Research Report

Motor Control Exercise for Chronic

Low Back Pain: A Randomized
Placebo-Controlled Trial
Leonardo O.P. Costa, Christopher G. Maher, Jane Latimer, Paul W. Hodges,

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

Robert D. Herbert, Kathryn M. Refshauge, James H. McAuley,
Matthew D. Jennings
L.O.P. Costa, PT, PhD, is Research
Fellow, Musculoskeletal Division,
Background. The evidence that exercise intervention is effective for treatment of The George Institute for Interna-
chronic low back pain comes from trials that are not placebo-controlled. tional Health, PO Box M201, Mis-
senden Rd, Sydney, New South
Objective. The purpose of this study was to investigate the efficacy of motor Wales 2050, Australia. Address all
control exercise for people with chronic low back pain. correspondence to Dr Costa at:
lcos3060@ gmail.com.
Design. This was a randomized, placebo-controlled trial. C.G. Maher, PT, PhD, is Director,
Musculoskeletal Division, The
Setting. The study was conducted in an outpatient physical therapy department George Institute for International
in Australia. Health, and Professor, Sydney
Medical School, The University of
Patients. The participants were 154 patients with chronic low back pain of more Sydney, Sydney, Australia.
than 12 weeks’ duration.
J. Latimer, PT, PhD, is Senior Re-
Intervention. Twelve sessions of motor control exercise (ie, exercises designed search Fellow, Musculoskeletal Di-
vision, The George Institute for In-
to improve function of specific muscles of the low back region and the control of ternational Health, and Associate
posture and movement) or placebo (ie, detuned ultrasound therapy and detuned Professor, Sydney Medical School,
short-wave therapy) were conducted over 8 weeks. The University of Sydney.

P.W. Hodges, PhD, Bphty(Hons), is

Measurements. Primary outcomes were pain intensity, activity (measured by Professor and NHMRC Principal Re-
the Patient-Specific Functional Scale), and patient’s global impression of recovery search Fellow/Professorial Research
measured at 2 months. Secondary outcomes were pain; activity (measured by the Fellow, Division of Physiotherapy,
Patient-Specific Functional Scale); patient’s global impression of recovery measured NHMRC Centre of Clinical Research
at 6 and 12 months; activity limitation (measured by the Roland-Morris Disability Excellence in Spinal Pain, Injury and
Health, School of Health and Reha-
Questionnaire) at 2, 6, and 12 months; and risk of persistent or recurrent pain at 12
bilitation Sciences, The University of
months. Queensland, Brisbane, Australia.

Results. The exercise intervention improved activity and patient’s global impres- R.D. Herbert, PT, PhD, is Senior
sion of recovery but did not clearly reduce pain at 2 months. The mean effect of Research Fellow, Musculoskeletal
Division, The George Institute for
exercise on activity (measured by the Patient-Specific Functional Scale) was 1.1
International Health, and Associ-
points (95% confidence interval [CI]⫽0.3 to 1.8), the mean effect on global impres- ate Professor, Sydney Medical
sion of recovery was 1.5 points (95% CI⫽0.4 to 2.5), and the mean effect on pain was School, The University of Sydney.
0.9 points (95% CI⫽⫺0.01 to 1.8), all measured on 11-point scales. Secondary
Author information continues on
outcomes also favored motor control exercise. next page.

Limitation. Clinicians could not be blinded to the intervention they provided.

Conclusions. Motor control exercise produced short-term improvements in
global impression of recovery and activity, but not pain, for people with chronic low
back pain. Most of the effects observed in the short term were maintained at the Post a Rapid Response or
6- and 12-month follow-ups. find The Bottom Line:
www.ptjournal.org

December 2009 Volume 89 Number 12 Physical Therapy f 1275

Motor Control Exercise for Chronic Low Back Pain

L
K.M. Refshauge, DipPhty, Grad ow back pain is a major health challenged in dynamic tasks.17,18
DipManipTher, MBiomedE, PhD, is Director, and socioeconomic problem and Morphologically, a lower cross-
Research & Innovation, and Deputy Dean,
is associated with high costs in sectional area19 and a larger percent-
Faculty of Health Sciences, The University of
Sydney, Sydney, Australia. care, work absenteeism, and disabil- age of intramuscular fat in the multi-
ity worldwide.1–3 A recent inception fidus muscle20 were found in
J.H. McAuley, PhD, is Research Manager,
cohort study demonstrated that 43% patients with low back pain com-
Musculoskeletal Division, The George Insti-
tute for International Health. of patients with acute low back pain pared with asymptomatic controls.
seen in primary care settings devel- Moreover, it was found that patients
M.D. Jennings, PT (Hons), is Deputy Direc-
oped chronic low back pain and that with low back pain tend to increase
tor, Physiotherapy Department, Liverpool
Hospital, Sydney South West and Western nearly a third of them did not re- the spinal stiffness to compensate for
Sydney Area Health Services, Sydney, cover within 1 year.4 the lack of stability from the deep
Australia. muscles by increasing the activity of

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

[Costa LOP, Maher CG, Latimer J, et al. Mo- Exercise is endorsed as an effective the superficial muscles.21 Finally, it
tor control exercise for chronic low back treatment for chronic low back pain was demonstrated that patients who
pain: a randomized placebo-controlled trial. in most clinical practice guide- recovered from an episode of acute
Phys Ther. 2009;89:1275–1286.] lines.1–3 However, at present, there low back pain are more susceptible
© 2009 American Physical Therapy Association are no placebo-controlled trials of to recurrence and chronicity if these
exercise for chronic low back changes were not treated with mo-
pain.5,6 The positive recommenda- tor control exercise.22
tions in guidelines are derived in-
stead from trials comparing exercise A large number of clinical trials on
with usual care,7,8 with a waiting this topic have been performed, and
list,9 or with no treatment.10 These 3 systematic reviews are now avail-
trials do not control for placebo ef- able.14 –16 The most recent system-
fects and potentially provide biased atic review was confined to clinical
estimates of the effect of exercise trials of motor control exercise for
because they do not control for patients with chronic low back
changes in patient and assessor be- pain15 and, as an advantage from the
havior caused by knowledge of treat- 2 previous systematic reviews,14,16 a
ment allocation.11,12 meta-analysis approach was used.
This review identified 13 random-
Motor control exercise (also known ized controlled trials and 1 quasi-
as specific stabilization exercise) randomized controlled trial, all of
was first considered as a treatment which compared motor control ex-
for low back pain about 13 years ago, ercise with other treatments (eg, spi-
when a group of researchers from nal manipulative therapy, other ex-
The University of Queensland in Aus- ercise regimens, education, surgery)
tralia published the first article on or with no treatment. Notably, no
this topic.13 Since then the number
of studies on this topic,14 –16 as well
as its popularity and use in clinical Available With
practice, have increased. This Article at
www.ptjournal.org
The biological rationale for motor
control exercise is fundamentally • Data Supplement: Motor
based on the idea that the stability Control Training Intervention
and control of the spine are altered Manual
in people with low back pain.13 • Invited Commentary by Julie M.
Physiological studies have demon- Fritz and the Author Response
strated that patients with low back • Audio Abstracts Podcast
pain may exhibit a delayed onset of
This article was published ahead of
activity of the deep trunk muscles print on November 5, 2009, at
(eg, transversus abdominis, multifi- www.ptjournal.org.
dus) when the stability of the spine is

