Content uploaded by Alice Kvåle
Author content
All content in this area was uploaded by Alice Kvåle on Nov 10, 2017
Content may be subject to copyright.
Original Article
Inter-examiner reliability of a classification system for patients
with non-specific low back pain
K. Vibe Fersum
a
,
*
, P.B. O’Sullivan
b
, A. Kvåle
a
, J.S. Skouen
a
,
c
a
Section for Physiotherapy Science, Department of Public Health and Primary Health Care, University of Bergen, Kalfarveien 31, 5018 Bergen, Norway
b
School of Physiotherapy, Curtin University, Bentley 6102, WA, Australia
c
The Outpatient Spine Clinic, Department of Physical Medicine and Rehabilitation, Haukeland University Hospital, Bergen, Norway
article info
Article history:
Received 20 February 2008
Received in revised form 10 July 2008
Accepted 1 August 2008
Keywords:
Agreement
Classification
Low back pain
Reliability
abstract
There is a lack of studies examining whether mechanism-based classification systems (CS) acknowl-
edging biological, psychological and social dimensions of long-lasting low back pain (LBP) disorders can
be performed in a reliable manner. The purpose of this paper was to examine the inter-tester reliability of
clinicians’ ability to independently classify patients with non-specific LBP (NSLBP), utilising a mecha-
nism-based classification method. Twenty-six patients with NSLBP underwent an interview and full
physical examination by four different physiotherapists. Percentage agreement and Kappa coefficients
were calculated for six different levels of decision making. For levels 1–4, percentage agreement had
a mean of 96% (range 75–100%). For the primary direction of provocation Kappa and percentage
agreement had a mean between the four testers of 0.82 (range 0.66–0.90) and 86% (range 73–92%)
respectively. At the final decision making level, the scores for detecting psychosocial influence gave
a mean Kappa coefficient of 0.65 (range 0.57–0.74) and 87% (range 85–92%). The findings suggest that the
inter-tester reliability of the system is moderate to substantial for a range of patients within the NSLBP
population in line with previous research.
Ó2008 Elsevier Ltd. All rights reserved.
1. Introduction
LBP represents a common and very costly health problem and
a definite diagnosis is difficult to achieve in most cases (85%)
(Waddell, 2004). As a result, uncertainty regarding treatment of
this group of patients is common (Cherkin et al., 1998).
A number of studies have shown little or no difference between
various physiotherapy treatments for chronic LBP (Delitto et al.,
1995; Petersen et al., 1999; Ferreira et al., 2007). Several authors
have suggested that these results may reflect the heterogeneity of
the NSLBP group, with several distinct subgroups, including
psychosocial problems, each with its own potential set of beneficial
treatments (O’Sullivan, 2000; Petersen et al., 2002; O’Sullivan,
2005; Dankaerts et al., 2006b). There is growing evidence sug-
gesting that sub-classifying patients and offering them tailored
interventions matching their disorder improves patient outcome
(Frymoyer et al., 1985; Main and Watson, 1996; O’Sullivan, 1997;
Nachemson, 1999; Linton, 2000; Skouen et al., 2002; Fritz et al.,
2003; Stuge et al., 2004). It has been proposed that a classification
system (CS) for NSLBP should identify the underlying mechanisms
driving the disorder within a bio-psycho-social framework,
enabling specific therapies to be applied so as to favourably influ-
ence the outcome of the disorder (O’Sullivan, 2005).
A number of CS have been proposed (McKenzie, 1981; Spitzer,
1987;Maluf et al., 2000; Sahrmann, 2001). However, only a few are
found sufficiently reliable and valid (Petersen et al.,1999), and even
fewer consider the disorder from a bio-psycho-social perspective
(Petersen et al., 1999; Ford et al., 2003; McCarthy et al., 2004;
O’Sullivan, 2005; Dankaerts et al., 2006b).
The Quebec Task Force system was designed to classify all LBP
patients to help with clinical decision making, establishing prog-
nosis and evaluating treatment effectiveness (Spitzer, 1987).
However, it has not been tested for reliability and does not consider
the underlying mechanism (Dankaerts et al., 2006b), except for
differentiating somatic from radicular pain. Within this system
there is no subgrouping of NSLBP except on the basis of pain area,
and no specific treatment is advocated for this large group of
patients other than general exercise, therefore limiting its use for
physiotherapy assessment and treatment (Padfield et al., 2002).
