Content uploaded by Audrey Long
Author content
All content in this area was uploaded by Audrey Long on Oct 07, 2017
Content may be subject to copyright.
SPINE Volume 29, Number 23, pp 2593–2602
©2004, Lippincott Williams & Wilkins, Inc.
Does it Matter Which Exercise?
A Randomized Control Trial of Exercise for Low Back Pain
Audrey Long, BScPT,* Ron Donelson, MD,† Tak Fung, PhD‡
Study Design. Multicentered randomized controlled
trial.
Objectives. To determine if previously validated low
back pain (LBP) subgroups respond differently to con-
trasting exercise prescriptions.
Summary of Background Data. The role of “patient-
specific” exercises in managing LBP is controversial.
Methods. A total of 312 acute, subacute, and chronic
patients, including LBP-only and sciatica, underwent a
standardized mechanical assessment classifying them by
their pain response, specifically eliciting either a “direc-
tional preference” (DP) (i.e., an immediate, lasting im-
provement in pain from performing either repeated lum-
bar flexion, extension, or sideglide/rotation tests), or no
DP. Only DP subjects were randomized to: 1) directional
exercises “matching” their preferred direction (DP), 2)
exercises directionally “opposite” their DP, or 3) “nondi-
rectional” exercises. Outcome measures included pain
intensity, location, disability, medication use, degree of
recovery, depression, and work interference.
Results. A DP was elicited in 74% (230) of subjects.
One third of both the opposite and non-directionally
treated subjects withdrew within 2 weeks because of no
improvement or worsening (no matched subject with-
drew). Significantly greater improvements occurred in
matched subjects compared with both other treatment
groups in every outcome (Pvalues ⬍0.001), including a
threefold decrease in medication use.
Conclusions. Consistent with prior evidence, a stan-
dardized mechanical assessment identified a large sub-
group of LBP patients with a DP. Regardless of subjects’
direction of preference, the response to contrasting exer-
cise prescriptions was significantly different: exercises
matching subjects’ DP significantly and rapidly decreased
pain and medication use and improved in all other out-
comes. If repeatable, such subgroup validation has im-
portant implications for LBP management.
Key words: exercise, centralization, low back pain sub-
groups, directional preference, McKenzie method. Spine
2004;29:2593–2602
A number of systematic reviews have raised important
questions regarding the role of exercise in the treatment
of low back pain (LBP) with a lack of evidence support-
ing any specific type of exercise, e.g., back or abdominal
strengthening, McKenzie, Williams, flexion, extension,
or stretching.
1–5
LBP clinical guidelines advocate advice
to stay active and an early return to normal activity as the
means to faster recovery with less disability.
6–8
These
guidelines challenge the popular clinical practice of pre-
scribing patient-specific exercises
9,10
determined by an
individual’s assessment findings and implies that nonspe-
cific exercise can be prescribed without consideration of
individual clinical signs, e.g., every LBP patient is given
the same exercises.
There is growing opinion that the equivocal or con-
flicting results among exercise trials can be attributed to
the faulty assumption that the “nonspecific” LBP popu-
lations studied were homogeneous.
6,11–14
Alternatively,
treatment efficacy has been demonstrated in trials that
studied defined LBP subgroups, although methodologic
weaknesses require cautious interpretation of re-
sults.
15–20
Meanwhile, identification of LBP subgroups
was listed as the top research priority by the Interna-
tional Forum Primary Care Research in Low Back
Pain.
21
The Cochrane Back Review Group has recently
referred to the identification of subgroups and predictors
of chronicity as “the Holy Grail” of LBP.
22
One subgroup classification method (McKenzie
Method),
23
often referred to as Mechanical Diagnosis
and Therapy (MDT), has demonstrated strong inter-
rater reliability (kappa values ranging from 0.79 to
1.0)
17,24–30
and other clinically useful properties: pre-
dicting outcome and discogenic pathology, and provid-
ing preliminary evidence of patient-specific treatments
based on assessment findings.
15,24,29,31–35
An important feature of the MDT assessment is the
identification of a patient’s “directional preference”
(DP).
23,36–39
DP is identified when posture or repeated
end-range movements in a single direction (flexion, ex-
tension, or side-glide/rotation) decrease or abolish lum-
bar midline pain, or cause referred pain emanating from
the spine to appear to progressively retreat in a proximal
direction back toward the lumbar midline (“centraliza-
tion”). There is often a rapid and concurrent restoration
of lumbar range of movement.
23
The inter-rater reliabil-
ity for identifying DP in the hands of qualified practitio-
ners (McKenzie Institute credentialed) is reported as ex-
cellent (agreement 90%, kappa 0.9).
26
The objective of this study is to determine if a subject-
specific exercise prescription concordant with a study
participant’s DP will achieve better outcomes than non-
concordant exercises.
From *Bonavista Physical Therapy, Calgary, Alberta, Canada; †Self-
Care First, Hanover, NH; and ‡Department of Psychology, University
of Calgary, Calgary, Alberta, Canada.
Supported by the Community Ethics Review Board of the Alberta Her-
itage Foundation for Medical Research, the Physiotherapy Foundation
of Canada, McKenzie Institute International, and Cambridge Physio-
therapy Associates.
The manuscript submitted does not contain information about medical
device(s)/drug(s).
Foundation funds were received to support this work. No benefits in
any form have been or will be received from a commercial party related
directly or indirectly to the subject of this manuscript.
Address correspondence and reprint requests to Audrey Long, BScPT,
620 Willesden Dr. S.E., Calgary, Alberta, Canada, T2J 2G1; E-mail:
longma@telusplanet.net
2593
Methods
Physical therapists (PTs) in 11 clinics volunteered to participate
in response to advertisements placed in McKenzie Institute
Branch newsletters. The Community Ethics Review Board of
the Alberta Heritage Foundation for Medical Research granted
approval for the study.
Consecutive patients presenting for treatment of their LBP
were asked to participate in a study of exercise for LBP. The
consent form stated: “previous studies have shown that exer-
cise helps promote healing, can help control pain, and improve
function. However, there are a variety of opinions regarding
which exercise(s) is the best.”
Baseline subject characteristics recorded included age, gen-
der, marital status, prior episode history, job demands, current
work status, mechanism of injury, current episode duration,
symptom location, and back and leg bothersome ratings.