1276 f Physical Therapy Volume 89 Number 12 December 2009

 Motor Control Exercise for Chronic Low Back Pain

placebo-controlled trials were identi- cated to treatment groups by the muscles (including transversus ab-
fied. In order to establish the efficacy physical therapist who opened the dominis and multifidus) and reduce
of motor control exercise for next-numbered envelope. overactivity of specific superficial
chronic low back pain, we con- muscles in an individualized
ducted the first placebo-controlled Participants in each group received manner.
trial of this intervention. 12 half-hour treatments over an • Stage 2. Implement precision of the
8-week period (2 sessions per week desired coordination and train
Method in the first month and 1 session per these skills in static tasks and incor-
Setting and Participants week in the second month). The pla- porate them into dynamic tasks and
This randomized, placebo-controlled cebo treatment was designed to be functional positions.
trial was conducted in an outpatient structurally equivalent26 to the active
physical therapy department of a intervention, providing similar con- Stage 1 of the exercise program in-

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

university teaching hospital in Syd- tact time with the physical therapist. volved retraining of the multifidus
ney, Australia. Consecutive patients Both interventions were provided by and transversus abdominis muscles.
seeking care for chronic low back 3 senior physical therapists who re- These exercises were supplemented
pain were screened for eligibility. To ceived training from experts in mo- with exercises for the pelvic-floor
be eligible for inclusion, participants tor control exercise and placebo in- muscles, breathing control, and con-
had to have nonspecific low back terventions. This training included a trol of spinal posture and movement.
pain (defined as pain and discom- 1-day workshop prior to the com- The specific muscles that were
fort) localized below the costal mar- mencement of the study and 3 half- trained depended on the initial as-
gin and above the inferior gluteal day follow-up sessions during the sessment. Participants were taught
folds, with or without referred leg trial period. Random audits and reg- how to contract these muscles inde-
pain of at least 3 months’ duration; ular meetings provided by the same pendently from the superficial trunk
be currently seeking care for low experts were conducted during the muscles.27,29 Physical therapists used
back pain; be aged between 18 and trial to monitor delivery of interven- real-time ultrasound biofeedback to
80 years; comprehend English; and tions. No deviations from the treat- enhance learning of the tasks. The
expect to continue residing in the ment protocol were observed during exercises were progressed until the
study region for the study duration. the audits. patient was able to maintain isolated
In addition, potential participants contractions of the target muscles
underwent a simple trunk muscle The motor control exercise program for 10 repetitions of 10 seconds each
test to determine that motor control was based on the treatment ap- while maintaining normal respira-
exercise treatment was indicat- proach described in previous publi- tion.27 When this level of compe-
ed.23,24 Exclusion criteria were sus- cations.7,8,27,28 At the first session, tence was achieved, patients were
pected or confirmed spinal pathol- participants were comprehensively considered ready to progress to
ogy (eg, tumor, infection, fracture, assessed by the physical therapist, stage 2.
inflammatory disease), pregnancy, who prescribed exercises that were
nerve root compromise, previous individualized based on the partici- Stage 2 of the exercise program in-
spinal surgery, major surgery sched- pant’s presentation. The exercises volved increasing the complexity of
uled during treatment or follow-up were designed to improve function the exercise by progressing through
period, and presence of any contra- of specific muscles of the low back a range of functional tasks and exer-
indication to exercise,25 ultrasound, region and control of posture and cises targeting coordination of trunk
or shortwave therapy. movement. and limb movement, maintenance of
optimal trunk stability, and improve-
Randomization and The motor control exercise program ment of posture and movement pat-
Interventions involved 2 stages. Each participant terns. Participants required the on-
The randomization sequence was was progressed through the stages going support of a trained physical
computer-generated by one of the according to specific criteria that therapist to ensure correct perfor-
investigators who was not involved should be met in each stage.23 The 2 mance of the exercises. The partici-
in recruitment of participants. The stages and their main objectives pants were instructed to perform a
sequence was blocked (block sizes were: daily set of home exercises. These
of 4, 6, and 8, in random order). exercises were performed at the
Allocation was concealed in sequen- • Stage 1. Train coordinated activity same level and in the same position
tially numbered, sealed, opaque en- of the trunk muscles, including in- as those demonstrated during the
velopes. Eligible patients were allo- dependent activation of the deeper treatment session. Session 12 was a

December 2009 Volume 89 Number 12 Physical Therapy f 1277

Motor Control Exercise for Chronic Low Back Pain

discharge session in which the pa- ments would not unblind partici- Baseline data were collected prior to
tient’s progress was reviewed and pants. The sham machines were randomization. The baseline data in-
exercises were prescribed to be con- identical to active machines (eg, the cluded all outcome measurements
tinued at home. A more comprehen- on and off lights illuminated, the out- and the participant’s characteristics
sive description of the motor control put dial moved), except that they did (age, sex, ethnicity, religion, weight,
intervention is presented online at not provide output. The nature of height, level of education, and em-
www.ptjournal.org. the interventions precluded blinding ployment status). In addition, we col-
of the treatment provider. lected information about depressive
The placebo intervention consisted symptoms (measured with the De-
of 20 minutes of detuned shortwave Outcomes and Follow-up pression Anxiety Stress Scales [DASS-
diathermy and 5 minutes of detuned Measurements of outcomes were ob- 21])39,40 to test whether the effect of
ultrasound for 12 sessions over an tained at baseline and at follow-up the exercise intervention on primary

