Phys Ther. 2016 (May); 96 (5): 623–630 ~ FULL TEXT
Alessandra Narciso Garcia, Luciola da Cunha Menezes Costa, Mark Hancock,
Leonardo Oliveira Pena Costa
A.N. Garcia, PT,
Masters and Doctoral Programs in Physical Therapy,
Universidade Cidade de São Paulo,
Rua Cesario Galeno 475,
São Paulo, Brazil, CEP 03071-000.
BACKGROUND: "Mechanical Diagnosis and Therapy" (MDT) (also known as the McKenzie method), like other interventions for low back pain (LBP), has been found to have small effects for people with LBP. It is possible that a group of patients respond best to MDT and have larger effects. Identification of patients who respond best to MDT compared with other interventions would be an important finding.
OBJECTIVE: The purpose of the study was to investigate whether baseline characteristics of patients with chronic LBP, already classified as derangement syndrome, can identify those who respond better to MDT compared with Back School.
METHODS This study was a secondary analysis of data from a previous trial comparing MDT with Back School in 148 patients with chronic LBP. Only patients classified at baseline assessment as being in the directional preference group (n=140) were included. The effect modifiers tested were: clear centralization versus directional preference only, baseline pain location, baseline pain intensity, and age. The primary outcome measures for this study were pain intensity and disability at the end of treatment (1 month). Treatment effect modification was evaluated by assessing the group versus predictor interaction terms from linear regression models. Interactions ≥1.0 for pain and ≥3 for disability were considered clinically important.
RESULTS: Being older met our criteria for being a potentially important effect modifier; however, the effect occurred in the opposite direction to our hypothesis. Older people had 1.27 points more benefit in pain reduction from MDT (compared with Back School) than younger participants after 1 month of treatment.
LIMITATIONS: The sample (n=140) was powered to detect the main effects of treatment but not to detect the interactions of the potential treatment effect modifiers.
CONCLUSIONS: The results of the study suggest older age may be an important factor that can be considered as a treatment effect modifier for patients with chronic LBP receiving MDT. As the main trial was not powered for the investigation of subgroups, the results of this secondary analysis have to be interpreted cautiously, and replication is needed.
From the FULL TEXT Article:
The cause of low back pain (LBP)
cannot be definitively identified in
the majority of cases, and as such
the label “nonspecific low back pain”
(NSLBP) is widely used. [1–3] Although
approximately 85% of patients with LBP
in primary care are considered to have
NSLBP,  most clinicians believe that
NSLBP includes several different subgroups
and is not one condition. [4–6]
Several subgrouping systems have been
developed to attempt to improve
outcomes for patients with NSLBP
rather than using a “one-size-fits-all”
approach. [5, 7–10]
“Mechanical Diagnosis and Therapy”
(MDT), also known as the McKenzie
method, is one such subgrouping
approach.  This method, proposed
by Robin McKenzie in 1981,  classifies
patients into 3 syndromes or
groups — derangement, dysfunction, and postural — based on responses to
repeated end-range movements performed
by the patient and assessed by
the clinician. Classification is based
largely on a patient’s response to sustained
postures and repeated movements. [11, 12]
Based on the classification, an
intervention approach is selected. [11, 12]
During the assessment, it is important to
identify the directional preference
response, which is characterized by a
reduction of pain intensity, centralization
(pain referred in a peripheral location
from the spine moves to the central
lower back and is progressively abolished),  or abolished pain. The MDT method emphasizes the use of simple
self-management strategies that require
adherence to home exercises and maintenance
of correct postures. [11, 12]
Although the MDT approach attempts to
improve outcomes in people with LBP
by targeting interventions to the 3 subgroups,
systematic reviews conclude
that MDT has relatively small effect sizes
and is not superior to other approaches
for nonspecific low back pain (NSLBP). [12, 13] One possible reason for
this conclusion is that the vast majority
of patients are classified as being in the
derangement group and, therefore, most
get a similar approach to treatment.  For
example, in our recent trial comparing
MDT with Back School, more than 85%
of participants were classified as having a
derangement. [15, 16] Anecdotally, some clinicians
report that within this large
derangement group, some patients
respond well to MDT, whereas others do
not. If a group of patients classified as
having a derangement who respond best
to MDT can be identified, the outcomes
of using the MDT approach in these
patients could be improved, and those
patients who are unlikely to respond
well could be treated with a different
The treatment-based classification system [7–9] is an example of a hybrid subgrouping
system for LBP that attempts to
identify a subset of patients with a directional
preference for specific exercise
direction, whereas other patients are recommended
for other interventions such
as manipulation or stabilization exercises.