40
Our inclusion/exclusion criteria are listed in Table 1. The
qualifying process also included a standardized MDT assess-
ment by credentialed or diplomaed McKenzie Institute-trained
therapists to identify the subgroup of subjects who demon-
strated a DP and to assign one of three directional labels to each
individual’s DP subset: extension DP (with or without hips
positioned off-center), flexion DP, or a lateral DP (left or right
side-glide or rotation in flexion).
23
Study participants were suc-
cessfully shielded from recognizing their own directional pref-
erence by not discussing this concept with them during their
baseline assessment. Their own directional label and the rele-
vance of changes in distal pain/symptom location (centraliza-
tion/peripheralization) were also not discussed. When pain lo-
cation changes do occur, without specific education, patients
routinely do not recognize it as beneficial, nor do they recognize
the directional theme so commonly present.
After giving informed consent, each member of these three
DP subsets (extension, flexion, and lateral) was randomized to
one of three treatments using treatment allocation cards drawn
by nonmedical staff from DP-labeled envelops. This ensured
equal representation of each directional subset in each treat-
ment group (Figure 1).
Treatments. The three treatment protocols, described in de-
tail in Appendix A, are summarized as follows: 1) Matched:
Subjects were taught unidirectional end-range lumbar exercises
matching the direction of their DP identified during baseline
assessment.
23
2) Opposite: Subjects were also taught unidirec-
Table 1. Inclusion and Exclusion Criteria
Inclusion Criteria Exclusion Criteria
Consecutive patients with:
Low back pain
With or without leg symptoms
With or without one
neurological sign
Age 18–65 years
Demonstrating a directional
preference during the
mechanical assessment
Cauda equina
2 or more neurological signs
Spinal fractures
Post-surgical
Off work for 1 year or more due to
LBP
Medical causes (e.g., severe
osteoporosis, inflammatory or
infectious conditions)
Uncontrolled medical conditions
(e.g., diabetes, angina,
hypertension)
Pregnancy
Inability to read English*
Patients with prior knowledge of,
or specific physician referral
for, the McKenzie method were
excluded to control for patient
bias
No directional preference elicited
*Study participants from Germany were provided with translated
questionnaires
Figure 1. Study flow.
2594 Spine •Volume 29 •Number 23 •2004
tional end-range exercises, but in a direction opposite to their
DP identified during baseline assessment. 3) Evidence-based
care (EBC): Subjects were taught commonly prescribed multi-
directional, midrange lumbar exercises, and stretches for the
hip and thigh muscles.
Both the opposite and EBC groups were provided education
consistent with LBP clinical guidelines, including advice aimed
at minimizing fear avoidance behavior
41
and to remain active
8
(Appendix B). Members of the matched group were likewise
instructed to remain active but also to avoid activities and po-
sitions that increase intensity or radiation of symptoms.
The baseline assessment in our study design required skills
(i.e., MDT) not part of standard PT training. The original de-
sign also called for the opposite and EBC treatments to be
provided by non-McKenzie trained PTs. However, 6 (30%) of
the first 20 subjects in our pilot project dropped out, 4 (67%)
because of dissatisfaction with having to change therapists af-
ter the initial assessment. This suggested either success by the
assessing therapists in convincing subjects there was no proven
difference between the three forms of exercises, or some attrac-
tive rapport established during the assessment procedures
(30– 45 minutes). Since all MDT-trained therapists were also
adequately trained to administer the other two exercise proto-
cols, and since all had expectations that subjects could improve
with all treatments (based at least on natural history and re-
gression to the mean), subjects were then allowed to continue
treatment with their assessing PT. Only 2 (9%) of the next 22
subjects dropped out. Furthermore, 2 years of clinician recruit-
ing failed to identify sufficient clinics with both MDT and non-
MDT trained clinicians working in the same clinic. Subse-
quently, all treating PTs were instructed to confidently present
all three exercise programs, encouraging subjects’ participa-
tion, with the intent of minimizing dropouts so as not to jeop-
ardize the study. Data from these 42 pilot subjects were in-
cluded in our analysis.
Treatment Protocols. For consistency, a minimum of three
and maximum of six visits over the 2-week study period was
imposed for all three treatment interventions. As in other stud-
ies,
29,31,32,35
our 2-week intervention period was deemed ade-
quate time to determine how these three interventions ad-
dressed pain control. We also anticipated difficulty maintaining
treatment compliance beyond 2 weeks (six visits) for those not
experiencing improvement with their assigned treatment. Ad-
ditionally, some clinics had contractual agreements to progress
clients into progressive strengthening/functional programs. As
part of their consent, subjects agreed not to participate in other
nonmedicinal treatments during the trial and were advised that
they were free to withdraw at any time. No attempts were made
to influence medication use.
Safety Guidelines. On recommendation of the Ethics Board,
the participation of subjects reporting a lasting increase in pain,
Figure 2. Exclusions.
Table 2. Origin of Subjects By Country and Clinician.
Country
No. of Practices
(all outpatient
departments, except
where stated
otherwise)
No. of
Patients
No. of Therapists
by Practice
(% by country)
Canada 2 Private practices 2 25/49 (24%)
United States 5 6 10/10/20/9/20 (22%)
Germany 1 1 88 (28%)
United Kingdom 2 2 14/61 (24%)
Kuwait 1 1 6 (2%)
Total 11 12 312 (100%)
Table 3. Subject Baseline Characteristics (N ⴝ312)
Characteristic No. (%)
Gender
Male 166 (53)
Female 146 (47)
Prior episodes
None 93 (30.0)
1 or 2 62 (19.9)
3 to 6 27 (11.9)
Many times, recover well in-between episodes 75 (24.0)
Many times, do not recover well between 31 (09.9)
Information not provided 24 (10.9)
Reported cause of injury*
No known reason 180 (34.6)
Injury at home 74 (23.7)
Injury at work 74 (23.7)
Injury during sports 24 (7.7)
Car accident 10 (3.2)
Lift/twist 74 (23.7)
Push/pull 13 (4.2)
Fall 8 (2.6)
Major job demands*
Student 5 (1.6)
Homemaker 30 (9.6)
Retired 14 (4.5)
Sitting 129 (41.3)
Light (⬍5 kg) 32 (10.30)
Medium (⬍23 kg) 43 (13.8)
Heavy (⬍50 kg) 49 (15.7)
Very heavy (⬎50 kg) 45 (14.4)
*Some study participants checked more than one category (e.g., truck driver:
sitting/heavy lifting).