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

8-week period. This form of placebo appointments 2, 6, and 12 months outcomes was influenced by the
was used because the detuned ma- after randomization. Primary out- DASS-21 depression score. We chose
chines do not provide a specific comes were nominated in the trial to investigate depression as a possi-
treatment effect, but it has been es- protocol.24 The primary outcomes ble effect modifier, first because de-
tablished in previous trials30 –32 that were pain intensity over the previ- pression is common in patients with
participants view this intervention as ous week (measured with a 0 –10 low back pain41 and second because
credible. To ensure the perceived numeric rating scale [NRS]),33 activ- there is evidence from cohort stud-
credibility of the placebo interven- ity (measured with the 0 –10 Patient- ies that depression is associated with
tion, physical therapists followed the Specific Functional Scale [PSFS]),34 poor outcomes in patients with low
usual clinical routine for the delivery and global impression of recovery back pain.42
of the active form of these 2 treat- (measured with the ⫺5 to ⫹5 Global
ments (ie, by checking for contrain- Perceived Effect Scale [GPE]) at 2 Participants rated treatment credibil-
dications, monitoring changes in months.35 Secondary outcomes were ity (measured with the 0 –24 Treat-
symptoms, adjusting the detuned de- pain intensity over the previous ment Credibility Scale)43 after the
vices, and appearing to progress the week, activity (measured with the first treatment session. They were
treatment). Each placebo treatment PSFS), patient’s global impression of asked about side effects at 2 months
session lasted 30 minutes to match recovery measured at 6 and 12 using open-ended questions.44 At 12
the duration of active treatment months, and activity limitation (mea- months, patients were asked about
sessions. sured with the 0 –24 Roland-Morris treatment satisfaction, measured
Disability Questionnaire [RMDQ])36 with a 4-item scale with questions
A careful explanation was provided at 2, 6, and 12 months. Table 1 pre- about the therapist (ie, how helpful,
to patients to ensure they remained sents the description of each of these friendly, and understanding the
blinded to treatment allocation. We outcome measures. Participants re- physical therapist was) and about
used the following description for ported their outcomes by telephone the treatment helpfulness in general.
the patients: “In this trial, normal interview to an investigator who was At 12 months, patients were asked
physical therapy treatment and pla- blinded to the treatment allocation. “Which treatment did you receive?
cebo physical therapy treatment will Patients were asked not to discuss Real physical therapy treatment? Or
be provided. A placebo treatment is any aspect of their treatment with a sham or pretend treatment?” to
a harmless treatment delivered at the assessor. check participant blinding.
less than the effective dose. We will
not tell you which type of treatment We also measured recovery and re- Data Analysis
you will receive, and it is unlikely currence at 12 months. Patients A sample size of 154 participants was
that you could distinguish them.” were considered to have recovered nominated in the trial protocol.24 We
The trial staff described the placebo if they reported that they had be- allowed for 15% nonadherence to
intervention as “pulsed ultrasound come pain-free and this pain-free pe- treatment and 15% loss to follow-up,
and pulsed shortwave” and ex- riod lasted for at least 1 month.37 and assumed a correlation of .5 be-
plained to patients that they proba- Recurrences could only occur in pa- tween baseline scores and out-
bly would not feel any sensation dur- tients who had recovered. Recur- comes. This sample size provides
ing treatment. The active forms of rence was defined as a new episode 80% power to detect an effect of
these treatments delivered in pulsed of low back pain that persisted for exercise of 1 unit on the pain inten-
mode do not produce heat; thus, pre- more than 24 hours.37,38 sity scale (estimated SD⫽2.0), 1 unit
vious experience with the treat- on the PSFS (estimated SD⫽1.8), 1

1278 f Physical Therapy Volume 89 Number 12 December 2009

 Motor Control Exercise for Chronic Low Back Pain

Table 1.
Outcome Measures

Measure Construct Description

Roland-Morris Disability Questionnaire (RMDQ)57 Activity limitation The RMDQ is a 24-item questionnaire related to normal activities of
daily living. Patients are asked to tick the items that they perceive
as difficult to perform due to low back pain. Each answer is scaled
either 0 (no difficulty) or 1 (difficulty), thus leaving a range of
scores from 0 to 24, with a higher score indicating higher levels of
activity limitation. This well-known questionnaire has proven to be
reliable,58 valid,59 and responsive35 in patients with low back pain.

Patient-Specific Functional Scale (PSFS)34 Activity In the PSFS, patients are asked to identify up to 3 important activities
that they are having difficulties with or are unable to perform due
to their condition (eg, low back pain). In addition, the patients are

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

asked to rate on an 11-point scale (ranging from 0 [“unable to
perform activity”] to 10 [“able to perform activity at preinjury
level”]) their current level of ability associated with each activity,
with a higher score indicating higher functional ability. This scale
has levels of reliability, validity, and responsiveness similar to those
of to the RMDQ.35,59

Pain numerical rating scale (NRS)60 Pain intensity The pain NRS involves asking patients to rate their pain intensity
levels over the previous week on an 11-point scale (ranging from 0
[“no pain”] to 10 [“pain as bad as could be”]). The number that
the patient states represents his or her pain intensity score. This
scale has good measurement properties.59

Global Perceived Effect Scale (GPE)61 Overall measure The GPE is an 11-point scale that ranges from ⫺5 (“vastly worse”) to
of change 0 (“no change”) to ⫹5 (“completely recovered”). For all measures
of global perceived effect (at baseline and all follow-ups),
participants were asked, “Compared to when this episode first
started, how would you describe your back these days?” A higher
score indicates greater recovery from the condition. This scale has
good measurement properties.62

unit on the GPE (estimated SD⫽1.7), treatment ⫻ time interactions pro- was tested in a post-hoc analysis and,
and 4 units on the RMDQ (estimated vided estimates of the effects of the therefore, was considered as second-
SD⫽4.9) when the alpha level is set exercise intervention. ary. We calculated confidence inter-
at .05. vals (CIs) for the risk difference us-
To determine whether baseline de- ing the method described by
Data were double-entered. The sta- pression scores modified the effect Newcombe based on Wilson’s score
tistical analysis was performed on an of exercise, a secondary analysis was method, without continuity
intention-to-treat basis. The statisti- conducted in which a higher-level correction.46
cian was given coded data and thus interaction term (baseline DASS-21
was blinded to which group re- depression score ⫻ group ⫻ time) Mixed-models analyses were per-
ceived the exercise intervention. was added to each of the regression formed with Stata 9.* Other analyses
models.45 were performed with SPSS version
The mean effects of intervention on 16.0 for Windows.†
pain intensity, activity (measured by As very few patients recovered, ac-
the PSFS and RDMQ), and global im- cording to our definition of being Role of Funding Sources
pression of recovery were calculated pain-free for 30 days during the study The study was funded by a Research
using linear mixed models (random period, only a small subset of partic- & Development grant from The Uni-
intercepts and fixed coefficients), ipants could experience a recur- versity of Sydney and the Physiother-
which incorporated terms for treat- rence. To provide a measure relevant apy Research Foundation–Australian
ment, time, and the treatment ⫻ to all participants, we created a new Physiotherapy Association. The
time interactions. The effect of time outcome called “persistent or recur- funding sources had no role in study
was nonlinear, so time was dummy rent pain,” which was coded as “no”
coded and analyzed as a categorical for participants who recovered and
variable (ie, 3 dummy variables were did not have a recurrent episode * StataCorp LP, 4905 Lakeway Dr, College Sta-
tion, TX 77845.
created for the categories 2, 6, and within 12 months and “yes” for all †
SPSS Inc, 233 S Wacker Dr, Chicago, IL
12 months). The coefficients of the other participants. This outcome 60606.