A limitation of the treatment-based
classification system is that patients
assigned to the specific exercise (MDT
category) are tested in one plane only
and are not subjected to the multitude of
planes of motion assessed and loading
and unloading strategies characteristic of
MDT. The recommended intervention
for the specific exercise subset is similar
to that used for the derangement syndrome
in the MDT approach. A previous
secondary analysis of a randomized controlled
trial (RCT) by Sheets et al  investigated
responders to MDT in a group of
patients with acute LBP. They investigated
6 potential effect modifiers (baseline
pain, pain changes with position or
movement, presence of leg pain, constant
pain, pain worse with flexion, and
patient expectation) that could influence
the clinical response of patients treated
with MDT.  This study showed that
these potential effect modifiers did not
predict more favorable response to
MDT.  A limitation of this study was that
only patients in the MDT group underwent
a MDT assessment, so it was not
possible to investigate whether clinical
examination findings such as centralization
predicted response to MDT. 
Another study  included only patients
with a changeable lumbar condition, (ie,
centralization or peripheralization) and
investigated those more likely to benefit
from the MDT method or spinal manipulation.
This study  included 350
patients with chronic LBP and showed
that patients with nerve root involvement
and peripheralization may have
clinically important differences in
response to MDT compared with spinal
Some patients classified by the MDT
method clearly centralize on assessment,
whereas others have a directional preference
but do not actually demonstrate
centralization. [11, 19, 20] To our knowledge,
no previous study has investigated
whether the presence of clear centralization
compared with directional preference
alone is an important effect modifier
for MDT. The purpose of this
hypothesis-setting, secondary analysis
was to investigate whether baseline characteristics
of patients with chronic LBP,
already classified as derangement syndrome,
can identify those who respond
better to MDT compared with Back
This study was a secondary analysis of
data from a 3–arm RCT that investigated
the effect of MDT compared with Back
School in patients with chronic LBP. [15, 16]
To increase the validity of the current
subgroup analysis, it was performed
using the approach recommended by
Sun et al,  which included investigating
a limited number of prespecified predictor
variables, prespecification of the
hypothesized direction of subgroup
effects, and use of interaction terms.
This study was conducted in the outpatient
physical therapy clinic of the Universidade
Cidade de Sa˜o Paulo, Sao
Paulo, Brazil, between July 2010 and July
2012. To be eligible, patients seeking
care had to have NSLBP with a duration
of at least 12 weeks  and be aged
between 18 and 80 years. Patients with
any contraindication to physical exercise,
based on the recommendations of
the guidelines of the American College of
Sports Medicine ; serious spinal pathology
(eg, tumors, fractures, inflammatory
diseases); previous spinal surgery; nerve
root compromise; cardiorespiratory illnesses;
or pregnancy were excluded.
Participants were allocated to 1 of 2
intervention groups—Back School (a
group-based treatment approach) or
MDT (an individually based treatment
approach)—by a simple randomization
sequence computer generated using
Microsoft Excel (Microsoft Corp, Redmond,
Washington). The randomization
sequence was conducted by one of the
investigators of the study who was not
directly involved with the assessments
and treatment of patients. The allocation
was concealed by using consecutively
numbered, sealed, opaque envelopes.
