Physiotherapy 97 (2011) 59–64

Reliability and diagnostic validity of the slump knee bend neurodynamic

test for upper/mid lumbar nerve root compression: a pilot study
Kate Trainor a,b,∗ , Mark A. Pinnington a
a Directorate of Physiotherapy, School of Health Sciences, University of Liverpool, Liverpool, UK
b Aintree University Hospitals NHS Foundation Trust, Physiotherapy Department, University Hospital Aintree, Liverpool, UK

Abstract
Objectives It has been proposed that neurodynamic examination can assist differential diagnosis of upper/mid lumbar nerve root compression;
however, the diagnostic validity of many of these tests has yet to be established. This pilot study aimed to establish the diagnostic validity of
the slump knee bend neurodynamic test for upper/mid lumbar nerve root compression in subjects with suspected lumbosacral radicular pain.
Design Two independent examiners performed the slump knee bend test on subjects with radicular leg pain. Inter-tester reliability was
calculated using the kappa coefficient. Slump knee bend test results were compared with magnetic resonance imaging findings, and diagnostic
accuracy measures were calculated including sensitivity, specificity, predictive values and likelihood ratios.
Setting Orthopaedic spinal clinic, secondary care.
Participants Sixteen patients with radicular leg pain.
Results All four subjects with mid lumbar nerve root compression on magnetic resonance imaging were correctly identified with the slump
knee bend test; however, it was falsely positive in two individuals without the condition. Inter-tester reliability for the slump knee bend test
using the kappa coefficient was 0.71 (95% confidence interval 0.33 to 1.0). Diagnostic validity calculations for the slump knee bend test (95%
confidence intervals) were: sensitivity, 100% (40 to 100%); specificity, 83% (52 to 98%); positive predictive value, 67% (22 to 96%); negative
predictive value, 100% (69 to 100%); positive likelihood ratio, 6.0 (1.58 to 19.4); and negative likelihood ratio, 0 (0 to 0.6).
Conclusions Results indicate good inter-tester reliability and suggest that the slump knee bend test has potential to be a useful clinical test
for identifying patients with mid lumbar nerve root compression. Further investigation is needed on larger numbers of patients to confirm
these findings.
© 2010 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved.

MeSH terms: Nerve compression syndromes; Radiculopathy; Physical examination

Keywords: Slump knee bend; Diagnostic accuracy; Neurodynamics

Introduction of spinal nerve/nerve root compression remains unclear, and

Accurate clinical diagnosis of upper/mid lumbar nerve
root compression (L2 to L4) is highly desirable to ensure that
appropriate surgical candidates are identified and referred on
for diagnostic imaging. The slump knee bend neurodynamic
test is not pathognomonic but is suggested to be useful in the
diagnosis of lumbar radicular pain among other pathologies
[1,2]. The diagnostic validity of this test for specific levels
∗ Correspondence: Directorate of Physiotherapy, School of Health Sci-
ences, Thompson Yates Building, University of Liverpool, Brownlow Hill,
Liverpool L69 3GB, UK. Tel.: +44 0151 794 5742; fax: +44 0151 794 5740.
E-mail address: K.L.Trainor@liverpool.ac.uk (K. Trainor).
this has implications for all clinicians who use it to assist with
differential diagnosis.
The slump knee bend test [1] or femoral slump test [2]
is an extension of the standard prone knee bend test [1] or
femoral nerve stretch test [3]. The prone knee bend test has
been shown to place a load on the upper/mid lumbar nerve
roots (L2 to L4) via movement of the femoral nerve during
knee flexion [3]. The slump knee bend test differs from the
prone knee bend test in that it is performed in side lying,
and combines knee flexion and hip extension with thoracic
and cervical flexion to further load the nervous system [1].