December 2009 Volume 89 Number 12 Physical Therapy f 1279

Motor Control Exercise for Chronic Low Back Pain

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

Figure 1.
Study flow diagram.

design, data collection, data analysis, tober 2005 and December 2007 sions due to distance (n⫽1), and ad-
interpretation of data, or writing of (Fig. 1). Seventeen patients chose vice from the trial therapists that the
the trial report. The investigators had not to participate, and 49 patients patient was not suitable for motor
final responsibility in the decision to were considered ineligible. The rea- control exercise treatment due to co-
submit the report for publication. sons for ineligibility were nerve root morbidities (n⫽5) (for reasons of bi-
The study was prospectively regis- compromise (n⫽9), previous spinal lateral knee replacement, substance
tered with the Australian Clinical Tri- surgery (n⫽8), serious spinal pathol- abuse, recent epilepsy collapse, vas-
als Registry (ACTRN01260500026- ogy (n⫽6), non-English speaker cular claudication, or Erdheim-
2606), and the protocol was (n⫽6), scheduled for major treat- Chester disease). Results of the sim-
published.24 ment or surgery during the follow-up ple trunk muscle task indicated that
period (n⫽5), low back pain of less motor control exercise was suitable
Results than 12 weeks’ duration (n⫽7), aged for all tested individuals, and thus no
In total, 220 participants seeking older than 80 years (n⫽1), contrain- participants were excluded based
care for low back pain were dication to exercise (n⫽1), unable to upon this criterion. Of the 154 par-
screened for eligibility between Oc- commit to attend the treatment ses- ticipants who were randomly as-

1280 f Physical Therapy Volume 89 Number 12 December 2009

 Motor Control Exercise for Chronic Low Back Pain

Table 2.
Baseline Characteristics

Exercise Group Placebo Group

Characteristic (nⴝ77) (nⴝ77)

Age (y), mean (SD) 54.6 (13.0) 52.8 (12.7)

Female, n (%) 45 (58) 48 (62)

Low back pain duration (wk), mean (SD) 334.8 (392.3) 328.2 (395.1)

Height (m), mean (SD) 1.65 (0.09) 1.64 (0.10)

Weight (kg), mean (SD) 74.5 (17.5) 75.9 (15.3)

Smoker, n (%) 21 (27) 19 (25)

Taking analgesics, n (%) 61 (79) 58 (75)

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

Participating in moderate exercise, n (%)a 41 (53) 51 (66)

Work status, n (%)

Working full-time 6 (8) 13 (17)

Working part-time 5 (7) 3 (4)

Not working 20 (26) 12 (16)

Not seeking employment 46 (60) 49 (64)

Education, n (%)

School certificate 19 (25) 17 (22)

High school certificate 19 (25) 18 (23)

Trade certificate, diploma, or advanced diploma 9 (11) 15 (20)

Bachelor’s degree or higher 15 (19) 12 (15)

Other (lower than school certificate) 15 (20) 15 (20)

General health status, n (%)

Excellent 3 (4) 8 (10)

Very good 18 (23) 12 (16)

Good 38 (49) 44 (57)

Fair 14 (18) 7 (9)

Poor 4 (5) 6 (8)

Depression Anxiety Stress Scales, mean (SD)

Depressionb 11.4 (12.9) 11.2 (13.4)

Anxietyc 11.9 (11.1) 11.8 (12.2)

d
Stress 14.1 (11.8) 14.4 (12.5)

Primary outcome scores, mean (SD)

Pain intensitye 6.8 (2.1) 6.6 (2.0)

Global impression of recoveryf ⫺1.9 (2.5) ⫺2.1 (2.4)

Activity (Patient-Specific Functional Scale)g 3.3 (1.7) 3.3 (1.8)

Secondary outcome scores, mean (SD)

Activity limitation (Roland-Morris Disability Questionnaire)h 13.1 (5.0) 13.4 (4.9)

a
Moderate exercise was any type of exercise of moderate intensity with a duration greater than 30 minutes at least 3 times per week.
b
Scores range from 0 (“no depression”) to 42 (“high depression”).
c
Scores range from 0 (“no anxiety”) to 42 (“high anxiety”).
d
Scores range from 0 (“no stress”) to 42 (“high stress”).
e
Scores range from 0 (“no pain”) to 10 (“worst pain possible”).
f
Scores range from ⫺5 (“vastly worse”) to 5 (“completely recovered), with 0 being “unchanged.”
g
Scores range from 0 (“cannot perform activity”) to 10 (“can perform activity at preinjury level”).
h
Scores range from 0 (“no disability”) to 24 (“high disability”).

December 2009 Volume 89 Number 12 Physical Therapy f 1281

Motor Control Exercise for Chronic Low Back Pain

Table 3.
Credibility and Treatment Evaluation Comparisons

Exercise Group Placebo Group

Characteristic (nⴝ77) (nⴝ77)

Median credibility scale (IQRa)

How confident do you feel that this treatment can help relieve your pain?b 5 (2) 4 (2)
b
How confident do you feel that this treatment will help you manage your pain? 5 (2) 4 (3)

How confident would you be in recommending this treatment to a friend who 5 (2) 4 (3)
suffered from similar complaints?b

How logical does this treatment seem to you?c 5 (2) 4 (3)

Median treatment evaluation (IQR)

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

Therapist helpfulnessd 5 (2) 5 (2)
e
Therapist understanding 6 (1) 6 (1)

Therapist friendlinessf 6 (0) 6 (0)

d
Treatment helpfulness 4 (2) 4 (3)
a
IQR⫽interquartile range.
b
Scores range from 0 (“not at all confident”) to 6 (“absolutely confident”).
c
Scores range from 0 (“not at all logical”) to 6 (“very logical”).
d
Scores range from 0 (“not at all helpful”) to 6 (“extremely helpful”).
e
Scores range from 0 (“not at all understanding”) to 6 (“extremely understanding”).
f
Scores range from 0 (“not at all friendly”) to 6 (“extremely friendly”).