The McKenzie (1981) system is based on information from
history taking, and symptom response to generated loading of the
lumbar spine. The system has been tested for reliability, and has
substantial inter-tester agreement when applied by trained
examiners (Kappa coefficients ranging from 0.6 to 0.7) (Kilpikoski
et al., 2002).
*Corresponding author: Tel.: þ47 55586711.
E-mail address: kjartan.fersum@isf.uib.no (K. Vibe Fersum).
Contents lists available at ScienceDirect
Manual Therapy
journal homepage: www.elsevier.com/math
1356-689X/$ – see front matter Ó2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.math.2008.08.003
Manual Therapy 14 (2009) 555–561
Petersen and co-workers (2004) have proposed a McKenzie-
based CS with good inter-tester reliability, but it has a patho-
anatomical orientation and lacks clear guidelines for management.
Sahrmann and co-workers have developed another CS,
comprising five categories based on testing of muscular stability,
alignment, asymmetry, flexibility of the lumbar spine, pelvis, and
hip (Maluf et al., 2000). Reliability of the individual tests used for
classification has been shown to vary from fair to almost perfect
(Van Dillen et al., 1998, 2003). However, there are no reports on
reliability in classification of the patients into the five categories,
nor does this system consider patho-anatomical or psychosocial
dimensions.
Since 1997 Peter O’Sullivan has developed a novel system, based
on the Quebeck Task Force, incorporating multiple dimensions in
the classification of patients into subgroups based on proposed
underlying pain mechanisms. Initially, this mainly targeted
a subgroup of patients with localised NSLBP where provocative
movement behaviours and positions of the spine, associated with
a loss of spinal control, represent a mechanism for ongoing pain.
These patients are classified as LBP patients with motor control
impairment (MCI). The evidence validating this subgroup is
growing (O’Sullivan et al., 1997, 2005; O’Sullivan, 1997, 200 0, 2003;
Dankaerts et al., 2006a) and the reliability of clinicians to identify
these different subgroups has been established (Dankaerts et al.,
2006b). Lately, this approach has also incorporated classification of
patients with lumbo-pelvic pain and a wider range of pain mech-
anisms linked to their disorder (O’Sullivan, 2005; O’Sullivan and
Beales, 2007a). This system differentiates between specific LBP
versus NSLBP. NSLBP is further split into subgroups based on the
proposed driving mechanism behind the disorder (Fig. 1). The
classification is based on a systematic examination process
(subjective history, objective examination and available medical
information). Within this system psychosocial factors are accoun-
ted for, acknowledging their potential to amplify pain and drive
disability. To date the ability of clinicians to agree on this broad
classification process has not been formally tested.
Validating the system has been a multi-step process, in which
establishing inter-tester reliability is crucial. The aim of this study
was therefore to examine the inter-tester reliability of clinicians’
ability to independently classify a wide range of patients with
NSLBP, utilising an extended mechanism-based classification
method lately developed by O’Sullivan.
2. Methods
The study was conducted from March 2006 to June 2006, and
was approved by the regional ethics committee of medical research
in western Norway.
2.1. Patients
Patients were recruited consecutively from physiotherapy
clinics around Bergen and from The Outpatient Multidisciplinary
Spine Clinic, Haukeland University Hospital. After recruitment,
a telephone screening was performed, and the first 30 patients that
fit the inclusion criteria, were tested (Table 1). Since the patients
were tested twice on each occasion, a 0–10 pain numerical rating
scale was conducted prior to each testing. If a patient’s pain score
changed 2 levels between two examinations on the same day, this
was considered to be a threat to the classification validity and the
patient would then be excluded. Four patients were excluded after
further examination: three did not fulfil the inclusion criteria and
one reported a two-level change in pain between examinations on
the given day.
This left 26 patients participating in the study. See Table 2 for the
patients’ characteristics. Prior to the study, design and possible
risks were fully explained to each subject, and all signed a consent
form.
2.2. Examiners
There were four physiotherapists, each with several years of
experience in examination and treatment of LBP patients (mean
12 years, range 7–20 years). Three of the four examiners were
physiotherapists with a masters degree in manual therapy. One was
the developer of the system.