2595Exercise for LBP •Long et al
symptoms radiating more distally, or demonstrating any dete-
rioration in neurologic signs/symptoms, would be discontinued
with outcome measures obtained and their data included in our
intention-to-treat analysis.
Therapists and Treatment Sites. Twelve PTs had a mean of
12 years experience (range, 3–18 years). All were credentialed
or diplomaed in MDT for a mean of 3 years (range, 1–8 years),
having passed at least the standardized, validated examination
from the McKenzie Institute International, a level of MDT
training shown to produce excellent reliability in patient assess-
ment.
26,27,29,42
Each therapist’s country of origin and study
subject contribution are listed in Table 2. Contributions were
skewed because of referral patterns, time committed to the
project, job changes, and a maternity leave.
Outcome Measures. Primary outcomes included back and
leg pain intensity ratings using an 11-point visual analogue
scale,
43
the 24-item Roland Morris Disability Questionnaire,
44
and medication use, classifying subjects first as takers or non-
takers of LBP medication. If taking medications, the total num-
ber of pills taken daily was recorded.
Secondary outcome measures included a rating of activity
interference at work and home (0–5),
40
the 21-item Beck De-
Table 4. Baseline Subject Characteristics by Treatment Group
Matched
(N ⫽80) (%)
Opposite
(N ⫽69) (%)
EBC*
(N ⫽80) (%)
2
(df)P
Categorical baseline variable
Gender
Male 39 (48.8) 35 (50.7) 46 (57.5) 1.339 (2) 0.5119
Female 41 (51.3) 34 (49.3) 34 (42.5)
QTF classification
1 39 (48.8) 34 (49.3) 42 (52.2) 2.660 (6) 0.8501
2 14 (17.5) 14 (20.3) 17 (21.3)
3 12 (15.0) 10 (14.5) 8 (10.0)
4† 15 (18.8) 11 (15.9) 8 (10.0)
QTF acute ⬍7 days 11 (14.1) 8 (11.4) 11 (14.7) 4.739 (4) 0.3151
QTF subacute ⬍7 weeks 31 (39.7) 18 (25.7) 22 (29.3)
QTF chronic ⬎7 weeks 36 (46.2) 44 (62.9) 42 (56.0) 4.739 (4) 0.3151
Episodes
History of prior episodes 59 (74.7) 48 (70.6) 55 (68.8) 0.7135 0.6999
This is a first episode 20 (25.3) 20 (29.4) 25 (31.3)
Off work
Yes 26 (36.6) 29 (44.6) 35 (46.7) 1.653 0.4376
No 45 (25.3) 36 (55.4) 40 (53.3)
Taking medication
Yes 28 (36.8) 33 (48.5) 32 (45.1) 2.139 (2) 0.3431
No 48 (63.2) 35 (51.5) 39 (54.9)
Rating of pain interference with usual work
(outside home and housework)
1 4 (5.1) 1 (1.4) 1 (1.3) 8.9168 (8) 0.3494
2 10 (12.8) 13 (18.8) 12 (15.0)
3 21 (26.9) 18 (26.1) 28 (35.0)
4 30 (38.5) 23 (33.3) 32 (40.0)
5 13 (16.7) 14 (20.3) 7 (8.8)
Bothersome scale
Back pain
1 2 (2.5) 1 (1.4) 1 (1.3) 3.4292 (8) 0.9046
2 10 (12.5) 5 (7.2) 10 (12.5)
3 22 (27.5) 24 (34.8) 34 (42.5)
4 32 (40.0) 31 (44.9) 34 (42.5)
5 14 (17.5) 8 (11.6) 11 (13.8)
Leg pain
1 25 (32.5) 22 (33.3) 29 (37.2) 2.052 (8) 0.9794
2 14 (18.2) 14 (21.2) 15 (19.2)
3 21 (27.3) 15 (22.7) 15 (19.2)
4 11 (14.3) 11 (16.2) 14 (17.9)
5 6 (7.8) 4 (6.1) 5 (6.4)
Mean (SD) Mean (SD) Mean (SD)
Continuous baseline variables
Age 42.86 (9.55) 42.19 (10.34) 41.51 (10.76) 0.3454 (2,224) 0.7083
Length of current episode‡ (weeks) 13.70 (19.84) 17.65 (21.82) 14.55 (17.60) 0.7181 (2,199) 0.4890
Back pain rating/10 6.01 (2.36) 6.06 (2.24) 6.01 (2.05) 0.0101 (2,226) 0.9900
Leg pain rating§/10 4.59 (N⫽41) (2.50) 4.74 (N⫽42) (2.48) 4.78 (N⫽41) (2.56) 0.07 (2,121) 0.934
Medication
Total pills per day 储
3.37 (N⫽27) (2.92) 3.65 (N⫽17) (2.74) 3.10 (N⫽19) (2.64) 0.17 (2,60) 0.845
BDI-II 10.16 (9.03) 9.88 (9.35) 8.05 (6.92) 1.4355 (2,225) 0.2402
*Evidence-based care group.
†These QTF 4 subjects represent a ⬘skewed⬘QTF 4 sample. Subjects with ⬎1 sign were excluded.
‡A total of 22 subjects reported ⬎100 week-long episode (SE skew, 0.163). These subjects were excluded from the above data.
§Note floor effect: those with leg pain ⫽zero at intake are excluded from this analysis.
储Note floor effect: those not taking medication at intake are excluded from this analysis.
2596 Spine •Volume 29 •Number 23 •2004
pression Inventory (BDI),
45
and the Quebec Task Force sever-
ity rating (QTF 1– 4).
3
The latter classifies patients by their pain
location and neurologic status. A newly generated satisfaction
questionnaire also rated subjects’ response to treatment, readi-
ness for discharge, need for further treatment, and the ability to
return to work and leisure activities.
Outcome measures were administered at baseline and 2
weeks by nonmedical reception staff blinded to the study de-
sign. Outcome measures requiring direct therapist measure-
ment (e.g., range of movement, strength, straight-leg raise)
were intentionally not used to minimize testing bias. However,
when no nonmedical staff was available, the treating PT some-
times handed questionnaires to the subjects.