Participants from both groups received 4
one-hour sessions over 1 month, once a
week. The number of sessions was chosen
following the recommendations
from the original Back School manual
method.  Therefore, the same number
of sessions was used for the MDT group.
Patients treated with Back School
method received advice about anatomy
and spinal biomechanics, epidemiology,
physiopathology of the most frequent
back disorders, posture, ergonomics,
and common treatment modalities and
practiced exercises (breathing, stretching
legs, trunk strengthening, and pelvic
mobility) for the maintenance of a
“healthy back.” [16, 24] Participants in the
MDT group practiced specific exercises
according to their mechanical diagnosis
and were instructed to follow the recommendations
of the book titled Treat Your
Own Back.  The care provider, who
treated the patients in both groups, completed
MDT training part A certified by
the McKenzie Institute of Brazil, has 1
year of experience, and received extensive
Back School training during her
undergraduate training program.
The outcome measures for this study
were (1) pain intensity, as measured with
the pain numerical rating scale (NRS), 
and (2) disability, as measured with the
Roland-Morris Disability Questionnaire
(RMDQ), [27, 28] at 1 month after randomization.
These were the same outcome
measures as in the primary study. [15, 16]
The NRS has good levels of reliability
(intraclass correlation coefficient [ICC
(2,1)] = .85; 95% confidence interval
[CI] = .77, .90), responsiveness (standardized
effect size = 1.16), and construct
validity.26 The RMDQ has good levels
of internal consistency (Cronbach
alpha = .90, reliability [ICC (2,1) = .94;
95% CI = .91, .96], responsiveness [standardized
effect size = 0.70], and construct
Variables of Interest
Four potential effect modifiers for treatment
were selected after consideration
of the theoretical rationale and consultation
with an educator in the MDT
approach.  The variables selected were:
(1) clear centralization versus directional
preference only, (2) baseline pain location,
(3) baseline pain intensity, and (4)
Clear centralization versus directional preference only.
hypothesized that patients with clear
centralization would respond better to
MDT than to Back School. Centralization
is considered a more “positive” response
to MDT assessment than directional prefence
alone. Patients were considered to
have centralization of symptoms if their
pain referred in a peripheral location
moved to the central lower back and was
progressively abolished, whereas if their
pain just decreased but did not move to
the central lower back, they were considered
to have directional preference
without centralization. Symptom diagram
and patient report of the location
of symptoms were used to determine
whether centralization had occurred.
Baseline pain location.
that patients with pain located
below the knee would respond better to
MDT than to Back School. Some preliminary
studies suggest people with leg
pain may respond well to the MDT
approach, [30–32] and there is little rationale
why Back School would specifically
help these people. The MDT approach
focuses on achieving centralization of
pain from the periphery into the low
back. Whether patients had pain extending
below the knee was determined
using a body chart and patient self-report
during the baseline assessment.
Baseline pain intensity.
that patients with higher baseline
pain intensity (using median split) would
respond better to MDT than to Back
School. This hypothesis was based on
clinical experience rather than any
strong existing evidence.
We hypothesized that patients
with younger age (using median split)
would respond better to MDT than to
Back School. The rationale was that they
might be able to move further into the
range of motion and, therefore, better
achieve an end-of-range position. Movement
to end of range is proposed to be
important to optimize response to MDT
in people classified as having a derangement
syndrome. [11, 20] This information
was determined during baseline
We investigated baseline patient characteristics
associated with greater effect of
MDT versus Back School separately for
outcomes of pain and disability. Each of
the 4 predictor variables was investigated
in separate univariate models.
The continuous effect modifiers of pain
intensity and age were dichotomized
using the median split method, as other
methods where optimal thresholds are
used have been shown to be substantially
biased and are recommended
against.  Thresholds dichotomizing pain
intensity and age have been used previously ;
however, these thresholds were
not specifically intended for our purpose,
and by using a median split, we
enhanced our statistical power by creating
equal-sized groups positive and negative
for the predictor. Each model
included terms for group, predictor, and
the interaction term, group X predictor.