signed to groups, 152 attended the fects were temporary exacerbations the placebo group recovered but
2-month follow-up (98.7%) and 145 of pain. None of the patients with- then experienced a recurrence
attended both 6- and 12-month drew from the trial due to adverse within 12 months. Consequently,
follow-ups (94.2%). No differences effects. Ten patients from the exer- 88% of the exercise group and 98%
were detected between the partici- cise group and 14 patients from the of the placebo group were catego-
pants who were lost to follow-up placebo group reported use of coint- rized at 12 months as having persis-
and the patients who were followed erventions during the study period. tent or recurrent pain (absolute risk
up. The characteristics of the partic- reduction⫽10%, 95% CI⫽1% to 19%,
ipants in the 2 groups were similar at The exercise intervention improved number needed to treat⫽10).
baseline (Tab. 2). activity and the patient’s global im-
pressions of recovery (Tab. 4 and Exercise improved activity limitation
Out of 12 planned treatment ses- Fig. 2). At 2 months, exercise im- (measured by the RMDQ) at 2
sions, the participants in the exer- proved activity by a mean of 1.1 months (⫺2.7 points, 95% CI⫽⫺4.4
cise group attended a mean of 8.8 points (95% CI⫽1.8 to 0.3) on the to ⫺0.9) and 6 months (⫺2.2 points,
sessions (SD⫽3.5) compared with PSFS and improved patient’s global 95% CI⫽⫺4.0 to ⫺0.5), but the dif-
9.6 sessions (SD⫽3.0) for patients impression of recovery by 1.5 points ferences were smaller and no longer
allocated to the placebo group. Most (95% CI⫽2.5 to 0.4). There was not significant at 12 months (differ-
of the participants believed that they a clear effect of exercise on pain in- ence⫽1.0 point, 95% CI⫽⫺2.8 to
were allocated to a “real or active” tensity at 2 months (⫺0.9 points, 0.8). Finally, there was no evidence
intervention (85% of patients from 95% CI⫽⫺1.8 to 0.0, P⫽.053) or 6 that depression was a predictor of
the exercise group versus 84% of pa- months (⫺0.5 points, 95% CI⫽⫺1.4 response to treatment at 2 months
tients from the placebo group). The to 0.5, P⫽.335), but there was a sta- for pain intensity (␤⫽⫺.03, 95%
ratings of treatment satisfaction were tistically significant effect at 12 CI⫽⫺0.10 to 0.04), global impres-
similar in both groups, with the me- months (⫺1.0 point, 95% CI⫽⫺1.9 sion of recovery (␤⫽⫺.05, 95%
dians ranging from 4 to 6 points (on to ⫺0.1, P⫽.030) in favor of the ex- CI⫽⫺0.23 to 0.13), or activity
a 0 – 6 scale) (Tab. 3). ercise group. During the study pe- (␤⫽.10, 95% CI⫽⫺0.07 to 0.27).
riod, few patients had become pain-
Five patients (2 from the placebo free (recovered): 22% of the patients Discussion
group and 3 from the exercise in the exercise group and 9% in the This is the first randomized, placebo-
group) reported mild adverse effects placebo group recovered. Ten per- controlled trial of motor control ex-
of the interventions. All adverse ef- cent of the exercise group and 7% of ercise for chronic low back pain. We

1282 f Physical Therapy Volume 89 Number 12 December 2009

 Motor Control Exercise for Chronic Low Back Pain

Table 4. considering motor control exercise

Effects of Exercise Versus Placeboa treatment, outlining the likely out-
Unadjusted Mean Exercise Group Versus
comes, and assisting them to decide
Outcome (SD) Placebo Group whether they want to pursue the
Adjusted treatment.
Exercise Placebo Treatment Effect
Variable Group Group (95% CI) P The mean effects of exercise treat-
b
Pain ment were smaller than has been re-
2 mo 4.6 (2.8) 5.6 (2.6) ⫺0.9 (⫺1.8 to 0.0) .053 ported in some trials5,11; however,
6 mo 5.0 (2.9) 5.6 (2.5) ⫺0.5 (⫺1.4 to 0.5) .335 these trials included features associ-
ated with exaggerated treatment ef-
12 mo 5.0 (2.9) 6.3 (2.3) ⫺1.0 (⫺1.9 to ⫺0.1) .030
fects, such as lack of patient blinding

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

c
Global impression of recovery
and absence of controlling for pla-
2 mo 1.3 (3.2) 0.0 (3.1) 1.5 (0.4 to 2.5) .005 cebo effects. Our use of a placebo-
6 mo 1.5 (2.6) 0.3 (3.0) 1.4 (0.3 to 2.4) .010 controlled design provides control
12 mo 1.2 (2.7) ⫺0.3 (2.9) 1.6 (0.6 to 2.6) .003 of potentially important sources of
Activity d bias, so the effects of treatment that
2 mo 5.2 (2.4) 4.1 (2.3) 1.1 (0.3 to 1.8) .004
we observed are less likely to be ex-
aggerated than the effects observed
6 mo 5.3 (2.7) 4.3 (2.6) 1.0 (0.3 to 1.8) .007
in non–placebo-controlled trials.11
12 mo 5.5 (2.6) 4.0 (2.6) 1.5 (0.7 to 2.2) ⬍.001