A take-home compliance-tracking sheet enabled subjects to
record the number of exercise sessions per day. Compliance
scores were rated as follows: 0 ⫽poor; 1–2 sessions ⫽fair; 3– 4
sessions ⫽good; 5⫹⫽excellent.
Data Analysis. All study sites mailed their subject data sheets
and questionnaires to our Alberta study center where a blinded
technician manually entered the data. Random, double-entry
methods documented entry accuracy.
Based on the effect sizes calculated, a minimal sample size of
38 per treatment group was required to have a power of 0.90
with an alpha level of 0.05 for hypothesis testing. A sample size
of 300 was chosen to allow for dropout and subgroup analysis.
Subjects unable/unwilling to continue for the full 2-week pro-
tocol were included in the analysis (intention to treat).
Descriptive statistics and frequencies distributions of all vari-
ables were determined. Specifically, two-way analysis of variance
for all continuous dependent variables determined whether there
was a 1) time by exercise group interaction effect, 2) time effect,
and 3) exercise group effect. Simple effects testing was performed
where appropriate.
2
tests determined relationships between
groups and the categorical/ordinal dependent variables at 2
weeks. Correlation analyses determined the association among
the interval/ratio variables. McNemar
2
was used to confirm the
heterogeneity of the changes within the treatment groups.
Results
Eight of the 11 clinics evaluated 503 consecutive LBP
patients with 257 excluded (Figure 2). The remaining
three clinics contributed 64 subjects, however, provided
no exclusion data. Baseline characteristics of the 312
recruited subjects are listed in Table 3. A DP was elicited in
230 (74%), with the remaining 82 (26%) excluded (Figure
1). After randomization, there were no differences among
the three treatment groups in any baseline demographic
characteristics or outcome measures (Table 4).
Of the 230 DP subjects, three directions of DP were
identified during the baseline assessment: 191 (83%) ex-
tension, 16 (7%) flexion, and 23 (10%) lateral respond-
ers (Figure 1). Twenty-four of those labeled as having an
extension DP first needed their pelvis offset from the mid-
line before their prone extension exercises produced cen-
tralization or reduction of pain. Figure 1 illustrates the
randomization process.
Dropouts
Twenty-nine subjects (12.6%) failed to return for treat-
ment appointments and did not provide data at 2 weeks,
equally distributed between the treatment groups. None
Table 5. Characteristics of Dropouts Compared With Those Who Completed Treatment
Baseline Characteristic Dropouts Completed Trial For
2
P
Age (years) 41.1 42.3 T(226) ⫽0.59 0.5530
Gender
Male 10 (34.5%) 156 (55.1%)
2
(1) ⫽3.71 0.05*
Female 19 (65.5%) 127 (44.9%)
Directional preference label
Extension 23 (79.3%) 168 (83.6%)
2
(2) ⫽.5333 0.7659
Flexion 2 (6.9%) 14 (7.0%)
Lateral 4 (13.8%) 19 (9.5%)
Treatment group allocation
Matched 10 (34.5%) 70 (34.8%) F(2) ⫽0.184 0.9122
Opposite 8 (27.6%) 62 (30.8%)
EBG 11 (37.9%) 69 (34.3%)
QTF classification
1 16 (55.2%) 100 (49.8%) F(1) ⫽2.53 0.4695
2 5 (17.2%) 40 (19.9%)
3 2 (6.9%) 33 (16.4%)
4 6 (20.7%) 28 (13.9%)
Mean length of current episode
Acute 2 (6.9%) 30 (13.0%)
2
(4) ⫽4.009 0.4047
Subacute 9 (31.0%) 71 (30.9%)
Chronic 17 (58.6%) 122 (53.0%)
Insufficient data 1 (3.5%) 7 (3.0%)
Off work due to LBP 11 (44.0%) 79 (42.2%) F(1) ⫽.0278 0.8677
LBP intensity rating 6.55 5.95 T(228) ⫽1.37 0.1710
Leg pain intensity rating 4.82 4.72 T(140) ⫽0.16 0.8760
RMDQ–Likert 74.00 66.60 T(228) ⫽1.17 0.2450
RMDQ–traditional 18.35 17.75 T(228) ⫽0.53 0.5940
BDI-II 14.86 8.57 T(31.5) ⫽3.08 0.004*
Prior episodes ⫽yes 15 (53.6%) 148 (74.0%)
2
(1) ⫽4.08 0.043*
*Significant at P⫽0.05.
2597Exercise for LBP •Long et al
of the 12 (41%) contacted by telephone stated that their
dropout (Figure 1) was related to unrelieved or worsened
pain. Only 3 of 16 baseline variables were statistically
different between dropouts and those supplying com-
plete data (Table 5). Dropouts were more likely to be
female (65.5% vs. 44.9%), have a higher mean depres-
sion score (14.86 vs. 8.57), and have a lower percentage
with prior episodes (53.6% vs. 74%). After dropouts,
201 subjects were eligible for analysis.
Complications Resulting in Early Withdrawal
As opposed to dropouts, “withdrawals” were unable/
unwilling to continue their exercises for the full 2-week
protocol (N ⫽36) because of no improvement, worsen-
ing, or increased distal radiation of symptoms. They pro-
vided their 2-week data early so they could move to alter-
native care. There was considerable variability in these
withdrawal rates in that no matched group subjects with-
drew compared with 16 (34.8%) of the opposite and 20
(32.8%) of the EBC groups (
2
⫽18.67, df ⫽2, P⬍
0.001). Withdrawal subjects were included in our analysis,
which accounts for the lower number of visits for the op-
posite (2.81) and EBC (3.05) groups compared with the
matched group (4.08) (F⫽16.15, df ⫽2203, P⬍0.001).
Treatment Effects
Sixty-eight percent of our 201 subjects returned their
compliance questionnaires. Their overall compliance
rating was good (3–4 sets/day) that did not differ signif-
icantly across treatment groups (P⫽0.121).
All three treatment groups improved in all outcome
measures over the 2-week trial. However, there was sta-
tistically significantly greater improvement in every out-
come variable for the matched group compared with the
opposite or EBC groups, with Pvalues ranging from
0.016 to ⬍0.001 (Table 6; and Figures 3, 4).