2.3. Training
All the examiners had been educated in the CS during several
workshops with the developer, and were using it in their clinical
practice. Prior to the study, O’Sullivan explained procedures and
classifications were discussed using a series of case studies. The
examiners also underwent a pilot training period where O’Sullivan
examined and classified six patients, while the three others
observed. The aim was to refine the specific criteria for assessment,
as well as making testers more familiar with the system. The esti-
mated training time for each therapist ranged from 69 to 140 h, the
average being 106.3 h (workshops and pilot study included).
2.4. Clinical procedure
A test–retest design was utilised. A classification manual was
developed by O’Sullivan prior to the study. The patients underwent
acomprehensive interview and full physical examination by each of
the four physiotherapists. Rather than assess the reliability of
individual tests, this system involved making a disorder classifica-
tion based on compilation of subjective and physical examination
findings in relation to other medical tests and radiological imaging.
The subjective assessment included pain area (pain drawing),
intensity and nature, pain behaviour (aggravating/easing move-
ments), identification of primary impairments, disability levels,
avoidance behaviours, pain coping and pain beliefs. The examina-
tion involved assessment of spinal range of movement, analysis of
the patient’s primary physical impairments (pain provocative and
easing postures, movements and functional tasks). Specific muscle
and movement tests were performed to identify the relationship
between the control of the lumbo-pelvic region and the pain
disorders (O’Sullivan, 2000), as well as specific articular tests for the
lumbar spine and pelvic region as indicated to identify the struc-
tural source of pain and the presence of movement impairments
(MI). These are important elements in the classification of the pain
disorder and in determining whether the habitual movements or
postures are provocative or protective (O’Sullivan, 2000, 2005;
O’Sullivan and Beales, 2007a,b). The process consists of several
stages before reaching a classification (Fig. 1):
1. The first part involves screening; determining if the condition
is specific LBP or NSLBP (O’Sullivan, 2005).
2. The second stage considers whether specific LBP disorders have
an adaptive or maladaptive response to the disorder (O’Sulli-
van, 2005). If the disorder is classified as non-specific, then
consideration of whether the disorder is predominantly cen-
trally or peripherally mediated is made. The presence of
localised and anatomically defined pain, associated with
specific and consistent mechanical aggravating and easing
factors, suggests that physical/mechanical factors are likely to
dominate the disorder resulting in a peripheral nociceptive
drive. Constant, non-remitting widespread pain, not influenced
by mechanical factors, could on the other hand indicate
inflammatory or centrally driven pain (O’Sullivan, 2005).
K. Vibe Fersum et al. / Manual Therapy 14 (2009) 555–561556
3. Centrally mediated pain can then be further sub-classified into
the presence of non-dominant or dominant psychosocial
factors. Peripherally mediated disorders are sub-classified into
either LBP or a pelvic girdle pain disorders.
4. Peripherally mediated lumbar spine pain disorders are divided
into MI or MCI disorders and peripherally mediated pelvic
girdle pain into excessive or deficit of force closure. Both these
classifications have been described in detail elsewhere
(O’Sullivan, 2005; O’Sullivan and Beales, 2007a,b).
5. If the lumbar spine is the source of pain, the primary directional
provocation bias as well as the symptomatic spinal level is
noted.
Classification process adapted from Peter O’ Sullivan
Chronic back pain disorders
- Spondylolisthesis
- disc herniation + radicular pain
- degenerative disc + modic
changes
- foraminal and central stenosis
- Multi-
disciplinary
management
Psychological
(CBT), medical,
functional
rehabilitation
Non-specific back pain disordersSpecific back pain disorders
Dominant
psycho-
social
factors
Non-
dominant
psycho-
social factors
- Medical
management
- Functional
rehabilitation
Control
impairment
(directional
subgroups)
- Motor
learning
within
cognitive
framework
(enhance
control)
- Functional
restoration
Red flag disorders
Cancer
Infection
Inflammatory disorder
Fracture
Adaptive response
Patients response to
disorder is adaptive /
protective
Mal-adaptive
Patients response to
disorder is mal-adaptive
Management
Advise, medical,
surgical – as
appropriate
Management
- Cognitive / Motor
learning
- Medical
Pelvic girdle pain Low Back Pain
Reduced
force closure
Excessive
force closure
- Motor
learning
within
cognitive
framework
(reduce force
closure/
relaxation)
- Functional
restoration
- Motor
learning
within
cognitive
framework
(enhance
force closure)
- Functional
restoration
+/- central pain modulation based on contribution of
psycho-social factors
Directional
subgroups
(+ level of
dysfunction)
Directional
subgroups
(+ level of
dysfunction)
Level 4
Level 3
Level 2
Level 5
Level 6
Movement
impairment
(directional
subgroups)
Level 1
Centrally mediated back pain Peripherally mediated back pain
Fig. 1. Classification process adapted from Peter O’Sullivan (O’Sullivan, 2005; O’Sullivan and Beales, 2007a,b).