Cervical extension is used when symptoms are reproduced
to ‘unload’ the nervous system and assist with structural dif-

0031-9406/$ – see front matter © 2010 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.physio.2010.05.004
60 K. Trainor, M.A. Pinnington / Physiotherapy 97 (2011) 59–64
ferentiation [1]. The addition of a spinal flexion component
to the slump knee bend test gives it suggested superiority
over the standard prone knee bend test in differentiation
between symptoms arising from neural and non-neural tis-
sues [1]. The prone knee bend and slump knee bend tests are
thought to be particularly useful to evoke symptoms in mid-
dle and upper lumbar radiculopathy, but it is also suggested
that they may have a role in the assessment of sensitive lower
lumbar radiculopathy [1]. Both tests are also advocated to
assist in the differential diagnosis of other disorders including
femoral-nerve-related pathologies [2].
An electronic search was performed of AMED (1985 to
2008), CINAHL (1982 to 2008), EMBASE (1980 to 2008),
Medline (1980 to 2008) and PEDro (1980 to 2008). The
search terms used were deliberately kept broad to allow
for the lack of literature relating to neurodynamic testing,
and were as follows: prone knee bend, crossed prone knee
bend, femoral nerve stretch test, slump knee bend, straight
leg raise, neurodynamics, adverse neural tension, thora-
columbar radiculopathy, lumbar radiculopathy and lumbar
nerve root compression. The search was further refined by
adding additional terms including ‘and Physiotherapy’, ‘and
Physi*therapy’, ‘and validity’, ‘and reliability’ and ‘and diag-
nosis’. The prone knee bend test [4–8] and crossed prone knee
bend test [9,10] have been investigated in relation to their
ability to diagnose upper/mid lumbar disc herniation, low
thoracic disc herniation and upper/mid lumbar radiculopa-
thy (L2 to L4 spinal levels), but these papers are descriptive
and do not report calculations of diagnostic accuracy. To the
authors’ knowledge, no published studies have investigated
the diagnostic validity of the slump knee bend test.
This paper reports the diagnostic validity of the slump
knee bend test for the diagnosis of L4 nerve root compression
in conjunction with other levels of nerve root compromise
in patients with radicular pain. Magnetic resonance imaging
(MRI) was used as the gold standard for diagnosis [11–14].
The inter-tester reliability of the slump knee bend test is also
reported.
Methods
Convenience sampling was used to recruit 16 patients
from an orthopaedic spinal clinic in an acute National Health
Service trust over a 6-month period.
Participants were included if they had symptoms synony-
mous with lumbosacral radicular pain, defined as having the
following features:
• intermittent or constant pain radiating into one or both legs
distal to the groin or gluteal crease; or
• distribution of pain in all or part of the dermatomal area
of one or two adjacent lumbosacral nerve roots (L1 to S2
levels).
Participants were only recruited if they were awaiting MRI
of the lumbar spine or returning to the spinal clinic following
MRI of this region. Exclusion criteria included pain in the
cervical or thoracic region likely to be exacerbated by spinal
slump, a history of central nervous system pathology, sys-
temic illness or symptoms indicating possible serious spinal
pathology or ‘red flags’ [15]. The neurodynamic test per-
formed in the study is presumed to load the nervous system;
for this reason, the presence of any of these characteristics
may have subjected the study participants to an unacceptable
level of risk. Patients with recent quadriceps injury (in the
last 6 months) and participants unable to lie in the test posi-
tion were also excluded. Finally, participants were excluded
if the anticipated wait for MRI was more than 6 weeks or if
their previous MRI was obtained more than 6 weeks previ-
ously.
Informed consent was obtained from suitable participants
who were then screened for the presence of serious spinal
pathology or ‘red flags’ [15]. Any participants found to
have signs of widespread neurology or other red flags were
highlighted to the medical team and withdrawn from the
study.