Satisfaction With Care
The matched group had statistically significantly greater
improvement than the other two treatment groups (P
values ⬍0.005) in all five areas of satisfaction with care:
return to work, home and recreational activity, perceived
need for further treatment, and self-rated improvement
(Table 7; Figure 3). The opposite group consistently
faired the poorest on all satisfaction parameters; indeed,
15% to 17% of both the opposite and the EBC groups
reported worsening, despite their expected favorable
prognosis (Figure 3).
Table 6. Outcomes by Treatment Groups (ANOVA)
Outcome Measure Pre/Post
Matched
关mean (SD)兴
Opposite
关mean (SD)兴
EBC*
关mean (SD)兴
Time Effect
F(df)关P兴
Group Effect
F(df)关P兴
Interaction
F(df)关P兴
Back pain/10 Pre
Post
5.86
(2.39)
2.51
(1.96)
6.08
(2.17)
4.65
(2.33)
5.97
(2.06)
4.34
(2.51)
138.50 (1,200) 关⬍0.001†兴8.09 (2,200) 关⬍0.001†兴11.55 (2,200) 关⬍0.001†兴
Leg pain/10‡ Pre
Post
4.58
(2.50)
1.61
(1.83)
N⫽41
4.74
(2.48)
3.29
(2.71)
N⫽42
4.78
(2.56)
3.56
(3.13)
N⫽41
71.80 (1,121) 关⬍0.001†兴2.73 (2,121) 关0.069兴6.11 (2,121) 关⬍0.003†兴
RMDQ–total no. of yes/24 Pre
Post
17.85
(5.66)
11.37
(7.55)
16.69
(5.97)
15.44
(6.92)
18.37
(5.34)
15.45
(7.34)
64.10 (1,197) 关⬍0.001†兴2.88 (2,199) 关0.059兴12.19 (2,197) 关⬍0.001†兴
Medication–total pills/day§ Pre
Post
3.37
(2.92)
0.81
(2.25)
N⫽27
3.29
(2.74)
2.57
(2.77)
N⫽21
2.65
(2.38)
1.73
1.73)
N⫽26
24.09 (1,71) 关⬍0.001†兴1.04 (2,72) 关0.357兴4.35 (2,71) 关0.016†兴
During the past week, pain
interference with usual
work, in and outside
home
Rating (0–5)
Pre
Post
3.41
(1.10)
2.24
(0.92)
3.49
(1.05)
3.06
(1.09)
3.39
(0.92)
2.88
(1.15)
89.80 (1,197) 关⬍0.001†兴4.38 (2,197) 关0.014†兴10.32 (2,197) 关⬍0.001†兴
BDI-11 Pre
Post
9.10
(8.13)
4.94
(6.11)
9.14
(8.74)
7.65
(8.50)
7.69
(6.98)
5.25
(5.25)
57.30 (1,195) 关⬍0.001†兴1.31 (2,195) 关0.273兴4.86 (2,195) 关0.009†兴
Taking medication (yes) Pre
Post
48
(63.2%)
7 (14.9%)
35
(51.5%)
15 (50.0%)
39
(54.9%)
14 (41.2%)
McNemar test used for time effect
Matched P⬍0.001†
Opposite P⫽0.250
EBC P⫽0.125
*Evidence-based care group.
†Significant at 0.05 level.
‡Those with no leg pain on admission are excluded from this analysis.
§Those with zero medication excluded from analysis.
2598 Spine •Volume 29 •Number 23 •2004
Changes in QTF Severity Classification
More than one third (36.8%) of the matched group re-
ported improvement in QTF severity (pain location/
neurologic status) classification compared with only
10.6% of the opposite and 19.3% of the EBC groups.
Further, no matched group subject reported any deteri-
oration in their severity classification compared with
12.8% of the opposite, and 17.5% of the EBC groups
(
2
ⴝ20.70, df ⫽4, P⬍0.001).
Discussion
Many trials describe exercise protocols extending over
many visits, sometimes for months, suggesting the intent
to address deconditioning.
6,46–48
Alternatively, patient-
specific DP exercises have a specific focus: pain control
and/or elimination.
23
Not surprisingly, gaining control
over pain that is inhibiting function improved every
other outcome measure in only four visits. Whether pro-
gression to cardiovascular or strength straining pro-
grams can further enhance these outcomes is worthy of
further investigation.
Consistent with current evidence-based guidelines,
the MDT method used in this study routinely provides
activity advice and patient education, but these are based
on DP principles. By study design, the mobilization and
manipulation components of the MDT method were not
included in the treatment protocol.
23
Another documented contribution of this form of as-
sessment is the ability to predict chronic pain and disabil-
ity at 1 year
54
by the early identification of those who fail
to demonstrate DP/centralization (“noncentralizers”).
Based on the many studies reporting high reliability
and predictive validity, this MDT form of assessment
was given the highest grade for scientific support as
Figure 3. Outcomes. A, back pain (1–10) P⬍.001. B, leg pain (0–10) P⬍.003. C, Roland-Morris Disability Questionnaire (0 –24) P⬍0.01.
D, medication total pills/day for those taking medication at baseline (56%) P⫽.016. E, Beck Depression Inventory (0 –21) P⫽.009. F,
activity interference (0 –5) P⬍.001.
Figure 4. Self-rated improvement at 2 weeks (P⬍.005).
2599Exercise for LBP •Long et al
“both a diagnostic tool and a prognostic indicator” by
Denmark’s LBP Clinical Guidelines.
50
Our results add
even further validation to that recommendation.
This is the first study that uses subgroups based solely
on DP/centralization and directly links this subgroup’s
outcome to a subgroup-specific treatment (i.e., individ-
ual exercise prescription based on DP). Previous studies
reporting a good prognosis in this subgroup had not
tested the role of treatment selection in the better out-
comes.
32–35
In contrast, other commonly used classifica-
tions, such as QTF 1, 2, 3, and 4, have shown little
clinical value beyond being descriptors of the popula-
tion.
51,52,58
Further, commonly used labels such as
“acute” and “chronic” only describe the current episode
while failing to recognize the recurrent nature of the con-
dition. Our high rate of prior episodes (70%), consistent
with others’ findings, illustrates the limitations of using
the simplistic descriptors of acute, subacute, and chron-
ic.