K. Vibe Fersum et al. / Manual Therapy 14 (2009) 555–561 557
6. The final decision is to indicate if significant psychosocial
factors are associated with the disorder, based on all informa-
tion from the examination process. The evaluation of psycho-
social factors considers the presence of underlying fear
avoidance behaviour, as well as psychological and social drivers
considered to contribute to the pain disorder. Within this
reasoning process, consideration is given to whether the
patient has adapted in a positive (confrontation, active coping
and minimal avoidance behaviours) or negative manner
(passive coping and fear avoidance).
Each testing took about 1 h. The patient was examined inde-
pendently twice on two days, within a 1-week period. Each ther-
apist filled out a classification form (see Supplementary Appendix
A.1) and put it in a sealed opaque envelope after their patient
assessment. After examination the patient completed several
questionnaires to formally assess their disorder. This included
a pain drawing, a functional assessment chart from the Dartmouth
Primary Care Cooperative Information Project (COOP/WONCA),
Oswestry Disability Index (ODI), Hopkins Symptoms Check List
(HSCL), Fear Avoidance Beliefs Questionnaire (FABQ) and Ørebro
Musculoskeletal Pain Screening Questionnaire (Ørebro MSPSQ).
2.5. Analysis
After completed examinations, the results were compared and
logged. The developer’s classification of each patient was used as
the gold standard to which the other results were compared. Kappa
coefficients and percentage of agreement were calculated using
SPSS 13.0 for Windows. Cohen’s Kappa statistic was used to
calculate inter-tester reliability and Landis and Koch’s (1977) values
for interpretation of the reliability scores were used. Kappa values
<0.20 indicate poor agreement, 0.21–0.40 fair, 0.41–0.60 moderate,
0.61–0.80 substantial, and 0.81–1.00 indicate almost perfect
agreement. The data was analysed based on agreement of overall
classification (specific LBP vs NSLBP), centrally or peripherally
mediated, adaptive or maladaptive movement disorders, and
whether it was considered to be a pelvic girdle pain or LBP disorder.
Kappa agreement of the primary directional pain provocation, the
spinal level of pain provocation and the presence of psychosocial
influence on their LBP disorder was calculated.
3. Results
In the first part of the classification process, all patients were
classified with NSLBP with 98% agreement for this level. All patients
in the study had pain arising from a peripheral pain source, with
99% agreement for this. One patient was classified by all four testers
as having pelvic girdle pain (100% agreement); the rest were clas-
sified as LBP disorders (99% agreement). The fourth level consid-
ered increased or decreased force closure for pelvic pain (one
patient, 100% agreement), MCI (24 patients, 99% agreement) or MI
(one patient, 75% agreement) for low back. In the fifth level, Kappa
agreement could be calculated, deciding the directional pattern of
provocation (Fig. 2). For the primary direction of provocation,
Kappa (K) and percentage agreement had a mean between the four
testers of 0.82 (range 0.66–0.90) and 86% (range 73–92%) respec-
tively. Increased familiarity with the system also increased the
reliability results (<100 h K¼0.66, >100 h K¼0.90). In the final
level of decision making, the mean Kappa coefficient for detecting
psychosocial influence was 0.65 (range 0.57–0.74) and the mean
agreement 87% (range 85–92%).
4. Discussion
The principal finding of our study suggests that therapists with
substantial training in this CS (O’Sullivan, 2005) demonstrated fair
to excellent agreement (Landis and Koch, 1977) in primary classi-
fication of the disorder, identification of directional patterns of
provocation and the presence of psychosocial factors associated
with the disorder, when applied to a range of NSLBP patients. Our
findings are in accordance with a recent study (Dankaerts et al.,
2006b), who also found moderate to excellent agreement between
testers examining patients within the MCI subgroup. Their study
consisted of two separate parts. The first part demonstrated almost
perfect agreement between two expert clinicians when classifying
35 patients with MCI identified from a clinical case load, into the
various directional patterns (K¼0.96, agreement 97%). In the
second part, 25 out of 35 patients with MCI in the first study were
randomly selected. These were videotaped and classified into
directional groups by 13 other therapists based on the video and
subjective complaints of the patients. The agreement between the
13 different raters was moderate to excellent (mean Kappa 0.61,
agreement 70%).
