Accepted Manuscript

A clinical test of lumbopelvic control: Development and reliability of a clinical test ofdissociation of lumbopelvic and thoracolumbar motion

Edith Elgueta-Cancino, PT Siobhan Schabrun, BPhysio(Hons), PhD Lieven Danneels,PhD Paul Hodges, BPhty(Hons), PhD

PII: S1356-689X(14)00043-5

DOI: 10.1016/j.math.2014.03.009

Reference: YMATH 1545

To appear in: Manual Therapy

Received Date: 17 December 2013

Revised Date: 19 March 2014

Accepted Date: 21 March 2014

Please cite this article as: Elgueta-Cancino E, Schabrun S, Danneels L, Hodges P, A clinical test oflumbopelvic control: Development and reliability of a clinical test of dissociation of lumbopelvic andthoracolumbar motion, Manual Therapy (2014), doi: 10.1016/j.math.2014.03.009.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

A clinical test of lumbopelvic control: Development and reliability of a

clinical test of dissociation of lumbopelvic and thoracolumbar motion

1Edith Elgueta-Cancino (PT), 1Siobhan Schabrun, (BPhysio(Hons), PhD), 2Lieven

Danneels, (PT, PhD), 1Paul Hodges (BPhty(Hons), PhD)

1. The University of Queensland, Centre of Clinical Research Excellence in Spinal Pain,

Injury and Health, Brisbane, Queensland, Australia.

2. Ghent University, Department of Rehabilitation Sciences and Physiotherapy, Ghent,

Belgium.

Ethical review:

Approved by the Medical Research Ethics Committee of the University of Queensland.

Corresponding author:

Dr Paul Hodges

School of Health & Rehabilitation Sciences,

The University of Queensland, QLD 4072, Australia

phone: +61 7 3365 2008

e-mail: p.hodges@uq.edu.au

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

A clinical test of lumbopelvic control: Development and reliability of a clinical test

of dissociation of lumbopelvic and thoracolumbar motion

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Abstract 1

LBP is often associated with changes in motor control. Some subgroups of LBP have been 2

argued to have a compromised ability to dissociate lumbopelvic movement from that of the 3

thoracolumbar junction. Clinical methods to evaluate this task may aid identification of this LBP 4

subgroup and determine the utility of this information to guide clinical interventions. The study aimed 5

to develop a clinical test to assess the ability to dissociate lumbopelvic movement from that of the 6

thoracolumbar junction, and to evaluate the inter-rater reliability of the test in individuals with and 7

without low back pain (LBP) when performed by experienced and novice therapists. A clinical scale 8

was developed to characterise quality of performance of lumbopelvic motion with limited motion at the 9

thoracolumbar junction. Inter-tester repeatability was measured in three experiments. Test outcomes for 10

pain-free controls were compared between three assessors with different amounts of clinical 11

experience. Test scores for LBP participants were compared between two assessors, and between 12

assessments undertaken from video recordings. Agreement between assessors was tested with weighted 13

Kappa Coefficient. The test had acceptable reliability in pain-free and LBP participants, but was better 14

when undertaken by experienced therapists. Kappa index ranged from 0.81-0.66 for live assessments, 15

and 0.62 for video assessments. The results showed that the test is reliable when performed by 16

experienced assessors. The test can assess thoracolumbar movements in different groups of individuals. 17

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Introduction 18

Individuals with persistent low back pain (LBP) exhibit changes in motor behaviour. These 19

include changes in posture (Claus et al., 2009; Consmuller et al., 2012; Dankaerts et al., 2006b; Roussouly et 20

al., 2005), muscle activation (Dankaerts et al., 2006a; van Dieën et al., 2003), and movement (MacDonald et 21

al., 2009; Mok et al., 2011; Scholtes et al., 2009; Wong & Lee, 2004). Such changes have been used as a 22

basis to identify homogeneous subgroups within the heterogeneous LBP population (Dankaerts & 23

O'Sullivan, 2011; Dankaerts et al., 2007; Dankaerts et al., 2009; Fersum et al., 2009; Harris-Hayes & Van 24

Dillen, 2009). However, examination of motor behaviour often requires sophisticated laboratory tests 25

that have limited utility in clinical practice. A critical step forward is the development of clinically 26

viable methods to identify changes in motor behaviour in LBP. 27

Altered motor coordination of thoracic, lumbar, and pelvic segments is a feature of motor 28

behaviour discussed in the clinical literature and studied using laboratory-based investigations. Specific 29

subgroups of individuals with LBP have been identified to have: poor ability to dissociate movement of 30

the lumbopelvic region spine from that of the thorax (poor intersegmental coordination) (Wallwork et al., 31

2009); poor capacity to voluntarily maintain a lumbar lordosis (Danneels et al., 2002); compromised 32

control of the multifidus muscle (Freeman et al., 2010; MacDonald et al., 2009; Nelson-Wong & Callaghan, 33