49,53
Further, 46.2% of our DP subjects were consid-
ered chronic and by conventional wisdom would be con-
sidered more difficult to treat; yet, as in many other
studies of centralization and DP,
29,31–35
good and excel-
lent outcomes are the rule when a pain-controlling exer-
cise prescription matches individual’s DP.
One of our treatment groups performed exercises in
the direction opposite to their baseline DP. While this
group’s mean pain ratings and Roland Morris Disability
Questionnaire scores improved, 34.8% of this group
withdrew early, 60.3% reported no improvement, and
15.3% reported worsening (Table 7, Figure 3). Six sub-
jects (12.8%) also reported “worsening” in their QTF
severity rating, i.e., their pain was farther down their leg
at 2 weeks than at baseline.
Given the many studies reporting good to excellent
prognoses for patients demonstrating a DP and central-
ization,
15,32–35,54–56
most of our study participants
would be expected to improve regardless of treatment,
based on being active, exercising, natural history, pla-
cebo and Hawthorne effects, and regression to the mean.
But even the EBC group, who exercised daily and re-
ceived both advice to remain active and reassurance of
likely recovery, did not fare nearly as well as anticipated
and significantly less well than those treated with the
matched exercises. Fifty-six percent of the EBC group re-
ported they were overall no better or were worse (Table 7),
with 17.5% reporting pain radiating further into their leg
after 2 weeks of treatment. The outcomes of many EBC
subjects, in whom improvement was anticipated, were ap-
Table 7. Satisfaction With Care/Beliefs Regarding Need for Further Treatment
Satisfaction/Beliefs Matched Opposite EBC*
2
(df)关P兴
Please tick the phrase that best describes how
you feel you have responded to treatment:
My back problem has resolved 32 (44.45) 4 (6.3%) 13 (19.1%) 74.63 (6) 关⬍0.001†兴
My back problem has improved 36 (50.0%) 12 (19.0%) 17 (25.0%)
My back has not changed 4 (5.6%) 38 (60.3%) 27 (39.7%)
My back problem is worse 0 (0.0%) 9 (15.3%) 11 (16.2%)
Please record your wishes for further treatment:
I do not need further treatment 30 (42.3%) 2 (3.5%) 10 (12.7%) 85.87 (4) 关⬍0.001†兴
I need a few more session of the same
treatment
31 (43.7%) 3 (5.3%) 14 (17.7%)
I prefer to change treatment to try to achieve
better pain control
10 (14.1%) 52 (91.2%) 55 (69.7%)
Regarding work and/or household chores, I feel
I am able to:
Return to all my previous activities. 28 (38.9%) 12 (19.0%) 14 (20.9%) 18.69 (6) 关0.005†兴
Return to all of my activities with minor
adjustments.
21 (29.2%) 18 (28.6%) 13 (19.4%)
Return to most but not all of my activities 18 (25.0%) 16 (25.4%) 26 (38.8%)
Return to few or none of my activities 5 (6.9%) 17 (27.0%) 14 (20.9%)
Regarding sports and recreational activities:
I can participate in the same activities as
before
28 (40.0%) 10 (15.9%) 11 (16.7%) 28.09 (6) 关⬍0.001†兴
I can participate in same activities with minor
adjustments
26 (37.1%) 20 (31.7%) 27 (40.9%)
I will have to change activities or make major
adjustments
13 (18.6%) 23 (36.5%) 11 (16.7%)
I will be unable to participate in sports and
recreation.
3 (4.2%) 10 (15.9%) 16 (25.8%)
Regarding work status, I will be returning to:
Same job, full time, full duties 53 (76.8%) 26 (58.1%) 30 (44.8%) 26.03 (10) 关0.004†兴
Same work, full time, modified 4 (5.8%) 5 (8.1%) 10 (14.9%)
Same work, part time 5 (7.2%) 9 (14.5%) 4 (6.0%)
I will have to find different work because of
back pain
1 (1.4%) 0 (0.0%) 4 (6.0%)
I am unable to work because of back pain 3 (4.3%) 0 (14.5%) 10 (14.9%)
Other 3 (4.3%) 3 (4.8%) 9 (13.4%)
*Evidence-based group.
†Significant at ⬍0.05 level.
2600 Spine •Volume 29 •Number 23 •2004
parently jeopardized by a counterproductive exercise pre-
scription.
That not a single subject in the matched group dem-
onstrated overall worsening or pain further into their leg
(Table 7) is not only an important finding; it also reflects
the overall safety of the DP-matched exercises when ap-
plied to patients in whom a DP was elicited by a properly
trained clinician.
Of our original 312 subjects who underwent the MDT
assessment, 53.5% demonstrated a DP for pure sagittal
extension, but the remainder either required a hips-off-
center start position or movements in other planes to
centralize/abolish their pain, or had no DP elicited. This
refutes the commonly held myth that the MDT (McKen-
zie) method consists only of extension exercises.
4,57
Strengths and Limitations
The authors hope our descriptions of our study sample
and treatment procedures will encourage replication of
this study.
40
No special recruitment techniques were
used to draw subjects who would not otherwise seek
care. Withdrawals were analyzed in an intention-to-treat
analysis, and dropouts are well documented.
Our primary limitation is the potential bias intro-
duced by the opposite and EBC treatments being pro-
vided by MDT-trained PTs. However, we think such bias
is minimal, if any, since all treating PTs were trained in
all three treatment interventions and were instructed to
confidently and enthusiastically present all exercises
based on the expectation that all three groups would
improve, as previously discussed. All PTs were also mo-
tivated to prevent dropouts and maintain the power of
the study.
As in other randomized controlled trials, care-seeking
subjects precluded the feasibility of using a “no treatment”
control group, and neither subjects nor treating PTs could
be blinded to their treatment.
4,47
Subjects were, however,
shielded from their directional preference identified at base-
line. The large number of subjects unable or unwilling to
complete the full 2 weeks of treatment confirmed the prag-
matic and ethical impossibility of maintaining the random-
ized treatments for a longer period.
A phenomenon of potential importance to this and
many other LBP clinical studies is illustrated by review-
ing the first subject randomized to the opposite treatment
group (classified as an extension DP assigned to flexion
exercises) who reported a 2-point improvement in her
pain intensity and then returned to modified work duties.