As in Dankaerts et al.’s study (2006b), familiarity with the CS
also influenced the reliability results, demonstrating higher
agreement among raters with more CS training. These findings are
in line with Strender’s study (1997), concluding that reliability of
clinical tests requires sufficient time for examination and confor-
mity of performance, definitions and evaluations. The protocol of
our study followed a similar examination procedure as the first part
of Dankaerts et al.’s (2006b) study. By including any patient with
localised low back pain in our study, we anticipated a more het-
erogenic NSLBP population with the inclusion of patients with back
pain associated with MI as well as pelvic girdle pain disorders.
However, 24 out of the 26 patients were classified as having MCI,
which is in line with the findings of Dankaerts et al. (2006b).
Furthermore, the current study involved four therapists examining
the patients versus two in the first part of Dankaerts study (2006b).
Table 1
Inclusion/exclusion criteria.
Inclusion criteria Exclusion criteria
Patients with non-specific LBP
(NSLBP) (6 weeks)
Sick-listed for more than 4 months
continuous duration during last year
Male or female Acute exacerbation of LBP
Age between 18 and 65 years Radicular pain. Positive neural tissue
provocation tests (primary peripheral
symptoms)
Localised LBP: primarily in the area
from T12 to gluteal folds
Any low limb surgery on the last 3 months
Moderate ongoing LBP, VAS >2/10
and Oswestry >14 %
Surgery involving the lumbar spine (fusion)
Mechanical provocation of pain:
postures, movement and activities
Pregnancy
Psychiatric disorders
Widespread non-specific pain disorder (no
primary LBP focus)
Specific diagnoses: active rheumatologic
disease, progressive neurological disease,
serious cardiac or other internal medical
disease
Table 2
Patients’ characteristics.
Number of patients 26
Female 11
Male 15
Mean age (years) 32.4
Mean pain intensity 6/10
Mean duration (years) 4.9
Mean Oswestry 21.2/100
Mean HSCL 1.53/4
Mean Ørebro score 87.5/210
K. Vibe Fersum et al. / Manual Therapy 14 (2009) 555–561558
Flexion pattern n =24
79,1
04,2 4,2 8,3
04,2
0
10
20
30
40
50
60
70
80
90
100
Flexion
Lateral shift
Active Extension
Passive Extension
Multidirectional
Sacroiliac
Movement
Impairment
Correct
Incorrect
Classification per
pattern (%)
Flexion
Lateral shift
Active Extension
Passive Extension
Multidirectional
Sacroiliac
Movement
Impairment
0
10
20
30
40
50
60
70
80
90
100
Classification per
pattern (%)
Lateral shift n= 8
75
00000
12,5 12,5
Active extension n=8
25
0
25
00
50
0
10
20
30
40
50
60
70
80
90
100
Flexion
Lateral Shift
Active Extension
Passive Extension
Multidirectional
Sacroiliac
Movement Impairment
Classification per
pattern (%)
Passive extension n=12
000 000
0
10
20
30
40
50
60
70
80
90
100 100
Flexion
Lateral Shift
Active Extension
Passive Extension
Multidirectional
Sacroiliac
Movement Impairment
Classification per
pattern (%)
Multidirectional n=44
75
11,4
04,5 2,3 2,3 4,5
0
10
20
30
40
50
60
70
80
90
100
Flexion
Lateral Shift
Active Extension
Passive Extension
Multidirectional
Sacroiliac
Movement Impairment
Classification per
pattern (%)
Classification per
pattern (%)
Classification per
pattern (%)
Sacroiliac n=4
75
25
00 00 0
0
10
20
30
40
50
60
70
80
90
100
Flexion
Lateral Shift
Active Extension
Passive Extension
Multidirectional
Sacroiliac
Movement Impairment
Movement impairment n=4
75
00
25
0
0
10
20
30
40
50
60
70
80
90
100
Flexion
Lateral Shift
Active Extension
Passive Extension
Multidirectional
Sacroiliac
Movement Impairment
Correct
Incorrect
Correct
Incorrect
Correct
Incorrect
Correct
Incorrect
Correct
Incorrect
Correct
Incorrect
0
00
Fig. 2. Classification per different pattern (in %) by all examiners; n¼total number of that specific pattern included 4 (total number of examiners).