2010); and “smudging” of the separate regions of the motor cortex involved in the activation of 34

short/deep and long/superficial muscles of the back which are responsible for coordinated motion of 35

these segments (Tsao et al., 2011). Although related, it has not been confirmed that all of these features 36

are present in the same individuals. Altered control of this aspect of motor behaviour may be relevant 37

for spine loading and characterise a specific LBP subgroup, but no clinical tools are available to 38

reliably measure dissociation of lumbopelvic from thoracic motion. Clinical methods are necessary to 39

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

enable wider evaluation of this movement behaviour in a broad LBP population, and to determine the 40

utility of this information to guide clinical interventions. 41

This study aimed to: (i) develop a clinical test to assess the ability to dissociate lumbopelvic 42

motion from that of the thoracolumbar junction; and (ii) evaluate the reliability of the test in individuals 43

with and without LBP when performed by experienced and novice therapists. 44

45

Methods 46

Study design 47

The study was divided into a test-development phase and three experiments. The test-48

development phase involved development and refinement of criteria to quantify movement 49

performance, and design of a standardised video to instruct participants for the movement task. The 50

first experiment investigated the inter-rater reliability of the clinical test when applied to healthy 51

participants and compared the ratings of three assessors; two with at least 1-year of clinical 52

physiotherapy experience and one novice physiotherapist (a new graduate with limited clinical 53

experience). The second experiment assessed the inter-rater reliability (two experienced assessors) in 54

patients with persistent LBP. A third experiment studied the inter-rater reliability (two experienced 55

assessors) of the test when assessments were undertaken from video recordings in the same group of 56

patients. 57

Test-Development 58

The development of the test involved a combination of judgmental and consensus opinion 59

methodology. The initial judgmental approach involved development of the criteria by the investigators on 60

the basis of an extensive literature review of the topic and through observation of patients with back pain. 61

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

This test was then refined using a consensus opinion method in which a panel of expert clinicians was 62

involved in testing and commenting on the appropriateness of the criteria and scoring descriptions. The 63

expert clinicians (n=3) involved in this phase were all experienced with rehabilitation of patients with back 64

pain. 65

Rating Scale 66

The “Clinical test of thoracolumbar dissociation” was devised to assess a patient’s ability to 67

perform anterior/posterior pelvic tilt in sitting while attempting to maintain a constant position of the 68

thorax/thoracolumbar junction. The rating scale of the test was structured with five criteria with a 69

maximum score of 10. Each criterion represented descriptive characteristics of a specific movement 70

feature. Assessors provided a numeric value, based on a criterion referenced assessment, as a measure 71

of the quality of movement performance using characteristics such as; muscle activity, timing and 72

coordination (Table1, Fig. 1). Criteria were arranged such that the assessor first focused attention to 73

lumbopelvic movement. The second focus was motion and muscle activity of adjacent regions, 74

including consideration of the ability of the participant to move the lumbopelvic region relatively 75

independently from the thoracolumbar junction. A final focus included more global features of 76

movement performance including; (i) consideration of whether the task could be performed with equal 77

quality between movement directions (anterior and posterior pelvic tilt), (ii) the ability to maintain 78

movement quality with repetition, and (iii) the ability to maintain breathing. 79

Process for test refinement 80

During the consensus opinion phase the panel of expert clinicians piloted the test and provided 81

feedback. This resulted in refinement of the wording of the criteria and the assessment sequence. 82

Factors considered in refinement of the test were that: (i) criteria should create a fluid and logical order 83

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

of assessment; (ii) discrimination of the quality of the movement during the test should be expressed as 84

a positive concept with different levels of performance, and (iii) language should be intuitive and non-85

ambiguous. This process resulted in modification of descriptions that were considered to be ambiguous, 86

addition of words to add clarity. 87

Test performance 88

Performance of the test was evaluated in two steps. The first evaluation was undertaken after 89

viewing standardised instruction video, and the second after a two minute individualised training session 90

with specific feedback of features that required correction to improve the performance of the task.. 91

Standardised video instructions: Instructions for each participant were standardised using a 92

video that contained; (i) written and verbal instructions of the task “tilt the pelvis forwards and 93

backwards in the sitting position”, (ii) demonstration of the movement by a model, and (iii) instruction 94

to perform 10 repetitions. After this training the assessor rated the performance. 95

Individualised training: Participants received a two-minute session of standardised feedback 96

or training to improve performance by a single trained experimenter (physiotherapist). To standardize 97

and optimise the learning process a list of “training strategies” was created. This guide indicated 98

training methods that could be used to improve task performance based on errors identified in the initial 99

assessment. Techniques included training of postural alignment in sitting, standing or four point 100

kneeling, manual guidance of pelvic tilt, guidance to correct alignment of the pelvis, and feedback to 101

avoid excessive muscle activity. Performance was reassessed after the completion of the two minutes of 102

training. 103

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Experiment 1: Inter-tester reliability – Healthy controls 104