The slump knee bend test was performed once on
each participant by two investigators, both experienced
physiotherapists who had been working in the field of mus-
culoskeletal physiotherapy for over 10 years. The first author
was named as the principal investigator. Tests were performed
with random examiner order. Repeated movement of a joint
to the end of its range during serial measurement may itself
change the range of movement available at that joint; for
this reason, each neurodynamic test was only carried out
once by each investigator. Any interval between repeated
testing was kept deliberately small (a few minutes), and indi-
vidual investigators were blinded to each other’s findings.
Investigators were also blinded to any previous MRI scan
results to avoid any conscious or subconscious ascertain-
ment bias. The slump knee bend test protocol was adapted
from a previously described protocol [1] and is shown in
Figs. 1 and 2. The slump knee bend test was performed in
an attempt to reproduce symptoms or to identify any changes
occurring in existing symptoms. Unnecessary pain provoca-
tion was avoided by using sensitising manoeuvres to reduce
rather than increase symptoms [2]. Each test was determined
to be either normal (negative) or abnormal (positive) for
each subject. No adequate criteria were found in the liter-
ature to describe normal or abnormal participant responses
to the slump knee bend test using the specific end-test mea-
sures chosen in this study, so criteria were established by the
authors. These criteria are shown in Boxes 1 and 2 and were
based on the suggestions of experts in the field of neurody-
namics [1,2]. The point of evoked symptoms, submaximal
pain or the onset of firm resistance (R1) were used as end-
test measures. The reliability of these end-test measures has
not been investigated previously for the slump knee bend
test.
Following neurodynamic testing, participants underwent
MRI if this had not been performed previously. All MRI scans
were evaluated routinely by hospital radiologists, and these
 K. Trainor, M.A. Pinnington / Physiotherapy 97 (2011) 59–64
Fig. 1. Protocol for the slump knee bend neurodynamic test. Adapted from:
[1].
Fig. 2. The slump knee bend neurodynamic test. Reproduced with per-
mission from Claire Molyneux, Senior Physiotherapist, Aintree University
Hospitals NHS Foundation Trust.
written reports were reviewed to establish:
• the presence of any spinal nerve/nerve root compression
and the level(s) it occurred in the lumbosacral spine; and
• the type of pathology present causing the spinal
nerve/nerve root compromise.
The radiologists evaluating the scans were not aware that a
research project was taking place. MRI was chosen as the gold
standard for diagnosis and is routinely ordered in the spinal
clinic where this study was undertaken if nerve root compres-
61
Box 1: Classification of a normal response (negative
result) to the slump knee bend neurodynamic test.
• Symptoms of stretching or discomfort on the side
being tested.
• These symptoms may be felt in the anterior thigh area.
• These normal symptoms may either decrease in inten-
sity or remain the same when cervical extension is
performed.
• Range of movement and normal symptom response
is the same on both sides.
Box 2: Classification of an abnormal response (pos-
itive result) to the slump knee bend neurodynamic
test.
• All or part of the subject’s reported symptoms are
reproduced or increased during the test procedure
(when testing on either or both sides). The symptoms
should be the same usually experienced by the subject
(e.g. pain, paraesthesia previously reported). The pro-
voked symptoms should diminish when the subject’s
cervical spine is extended.
OR
• Associated symptoms (but not the subject’s reported
symptoms) are reproduced during the test procedure.
If the associated symptoms are not in a normal dis-
tribution and symptoms diminish when the subject’s
cervical spine is extended, this can be considered
a positive test. If the associated symptoms are in
a normal distribution, there must be evidence of
asymmetry between sides, and the symptoms should
diminish when extending the subject’s cervical spine.
OR
• The onset of firm resistance is perceived to occur ear-
lier in range during the testing procedure when one
side is tested compared with the other. This may or
may not be accompanied by symptom reproduction
or associated symptom production. There should be
a decrease in perceived resistance at the hip when
extending the subject’s cervical spine.
sion is suspected. It is a reliable and valid tool for diagnosing
lumbosacral nerve root compression and intervertebral disc
injury [11,12], and determination of nerve root compromise
by MRI appears to be well correlated with surgical findings
[13]. The sensitivity, specificity, accuracy, and positive and
negative predictive values for standard MRI and nerve root
compression have been shown to be 81%, 100%, 82%, 100%
and 25%, respectively, when following the protocol used in
this study [14].