K. Vibe Fersum et al. / Manual Therapy 14 (2009) 555–561 559
This may explain the greater reliability in this aspect of Dankaerts
study, in comparison to ours. With regards to the second part of
Dankaerts et al.’s study (2006b), it was acknowledged that the use
of previously collected information (both subjective and video)
represented a bias for the 13 clinicians. In our study, the testers did
not have any prior information regarding the patient’s disorder as
this could influence the classification reliability, as different raters
may gather information from patients in different ways.
Eight subjects in our study out of 26 with disorders classified as
peripherally mediated NSLBP were also identified as having
moderate, but significant psychosocial factors contributing to their
disorders. Analysis of the questionnaire data collected after all
assessments, confirmed that these eight patients scored signifi-
cantly higher on HSCL and Ørebro MSPSQ (p<0.05). Linton and
Hallden (1998) identified potential psychosocial risk factors asso-
ciated with future sick absenteeism in a study, using the Ørebro as
the screening instrument. High total scores were related to
outcome and to cut-off points that correctly identified the prog-
nosis of nearly 80% of the patients. Psychosocial factors can
modulate pain behaviour, which then can increase disability via
fear avoidance, as well as promoting pain levels via central mech-
anisms (Vlaeyen and Linton, 2000). However there is little evidence
to date that physiotherapists can identify these subjects at risk,
based on subjective examination.
It has been emphasised (Dankaerts et al., 2006b) that the
development of a multi-dimensional mechanism-based CS based
on a bio-psycho-social framework should be seen as a critical
development of a CS. The Quebeck Task Force has been considered
by many as the first CS that included biomedical, psychological and
social considerations in the classification process (McCarthy et al.,
2004). The system used in our study, developed by O’Sullivan,
utilises the Quebeck Task Force as an underlying framework, by
classifying specific LBP versus NSLBP, the stage of the disorder, and
the presence of red and dominant yellow flags. However, patients
are sub-classified further, identifying the primary direction of
provocation and the proposed underlying mechanism of the
disorder. Furthermore, very specific interventions are indicated for
the different classifications (O’Sullivan, 2005; O’Sullivan and Beales,
2007a,b).
In contrast, the McKenzie CS is a bio-system that lacks validity
within a chronic LBP population, as only about 40% of patients have
a directional pain preference (Donelson et al., 1990). Consistent
with our findings, 45% of the subjects were classified as having MCI
with multi-directional pain provocation, suggesting that a uni-
directional preference was not present. This lack of uni-directional
preference limits the use of directional treatment methods as
advocated by McKenzie.
Interestingly, 25 of the patients in our study had MCI, and only
one had MI. This finding is consistent with reports that impair-
ments of range of motion are often not present in chronic low back
pain disorders (Nattrass et al., 1999). However the lack of subjects
with MI disorders in this study limits the ability to confirm the
reliability of physiotherapists when identifying this subgroup.
The Sahrmann CS for NSLBP proposes a single mechanism for
LBP (movement dysfunction), but does not consider specific diag-
nosis of LBP, CNS mediated pain, MIs or psychosocial factors,
limiting its application within a chronic LBP setting. Petersen et al.
(2004) in contrast proposed a system that demonstrated substan-
tial reliability, but it lacked clear guidelines for management.
Reliability can be influenced by many different factors. The
participants seemed representative of the population normally
seen in primary health care, but the small sample may not be
representative of the chronic LBP population. The first part of the
classification process in this study was to determine whether the
patient’s condition was specific or non-specific. Secondly, an
assessment was made to classify the source of the underlying
mechanism as being centrally or peripherally driven. Our study’s
inclusion criteria were aimed at subjects with localised NSLBP that
was mechanically provoked, making it more likely that they had
a peripheral pain disorder. None of our subjects were classified
with neurogenic pain. This fits with Bogduk’s study (1995), which
concluded that most NSLBP disorders are peripherally mediated,
having a pain source that most likely is discogenic or from the facet
joint and less commonly from the sacroiliac joint.