Participants: Eleven healthy individuals with no history of low back or pelvic pain (27[3] 105

years (mean [standard deviation]); 6 males, 5 females) were recruited. Participants were excluded if 106

they had suspected or confirmed spinal pathology, pregnancy, nerve root compromise, and previous 107

spinal surgery, or major surgery scheduled. The institutional Medical Research Ethics Committee 108

approved the study and all participants provided written, informed consent. 109

Assessors: Three independent assessors, one with postgraduate training in manual therapy (R1), 110

one working in clinical practice (R2, >1-year of clinical experience), and a novice new graduate 111

physiotherapist with limited clinical experience (R3) were trained to perform the assessment. Training 112

involved: (i) explanation and discussion of the criteria; (ii) familiarization with key concepts and 113

definitions, (iii) guided assessment of the test in pilot subjects, and (iv) review and discussion of the 114

rating of several individuals. 115

Procedure: The assessment was performed in one session for each participant. Each session 116

was divided into three phases: (i) pre-training assessment – basic training of the movement task 117

provided using the video and performance simultaneously assessed by 3 assessors; (ii) training - 2-118

minutes of standardised training provided by the experimenter based on the standardised training 119

strategies; and (iii) post-training assessment – same procedure as the pre-training assessment. 120

Participants sat on a chair adjusted to 110% of the height of fibula head with the hands relaxed 121

on the thighs, eyes open and looking forward to a screen. The back was exposed and the participant 122

wore shorts that did not restrict hip motion. 123

The test was performed once (ten repetitions of the test movement) with all assessors able to 124

evaluate the same participant at the same time. Assessors rated the participant’s best performance, 125

without interaction. The best performance was defined as the highest score that could be awarded based 126

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

on the assessed repetitions. Assessors were allowed and encouraged to palpate the participant to clarify 127

observations (muscle activation). It was considered necessary for all assessors to evaluate the same 128

performance so that inter-rater reliability could be determined without confounding by differences in 129

performance by the participant. 130

Assessors left the room after pre-training assessment and the experimenter provided 2-minutes 131

of training. All assessors then returned and scored the post-training performance concurrently. 132

Experiment 2: Inter-tester reliability – LBP participants 133

Participants: Twenty individuals with non-specific persistent LBP were recruited (31[9] years, 6 134

males, 14 females). Participants were excluded if they had suspected or confirmed spinal pathology of 135

non-musculoskeletal origin (e.g., tumor, infection, fracture, inflammatory disease), pregnancy, nerve 136

root compromise, or previous spinal surgery. 137

Assessors and Procedure: The experienced assessors from Experiment 1 (R1, R2) assessed the 138

participants as for Experiment 1. 139

Experiment 3: Inter-tester reliability - LBP participants by video assessment 140

Participants: Sixteen individuals with non-specific persistent LBP selected randomly from the 141

sample used in Experiment 2 were videoed from the front, side and back during the task, pre- and post-142

training. A research assistant edited the videos in a blinded manner. Participants were excluded if the 143

movement could not be observed clearly in all three views. Video recordings of ten individuals (28[9] 144

years, 2 males, 8 females) could be included. 145

Assessors and procedure: An experienced assessor (R1) from Experiment 1 and an additional 146

experienced assessor (R4) with postgraduate training in manual therapy participated. Assessors were 147

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

instructed to rate the performance after watching the video only once. They rated the pre-training video, 148

waited 2 minutes, and then rate the post-training performance to mimic the clinical situation. 149

Intra-tester reliability 150

Intra-tester reliability was assessed using the data for assessor R1. It was considered that 151

assessment of intra-tester reliability would be confounded by improvement in performance of the 152

participant after training. For this reason intra-tester reliability was assessed by comparison of the scores 153

provided by a single assessor on separate days, but with one assessment performed as a live assessment, 154

and the second assessment as a video assessment of the exact same performance by the participant. This 155

was achieved by using the data from assessor R1 in Experiments 2 and 3. 156

Statistical Analyses 157

Inter-rater reliability was assessed in two ways: general analysis of the total score, and detailed 158

analysis of each criterion separately. For both analysis the scores (total and each criterion) of the test 159

pre-training and post-training were compared between each combination of assessor pairs (R1 vs. R2; 160

R1 vs. R3; R2 vs. R3) using the weighted Cohen's Kappa Coefficient. This method measures the 161

agreement between assessors for categorical/ordinal data. Values of this index (k) range from 0 (no 162

agreement) to 1 (100% agreement) or -1 (100% disagreement) and the strength of agreement has been 163

defined by Landis and Koch (Landis & Koch) and Sim and Wright (Sim & Wright, 2005) (see Table 2). 164

Analyses were performed with a confidence interval (CI) of 95%. 165

166

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Results 167

Inter-tester reliability – Healthy controls 168

The repeatability of the total scores revealed good agreement between assessors but differences 169

in the level of agreement between pairs of assessors based on experience and between measures made 170

before- and after-training. For pre-training measures the inter-rater agreement between R1 and R2 was 171

“almost perfect/very good”, but was less (“substantial/good”) for comparison between experienced 172

assessors (R1 and R2) and the less experienced assessor (R3) (Table 3). Agreement between assessors 173

was less after 2 minutes training. Agreement between R1 and R2, and R1 and R3 was “substantial” but 174

only “moderate” between R2 and R3 (Table 3). 175

Analysis of inter-rater reliability for each criterion revealed higher variability, but with the same 176

tendency for less reliable post-training measures and lower agreement with the less experienced 177

assessor. “Breathing”, was scored identically by all assessors at both measurement times. In the pre-178

training measures, all criteria showed at least a “moderate” level of agreement for the assessor with 179

postgraduate training (R1) and the other two assessors (R2 and R3) (Table 4). For the post-training 180

measures “moderate” agreement was only achieved for the experienced assessors (R1 and R2). 181