62 K. Trainor, M.A. Pinnington / Physiotherapy 97 (2011) 59–64
Data analysis
Inter-tester reliability for the slump knee bend neurody-
namic test was calculated using the kappa coefficient [16–18].
Strength of agreement was ascertained using standards pro-
posed by Brennan and Silman [19]. The diagnostic validity
of the slump knee bend test was established for upper/mid
lumbar nerve root compression (L2 to L4) by calculating
sensitivity, specificity, and positive and negative predictive
values. Positive and negative likelihood ratios were also cal-
culated, defined as how many times more (or less) likely
patients with the disease are to have a particular result than
patients who do not have the disease [20].
All diagnostic accuracy measurements were calculated
using data collected by the principal investigator, and 95%
confidence intervals were calculated for each value to estab-
lish the precision of the estimates [21,22].
Results
The study sample was taken from an area of the North
West of the UK. Of the 16 patients approached, all agreed to
participate. Nine females (56%) and seven males (44%) were
recruited, aged between 35 and 72 years with a mean age of
49 years (standard deviation 12.6). Symptom duration was
reported as subacute (between 3 weeks and 3 months) for 11
participants (69%) and chronic (more than 3 months) for the
remaining five participants (31%).
Reliability testing
For the slump knee bend test, kappa was 0.71 (95% con-
fidence interval 0.33 to 1.0). This indicates that the strength
of agreement using the kappa coefficient was ‘good’ for this
test (between 0.61 and 0.8) using guidelines suggested by
Brennan and Silman [19].
MRI results
MRI findings can be found in Table 1. Four of the 16
participants had mid lumbar nerve compression on MRI, with
only two of these participants showing L4 compression in
isolation.
Diagnostic validity of the slump knee bend test for
upper/mid lumbar nerve root compression
Neurodynamic test findings were reported as either posi-
tive or negative, and all positive tests reproduced symptoms
or associated symptoms in test participants (Boxes 1 and 2).
These findings were compared with MRI findings as the gold
standard, and the results are shown in Table 2. The slump
knee bend test correctly confirmed or negated the presence
of mid lumbar nerve root compression in 14 of the 16 partic-
ipants (88%). Diagnostic validity calculations for the slump
knee bend test (95% confidence intervals) were: sensitivity,
100% (40 to 100%); specificity, 83% (52 to 98%); positive
predictive value, 67% (22 to 96%); negative predictive value,
100% (69 to 100%); positive likelihood ratio, 6.0 (1.58 to
19.4); and negative likelihood ratio, 0 (0 to 0.6).
Discussion
Inter-tester reliability was found to be ‘good’ for the slump
knee bend test using the kappa coefficient. This suggests
that clinicians can reliably agree on the findings of this test
using the operational definitions described. All subjects with
mid lumbar nerve root compression confirmed on MRI had
a positive slump knee bend test, and over 80% of partici-
pants without the disorder had a negative test. The slump
knee bend test had some clinical value in predicting the dis-
order, and was found to be excellent at negating the disorder.
The rate of MRI-confirmed cases being correctly diagnosed
with a positive slump knee bend test was found to be six
times greater than that of negative MRI cases being incor-
rectly diagnosed with a positive slump knee bend test. The
lower confidence interval of the positive likelihood ratio for
the slump knee bend test is greater than 1 (1.6), so it can
therefore be suggested with 95% confidence that a patient
classified with a positive slump knee bend test is more likely
to have mid lumbar nerve root compression than to be falsely
positive for the disorder.
The authors were unable to find any published literature
investigating the diagnostic validity of the slump knee bend
test in relation to any pathology; for this reason, no direct
comparison can be made with other work in this particu-
lar area. The only work previously undertaken in this area
has been in relation to the standard [4–8] and crossed [9,10]
versions of the prone knee bend/femoral nerve stretch tests.