The interaction term was used to quantify
size of the effect modification.
It has been estimated that the detection
of a statistically significant subgroup
interaction effect in an RCT requires a
sample size approximately 4 times that
required to detect a main effect of the
same size.  Previous authors have suggested
secondary analysis of RCTs as an
approach to develop hypotheses for
potentially important effect modifiers
that can then be tested in suitably large
trials. [35, 36]
As the current hypothesis-setting study was underpowered, we
focused on the estimated effect size
rather than statistical significance. If the
interaction was greater than 1.0 points
on the NRS or 3 points on the RMDQ, we
proceeded to investigate the potential
clinical importance by assessing the
effect of intervention (MDT compared
with Back School) separately for those
positive for the subgroup and those negative
for the subgroup. This was done by
calculating the marginal means for the
subgroups.  These thresholds are somewhat
arbitrary, as it is difficult to determine
exactly what a clinically important
interaction effect is. Previous work
suggests this is influenced by the main
treatment effect [18, 37] as well as the potential
harms and benefits of the
This secondary analysis was based on
existing data collected for an RCT15
approved by the Ethics Committee in
Research of the Universidade Cidade de
Sa˜o Paulo (number 134699394). The
RCT also was prospectively registered
with the Australian and New Zealand
Clinical Trials Registry
(ACTRN12610000435088). The research
protocol was published elsewhere.16
Role of the Funding Source
This study was funded by Fundacao de
Amparo a Pesquisa do Estado de Sao
Paulo (FAPESP), Coordenacao de Aperfeicoamento de Pessoal de Nivel Superior
(CAPES), and International Mechanical
Diagnosis and Therapy Research
Between July 2010 and January 2012, a
total of 182 patients who were seeking
care for LBP in the physical therapy
clinic of the Universidade Cidade de Sao
Paulo were screened for potential entry
into the study. Of these patients, 148
were considered eligible and randomized
(74 to each treatment group). The
reasons for ineligibility were cardiorespiratory
illnesses (n = 8), age over 80 years
(n = 5), acute LBP (n = 4), nerve root compromise
(n = 4), neck pain instead LBP
(n = 3), grade II spondylolisthesis (n = 2),
vertebral fracture (n = 1), rib fracture
(n = 1), deep vein thrombosis (n = 1),
abdominal tumor (n = 1), advanced osteoporosis
(n = 1), metabolic myopathy
(n = 1), colitis (n = 1), and urinary tract
infection (n = 1). All patients received the
treatments as allocated. A total of 148
patients were included in the main RCT.
However, 8 patients were not classified
as having directional preference and
were excluded from the analysis of this
study. Therefore, 140 patients were
included in this secondary analysis
Flow of patients within the study.
MDT = Mechanical Diagnosis and Therapy.
After dichotomizing, the average age of
older patients was 64.81 years
(SD = 5.95) (MDT group: mean
age = 65.16 years, SD = 6.23; Back School
group: mean age = 64.54 years,
SD = 7.78), and the average age of
younger people was 43.47 years
(SD = 9.60) (MDT group: mean
age = 44.28 years, SD = 9.50; Back School
group: mean age = 42.45 years,
SD = 9.78). The average score for higher
pain intensity was 8.07 points (SD = 1.0)
(MDT group: mean = 8.07 points,
SD = 1.0; Back School group: mean = 8.07
points, SD = 1.16), and the average score
for low level of pain intensity 4.35 points
(SD = 1.69) (MDT group: mean = 4.40
points, SD = 1.52; Back School group:
mean = 4.3 points, SD = 1.87). The baseline
characteristics of both groups,
including effect modifiers investigated in
this study, were similar. Nearly 75% of
the patients were women, and the average
symptom duration was approximately
2 years. Patients generally had
moderate levels of pain and disability
The results of the linear regression analyses
for the outcomes of pain and disability
are shown in Tables 2 and 3, respectively.