Inter-rater reliability – LBP participants 182

Analysis of total scores for LBP participants demonstrated “substantial/good” inter-rater 183

reliability in the pre-training phase and “moderate” agreement in the post-training phase (Table 3). 184

Analysis of individual criteria showed lower agreement for pre- and post-training assessments. 185

Three criteria had “moderate” to “very good” agreement in the pre-training phase, but higher variability 186

in the post-training phase (Table 4). “Breathing” was scored with perfectly agreement between 187

assessors. 188

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Inter-rater reliability – LBP participants by video assessment 189

Inter-rater reliability of the total scores from video analysis revealed that assessment remains 190

reliable with a “substantial” agreement in pre-training and “moderate” agreement post-training (Table 191

3). Analysis of each criterion demonstrated at least a “moderate” agreement in four of five criteria. 192

Again “breathing” was perfectly scored. This was also apparent for Criterion 2 (control of adjacent 193

regions) in the pre-training phase and Criterion 5 (repetition) in the post-training phase. One criterion 194

(Criterion 3 (performance in opposite directions)) was less reliable, with “slight” and “poor” levels of 195

agreement pre- and post-training, respectively (Table 4). 196

Intra-tester reliability 197

The intra-tester reliability of the total scores of R1 from LBP participants in vivo and video 198

analysis revealed a “substantial” and “moderate” agreement between scores provided pre-training and 199

post-training, respectively (Table 5). 200

201

Discussion 202

The results demonstrate that the “Clinical test of thoracolumbar dissociation” has acceptable 203

inter-rater reliability when used by trained physiotherapists. Although the agreement between assessors 204

was acceptable, it depended on the assessor’s experience and the degree of task training. Agreement 205

between assessors was better when participants’ performance was judged after minimal instruction than 206

after the training period. The reliability was also acceptable, although with marginally less agreement, 207

for participants with LBP, even when the test was undertaken using video recordings of movement. 208

These data confirm the potential utility of this test in clinical evaluation of patients with LBP. 209

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Intra and inter-rater reliability of the “Clinical test of thoracolumbar dissociation” 210

The total test score had good reliability between sessions for a single trained assessor, and 211

between assessors, but with some variation between criteria. This latter observation could be explained 212

by differences in the components evaluated in each criterion. For instance, Criterion 4, which assesses 213

breathing, had a perfect agreement. This criterion is rated as a dichotomous “yes” or “no” pertaining to 214

the ability to breathe during the movement, which is relatively straightforward to assess. High 215

agreement was also identified for Criteria 1 and 2, which describe key movement components of the 216

test and are defined by detailed criteria that are assessed from the “best” performance of the task. In 217

contrast, agreement between assessors was lower for Criteria 3 and 5 (healthy control – “moderate”; 218

LBP – “slight” – “poor”), which assess differences in the performance of the task between directions of 219

movement (i.e. anterior and posterior pelvic tile) and consistency of the ability to perform the task with 220

repetition. Greater difficulty with assessment of these criteria might reflect subjectivity of assessment 221

of features that rely on relative quality between directions/repetitions. In addition, the determination of 222

quality of performance in these criteria involves consideration of multiple components including 223

movement features and muscle activation that are separately assessed in Criteria 1 and 2. 224

Greater reliability for experienced assessors 225

Our data show a “high” level of agreement (Landis et al.; Sim et al., 2005) between assessors who 226

have several years of clinical experience and less agreement between assessors when one is a novice 227

clinician. This concurs with data from other studies (Alqarni, 2011; Dankaerts et al., 2011; Dankaerts et al., 228

2006b; Dankaerts et al., 2006c; Luomajoki et al., 2007; 2008; Tidstrand & Horneij, 2009; Van Dillen et al., 229

1998) and the conclusion of a recent systematic review that identified assessor experience as a key 230

factor in determination of reliability and minimization of the risk of bias for clinical assessments of 231

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

movement control in LBP (Alqarni, 2011; Carlsson & Rasmussen-Barr, 2013). Our study supports this 232

observation and indicates that the reliability of the “Clinical test of thoracolumbar dissociation” 233

depends on the experience/skill of the assessors. Although some studies suggest experience is not 234

important, this is generally identified for tests that rely on simple criteria that involve limited clinical 235

judgment such as examination forms and videoed assessments (Fritz et al., 2000) 236