A post-study interview revealed that, when not perform-
ing her flexion exercises, she avoided forward bending
and sitting (flexion) because they consistently increased
her pain, leading her to do more standing and walking
(extension activities) than usual, simply because this “felt
better.” Given her assignment to flexion treatment, this
self-imposed activity modification of avoiding flexion ac-
tivities and positions in favor of extension is a form of
directional “contamination” of her treatment and may
well have been more influential in her final outcome, i.e.,
overall improvement, despite her actual study assign-
ment of potentially aggravating flexion exercises. Future
research might do well to focus on the role of subject’s
directional self-modification of daily activities in re-
sponse to pain monitoring that would influence the out-
come of any therapeutic intervention or recovery by
“natural history.”
Conclusion
Exercises concordant with patients’ DP significantly im-
proved outcomes compared with nonconcordant exer-
cises and advice, and appear to be an effective pain con-
trol/elimination treatment strategy. This refutes prior
systematic reviews concluding that specific exercises are
not warranted.
While highly favorable outcomes are well docu-
mented in many studies for the DP subgroup of LBP, this
study adds further validity by demonstrating that a sub-
ject-specific treatment is superior to others in creating
good outcomes for this subgroup.
Further, the literature-predicted favorable prognosis
for the DP subgroup was actually compromised by our
version of “evidence-based care.” Our data also intro-
duce evidence for the existence of counterproductive ex-
ercises in many LBP patients.
Key Points
●A subgroup defined by the presence of a direc-
tional preference has better outcomes when pre-
scribed exercises that match the individual’s direc-
tional preference than with nonindividualized
evidence-based care.
●The good prognosis documented in other studies
for study participants with directional preference
and centralization findings at baseline assessment is
significantly diminished if exercise prescriptions do
not match their directional preference findings.
●Reliable identification of a previously validated
low back pain subgroup before our randomization
provided homogeneity to our study sample, which
likely contributed to the significant differences
found between treatment group outcomes.
References
1. Faas A. Exercises: which ones are worth trying, for which patients, and
when? Spine 1996;21:2874–2878; discussion 2878 –2879.
2. Koes B, Boutar L, Beckerman H, et al. Physiotherapy exercises and back
pain: a blinded review. Br Med J 1991;302:1572–1576.
3. Spitzer W, LeBlanc F, Dupuis M. Scientific approach to the assessment and
management of activity-related spinal disorders (the Quebec Task Force).
Spine 1987;12(suppl):16–21.
4. van Tulder M, Malmivaara A, Esmail R, et al. Exercise therapy for low back
pain. Spine 2000;25:2784–2796.
5. van Tulder M, Koes B, Bouter L. Conservative treatment of acute and
chronic nonspecific low back pain: a systematic review of randomized con-
trolled trials of the most common interventions. Spine 1997;22:2128–2156.
6. Abenhaim L, Rossignol M, Valat JP, et al. The role of activity in the thera-
peutic managment of back pain: report of the International Paris Task Force
on Back Pain. Spine 2000;25(suppl):1–33.
2601Exercise for LBP •Long et al
7. Bigos S, Bowyer O, Braen G, et al. Acute Low Back Problems in Adults:
Clinical Practice Guidelines [quick reference guide no. 14]. Rockville, MD:
U.S. Department of Health and Human Services, Agency for Health Care
Policy and Research, 1994.
8. Practitioners RCoG. Clinical guidelines for the management of acute low
back pain. London:Royal College of General Practitioners, 1996, 1999.
9. Battie M, Cherkin D, Dunn R, et al. Managing low back pain: attitudes and
treatment preferences of physical therapists. Phys Ther 1994;74:219–226.
10. Foster N, Thompson K, Baxter D, et al. Management of nonspecific low back
pain by physiotherapists in Britain and Ireland. Spine 1999;24:1332–1342.
11. Hsieh C, Adams A, Tobis J, et al. Effectiveness of four conservative treat-
ments for subacute low back pain: a randomized clinical trial. Spine 2002;
27:1142–1148.
12. Dettori JR, Bullock SH, Sutlive TG, et al. The effects of spinal flexion and
extension exercises and their associated postures in patients with acute low
back pain [see comments]. Spine 1995;20:2303–2312.
13. Elnaggar I, Nordin M, Sheikhzadeh A, et al. Effects of spinal flexion and
extension exercises on low-back pain and spinal mobility in chronic mechan-
ical low-back pain patients. Spine 1991;16:967–972.
14. Koes B, Bouter L, van der Heijden G. Methodological quality of randomized
clinical trials on treatment efficacy in low back pain. Spine 1995;20:228–
235.
15. Delitto A, Cibulka M, Erhard R, et al. Evidence for an extension-
mobilization category in acute low back syndrome: a prescriptive validation
pilot study. Phys Ther 1993;73:216–228.
16. Erhard R, Delitto A, Cibulka M. Relative effectiveness of an extension pro-
gram and a combined program of manipulation and flexion and extension
exercises in patients with acute low back syndrome. Phys Ther 1994;74:
1093–1100.
17. Fritz J, George S. The use of a classification approach to identify subgroups
of patients with acute low back pain. Spine 2000;25:106–114.
18. Kopp JR, Alexander AH, Turocy RH, et al. The use of lumbar extension in
the evaluation and treatment of patients with acute herniated nucleus pul-
posus: a preliminary report. Clin Orthop 1986;202:211–218.
19. O’Sullivan P, Twomey L, Allison G. Evaluation of specific stabilizing exer-
cise in the treatment of chronic low back pain with radiologic diagnosis of
spondylosis or spondylolisthesis. Spine 1997;22:2959–2967.
20. Spratt K, Lehrmann T, Weinstein J, et al. A new approach to the low back
physical examination: behavioral assessment of mechanical signs. Spine
1990;15:96–102.
21. Borkan J, Koes B, Reis S, et al. A report from the second international forum
for primary care research on low back pain: reexamining priorities. Spine
1998;23:1992–1996.
22. Bouter L, Pennick V, Bombardier C. Cochrane back review group. Spine
2003;28:1215–1218.
23. McKenzie R, May S. Mechanical Diagnosis and Therapy, 2nd ed. Waikanae,
New Zealand: Spinal Publications New Zealand Ltd., 2003.