It can be argued that the Kappa scores could have been higher if
all the testing procedures had been standardised. However, the
study’s intention was to evaluate the reliability as a result of the
whole examination as performed in clinical practice, and stand-
ardising the examination for this heterogenic group of patients
could have influenced the validity.
5. Conclusion
The findings provide evidence that the inter-tester reliability of
O’Sullivan’s CS is substantial for a range of patients within the
NSLBP population in line with previous research. Using a mecha-
nism-based CS has implications in terms of treatment being
directed towards identified subgroups. The use of the CS is
currently being evaluated in a randomised controlled trial in order
to compare the efficacy of different interventions for any given
category.
Appendix A. Supplemental material
Supplementary information for this manuscript can be down-
loaded at doi: 10.1016/j.math.2008.08.003.
References
Bogduk N. The anatomical basis for spinal pain syndromes. Journal of Manipulative
and Physiological Therapeutics 1995;18(9):603–5.
Cherkin D, Deyo R, Battie M, Street J, Barlow W. A comparison of physical therapy,
chiropractic manipulation, and provision of an educational booklet for the
treatment of patients with low back pain. New England Journal of Medicine
1998;339:1021–9.
Dankaerts W, O’Sullivan PB, Burnett AF, Straker LM. Differences in sitting posture
are associated with non-specific chronic low back pain disorders when patients
are sub-classified. Spine 2006a;31(6):698–704.
Dankaerts W, O’Sullivan PB, Straker LM, Burnett AF, Skouen JS. The inter-examiner
reliability of a classification method for non-specific chronic low back pain
patients with motor control impairment. Manual Therapy 2006b;11(1):28–39.
Delitto A, Erhard RE, Bowling RW. A treatment-based classification approach to low
back syndrome: identifying and staging patients for conservative treatment.
Physical Therapy 1995;75(6):470–85.
Donelson R, Silva G, Murphy K. Centralization phenomenon. Its usefulness in
evaluating and treating referred pain. Spine 1990;15(3):211–3.
Ferreira ML, Ferreira PH, Latimer J, Herbert RD, Hodges PW, Jennings MD, et al.
Comparison of general exercise, motor control exercise and spinal manipulative
therapy for chronic low back pain: a randomized trial. Pain 2007;131(1-2):31–7.
Ford J, Story I, McKeenen J. A systematic review on methodology of classification
system research for low back pain. Musculoskeletal Physiotherapy Australia
13th Biennial Conference, Sydney, Australia, 2003.
Fritz JM, Delitto A, Erhard RE. Comparison of classification-based physical therapy
with therapy based on clinical practice guidelines for patients with acute low
back pain – a randomized clinical trial. Spine 2003;28(13):1363–71.
Frymoyer J, Rosen J, Clements J, Pope M. Psychological factors in low back pain
disability. Clinical Orthopaedics and Related Research 1985;May;(195):178–84.
Kilpikoski S, Airaksinen O, Kankaanpaa M, Leminien P, Viderman T, Alen M. Inter-
examiner reliability of low back pain assessment using the Mckenzie method.
Spine 2002;27(8):207–14.
Landis JR, Koch GG. The measurement of observer agreement for categorical data.
Biometrics 1977;33(1):159–74.
Linton SJ. A review of psychological risk factors in back and neck pain. Spine
2000;25(9):1148–56.
Linton SJ, Hallden K. Can we screen for problematic back pain? A screening ques-
tionnaire for predicting outcome in acute and subacute back pain. Clinical
Journal of Pain 1998;14(3):209–15.
Main C, Watson P. Guarded movements: development of chronicity. Journal of
Musculoskeletal Pain 1996;4(4):163–70.
Maluf KS, Sahrmann SA, Van Dillen LR. Use of a classification system to guide
nonsurgical management of a patient with chronic low back pain. Physical
Therapy 2000;80(11):1097–111.
K. Vibe Fersum et al. / Manual Therapy 14 (2009) 555–561560
McCarthy C, Arnall F, Strimpakos N, Freemont A, Oldham J. The biopsychosocial
classification of non-specific low back pain: a systematic review. Physical
Therapy Reviews 2004;9:17–30.
McKenzie R. The lumbar spine, mechanical diagnosis and treatment. Waikanae,
New Zealand: Spinal Publications Ltd; 1981.