Assessment can be biased by training of the task 237

The two minutes of standardized training appears to have differentially biased the observations 238

of the assessors. This is evidenced by greater agreement between assessors for assessment conducted 239

after the video training than for the second assessment after training by the therapist. This cannot be 240

explained by differences in task performance as the assessors examined patients concurrently. The 241

difference in rating after training indicates additional factors influenced the assessor’s judgment. There 242

are several possibilities. First, assessors may have been influenced by an expectation of improved 243

performance after training. Some assessors may have rated performance poorer if they expected greater 244

improvement from training. Second, recollection of the pre-training assessment and subsequent 245

comparison with the post-training performance may have affected the rating. The 2-minute period 246

between assessments is unlikely to prevent assessors being influenced by their initial assessment. Other 247

studies have included a 15 minute “washout” period and reported this time period to be sufficient to 248

reduce the influence on the assessment (Dankaerts et al., 2011). 249

Reduced agreement (change of Kappa value to a lower category of strength of agreement or 250

lower Kappa value with in a strength of agreement category) after training was limited to Criterion 3 251

(ability to perform the task in opposite directions) for healthy controls. For LBP participants, agreement 252

between assessors and between sessions for a single assessor declined for both Criteria 3 and 5 after 253

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

training. As indicated above, these scores involve assignment of a score on the basis of comparison of 254

performance between movement directions, and the maintenance of quality of performance between 255

repetitions. Different assessors not only score individuals differently on these items, but agreement on 256

these items declined after training. As these features require reflection of performance over a number of 257

repetitions and require some qualitative judgement, they may be influenced to a greater degree by the 258

expectation of the assessor, and this may differ between assessors and between patients. Alternatively, 259

it is possible that although training improved performance, variation between repetitions increased as 260

participants trialled various methods they had been taught to optimize task performance. This may have 261

been expressed as an initial deterioration of performance. As people with LBP have less precise control 262

of movement (Willigenburg et al., 2013) the attempt to voluntarily correct or improve movement based 263

on learning of a new task could have led to disorganization of the movement as they attempted to 264

modify their established pattern. Previous studies have shown that specific training of voluntary actions 265

and skills change motor patterns (Adkins et al., 2005; Laird et al., 2012; Perez et al., 2004) of muscle 266

coordination (Jensen, 2005; van Dieën et al., 2003) even after a short period of training time (Tsao et al., 267

2010). However, the short duration of practice is unlikely to be sufficient to change behaviour and the 268

mode of training might have interfered with the initial performance, as has been shown for some 269

training paradigms (Shea & Morgan, 1979). Reliability could be improved in future by; (i) use of only 270

the pre-training measure, (ii) provision of a longer period between assessments, (iii) greater period of 271

training longer to allow performance to stabilise, or (iv) tighter definition of Criteria 3 and 5. 272

Inter-rater reliability for LBP vs. healthy control participants 273

Although inter-rater reliability for both healthy and LBP groups was acceptable, agreement was 274

better for the healthy group. It is possible that worse agreement for the LBP group was the result of 275

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

greater variability in task performance by the participants, leading to different scores by the assessors 276

despite both scoring the performance at the same time. However, other possible explanations require 277

consideration. In this study the assessors knew that the participants had LBP and this may have biased 278

assessment in a differential manner. Perhaps some therapists rated LBP participants differently if they 279

“expected” the performance to be better or worse rather than attributing a rating purely based on 280

observation. In some ways this concurs with the observation that agreement was worse after the 2-281

minutes training for the pain-free participants. This implies some potential for differential bias of 282

therapists based on expectation. Despite this possibility, the ratings maintained acceptable reliability, 283

but future use of the test should highlight that this may be possible and assessors should be encouraged 284

to faithfully report the performance without bias. 285

Reliability of video assessments 286

Reliability of many clinical tests has been evaluated using video analysis (Henry, 2013, 287

Luomajoki, 2007). Videoed assessment has advantages as it allows the test to be viewed by multiple 288

assessors, at any time, under identical conditions with a single performance by the participant. A 289

disadvantage is that it may reduce the quality of clinical judgment, as the assessor is not able to interact 290

with the participant. This is important for the clinical test under investigation as one component of the 291

criteria involved the qualification of muscle activity and palpation of muscle could provide useful 292

information. Despite this limitation, agreement between assessors was acceptable. The small reduction 293

in reliability is explained by either the absence of interaction with the participant or because assessors 294

were unable to see the features of the task performance with sufficient detail as a result of the video 295

recording. 296

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Intra-tester reliability 297

Intra-rater reliability is a complex measure for clinical test that incorporate training in their format, 298

as the patient's performance is likely to improve with the training component. So to more purely evaluate 299

the contribution of measurement on different days but without the variation in patient performance we did 300

it by comparison of the measures of the exact same trials but first live and second by video. Even though we 301

expect that there will be some loss of accuracy by the different modes we found good intra-rater reliability, 302

which indicate that test has consistency even when it is used by different approaches. 303

Clinical issues/implications 304

The test described here is the first to assess the ability to dissociate movement of the 305

lumbopelvic region from that of the thoracolumbar region. The reliability of the test confirms it may be 306

used clinically to aid identification of subgroups within the LBP population. Further work is required to 307

determine whether subgrouping based on this test, and subsequent directed treatment, influences 308

clinical outcomes; whether the measure can be changed with treatment; whether the movement pattern 309

relates to other motor control parameters such as muscle activation; and whether the test can 310

discriminate individuals in this subgroup from the healthy population. 311

312

313

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

17

References 314

Adkins, DL, Boychuk, J, Remple, MS, Kleim, JA. Motor training induces experience-315

specific patterns of plasticity across motor cortex and spinal cord. Journal of 316