As expected in this hypothesissetting
study, none of the interaction
terms for any findings for these outcomes
were statistically significant. The
interaction term (–1.27) for age
exceeded our prespecified threshold of
> 1 for the outcome of pain (Tab. 2).
However, the direction of effect was
opposite to our hypothesis. Older people
appeared to benefit more from MDT
compared with Back School. For the outcome
of disability, the interaction term
for age was –2.39 and, therefore, did not
meet out threshold for clinical importance
of > 3 (Tab. 3). Interaction terms
for the other 3 effect modifiers (clear
centralization, pain below knee, and
high pain intensity) were below our
thresholds for potential clinical
Table 4 shows the effect of MDT compared
with Back School separately for
patients younger and older than 54 years.
Older people improved 1.42 points for
pain intensity and 3.51 points for disability,
more than younger people, from
MDT compared with Back School.
Statement of Principal Findings
The purpose of this hypothesis-setting,
secondary analysis was to investigate
whether 4 baseline characteristics (presence
of clear centralization, pain location,
pain intensity, and age) of patients
with chronic LBP can identify those who
respond better to MDT compared with
Back School. Based on our results, the
interaction term of –1.27 for age
exceeded our prespecified threshold of
> 1 for the outcome of pain, suggesting
older people may benefit more from
MDT compared with Back School. However,
the direction of effect was opposite
to our initial hypothesis, and as expected
the effect was not statistically significant,
so caution is required when interpreting
this finding. The presence of clear centralization,
pain located below the knee
and, pain intensity were not found to be
effect modifiers for response to MDT
compared with Back School.
Strengths and Weaknesses of the Study
A strength of our study was that the data
were derived from a high-quality RCT. 
The trial collected important potential
treatment effect modifiers from all
patients that are likely to influence the
response to MDT. Only 4 potential effect
modifiers were selected a priori after
consideration of the theoretical rationale
and consultation with a specialist musculoskeletal
physical therapist who is a credentialed
MDT therapist.  This therapist
has been an educator of the McKenzie
Institute since 1986 and has been the
International Director of Education for
the McKenzie Institute International
since 1999.  She is currently the only
teaching member of the McKenzie Institute
in Australia and teaches MDT
courses internationally.  We avoided
selecting a higher number of effect modifiers
in order to minimize the chance of
spurious findings.  The methodological
approach of this study, based on recommendations
in the literature, can act as a
model for subgroup studies within RCTs
in the rehabilitation field. Another contribution
of this study is the interpretation
of the actual findings leading to
hypotheses that can be tested in future
trials. The main limitation of this study
was the lack of statistical power for an
ideal subgroup analysis. Our sample
(n = 140) was powered to detect the
main effects of treatment but not to
detect the interactions of the potential
treatment effect modifiers.  For this reason,
this study was set up as a
hypothesis-generating study, and the
focus was on estimated effect size rather
than statistical significance.
We used a simple analysis that did not
contain any covariates. This approach
was chosen to minimize the risk of overfitting
the model and because the resulting
interaction effect size is exactly equal
to the difference between treatment
effect in one subgroup (eg, older age)
and treatment effect in the other subgroup
(eg, younger age), [18, 35–37] which
makes the finding easier to interpret.
However, to test potential confounding,
we conducted a post hoc analysis for the
predictor of age in which we added sex
and duration of symptoms to the models.
For both pain and disability models, this
analysis resulted in the interaction effect
becoming marginally greater (–2.85;
95% CI = –5.90, 0.20 for pain and –1.35;
95% CI = –3.45, 0.74 for disability) than
in the simple models.