Only one paper was found that reported the diagnostic valid-
ity of these tests for L2 to L4 nerve root compression rather
than disc pathology [10]. This paper cannot be used for com-
parison with the present research as it only details two case
studies, and the data presented were insufficient for the cal-
culation of sensitivity, specificity or predictive values for
diagnostic accuracy.
Bias can potentially limit diagnostic validity studies, so
attempts were made to eliminate this where possible. Inter-
pretation biases, which can occur when the results of the test
under investigation are known when the diagnosis is made
using the gold standard (diagnostic review bias) or vice versa
(test review bias) [23], were eliminated by the blinding of
investigators. Observer variability bias [23], which can have
an adverse effect on external validity if not adequately con-
trolled, was controlled for using a highly experienced tester
using standardised protocols for all neurodynamic testing.
Similarly, potential bias arising from the order in which the
tests were performed was controlled for by the random test-
ing procedure. As the participants in this study were patients
referred to a spinal clinic due to the suspicion of lumbosacral
 K. Trainor, M.A. Pinnington / Physiotherapy 97 (2011) 59–64 63

Table 1
Magnetic resonance imaging results indicating level and cause of spinal nerve/nerve root compression.
Subject Levels of SN/NR compression evident with MRI Significant pathology present causing spinal
SN/NR compression
1 No SN/NR compression N/A
2 No SN/NR compression N/A
3 Compression of the left L4 and L5 nerve roots Disc herniation
4 Compression of the left L5 nerve root Degenerative disc disease with herniation
5 Compression of the left L5 nerve root Disc herniation
6 Compression of the right L5 nerve root Disc herniation
7 L4 nerve root compression L3/L4 disc bulge, L4/L5 right lateral recess
stenosis and narrowing of the left
neuroforamina
8 No SN/NR compression N/A
9 Compression of the right S1 nerve root Narrowing of the L4/L5 and L5/S1 discs,
and small disc prolapse at L5/S1
10 No SN/NR compression N/A
11 No SN/NR compression N/A
12 Compression of the left S1 nerve root Disc herniation
13 No SN/NR compression N/A
14 Compression of the left L4 and L5 nerve roots L3/L4 disc dehydration. Posterior and lateral
disc bulging together with facet joint
hypertrophy causing a degree of
anteroposterior and transverse canal stenosis
and narrowing of the lateral recess exit
foramina. Minor disc bulge at L4/L5
impinging on the left exit foramina
15 Compression of the L4 nerve root Disc degeneration and bulging
16 Compression of the S1 nerve root Disc herniation
SN/NR, spinal nerve/nerve root; MRI, magnetic resonance imaging; N/A, not applicable.

nerve root compression, an element of patient cohort bias may
have been present (in particular, population bias). The authors
acknowledge that this homogenous sample may affect the
external validity of the findings to the general population, but
suggest that the findings remain valid for individuals who
would be likely to undergo these tests in clinical practice (i.e.
patients with suspected radicular pain).
Clinicians should be aware that predictive values are
dependent on the prevalence of disease in a study population,
whereas sensitivity and specificity findings are not. Increas-
ing prevalence of disease makes it more likely that a person
with a positive test result has the disease, and less likely that a
positive result is a false positive. It should therefore be empha-
sised that clinicians may not be able to apply these predictive
test values directly to their own patient populations unless
disease prevalence is similar.
Other limitations to the study are in relation to the opera-
tional definitions and protocol for the slump knee bend test.
The use of R1 as an end-test measure is contentious in the neu-
rodynamic literature, both in relation to its reliability and its
validity [24]. All positive slump knee bend tests in this study
reproduced symptoms or associated symptoms; for this rea-
son, the use of R1 would not have affected the internal validity
of the study findings. The use of M1, the onset of reflex mus-
cle activity, may be a more valid end-test measure for future
neurodynamic research. However, it remains unclear whether
clinicians can ascertain M1 reliably during clinical testing.