Applied Physiology 2005; 101(6): 1776-1782. 317

Alqarni, AM. Clinical Tests to Diagnose Lumbar Segmental Instability: A Systematic 318

Review. Journal of Orthopaedic and Sports Physical Therapy 2011. 319

Carlsson, H, Rasmussen-Barr, E. Clinical screening tests for assessing movement 320

control in non-specific low-back pain. A systematic review of intra- and inter-321

observer reliability studies. Manual therapy 2013; 18(2): 103-110. 322

Claus, AP, Hides, JA, Moseley, GL, Hodges, PW. Is ‘ideal’ sitting posture real?: 323

Measurement of spinal curves in four sitting postures. Manual therapy 2009; 324

14(4): 404-408. 325

Consmuller, T, Rohlmann, A, Weinland, D, Druschel, C, Duda, G, Taylor, W. 326

Comparative evaluation of a novel measurement tool to assess lumbar spine 327

posture and range of motion. European Spine Journal 2012; 21(11): 2170-2180. 328

Dankaerts, W, O'Sullivan, P. The validity of O'Sullivan's classification system (CS) for a 329

sub-group of NS-CLBP with motor control impairment (MCI): Overview of a 330

series of studies and review of the literature. Manual therapy 2011; 16(1): 9-14. 331

Dankaerts, W, O'Sullivan, P, Burnett, A, Straker, L. Altered patterns of superficial trunk 332

muscle activation during sitting in nonspecific chronic low back pain patients: 333

importance of subclassification. Spine 2006a; 31(17): 2017-2023. 334

Dankaerts, W, O'Sullivan, P, Burnett, A, Straker, L. Differences in sitting postures are 335

associated with nonspecific chronic low back pain disorders when patients are 336

subclassified. Spine 2006b; 31(6): 698-704. 337

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

18

Dankaerts, W, O'Sullivan, P, Burnett, A, Straker, L. The use of a mechanism-based 338

classification system to evaluate and direct management of a patient with non-339

specific chronic low back pain and motor control impairment-a case report. Man 340

Ther 2007; 12: 181 - 191. 341

Dankaerts, W, O'Sullivan, P, Burnett, A, Straker, L, Davey, P, Gupta, R. Discriminating 342

Healthy Controls and Two Clinical Subgroups of Nonspecific Chronic Low 343

Back Pain Patients Using Trunk Muscle Activation and Lumbosacral Kinematics 344

of Postures and Movements A Statistical Classification Model. Spine 2009; 345

34(15): 1610-1618. 346

Dankaerts, W, O'Sullivan, PB, Straker, LM, Burnett, AF, Skouen, JS. The inter-examiner 347

reliability of a classification method for non-specific chronic low back pain 348

patients with motor control impairment. Manual therapy 2006c; 11(1): 28-39. 349

Danneels, LD, Coorevits, PC, Cools, AC, Vanderstraeten, GV, Cambier, DC, Witvrouw, 350

EW, De Cuyper, HDC. Differences in electromyographic activity in the 351

multifidus muscle and the iliocostalis lumborum between healthy subjects and 352

patients with sub-acute and chronic low back pain. European Spine Journal 353

2002; 11(1): 13-19. 354

Fersum, KV, Sullivan, PBO, Kvaringle, A, Skouen, JS. Inter-examiner reliability of a 355

classification system for patients with non-specific low back pain. Manual 356

therapy 2009; 14(5): 555-561. 357

Freeman, M, Woodham, M, Woodham, A. The role of the lumbar multifidus in chronic 358

low back pain: a review. Pm & R 2010; 2(2)): 142 - 146. 359

Fritz, JM, Delitto, A, Vignovic, M, Busse, RG. Interrater reliability of judgments of the 360

centralization phenomenon and status change during movement testing in 361

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

19

patients with low back pain. Archives of Physical Medicine and Rehabilitation 362

2000; 81(1): 57-61. 363

Harris-Hayes, M, Van Dillen, LR. The Inter-Tester Reliability of Physical Therapists 364

Classifying Low Back Pain Problems Based on the Movement System 365

Impairment Classification System. PM&R 2009; 1(2): 117-126. 366

Henry, SM, Van Dillen, LR, Trombley, AR, Dee, JM, Bunn, JY. Reliability of novice 367

raters in using the movement system impairment approach to classify people 368

with low back pain. Manual therapy 2013; 18(1): 35-40. 369

Jensen, JL. Motor skill training and strength training are associated with different plastic 370

changes in the central nervous system. Journal of Applied Physiology 2005; 371

99(4): 1558-1568. 372

Laird, R, Kent, P, Keating, J. Modifying patterns of movement in people with low back 373

pain -does it help? A systematic review. BMC Musculoskeletal Disorders 2012; 374

13(1): 169. 375

Landis, JR, Koch, GG. The Measurement of Observer Agreement for Categorical Data. 376