Activity limitatione The exact biological basis for the ef-

2 mo 9.6 (6.5) 11.9 (5.9) ⫺2.7 (⫺4.4 to ⫺0.9) .003 ficacy of motor control exercise in
6 mo 10.3 (7.0) 12.2 (6.7) ⫺2.2 (⫺4.0 to ⫺0.5) .014 patients with low back pain is still
12 mo 11.4 (7.8) 12.3 (6.4) ⫺1.0 (⫺2.8 to 0.8) .271
unclear,50 but if subjects can be
taught to control their trunk muscles
a
Primary outcomes are highlighted. CI⫽confidence interval.
b
Measured with a numerical rating scale, with scores ranging from 0 (“no pain”) to 10 (“worst pain
while performing functional activi-
possible”). ties, then this may explain the im-
c
Scores ranged from ⫺5 (“vastly worse”) to 5 (“completely recovered”), with 0 being “unchanged.”
d
Measured with Patient-Specific Functional Scale, with participant selecting 3 activities and rating his
provements seen in activity, activity
or her ability to perform the activity on from 0 (“cannot perform activity”) to 10 (“can perform activity limitation, and global impression of
at preinjury level”). Summary score is the mean of the 3 activities.
e
Measured with Roland-Morris Disability Questionnaire, with scores ranging from 0 (“no disability”) to
recovery.19,22 There is some evi-
24 (“high disability”). dence that this training can change
trunk muscle behavior during func-
tional tasks.51,52 A range of mecha-
found evidence of a beneficial, but ical improvements, but complete re- nisms have been proposed to ex-
small, effect of motor control exer- covery is unlikely in this nonspecific plain the effect of motor control
cise on global impression of recov- population. Some patients and clini- training on pain. These mechanisms
ery, activity, and activity limitation cians may not consider these effects include reduced load and improved
(measured by the PSFS and RDMQ, clinically worthwhile. The effects quality of movement53 as a result of
respectively) at 2 months and on are smaller than benchmarks for clin- improved coordination of trunk mus-
“persistent or recurrent pain” at 12 ically important effects suggested by cles. Such changes in control may be
months, but not pain intensity at 2 expert researchers in the low back mediated by plastic changes at the
and 6 months and activity limitation pain field47,48 and in recent clinical motor cortex or elsewhere in the
(measured by the RMDQ) at 12 practice guidelines.2 However, we motor system.54
months. Most of the effects observed acknowledge that consensus has not
at short-term follow-up were main- been reached on this issue among Our study demonstrated that motor
tained 12 months after randomiza- back pain researchers, and one study control exercise produced a small re-
tion. We also found that the effect of of patients with low back pain49 re- duction in the risk for persistent pain
motor control exercise was not influ- vealed an even wider range of views at 12 months. This finding is sup-
enced by the level of depressive on how big an improvement in out- ported by earlier work22 suggesting
symptoms. comes needs to be before it is con- that patients who have continuing
sidered worthwhile. Given this di- impairment of the deep trunk mus-
Our interpretation of the trial results versity of views, clinicians may need cles experience more recurrent low
is that exercise produces small clin- to spend some time with patients back pain episodes. This earlier

December 2009 Volume 89 Number 12 Physical Therapy f 1283

Motor Control Exercise for Chronic Low Back Pain

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

Figure 2.
Outcomes in the 2 treatment groups. Values shown are unadjusted means (SDs). Measurements were obtained at baseline and at
2, 6, and 12 months, but the data are slightly offset in the figure for clarity. Higher scores represent better outcomes for global
impression of recovery and disability, and lower scores represent better outcomes for pain and function. PSFS⫽Patient-Specific
Functional Scale, RMDQ⫽Roland-Morris Disability Questionnaire.

work22 provides a rationale for why propriate training would be able to The main limitation of our study was
those in the exercise group, who re- perform this intervention similarly. that the trial therapists were not
trained the deep trunk muscles, ex- blinded to the treatment allocation.
perienced less resistant or recurrent Although systematic reviews of the We are unaware of a method to blind
pain than those in the placebo efficacy of exercise for chronic low therapists in trials of exercise. We
group, who had no such training. back pain5 have generally concluded tried to minimize the effect of un-
that exercise is effective, most re- blinding by training the trial thera-
This study was performed in an out- views also signal some uncertainty in pists to provide a credible placebo
patient physical therapy department their conclusions because of meth- treatment and by auditing placebo
of a public hospital, and the results odological concerns in the available treatment sessions. We believe that
of this study should be generalizable trials. Our trial avoided the main these steps were effective because
to groups of patients with similar methodological problems of previ- scores on credibility and treatment
characteristics (ie, patients with ous trials by using a placebo control satisfaction were similar in both
chronic low back pain for a long and blinding patients and assessors. treatment groups. Nevertheless, we
time, seeking care for their low back In addition, the trial was prospec- cannot exclude the possibility that
pain problems, with moderate levels tively registered and the trial proto- the lack of therapist blinding intro-
of depression and not working). In col was published.24 Lastly, we took duced some degree of bias into our
terms of the intervention, we believe steps to ensure treatment quality by results. Another potential limitation
that the motor control exercise inter- using experienced clinicians who of this study was that we were not
vention implemented in our study were trained to deliver the treat- able to monitor adherence to the
was well defined (as described in the ments according to the protocol, and home exercise program for the pa-
Data Supplement), and we are confi- we monitored treatment delivery. tients allocated to the motor control
dent that physical therapists with ap- exercise intervention.

1284 f Physical Therapy Volume 89 Number 12 December 2009

 Motor Control Exercise for Chronic Low Back Pain

Although it could be argued that our tional liaisons and clerical support. Dr Ma- 8 Moseley L. Combined physiotherapy and
her, Dr Latimer, Dr Herbert, Dr Refshauge, education is efficacious for chronic low
choice of placebo was not perfect, back pain. Aust J Physiother. 2002;48:
and Mr Jennings provided consultation
we believe that this choice was the 297–302.
(including review of manuscript before
best possible. We do not know of a submission).
9 Risch SV, Norvell NK, Pollock ML, et al.
Lumbar strengthening in chronic low-
“placebo exercise” that is both cred- back-pain patients: physiological and psy-
The study protocol was approved by The
ible and inert. This problem is not chological benefits. Spine. 1993;18:
University of Sydney Human Research Ethics 232–238.
unique to the study of exercise, and Committee. 10 Kuukkanen T, Mälkiä E, Kautiainen H,
similar problems with developing an et al. Effectiveness of a home exercise pro-
This study was funded by a Research & De-
appropriate placebo were found in gramme in low back pain: a randomized
velopment grant from The University of Syd- five-year follow-up study. Physiother Res
trials of complex nonpharmaceutical ney and by the Physiotherapy Research Int. 2007;12:213–224.
interventions such as spinal manipu- Foundation–Australian Physiotherapy Associ- 11 Schulz KF, Chalmers I, Hayes RJ, et al.
lative therapy32,55 and acupunc- ation. Dr Costa had his PhD supported by Empirical-evidence of bias - dimensions of