24. Fritz J, Delitto A, Vignovic M, et al. Interrater reliability of judgments of the
centralization phenomenon and status change during movement testing in
patients with low back pain. Arch Phys Med Rehabil 2000;81:57–61.
25. Kilby J, Stigant M, Roberts A. The reliability of back pain assessment by
physiotherapists, using a ‘McKenzie algorithm.’ Physiotherapy 1990;76:
579–583.
26. Kilpikoski S, Airaksinen O, Kankaanpa¨a¨ M, et al. Interexaminer reliability in
low back pain assessment using the McKenzie method. Spine 2002;27:E207–
E214.
27. Razmjou H, Kramer J, Yamada R. Inter-tester reliability of the McKenzie
evaluation of mechanical low back pain. J Orthop Sports Phys Ther 2000;
30:368–383.
28. Spratt K, Weinstein J, Lehmann T, et al. Efficacy of flexion and extension
treatments incorporating braces for low-back pain patients with retrodis-
placement, spondylolisthesis, or normal sagittal translation. Spine 1993;18:
1839–1849.
29. Werneke M, Hart DL, Cook D. A descriptive study of the centralization
phenomenon: a prospective analysis. Spine 1999;24:676– 683.
30. Wilson L, Hall H, McIntosh G, et al. Intertester reliability of a low back pain
classification system. Spine 1999;24:248–254.
31. Donelson R, Aprill C, Medcalf R, et al. A prospective study of centralization
of lumbar and referred pain: a predictor of symptomatic discs and anular
competence. Spine 1997;22:1115–1122.
32. Donelson R, Silva G, Murphy K. The centralization phenomenon: its useful-
ness in evaluating and treating referred pain. Spine 1990;15:211–213.
33. Karas R, McIntosh G, Hall H, et al. The relationship between nonorganic
signs and centralization of symptoms in the prediction of return to work for
patients with low back pain. Phys Ther 1997;77:354–360.
34. Long A. The centralization phenomenon: its usefulness as a predictor of
outcome in conservative treatment of chronic low back pain. Spine 1995;20:
2513–2521.
35. Sufka A, Hauger B, Trenary M, et al. Centralization of low back pain and
perceived functional outcome. J Orthop Sports Phys Ther 1998;27:205–212.
36. Donelson R, Grant W, Kamps C, et al. Pain response to end-range spinal
motion in the frontal plane: a multi-centered, prospective trial. International
Society for the Study of the Lumbar Spine, Heidelberg, Germany, 1991.
37. Donelson R, Grant W, Kamps C, et al. Pain response to repeated end-range
sagittal spinal motion: a prospective, randomized, multi-centered trial. Spine
1991;16:206–212.
38. Snook S, Webster B, McGorry R, et al. The reduction of chronic nonspecific
low back pain through the control of early morning lumbar flexion: a ran-
domized controlled trial. Spine 1998;23:2601–2607.
39. Williams M, Hawley J, McKenzie R, et al. A comparison of the effects of two
sitting postures on back and referred pain. Spine 1991;16:1185–1191.
40. Deyo RA, Battie M, Beurskens AJ, et al. Outcome measures for low back
pain research: a proposal for standardized use. Spine 1998;23:2003–2013.
41. Indahl A, Velund L, Reikeraas O. Good prognosis for low back pain when
left untampered: a randomized clinical trial. Spine 1995;20:473–477.
42. Riddle D, Rothstein J. Intertester reliability of McKenzie’s classifications of
the syndrome types present in patients with low-back pain. Spine 1994;18:
1333–1344.
43. Jensen M, Karoly P, Braver S. The measurement of clinical pain intensity: a
comparison of six methods. Pain 1986;27:117–126.
44. Roland M, Morris R. A study of the natural history of low back pain: Part 1.
Development of a reliable and sensitive measure of disability in low back
pain. Spine 1983;8:141–144.
45. Beck A, Steer R, Brown, G. Beck Depression Inventory, vol. II, Manual. San
Antonio, TX: Psychological Corporation, 1996.
46. Beurskens AJ, de Vet HC, Koke AJ, et al. Measuring the functional status of
patients with low back pain. Spine 1995;20:1017–1028.
47. Mannion A, Mu¨ ntener M, Taimela S, et al. A randomized clinical trial of
three active therapies for chronic low back pain. Spine 1999;24:2435–
2448.
48. Soukup MG, Glomsrod B, Lonn JH, et al. The effect of a Mensendieck
exercise program as secondary prophylaxis for recurrent low back pain.
Spine 1999;24:1585–1591.
49. Von Korff M, Ormel J, Keefe F, et al. Grading the severity of chronic pain.
Pain 1992;21:2833–2839.
50. Danish Institute for Health Technology Assessment. Low Back Pain:Fre-
quency, Management, and Prevention From an HTA Perspective. Danish
Health Technology Assessment 1999;1(1).
51. Loisel P, Vachon B, Lamaire J, et al. Discriminiative and predictive validity
assessment of the Quebec Task Force classification. Spine 2002;27:851–857.
52. Werneke M, Hart D. Categorizing patients with occupational low back pain
by use of the Quebec Task Force classification system versus pain pattern
classification procedures: discriminant and predictive validity. Phys Ther
2004;84:243–254.
53. McGorry R, Webster B, Snook S, et al. The relation between pain intensity,
disability, and the episodic nature of chronic and recurrent low back pain.
Spine 2000;25:834– 841.
54. Werneke M, Hart DL. Centralization phenomenon as a prognostic factor for
chronic low back pain and disability. Spine 2001;26:758–765.
55. Werneke M, Hart D. Discriminant validity and relative precision for classi-
fying patients with nonspecific neck and back pain by anatomic pain pat-
terns. Spine 2003;28:161–166.
56. Aina S, May S, Clare H. The centralization phenomenon of spinal symptoms:
a systematic review. Man Ther 2004;9:134–143.
57. Donelson R. Letter to the editor. Spine 2001;26:1827–1829.
58. Padfield B, Chesworth B, Butler R. Use of an outcome measurement system to
answer a clinical question: Is the Quebec Task Force classification system useful
in an outpatient setting? Physiotherapy Canada 2002;54(4):258– 64.
2602 Spine •Volume 29 •Number 23 •2004