, Ferri CP, FevreEM, Finucane MM, Flaxman S, Flood L, Foreman K, Forouzanfar MH, FowkesGR, Franklin R, Fransen M, Freeman MK, Gabbe BJ, Gabriel SE, Gakidou E,Ganatra HA, Garcia B, Gaspari F, Gillum RF, Gmel G, Gosselin R, Grainger R,Groeger J, Guillemin F, Gunnell D, Gupta R, Haagsma J, Hagan H, Halasa YA,Hall W, Haring D, Haro JM, Harrison JE, Havmoeller R, Hay RJ, Higashi H, HillC, Hoen B, Hoffman H, Hotez PJ, Hoy D, Huang JJ, Ibeanusi SE, Jacobsen KH,James SL, Javis D, Jarsrasaria R, Jayaraman S, Johns N, Jonas JB, Karthikeyan G,Kassebaum N, Kawakami N, Keren A, Khoo JP, King CH, Knowlton LM,Kobusingye O, Koranteng A, Krishnamurthi R, Lalloo R, Laslett LL, Lathlean T,Leasher JL, Lee YY, Leigh J, Lim SS, Limb E, Lin JK, Lipnick M, Lipshultz SE, LiuW, Loane M, Lockett Ohno S, Lyons R, Ma J, Mabweijano J, Maclntyre MF,Malekzadeh R, Mallinger L, Manivannan S, Marcenes W, March L, Margolis DJ,Marks GB, Marks R, Matsumori A, Matzopoulos R, Mayosi BM, McAnulty JH,McDermott MM, McGill N, McGrath J, Medina-Mora ME, Meltzer M, MensahGA, Merriman TR, Meyer AC, Miglioli V, Miller M, Miller TR, Mitchell PB,Mocumbi AO, Moffitt TE, Mokdad AA, Monsata L, Montico M, Moradi-LakehM, Moran A, Morawska L, Mori R, Murdoch ME, Mwaniki MK, Naidoo K, NairMN, Naldi L, Narayan KMV, Nelson PK, Nelson RG, Nevitt M, Newton CR, NolteS, Norman P, Norman R, O’Donnell M, O’Hanlon S, Olives C, Omer SB, OrtbladK, Osborne R, Ozgediz D, Page A, Pahari B, Pandian JD, Rivero AP, Patten SB,Pearce N, Padilla RP, Perez-Ruiz F, Perico N, Pesudovs K, Phillips D, PhillipsMR, Pierce K, Pion S, Polanczyk GV, Polinder S, Pope CA III, Popova S, Porrini E,Pourmalek F, Prince M, Pullan RL, Ramaiah KD, Ranganathan D, Razavi H,Regan M, Rehm JT, Rein DB, Remuzzi G, Richardson K, Rivara FP, Roberts T,Robinson C, Rodriguez De Leon F, Ronfani L, Room R, Rosenfeld LC, Rushton L,Sacco RL, Saha S, Sampson U, Sanchez-Riera L, Sanman E, Schwebel DC, ScottJG, Segui-Gomez M, Shahraz S, Shepard DS, Shin H, Shivakoti R, Singh D, Singh


�(2013), http:

GM, Singh JA, Singleton J, Sleet DA, Sliwa K, Smith E, Smith JL, Stapelberg NJC,Steer A, Steiner T, Stalk WA, Stovner LJ, Sudfeld C, Syed S, Tamburlini G,Tavakkoli M, Taylor HR, Taylor JA, Taylor WJ, Thomas B, Thomson WM,Thurston GD, Tleyjeh IM, Tonelli M, Towbin JA, Truelsen T, Tsilimbaris MK,Ubeda C, Undurranga EA, van der Werf MJ, van Os J, Vavilala MS,Venketasubramanian N, Wang M, Wang W, Watt K, Weatherall DJ,Weinstock MA, Weintraub R, Weisskopf MG, Weissman MM, White RA,Whiteford H, Wiersma ST, Wilkinson JD, Williams HC, Williams SRM, Witt E,Wolfe F, Woolf AD, Wulf S, Yeh PH, Zaidi AKM, Zheng ZJ, Zonies D, Lopez AD,Murray CJL. Years lived with disability (YLDs) for 1160 sequelae of 289diseases and injuries, 1990–2010: a systematic analysis for the Global Burdenof Disease Study 2010. Lancet 2012;380:2163–96.

[92] Wallwork TL, Stanton WR, Freke M, Hides JA. The effect of chronic low backpain on size and contraction of the lumbar multifidus muscle. Man Ther2009;14:496–500.

[93] Wessels T, van Tulder M, Sigl T, Ewert T, Limm H, Stucki G. What predictsoutcome in non-operative treatments of chronic low back pain? A systematicreview. Eur Spine J 2006;15:1633–44.

[94] Westad C, Mork PJ, Vasseljen O. Location and sequence of muscle onset indeep abdominal muscles measured by different modes of ultrasoundimaging. J Electromyogr Kinesiol 2010;20:994–9.

[95] Whisman MA, McClelland GH. Designing, testing, and interpretinginteractions and moderator effects in family research. J Fam Psychol2005;19:111–20.

[96] Whittaker JL, McLean L, Hodder J, Warner MB, Stokes MJ. Association betweenchanges in electromyographic signal amplitude and abdominal musclethickness in individuals with and without lumbopelvic pain. J OrthopSports Phys Ther 2013;43:466–77.

[97] Whittaker JL, Stokes M. Ultrasound imaging and muscle function. J OrthopSports Phys Ther 2011;41:572–80.

[98] Whittaker JL, Teyhen DS, Elliott JM, Cook K, Langevin HM, Dahl HH, Stokes M.Rehabilitative ultrasound imaging: understanding the technology and itsapplications. J Orthop Sports Phys Ther 2007;37:434–49.

[99] Whittaker JL, Warner MB, Stokes M. Comparison of the sonographic featuresof the abdominal wall muscles and connective tissues in individuals with andwithout lumbopelvic pain. J Orthop Sports Phys Ther 2013;43:11–9.

[100] Wilke HJ, Wolf S, Claes LE, Arand M, Wiesend A. Stability increase of thelumbar spine with different muscle groups. A biomechanical in vitro study.Spine 1995;20:192–8.

[101] Wong AYL, Parent E, Kawchuk G. Reliability of 2 ultrasonic imaging analysismethods in quantifying lumbar multifidus thickness. J Orthop Sports PhysTher 2013;43:251–62.

[102] Zielinski KA, Henry SM, Ouellette-Morton RH, DeSamo MJ. Lumbar multifidusmuscle thickness does not predict patients with low back pain who improvewith trunk stabilization exercises. Arch Phys Med Rehabil 2013;94:1132–8.




































Do various baseline characteristics of transversus abdominis and lumbar multifidus predict clinical...

Documents

Transcript of Do various baseline characteristics of transversus abdominis and lumbar multifidus predict clinical...