Longer time intervals between repeated tests on individuals
would have allowed any symptoms and potential biomechan-
ical effects on tissues to settle, as this could have affected the
validity of the results. Both intra- and inter-tester reliabil-
ity could also have been established for the individual test
measures used within the protocols.
The wide confidence intervals presented for the slump
knee bend test mean that further data are needed from a larger
sample in order to clarify the usefulness of the test for this dis-
order. In particular, more patients with upper lumbar nerve
root compression (L2 to L3) would have strengthened the
conclusions, as this study has only shown the slump knee
bend test to be valid for L4 spinal nerve/nerve root compres-

Table 2
Accuracy of the slump knee bend test for the diagnosis of L2 to L4 spinal nerve/nerve root compression using data collected by the principal investigator.
L2 to L4 SN/NR compression L2 to L4 SN/NR compression Total number of subjects
confirmed with MRI negated with MRI
Slump knee bend positive 4 True positives 2 False positives 6
Slump knee bend negative 0 False negatives 10 True negatives 10
Total number of subjects 4 12 16
SN/NR, spinal nerve/nerve root; MRI, magnetic resonance imaging.
64 K. Trainor, M.A. Pinnington / Physiotherapy 97 (2011) 59–64
sion. The practicalities of MRI waiting times in the National
Health Service trust for non-urgent imaging resulted in a time
delay between neurodynamic testing and scanning. Times
between testing and MRI would ideally have been shorter to
strengthen the study’s internal validity.
Conclusion
The results of this study suggest that the slump knee bend
test has value in predicting the presence or absence of L4
lumbar nerve root compression, and as such, may be useful
to assist with the differential diagnosis of leg pain suspected
to be due to this condition. It should be highlighted, however,
that the individuals correctly identified with this level of nerve
root compromise also had concurrent lower lumbar nerve root
pathology. It therefore remains unclear whether the slump
knee bend test can identify L4 nerve root pathology reliably
in the absence of other levels of nerve root pathology. It also
remains unclear whether the test can identify L2 and L3 levels
of nerve root compression, as these levels were not found to
be affected in the study participants.
A larger study would help to confirm the diagnostic use-
fulness of the slump knee bend test more definitively. Further
research should also investigate the diagnostic value of clus-
ters of tests for upper/mid lumbar nerve root compression,
including the slump knee bend test, as this is how it is likely
to be used clinically.
Acknowledgements
The authors would like to thank the Orthopaedic Depart-
ment at Aintree University Hospitals NHS Foundation Trust
for allowing the collection of data, and Margaret Beswick,
Extended Scope Physiotherapy Practitioner for her assistance
with data collection. Thanks are extended to Dr Hilary Chat-
terton at Manchester Metropolitan University for her advice
and support during this project. Permission was obtained
from Claire Molyneux, Senior Physiotherapist, Aintree Uni-
versity Hospitals NHS Foundation Trust for the use of her
image in Fig. 2.
Ethical approval: Manchester Metropolitan University Ethics
Committee and Sefton Local Research Ethics Committee
(Ref. No. 04/Q1501/20).
Conflict of interest: None declared.
References
[1] Butler DS. The sensitive nervous system. Adelaide: Noigroup Publica-
tions; 2000. p. 303–304.
Available online at www.sciencedirect.com
[2] Shacklock MS. Clinical neurodynamics: a new system of muscu-
loskeletal treatment. London: Butterworth Heinemann; 2005.
[3] Kobayashi S, Suzuki Y, Asai T, Yoshizawa H. Changes in nerve
root motion and intraradicular blood flow during intraoperative
femoral nerve stretch test. Report of four cases. J Neurosurg
2003;99(Suppl):298–305.
[4] Epstein NE. Foraminal and far lateral lumbar disc herniations: surgical
alternatives and outcome measures. Spinal Cord 2002;40:491–500.