Meaning of the Study and Comparison With Other Studies
Our study showed that patients who
were older than 54 years and received
MDT compared with Back School experienced
an additional reduction in pain
of 1.27 points compared with young
people. It is important to note that the
interaction does not define the main
effect of the interventions, but rather the
difference in effect of treatment for older
patients compared with young patients.6
The treatment effect within a subgroup
is a combination of the interaction and
the main treatment effect.  The findings
presented in Table 4 show that MDT was
statistically more effective for pain (1.42;
95% CI = 0.02, 2.83) compared with Back
School in the subgroup of older people,
whereas there was no difference
between the 2 interventions in younger
people (0.15; 95% CI = –1.38, 1.68). Similarly,
for disability, MDT was statistically
more effective for disability in the subgroup
of older people (3.51; 95%
CI = 1.42, 5.60) compared with Back
School, whereas there was no difference
between the 2 interventions in younger
people (1.12; 95% CI = –1.11, 3.35). We
only investigated treatment effects (MDT
versus Back School) within subgroups
where the interaction met our criteria for
clinical importance to reduce the chance
of spurious findings. 
We initially hypothesized that MDT
would be more effective in younger
patients, as they might be able to move
further into range of lumbar spine
motion and, therefore, gain more benefit.
Our findings suggesting that MDT
may be more effective in older people
raise the question of the potential mechanism
underlying this hypothesis. One
possibility is that older people were
more adherent to the approach that is
almost entirely a self-management intervention.
Another possibility is that pain
has a somewhat different physiological
basis in older people and responds better
to MDT. However, we do not have data
to support these theories. These results
may simply be spurious findings due to
lack of statistical power. We recommend
the investigation of age as a potential
effect modifier in future rehabilitation trials,
including but not limited to those
investigating MDT for spinal pain.
There are 2 previous studies that tested
possible treatment effect modifiers for
MDT. [17, 18] The first study recruited
patients with acute LBP who received
either MDT or usual care.  This study
tested baseline pain, pain changes with
position or movement, and presence of
leg pain as potential effect modifiers. The
authors found that these potential effect
modifiers did not predict a more favorable
response to MDT. The second
study18 compared MDT with spinal
manipulation in 350 patients with
chronic back pain. The authors included
6 predictor variables: centralization, age
below 40 years, duration of symptoms
more than 1 year, leg pain, pain below
the knee, signs of nerve root involvement.
and pain response.  They concluded
that it was not possible to find
any statistically significant predictive factor
that identified a better response to
either MDT or spinal manipulation. The
difference in our findings regarding age
as an effect modifier may be due to a
different control intervention or population or may simply be a spurious finding.
Subgroup effects within trials are always
specific to the control group.  We recommend
that our results be interpreted
carefully and that adequately powered
replication studies are needed. These
studies together suggest that it is difficult
to identify powerful effect modifiers for
MDT. This difficulty may be due to the
fact that the MDT approach already uses
a stratified approach to care. Interestingly,
the existing studies have not investigated psychosocial characteristics as
effect modifiers, and this is an area of
future investigation that we would
In conclusion, we conducted a secondary
analysis of an RCT to determine
whether potential treatment effect modifiers
for MDT could be identified in
patients with chronic LBP and with a
directional preference. We found that
patients who were older appeared to
respond better to MDT compared with
Back School (the direction of effect was
opposite to our initial hypothesis). Clear
centralization, pain below the knee, and
high pain intensity do not appear to be
useful effect modifiers. The results of this
hypothesis-setting, secondary analysis
have to be interpreted cautiously
because of the small sample size. These
findings, particularly of the potential
effect modification effect of age, need
testing in larger trials and with different
All authors provided concept/idea/research
design. Ms Garcia, Dr Hancock, and Dr Leonardo
Costa provided writing and data analysis.
Ms Garcia and Dr Luciola Costa provided
data collection, project management,
and consultation (including review of manuscript
before submission). Dr Hancock and
Dr Leonardo Costa provided fund procurement.
Dr Leonardo Costa provided participants
and institutional liaisons.