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

CAPES – Ministério da Educação–Brazil and methodological quality associated with es-
ture.56 Our selection of sham electro- timates of treatment effects in controlled
Pontifı́cia Universidade Católica de Minas
therapy as a placebo was primarily trials. JAMA. 1995;273:408 – 412.
Gerais–Brazil; Dr Maher, Dr Hodges, and Dr
based upon the knowledge that Herbert hold research fellowships funded by
12 Juni P, Altman DG, Egger M. Systematic
reviews in health care: assessing the qual-
these machines do not share the the National Health and Medical Research ity of controlled clinical trials. BMJ. 2001;
same specific components of the ex- Council of Australia. 323:42– 46.
ercise intervention and that they The study was prospectively registered with
13 Hodges PW, Richardson CA. Inefficient
muscular stabilization of the lumbar spine
have been used successfully in pre- the Australian Clinical Trials Registry (AC- associated with low back pain: a motor
vious randomized controlled TRN012605000262606), and the protocol control evaluation of transversus abdomi-
was published.24 nis. Spine. 1996;21:2640 –2650,
trials.30,32
14 Ferreira PH, Ferreira ML, Maher CG, et al.
This article was received July 2, 2009, and was Specific stabilisation exercise for spinal
Our study provides evidence that accepted September 6, 2009. and pelvic pain: a systematic review. Aust
J Physiother. 2006;52:79 – 88.
motor control exercise was better DOI: 10.2522/ptj.20090218 15 Macedo LG, Maher CG, Latimer J, et al.
than placebo in patients with Motor control exercises for persistent
chronic low back pain. Most of the nonspecific low back pain: a systematic
effects observed in the short term References review. Phys Ther. 2009;89:9 –25.
1 Cost B13 Working Group. European guide- 16 Rackwitz B, de Bie R, Limm H, et al. Seg-
were maintained at 6- and 12-month lines for the management of chronic non- mental stabilizing exercises and low back
follow-ups, but the magnitude of the specific low back pain. Eur Spine J. 2006; pain. What is the evidence? A systematic
15:S192–S300. review of randomized controlled trials.
effects was small in this population, Clin Rehabil. 2006;20:553–567.
2 Chou R, Qaseem A, Snow V, et al. Diagno-
who have aspects associated with sis and treatment of low back pain: a joint 17 Hodges PW, Richardson CA. Delayed pos-
poor outcome. Our results suggest clinical practice guideline from the Amer- tural contraction of transversus abdominis
ican College of Physicians and the Ameri- in low back pain associated with move-
that this intervention should be con- can Pain Society. Ann Intern Med. 2007; ment of the lower limb. J Spinal Disord.
sidered for patients with chronic low 147:478 – 491. 1998;11:46 –56.
back pain in order to improve activ- 3 Chou R, Huffman LH. Nonpharmacologic 18 Hodges PW, Richardson CA. Altered trunk
therapies for acute and chronic low back muscle recruitment in people with low
ity and global impression of recovery pain: a review of the evidence for an back pain with upper limb movement at
and to improve pain intensity in the American Pain Society/American College different speeds. Arch Phys Med Rehabil.
of Physicians clinical practice guideline. 1999;80:1005–1012.
long term but not the short term. Ann Intern Med. 2007;147:492–504.
19 Hides JA, Stokes MJ, Saide M, et al. Evi-
4 Henschke N, Maher CG, Refshauge KM, dence of lumbar multifidus muscle wast-
et al. Prognosis in patients with recent on- ing ipsilateral to symptoms in patients
Dr Costa, Dr Maher, Dr Latimer, Dr Hodges, set low back pain in Australian primary with acute/subacute low back pain. Spine.
Dr Refshauge, and Dr McAuley provided care: inception cohort study. BMJ. 2008; 1994;19:165–172.
concept/idea/research design. Dr Costa, Dr 337:1–7.
20 Alaranta H, Tallroth K, Soukka A, et al. Fat-
Maher, Dr Latimer, Dr Hodges, Dr Herbert, 5 Hayden JA, van Tulder MV, Malmivaara A, content of lumbar extensor muscles and
and Dr McAuley provided writing. Dr Costa, et al. Exercise therapy for treatment of low-back disability: a radiographic and
non-specific low back pain. Cochrane Da- clinical comparison. J Spinal Disord.
Dr Maher, and Dr Hodges provided data tabase Syst Rev. 2005;3:CD000335. 1993;6:137–140.
collection. Dr Costa, Dr Maher, Dr Hodges,
6 Machado LAC, Kamper S, Herbert RD, 21 van Dieen JH, Selen LP, Cholowicki J.
and Dr Herbert provided data analysis. Dr et al. Analgesic effects of treatments for Trunk muscle activation in low back pain
Costa, Dr Maher, Dr Latimer, Dr Refshauge, non-specific low back pain: a meta-analysis patients, an analysis of literature. J Electro-
Dr McAuley, and Mr Jennings provided of placebo-controlled randomized trials. myogr Kinesiol. 2003;13:333–351.
project management. Dr Costa, Dr Maher, Rheumatology (Oxford). 2009;48:
520 –527. 22 Hides JA, Jull GA, Richardson CA. Long-
Dr Latimer, and Dr Refshauge provided fund term effects of specific stabilizing exer-
procurement. Dr Maher and Mr Jennings 7 O’Sullivan PB, Twomey LT, Allison GT. cises for first-episode low back pain.
Evaluation of specific stabilizing exercise Spine. 2001;26:E243–E248.
provided participants. Dr Maher, Dr Ref- in the treatment of chronic low back pain
shauge, Dr McAuley, and Mr Jennings pro- with radiologic diagnosis of spondylolysis
vided facilities/equipment. Dr Maher, Dr or spondylolisthesis. Spine. 1997;22:
McAuley, and Mr Jennings provided institu- 2959 –2967.

December 2009 Volume 89 Number 12 Physical Therapy f 1285

Motor Control Exercise for Chronic Low Back Pain

23 Ferreira PH. Effectiveness of Specific Sta- 36 Roland M, Morris R. A study of natural 50 Hodges PW. Transversus abdominis: a dif-
bilisation Exercises for Chronic Low history of back pain, part 1: development ferent view of the elephant. Br J Sports
Back Pain [PhD thesis]. Sydney, New of a reliable and sensitive measure of dis- Med. 2008;42:941–944.
South Wales, Australia: School of Physio- ability in low back pain. Spine. 1983;8: 51 Tsao H, Hodges PW. Specific abdominal
therapy, The University of Sydney; 2004. 145–150. retraining alters motor coordination in
24 Maher CG, Latimer J, Hodges PW, et al. 37 de Vet HCW, Heymans MW, Dunn KM, people with persistent low back pain. Pre-
The effect of motor control exercises ver- et al. Episodes of low back pain: a pro- sented at: 11th World Congress on Pain of
sus placebo in patients with chronic low posal for uniform definitions to be used in the International Association for the Study
back pain. BMC Musculoskelet Disord. research. Spine. 2002;27:2409 –2416. of Pain; August 21–26, 2005; Sydney, New
2005;6:1– 8. South Wales, Australia.
38 Stanton TR, Henschke N, Maher CG, et al.
25 American College of Sports Medicine. AC- After an episode of acute low back pain, 52 Tsao H, Hodges PW. Immediate changes in
SM’s Guidelines for Exercise Testing and recurrence is unpredictable and not as feedforward postural adjustments follow-
Prescription. 5th ed. Baltimore, MD: Wil- common as previously thought. Spine. ing voluntary motor training. Exp Brain
liams & Wilkins; 1995. 2008;33:2923–2928. Res. 2007;181:537–546.
26 Machado LAC, Kamper SJ, Herbert RD, 39 Henry JD, Crawford JR. The short-form 53 Hodges PW, Moseley GL. Pain and motor
et al. Imperfect placebos are common in version of the Depression Anxiety Stress control of the lumbopelvic region: effect