Nachemson A. Back pain; delimiting the problem in the next millennium. Inter-
national Journal of Law Psychiatry 1999;22(5-6):473–80.
Nattrass CL, Nitsche JE, Disler PB, Chou MJ, Ooi KT. Lumbar spine range of motion as
a measure of physical and functional impairment: an investigation of validity.
Clinical Rehabilitation 1999;13:211–8.
O’Sullivan PB. Evaluation of specific stabilizing exercise in the treatment of chronic
low back pain with radiologic diagnosis of spondylolysis or spondylolisthesis.
Spine 1997;22(24):2959–67.
O’Sullivan PB. Lumbar segmental ’instability’: clinical presentation and specific
stabilizing exercise management. Manual Therapy 2000;5(1):2–12.
O’Sullivan PB. Lumbar repositioning deficit in a specific low back pain population.
Spine 2003;28(10):1074–9.
O’Sullivan P. Diagnosis and classification of chronic low back pain disorders: mal-
adaptive movement and motor control impairments as underlying mechanism.
Manual Therapy 2005;10(4):242–55.
O’Sullivan PB, Beales DJ. Diagnosis and classification of pelvic girdle pain disorders –
Part 1: a mechanism based approach within a biopsychosocial framework.
Manual Therapy 2007a;12(2):86–97.
O’Sullivan PB, Beales DJ. Diagnosis and classification of pelvic girdle pain disorders –
Part 2: illustration of the utility of a classification system via case studies.
Manual Therapy 2007b;12(2):1–12.
O’Sullivan P, Twomey L, Allison G, Sinclair J, Miller K, Knox J. Altered patterns of
abdominal muscle activation in patients with chronic back pain. Australian
Journal of Physiotherapy 1997;43(2):91–8.
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 2002:254–60.
Petersen T, Kryger P, Ekdahl C, Olsen S, Jacobsen S. Theeffect of McKenzie therapy as
compared with that of intensive strengthening training for the treatment of
patients with subacute or chronic low back pain: a randomized controlled trial.
Spine 2002;27(16):1702–9.
Petersen T, Olsen S, Laslett M, Thorse n H, Manniche C, Ekdahl C, et al . Inter-
tester reliability of a new diag nostic classificati on system for patients with
non-specific low back pain. Australian Journ al of Physiotherapy 2004; 50(2):
85–94.
Petersen T, Thorsen H, Manniche C, Ekhdahl C. Classification of non-specific low
back pain: a review of the literature on classification systems relevant to
physiotherapy. Physical Therapy Reviews 1999;4:265–81.
Sahrmann SA. Diagnosis and treatment of movement impairment syndromes.
Mosby: St Louis; 2001.
Skouen JS, Grasdal AL, Haldorsen EM, Ursin H. Relative cost-effectiveness of
extensive and light multidisciplinary treatment programs versus treatment as
usual for patients with chronic low back pain on long-term sick leave:
randomized controlled study. Spine 2002;27(9):901–9.
Spitzer WO. Scientific approach to the assessment and management of activity-
related spinal disorders. Spine 1987;7S:S1–55.
Strender LE, Sjoblom A, Sundell K, Ludwig R, Taube A. Interexaminer reliability in
physical examination of patients with low back pain. Spine 1997;22(7):
814–20.
Stuge B, Laerum E, Kirkesola G, Vollestad N. The efficacy of a treatment program
focusing on specific stabilizing exercises for pelvic girdle pain after pregnancy:
a randomized controlled trial. Spine 2004;29(4):351–9.
Van Dillen LR, Sahrmann SA, Norton BJ, Caldwell CA, Fleming DA, McDonnell MK,
et al. Reliability of physical examination items used for classification of patients
with low back pain. Physical Therapy 1998;78(9):979–88.
Van Dillen LR, Sahrmann SA, Norton BJ, Caldwell CA, McDonnell MK, Bloom NJ.
Movement system impairment-based categories for low back pain: stage 1
validation. Journal of Orthopaedic and Sports Physical Therapy 2003;33(3):
126–42.
Vlaeyen JW, Linton SJ. Fear-avoidance and its consequences in chronic musculo-
skeletal pain: a state of the art. Pain 200 0;85(3):317–32.
Waddell G. The back pain revolution. 2nd ed. Edinburgh: Churchill Livingstone;
2004.
K. Vibe Fersum et al. / Manual Therapy 14 (2009) 555–561 561