This study was funded by Fundac¸ao de
Amparo a Pesquisa do Estado de Sao Paulo
(FAPESP), Coordenac¸ao de Aperfeic¸oamento
de Pessoal de Nivel Superior (CAPES), and
International Mechanical Diagnosis and
Therapy Research Foundation (IMDTRF).
Chapter 1. European guidelines.
Eur Spine J. 2006;15:134–135.
The Back Pain Revolution. 2nd ed.
Edinburgh, United Kingdom:
Churchill Livingstone; 2004.
Menezes Costa Lda C, Maher CG, Hancock MJ, et al.
The prognosis of acute and persistent low-back pain: a meta-analysis.
Kent P, Hancock M, Petersen DH, Mjosund HL.
Clinimetrics corner: choosing appropriate study designs for particular questions about treatment subgroups.
J Man Manip Ther. 2011;18:147–152.
Kent P, Keating JL.
Classification in nonspecific low back pain: what methods do primary care clinicians currently use?
Spine (Phila Pa 1976). 2005;30:1433–1440.
Kent P, Keating JL, Leboeuf-Yde C.
Research methods for subgrouping low back pain.
BMC Med Res Methodol. 2010; 10:62.
Henry SM, Fritz JM, Trombley AR, Bunn JY.
Reliability of a treatment-based classification system for subgrouping people with low back pain.
J Orthop Sports Phys Ther. 2012;42:797–805.
Stanton TR, Hancock MJ, Appeldoorn AT, et al.
What characterizes people who have an unclear classification using a treatmentbased classification algorithm for low back pain? A cross-sectional study.
Phys Ther. 2013;93:345–355.
Hebert JJ, Koppenhaver SL, Walker BF.
Subgrouping patients with low back pain: a treatment-based approach to classification.
Sports Health. 2011;3:534 –542.
Beneciuk JM, Robinson ME, George SZ.
Subgrouping for patients with low back pain: a multidimensional approach incorporating cluster analysis and the STarT Back Screening Tool.
J Pain. 2015;16:19–
McKenzie R, May S.
The Lumbar Spine: Mechanical Diagnosis & Therapy. Vol 1. 2nd ed.
Waikanae, New Zealand: Spinal Publications Ltd; 2003.
Clare HA, Adams R, Maher CG.
A systematic review of efficacy of McKenzie therapy for spinal pain.
Aust J Physiother. 2004;50:209–216.
Machado LA, de Souza MS, Ferreira PH, Ferreira ML.
The McKenzie method for low back pain: a systematic review of the literature
with a meta-analysis approach.
Spine (Phila Pa 1976). 2006;31:254–262.
McKenzie classification of mechanical spinal pain: profile of syndromes and directions of preference.
Man Ther. 2008;13:75–81.
Garcia AN, Costa Lda C, da Silva TM, et al.
Effectiveness of Back School versus McKenzie exercises in patients with chronic nonspecific low back pain: a randomized controlled trial.
Phys Ther. 2013; 93:729–747.
Garcia AN, Gondo FL, Costa RA, et al.
Effectiveness of the Back School and McKenzie techniques in patients with chronic non-specific low back pain: a protocol of a randomised controlled trial.
BMC Musculoskelet Disord. 2011;12:179.
Sheets C, Machado LA, Hancock M, Maher C.
Can we predict response to the McKenzie method in patients with acute low back pain? A secondary analysis of a
randomized controlled trial.
Eur Spine J. 2011;21:1250–1256.
Petersen T, Christensen R, Juhl C.
Predicting a clinically important outcome in patients with low back pain following McKenzie therapy or spinal manipulation: a stratified analysis in a randomized controlled trial.
BMC Musculoskelet Disord. 2015;16:74–80.
Machado L, van Tulder MW, Lin CW, et al.
The McKenzie method for chronic nonspecific low back pain: study protocol.
Cochrane Database Syst Rev. 2012;3: CD009712.
McKenzie R, May S.
The Lumbar Spine: Mechanical Diagnosis & Therapy. Vol 2. 2nd ed.