Downloaded from https://academic.oup.com/ptj/article/89/12/1275/2737561 by guest on 08 April 2024

low back pain trials: a systematic review of Scales (DASS-21): construct validity and and possible mechanisms. J Electromyogr
the literature. Eur Spine J. 2008;17: normative data in a large non-clinical sam- Kinesiol. 2003;13:361–370.
889 –904. ple. Br J Clin Psychol. 2005;44:227–239. 54 Tsao H, Galea M, Hodges PW. Skilled mo-
27 Richardson CA, Jull GA, Hodges PW, Hides 40 Haggman S, Maher CG, Refshauge KM. tor training induces reorganisation of the
J. Therapeutic Exercise for Spinal Seg- Screening for symptoms of depression by motor cortex and is associated with im-
mental Stabilization in Low Back Pain. physical therapists managing low back proved postural control in chronic low
Edinburgh, United Kingdom: Churchill pain. Phys Ther. 2004;84:1157–1166. back pain. Presented at: 12th World Con-
Livingstone; 1999. gress on Pain of the International Associa-
41 Urquhart DM, Hoving JL, Assendelft WJJ, tion for the Study of Pain; August 17–22,
28 Hodges PW, Ferreira PH, Ferreira ML. et al. Antidepressants for non-specific low 2008; Glasgow, United Kingdom.
Lumbar spine: treatment of instability and back pain. Cochrane Database Syst Rev.
disorders of movement control. In: Magee 2009;1:CD001703. 55 Hancock MJ, Maher CG, Latimer J, McAu-
DJ, Zachazewski JE, Quillen WS, eds. Sci- ley JH. Selecting an appropriate placebo
42 Pincus T, Burton AK, Vogel S, et al. A sys-
entific Foundations and Principles of trial of spinal manipulative therapy. Aust J
tematic review of psychological factors as
Practice in Musculoskeletal Rehabilita- Physiother. 2006;52:135–138.
predictors of chronicity/disability in pro-
tion: Pathology and Intervention in Mus- spective cohorts of low back pain. Spine. 56 Paterson C, Dieppe P. Characteristic and
culoskeletal Rehabilitation. Philadelphia, 2002;27:E109 –E120. incidental (placebo) effects in complex in-
PA: WB Saunders Co; 2009:398 – 425. terventions such as acupuncture. BMJ.
43 Borkovec TD, Nau S. Credibility of ana-
29 Twomey LT, Taylor JR, eds. Physical Ther- 2005;330:1202–1205.
logue therapy rationales. J Behav Ther
apy of the Low Back. New York, NY: Exp Psych. 1972;3:257–260. 57 Roland M, Morris R. A study of natural
Churchill Livingstone Inc; 1994. Clinics in history of back pain, part 1: development
Physical Therapy, vol 18. 44 Bent S, Padula A, Avins AL. Better ways to of a reliable and sensitive measure of dis-
question patients about adverse medical
30 Pengel LH, Refshauge KM, Maher CG, ability in low back pain. Spine
events: a randomized, controlled trial.
et al. Physiotherapist-directed exercise, 1983;8:145–150.
Ann Intern Med. 2006;144:257–261.
advice or both for subacute low back pain: 58 Brouwer S, Kuijer W, Dijkstra PU, et al.
a randomized trial. Ann Intern Med. 2007; 45 Pocock SJ, Assmann SE, Enos LE, et al. Sub- Reliability and stability of the Roland-
146:787–796. group analysis, covariate adjustment and Morris Disability Questionnaire: intraclass
baseline comparisons in clinical trial re-
31 Ferreira ML, Ferreira PH, Latimer J, et al. correlation and limits of agreement. Dis-
porting: current practice and problems.
Comparison of general exercise, motor abil Rehabil. 2004;26:162–165.
Stat Med. 2002;21:2917–2930.
control exercise and spinal manipulative 59 Costa LO, Maher CG, Latimer J, et al. Clini-
therapy for chronic low back pain: a ran- 46 Newcombe RG. Interval estimation for the metric testing of three self-report outcome
domized trial. Pain. 2007;131:31–37. difference between independent propor- measures for low back pain patients in
tions: comparison of eleven methods. Stat
32 Hancock MJ, Maher CG, Latimer J, et al. Brazil. Which one is the best? Spine. 2008;
Med. 1998;17:873– 890.
Assessment of diclofenac or spinal manip- 33:2459 –2463.
ulative therapy, or both, in addition to rec- 47 van Tulder M, Malmivaara A, Hayden JA, 60 Turk DC, Melzack R. Handbook of Pain
ommended first-line treatment for acute et al. Statistical significance versus clinical Assessment. New York, NY: The Guilford
low back pain: a randomised controlled importance: trials on exercise therapy for Press; 1992.
trial. Lancet. 2007;370:1638 –1643. chronic low back pain as example. Spine.
2007;32:1785–1790. 61 Feinstein AR. Clinimetrics. New Haven,
33 Scrimshaw SV, Maher CG. Responsiveness CT: Yale University Press; 1987:91–103.
of visual analogue and McGill pain scale 48 Ostelo R, Deyo RA, Stratford PW, et al.
measures. J Manipulative Physiol Ther. Interpreting change scores for pain and 62 Kamper SJ, Maher CG, Mackay G. Global
2001;24:501–504. functional status in low back pain: to- rating of change scales: a review of
wards international consensus regarding strengths and weaknesses and consider-
34 Stratford PW, Gill C, Westaway M, et al. minimal important change. Spine. 2008; ations for design. J Manipulative Physiol
Assessing disability and change on individ- 33:90 –94. Ther. 2009;17:163–170.
ual patients: a report of a patient-specific
measure. Physiother Can. 1995;47: 49 Farrar JT, Portenoy RK, Berlin JA, et al.
258 –263. Defining the clinically important differ-
ence in pain outcome measures. Pain.
35 Pengel LHM, Refshauge KM, Maher CG. 2000;88:287–294.
Responsiveness of pain, disability and
physical impairment outcomes in patients
with low back pain. Spine. 2004;29:
879 – 883.

1286 f Physical Therapy Volume 89 Number 12 December 2009