[5] Christodoulides AN. Ipsilateral sciatica on femoral nerve stretch test
is pathognomonic of an L4/5 disc protrusion. J Bone Joint Surg Br
1989;71:88–9.
[6] Porchet F, Fankhauser H, de Tribolet N. Extreme lateral lumbar disc
herniation: clinical presentation in 178 patients. Acta Neurochir (Wien)
1994;127:203–9.
[7] Asquier C, Trousssier B, Chirossel JP, Chardonnet E, Mouries D, Juvin
R, et al. Femoral neuralgia due to degenerative spinal disease. A retro-
spective clinical and radio-anatomical study of one hundred cases. Rev
Rhemat English Ed 1996;63:278–84.
[8] Tokuhashi Y, Matsuzaki H, Uematsu Y, Oda H. Symptoms of thora-
columbar junction disc herniation. Spine 2001;26:512–8.
[9] Kreitz BG, Cote P, Yong-Hing K. Crossed femoral stretching test. A
case report. Spine 1996;21:1581–6.
[10] Nadler SF, Malanga GA, Stitik TP, Keswani R, Foye PM. The crossed
femoral nerve stretch test to improve diagnostic sensitivity for the
high lumbar radiculopathy: 2 case reports. Arch Phys Med Rehabil
2001;82:522–3.
[11] Kim KY, Kim YT, Lee CS, Kang JS, Kim YJ. Magnetic resonance
imaging in the evaluation of the lumbar herniated intervertebral disc.
Int Orthop 1993;17:241–4.
[12] Schellhas KP, Pollei SR, Gundry CR, Heithoff KB. Lumbar disc
high-intensity zone. Correlation of magnetic resonance imaging and
discography. Spine 1996;21:79–86.
[13] Pfirrmann CW, Dora C, Schmid MR, Zanetti M, Hodler J, Boos N.
MR image-based grading of lumbar nerve root compromise due to
disk herniation: reliability study with surgical correlation. Radiology
2004;230:583–8.
[14] Chawalparit O, Churojana A, Chiewvit P, Thanapipatsir S, Vamvanij
V, Charnchaowanish P. The limited protocol MRI in the diagnosis of
lumbar disc herniation. J Med Assoc Thail 2006;89:182–9.
[15] Greenhalgh S, Selfe J. Red flags: a guide to identifying serious pathol-
ogy of the spine. Edinburgh: Churchill Livingstone; 2009.
[16] Sim J, Wright C. Research in health care: concepts, designs and meth-
ods. Cheltenham: Stanley Thornes (Publishers) Ltd; 2000.
[17] Tooth LR, Ottenbacher KJ. The Kappa statistic in rehabilitation
research: an examination. Arch Phys Med Rehabil 2004;85:1371–6.
[18] Sim J, Wright CC. The kappa statistic in reliability studies: use, inter-
pretation, and sample size requirements. Phys Ther 2005;85:257–68.
[19] Brennan P, Silman A. Statistical methods for assessing observer vari-
ability in clinical measures. BMJ 1992;304:1491–4.
[20] Deeks JJ, Altman DG. Diagnostic tests 4: likelihood ratios. BMJ
2004;329:168–9.
[21] Petrie A, Sabin C. Medical statistics at a glance. 3rd ed. Oxford: Wiley
Blackwell; 2009.
[22] Altman DG, Machin D, Bryant TN, Gardner MJ. Statistics with confi-
dence. 2nd ed. London: BMJ Books; 2000.
[23] Whiting P, Rutjes AWS, Reitsma JB, Glas AS, Bossuyt PMM, Kleijnen
J. Sources of variation and bias in studies of diagnostic accuracy: a
systematic review. Ann Intern Med 2004;140:189–202.
[24] Hall T, Zusman M, Elvey R. Adverse mechanical tension in the nervous
system? Analysis of straight leg raise. Man Ther 1998;3:140–6.