Biometrics 1977; 33(1): 159-174. 377

Luomajoki, H, Kool, J, de Bruin, ED, Airaksinen, O. Reliability of movement control 378

tests in the lumbar spine. BMC Musculoskeletal Disorders 2007; 8(1): 90. 379

Luomajoki, H, Kool, J, de Bruin, ED, Airaksinen, O. Movement control tests of the low 380

back; evaluation of the difference between patients with low back pain and 381

healthy controls. BMC Musculoskeletal Disorders 2008; 9(1): 170. 382

MacDonald, D, Moseley, GL, Hodges, PW. Why do some patients keep hurting their 383

back? Evidence of ongoing back muscle dysfunction during remission from 384

recurrent back pain. Pain 2009; 142(3): 183-188. 385

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

20

Mok, NW, Brauer, SG, Hodges, PW. Changes in Lumbar Movement in People With 386

Low Back Pain Are Related to Compromised Balance. Spine 2011; 36(1): E45-387

E52. 388

Nelson-Wong, E, Callaghan, JP. Is muscle co-activation a predisposing factor for low 389

back pain development during standing? A multifactorial approach for early 390

identification of at-risk individuals. Journal of Electromyography and 391

Kinesiology 2010; 20(2): 256-263. 392

Perez, MA, Lungholt, BKS, Nyborg, K, Nielsen, JB. Motor skill training induces 393

changes in the excitability of the leg cortical area in healthy humans. 394

Experimental Brain Research 2004; 159(2): 197-205. 395

Roussouly, P, Gollogly, S, Berthonnaud, E, Dimnet, J. Classification of the normal 396

variation in the sagittal alignment of the human lumbar spine and pelvis in the 397

standing position. Spine 2005; 30(3): 346-353. 398

Scholtes, SA, Gombatto, SP, Van Dillen, LR. Differences in lumbopelvic motion 399

between people with and people without low back pain during two lower limb 400

movement tests. Clinical Biomechanics 2009; 24(1): 7-12. 401

Shea, JB, Morgan, RL. Contextual Interference Effects on the Acquisition, Retention, 402

and Transfer of a Motor Skill. Journal of Experimental Psychology: Human 403

Learning and Memory 1979; 5(2): 179-87. 404

Sim, J, Wright, CC. The Kappa Statistic in Reliability Studies: Use, Interpretation, and 405

Sample Size Requirements. Physical Therapy 2005; 85(3): 257-268. 406

Tidstrand, J, Horneij, E. Inter-rater reliability of three standardized functional tests in 407

patients with low back pain. BMC Musculoskeletal Disorders 2009; 10(1): 58. 408

Tsao, H, Danneels, LA, Hodges, PW. ISSLS prize winner: smudging the motor brain in 409

young adults with recurrent low back pain. Spine 2011; 36(21): 1721-1727. 410

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

21

Tsao, H, Druitt, TR, Schollum, TM, Hodges, PW. Motor Training of the Lumbar 411

Paraspinal Muscles Induces Immediate Changes in Motor Coordination in 412

Patients With Recurrent Low Back Pain. The Journal of Pain 2010; 11(11): 413

1120-1128. 414

van Dieën, JH, Selen, LPJ, Cholewicki, J. Trunk muscle activation in low-back pain 415

patients, an analysis of the literature. Journal of Electromyography and 416

Kinesiology 2003; 13(4): 333-351. 417

Van Dillen, LR, Sahrmann, SA, Norton, BJ, Caldwell, CA, Fleming, DA, McDonnell, 418

MK, Woolsey, NB. Reliability of Physical Examination Items Used for 419

Classification of Patients With Low Back Pain. Physical Therapy 1998; 78(9): 420

979-988. 421

Wallwork, T, Stanton, W, Freke, M, Hides, J. The effect of chronic low back pain on 422

size and contraction of the lumbar multifidus muscle. Man Ther 2009; 14(5): 423

496 - 500. 424

Willigenburg, NW, Kingma, I, Hoozemans, MJM, van Dieën, JH. Precision control of 425

trunk movement in low back pain patients. Human Movement Science 2013. 426

Wong, TKT, Lee, RYW. Effects of low back pain on the relationship between the 427

movements of the lumbar spine and hip. Human Movement Science 2004; 428

23(1): 21-34. 429

430

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Acknowledgements

The authors thank Annina Schmidt, David Klyne and Michael Bergin for their

assistance with data collection and Asad Khan for his assistance with statistical

analysis.

Funding:

This project was supported by funds from the National Health and Medical

Research Council of Australia (Fellowships [PH: APP1002190, SS: ID631612],

Program Grant [ID631717]).