Waikanae, New Zealand: Spinal Publications Ltd; 2003.
Sun X, Briel M, Busse JW, et al.
Subgroup Analysis of Trials Is Rarely Easy (SATIRE): a study protocol for a systematic review to characterize the analysis, reporting, and claim of subgroup effects in randomized trials.
Airaksinen O, Brox JI, Cedraschi C, et al.
Chapter 4. European guidelines for the management of chronic nonspecific low back pain.
Eur Spine J. 2006;15:192–300.
Kenney LW, Humphrey RH, Mahler DA.
ACSM’s Guidelines for Exercise Testing and Prescription.
Baltimore, MD: Williams & Wilkins; 1995.
Heymans MW, van Tulder MW, Esmail R, et al.
Back schools for nonspecific low back pain: a systematic review within the framework of the Cochrane Collaboration Back Review Group.
Spine (Phila Pa 1976). 2005;30:2153–2163.
Trate Voceˆ Mesmo a Sua Coluna.
Waikanae, New Zealand: Spinal Publication Ltd; 1998.
Costa LO, Maher CG, Latimer J, et al.
Clinimetric testing of three self-report outcome measures for low back pain patients in Brazil: which one is the best?
Spine (Phila Pa 1976). 2008;33:2459–2463.
Costa LO, Maher CG, Latimer J, et al.
Psychometric characteristics of the Brazilian-Portuguese versions of the Functional Rating Index and the Roland-Morris Disability Questionnaire.
Spine (Phila Pa 1976). 2007;32:1902–1907.
Nusbaum L, Natour J, Ferraz MB, Goldenberg J.
Translation, adaptation and validation of the Roland-Morris questionnaire: Brazil Roland-Morris.
Braz J Med Biol Res. 2001;34:203–210.
Dr Helen Clare.
McKenzie Institute Australia. Available at:
Accessed September 9, 2015.
Paatelma M, Kilpikoski S, Simonen R, et al.
Orthopaedic manual therapy, McKenzie method or advice only for low back pain in working adults: a randomized controlled trial with one year follow-up.
J Rehabil Med. 2008;40:858–863 .
Albert HB, Hauge E, Manniche C.
Centralization in patients with sciatica: are pain responses to repeated movement and positioning associated with outcome or types of disc lesions?
Eur Spine J. 2012; 21:630–636.
Long A, May S, Fung T.
The comparative prognostic value of directional preference and centralization: a useful tool for frontline
J Man Manip Ther. 2008; 16:248–254.
Altman DG, Royston P.
The cost of dichotomising continuous variables.
Brookes ST, Whitely E, Egger M, et al.
Subgroup analyses in randomized trials: risks of subgroup-specific analyses; power and sample size for the interaction test.
J Clin Epidemiol. 2004;57:229–236.
Hancock M, Herbert RD, Maher CG.
A guide to interpretation of studies investigating subgroups of responders to physical therapy interventions.
Phys Ther. 2009;89:698–704.
Kamper SJ, Maher CG, Hancock MJ, et al.
Treatment-based subgroups of low back pain: a guide to appraisal of research studies and a summary of current evidence.
Best Pract Res Clin Rheumatol. 2010;24: 181–191.
Hancock MJ, Kjaer P, Korsholm L, Kent P.
Interpretation of subgroup effects in published trials.
Phys Ther. 2013;93:852–859.
Ferreira ML, Herbert RD, Ferreira PH, et al.
A critical review of methods used to determine the smallest worthwhile effect of interventions for low back pain.
J Clin Epidemiol. 2012;65:253–261.
Sun X, Briel M, Walter SD, Guyatt GH.
Is a subgroup effect believable? Updating criteria to evaluate the credibility of subgroup analyses.
Brookes ST, Whitley E, Peters TJ, et al.
Subgroup analyses in randomised controlled trials: quantifying the risks of falsepositives and false-negatives.
Health Technol Assess. 2001;5:1–56
Return to the LOW BACK PAIN Section