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

1

Table1. Rating scale for the “Clinical test of thoracolumbar dissociation”

Criteria Initial performance Initial performance

1. Quality of pelvic motion

– direction with best

quality

Moves the pelvis in to

anterior/posterior tilt in

manner that is smooth,

controlled and to the end

3 3

Moves the pelvis but not

ideal in terms of

smoothness, range or with

too much muscle activity

2 2

Jerky, uncoordinated,

limited motion of the

pelvis +/- excessive

muscle activity

1 1

Inability to move the

pelvis into anterior or

posterior tilt

0 0

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

2

2. Control of adjacent

regions – direction with

best quality

Moves pelvis with gentle

lordosis/kyphosis of the

lumbar spine but limited

movement at the

3 3

Moves pelvis mild

thoracolumbar extension

or activity of

thoracolumbar erector

2 2

Moves pelvis but with

excessive thoracolumbar

extension and activation

of thoracolumbar erector

1 1

Moves into thoracolumbar

extension/flexion with no

motion of pelvis

0 0

3. Directional preference

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

3

Moves with high quality a

into anterior and posterior

tilt

2 2

Moves with high quality

in one direction 1 1

Does not moves in high

quality in either direction 0 0

4. Breathing

Able to maintain quiet

breathing during the task 1 1

Unable to maintain quiet

breathing during the task 0 0

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

4

5. Repetition

Able to perform 10

repetitions with high

quality a

1 1

Able to perform some

(between 1-10) repetitions

with higha quality

0.5 0.5

Unable to perform any

repetitions with high

quality a

0 0

TOTAL SCORE /10 /10

a High quality pelvic control

- Moves into

anterior/posterior tilt in

manner that is initiated with

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

5

Table 2. Strength of agreement for the Kappa coefficients (Landis et al., 1977)

Strength of agreement Kappa

coefficients

Almost perfect/Very Good 0.81 - 1

Substantial/Good 0.61 – 0.80

Moderate 0.41 – 0.60

Fail 0.21 – 0.40

Slight 0.01- 0.20

Poor < 0

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

6

Table 3. Inter-rater agreements of total test scores in Control and LBP participants

Assessors Pre-training Post-training

k 95%CI k 95%CI

Control

R1/R2 0.81 (0.68 - 0.93) 0.75 (0.29 - 0.95)

R1/R3 0.70 (0.34 - 0.93) 0.71 (0.31 - 0.95)

R2/R3 0.63 (0.29 - 0.89) 0.48 (-0.05 - 0.92)

LBP

R1/R2 0.66 (0.33 - 0.84) 0.51 (-0.03 - 0.79)

LBP (Video)

R1/R4 0.62 (0.19-0.81) 0.60 (0.11-0.90)

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

7

Table 4. Inter-rater agreement of each criteria score for control and LBP participants

Control Control Control LBP LBP

(Video)

Assessors R1/R2 R1/R3 R2/R3 R1/R2 R1t/R4

Criteria 1 Pre k 0.94 0.77 0.66 0.60 0.44

CI (0.77-1.00) (0.441-0.97) (0.37 - 0.87) (0.07 - 0.82) (0.00-1.00)

Criteria 1 Post k 0.65 0.76 0.49 0.63 0.67

CI (0.24 -0.89) (0.3 -1.00) (0.06 - 0.78) (0.11 - 0.87) (0.00-0.93)

Criteria 2 Pre

k 0.43 0.70 0.09 0.57 NaN

CI (0.0 - 1.0) (0.27 - 0.92) (-0.71 - 0.89)

(0.130 -

0.85) NaN

Criteria 2 Post k 0.61 0.18 0.35 0.34 0.68

CI (0.12 - 0.93) (-0.34 - 0.60) (-0.07 - 0.67) (0.12 - 0.54) (0.17-1.00)

Criteria 3 Pre k 0.57 0.58 0.44 0.15 0.29

CI (0.18- 0.87) (0.17 - 0.83) (-0.11 - 0.87) (-0.14 - 0.43) (-0.46-1.0)

Cirteria 3 Post

k 0.40 0.52 0.36 0.41 0.06

CI (0.05 - 0.86) (0.22 - 0.86) (0.00 - 0.75) (0.00 - 0.76)

(-0.23-

0.36)

Criteria 4 Pre k NaN NaN NaN NaN NaN

CI NaN NaN NaN NaN NaN

Criteria 4 Post k NaN NaN NaN NaN NaN

CI NaN NaN NaN NaN NaN

Criteria 5 Pre k 0.42 0.67 0.33 0.21 0.62

CI (-0.03 - 0.88) (0.21 - 1.00) (-0.13 - 0.80) (-0.15 - 0.56) (-0.05-1.0)

Criteria 5 Post k 0.51 0.49 0.35 -0.03 NaN

CI (0.00 - 0.89) (-0.08 - 0.92) (-0.83 - 1.00) NaN

*NaN= it is not possible to calculate k because there is a perfect match between

scores

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

8

Figure 1. Demonstration of Clinical test of thoracolumbar dissociation. (A) Expected

performance anterior/posterior tilt of the pelvis in sitting while attempting to maintain a

constant position of the thorax and thoracolumbar junction. (B) Spine curvatures

described for each position. (C) Pattern of movement that combines lumbar and

thoracolumbar junction. (D) Spine curvatures described by the altered pattern.

Table 5. Rater 1 intra-rater agreement of total test scores in LBP participants live and

videoed assessments.

Assessors Pre-training Post-training

k 95%CI k 95%CI

Total Scores

R1/R1v 0.56 (0.17-0.86) 0.78 (0.58-0.93)

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT