Content uploaded by Ulrich Gockel
Author content
All content in this area was uploaded by Ulrich Gockel on Feb 26, 2014
Content may be subject to copyright.
It is the responsibility of the
corresponding author to check fully
against the original manuscript and
approve these proofs
LibraPharm cannot be held responsible for
uncorrected errors, even if introduced in the
course of the publishing process. The journal
reserves the right to charge for excessive
author alterations, or for changes requested at
later proof stages
ORIGINAL ARTICLE
painDETECT: a new screening
questionnaire to identify
neuropathic components in
patients with back pain
Rainer Freynhagen
a
, Ralf Baron
b
, Ulrich Gockel
c
and
Thomas R. Tölle
d
a
Klinik für Anästhesiologie, Universitätsklinikum Düsseldorf, Düsseldorf,
Germany
b
Abteilung für Neurologische Schmerzforschung und –therapie, Universität
Kiel, Kiel, Germany
c
Pfizer Pharma GmbH, Karlsruhe, Germany
d
Klinik für Neurologie, Technische Universität München, München, Germany
Address for correspondence: Dr R. Freynhagen, MD, DEAA Klinik für Anästhesiologie,
Universitätsklinikum Düsseldorf, Moorenstraße 5, 40225 Düsseldorf, Germany.
Tel.: +49 211 811 6157; email: freynhagen@med.uni-duesseldorf.de; and Professor R. Baron, MD,
Abteilung für Neurologische Schmerzforschung und therapie, Universität Kiel, Schittenhelmstr. 10,
24105 Kiel, Germany; Tel.: +49 431 597 8504; email: r.baron@neurologie.uni-kiel.de
Key words: Back pain – Co-morbidity – Neuropathic pain – Nociceptive pain – painDETECT
– Prevalence – Questionnaire
0300-7995
doi:10.1185/030079906X132488
All rights reserved: reproduction in whole or part not permitted
CURRENT MEDICAL RESEARCH AND OPINION®
VOL. 22, NO. ••, 2006, Fast Track PREPRINT: P1–P10
© 2006 LIBRAPHARM LIMITED
Paper 3586 P1
Objective: Nociceptive and neuropathic
components both contribute to pain. Since these
components require different pain management
strategies, correct pain diagnosis before and
during treatment is highly desirable. As low
back pain (LBP) patients constitute an important
subgroup of chronic pain patients, we addressed
the following issues: (i) to establish a simple,
validated screening tool to detect neuropathic
pain (NeP) components in chronic LBP patients,
(ii) to determine the prevalence of neuropathic
pain components in LBP in a large-scale survey,
and (iii) to determine whether LBP patients with
an NeP component suffer from worse, or different,
co-morbidities.
Methods: In co-operation with the German
Research Network on Neuropathic Pain we
developed and validated the painDETECT
questionnaire (PD-Q) in a prospective, multicentre
study and subsequently applied it to approximately
8000 LBP patients.
Results: The PD-Q is a reliable screening tool
with high sensitivity, specificity and positive
predictive accuracy; these were 84% in a palm-
top computerised version and 85%, 80% and
83%, respectively, in a corresponding pencil-and-
paper questionnaire. In an unselected cohort of
chronic LBP patients, 37% were found to have
predominantly neuropathic pain. Patients with
NeP showed higher ratings of pain intensity, with
more (and more severe) co-morbidities such as
depression, panic/anxiety and sleep disorders.
This also affected functionality and use of health-
care resources. On the basis of given prevalence
of LBP in the general population, we calculated
that 14.5% of all female and 11.4% of all male
Germans suffer from LBP with a predominant
neuropathic pain component.
Conclusion: Simple, patient-based, easy-to-
use screening questionnaires can determine the
prevalence of neuropathic pain components both
in individual LBP patients and in heterogeneous
cohorts of such patients. Since NeP correlates
with more intense pain, more severe co-morbidity
and poorer quality of life, accurate diagnosis is a
milestone in choosing appropriate therapy.
A B S T R A C T
P2 painDETECT: screening for neuropathic pain © 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••)
Introduction
Low back pain (LBP) is a common public health
problem world-wide
1–3
. For quite a number of patients
affected by the development of chronic back pain,
the disease is extremely disabling; it is likely to show
co-morbidity with mood disorders and is associated
with serious risk of ending in early retirement or an
inability to cope with activities of daily living. The
pathophysiological mechanisms of pain generation
in back pain are complex, current diagnostic tools do
not meet the patients’ needs in many respects, and
therapeutic results are frequently poor. However,
there is a consensus among many disciplines involved
in the treatment of back pain that the classification of
LBP disorders has to be based on the underlying mech-
anism, in order to allow formulation of clear treatment
recommendations and to ensure appropriate manage-
ment of the individual patient
1,4,5
. At the same time
there is a strong belief that the early inception of the
most appropriate treatment is most likely to avoid the
risk of developing chronic back pain.
In recent concepts for the understanding of back
pain, it is hypothesised that different components of
chronic pain, i.e. nociceptive and neuropathic com-
ponents, may both contribute to an individual’s pain.
In chronic back pain the nociceptive component
results from activation of nociceptors that innervate
ligaments, small joints, muscle and tendons. However,
at the same time, neuropathic processes may also be
involved
6
. The current International Association for
the Study of Pain taxonomy defines neuropathic pain
as ‘Pain resulting from a lesion or dysfunction of the
peripheral or central nervous system’. In a broad variety
of back-pain diagnoses it seems very likely that this is
indeed the case. Mechanical compression of radicular
nervous tissue may occur within or adjacent to the inter-
vertebral foramina, within the lateral recess and within
the spinal canal itself. In addition, the action of inflam-
matory mediators (cytokines, chemokines) originating
from the degenerative disc has been implicated in the
chemical pathomechanism of radicular neuropathic
pain.
Hence, a variety of nerve-damaging stimuli are likely
to generate a neuropathic pain component in patients
presenting with a chronic LBP. Significantly, many
long-term LBP patients report neurological deficits
and/or somatosensory positive symptoms that can
be described as characteristic of a neuropathic pain
component, i.e. hypoesthesia and allodynia. However,
this is only true if the clinical investigation is done
carefully and comprehensively, as these changes may
be subtle and escape routine examination. Since the
rationale for separating neuropathic pain from non-
neuropathic pain has direct implications for treatment,
it is of the utmost importance for the clinician to
identify the severity of the different pain components
in each individual patient. It is assumed that treatment
interventions based on particular pain mechanisms
(rather than treating pain as a uniform phenomenon)
will lead to increased ability to tailor specific therapies
to particular subgroups of patients.
Thus, primary care physicians have a key diagnostic
position since they guide the therapeutic management
from early on and have a pivotal role in triaging LBP
patients for specific treatment approaches
7
. Beside the
matter of pain treatment, LBP is in general associated
with numerous co-morbidities such as sleep disturb-
ance, depression and anxiety. Furthermore, other
fundamental questions have not yet been investi-
gated, and – although such knowledge is essential
– it is not known whether patients suffering from
pain of predominantly nociceptive origin differ from
those suffering from neuropathic pain components
(e.g. in respect of socio-demographic characteristics,
the intensity, duration or management of pain, co-
morbidities, functional disorders, pension applications
etc.). An effort to move forward towards a rational and
symptom-based treatment seems to be timely for the
improvement of LBP therapy. Therefore, physicians
should abandon any view of LBP as a monolithic entity
and, instead, distinguish the nociceptive and neuro-
pathic pain components and realise the co-morbid
conditions of their patients.
The purpose of screening tools is to identify patients
using a few key symptoms and signs. To encourage
their broad application such tools should be short and
easy to administer and most importantly independent
from physical examination. Several screening tools
for neuropathic pain have been developed and tested
with different patient populations but up to the
date when we started to develop our new questionnaire,
none of them met this need of primary care
physicians and moreover, none specifically targeted
LBP.
Consequently, it is essential to focus on three unmet
needs:
(i) to establish a validated easy screening tool to
discriminate between neuropathic and noci-
ceptive pain components in a large-scale survey
of chronic LBP patients,
(ii) to provide the prevalence of neuropathic pain
components in LBP in the general population,
and
(iii) to answer the question of whether LBP
patients with a neuropathic pain component
suffer more severely than those without and,
if so, to substantiate the differences between
these two groups.
© 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••) painDETECT: screening for neuropathic pain Freynhagen et al. P3
We therefore set out to develop a screening tool
for neuropathic pain components specifically in LBP
patients using a simple patient-based questionnaire
(painDETECT questionnaire; PD-Q), that is both
reliable and easy to apply, either programmed into
hand-held ‘palmtop’ computers (personal digital
assistant, PDA; e.g. in the context of a clinical trial),
or for simple routine use as a pencil-and-paper
questionnaire.
Methods
painDETECT questionnaire
The aim was to develop an easy questionnaire to
detect neuropathic pain components, especially in
chronic LBP patients. Such a questionnaire should rely
on the characteristic clinical neuropathic symptoms.
Therefore, we performed a systematic review of the
literature, interviewed several national and international
pain experts and reviewed files of patients included
in the data base of the German Research Network
on Neuropathic Pain (Deutscher Forschungsverbund
Neuropathischer Schmerz, DFNS) to find out how
neuropathic pain symptoms are perceived by the
patient. As a result, seven questions that address the
quality of neuropathic pain symptoms (Table 1) were
included in the questionnaire. Different questions were
not tested or discarded. Furthermore, it became clear
that many patients described distinct pain patterns
during the course of the disease that might be of
predictive value to distinguish neuropathic from noci-
ceptive pain. Lastly, the description ‘radiating pain’ was
found to be a neuropathic descriptor, in particular when
LBP patients were interviewed. Following selection of
the most appropriate items, a principal-components
analysis (PCA) was performed. Kaiser’s criterion was
then applied, thus retaining for further analysis only
those factors with an eigenvalue of 1.0 or greater.
Validation study
In a prospective multicentre study, 411 pain patients
were recruited consecutively at ten specialised pain
centres and screened for participation. In accordance
with the current gold standard (diagnoses by expert
pain physicians to assess the predominant pain type),
each patient was examined by two experienced pain
specialists, working independently of each other, at
each centre. Both specialists assessed the predominant
pain type on the basis of his/her experience, using
whatever diagnostic methods were considered appro-
priate (neurological or electrophysiological examination,
imaging, etc.). The assessments were not performed until
all additional data were available. They were required to
rate the pain as predominantly neuropathic, nociceptive
or unclear. To reduce the error ratio within the validation
study and to enlarge the reliability of a diagnosis of
Item Score
Gradation of pain*
Do you suffer from a burning sensation (e.g. stinging nettles) in the marked areas?
0–5
Do you have a tingling or prickling sensation in the area of your pain (like crawling ants or electrical tingling)?
0–5
Is light touching (clothing, a blanket) in this area painful?
0–5
Do you have sudden pain attacks in the area of your pain, like electric shocks?
0–5
Is cold or heat (bath water) in this area occasionally painful?
0–5
Do you suffer from a sensation of numbness in the areas that you marked?
0–5
Does slight pressure in this area, e.g. with a finger, trigger pain?
0–5
Pain course pattern
Please select the picture that best describes the course of your pain:
Persistent pain with slight fluctuations
0
Persistent pain with pain attacks
–1
Pain attacks without pain between them
+1
Pain attacks with pain between them
+1
Radiating pain
Does your pain radiate to other regions of your body? Yes/No +2/0
*For each question: never, 0; hardly noticed, 1; slightly, 2; moderately, 3; strongly, 4; very strongly, 5
Questions used to document pain, but which were not used in the scoring, are not shown
Table 1. painDETECT questionnaire
P4 painDETECT: screening for neuropathic pain © 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••)
either predominantly neuropathic or nociceptive pain
only patients diagnosed with ‘typical’ neuropathic or
nociceptive entities were included. Pain of predominantly
neuropathic origin (n = 228) included post-herpetic
neuralgia (PHN), painful polyneuropathy (PNP), nerve
trauma and low back pain (solely of the lumbar vertebrae,
sacrum and coccyx) diagnosed as predominantly
neuropathic. Pain of predominantly nociceptive origin
(n = 164) included visceral pain, osteoarthritis,
inflammatory arthropathies and mechanical LBP (source
of pain is in the spine and/or its supporting structures
such as surrounding muscles, ligaments, joints or soft
tissues). A patient was included in the validation cohort
(i) if, and only if, the two specialists concurred in
the diagnosis,
(ii) the diagnosis was not ‘unclear’,
(iii) the patient’s average pain was stated by the
patient to be at least 40% on a visual analogue
scale (VAS) from 0 (no pain) to 100% (worst
possible pain), and
(iv) the patient was able to answer unaided
the German-language questionnaire on the
PDA.
If a patient was diagnosed with pain of assumed
mixed origin (e.g. malignancy, compression fractures,
fibromyalgia, ancylosing spondylitis) or if the
symptomatology was unclear the patient was explicitly
excluded.
The result of the experts’ opinion was then compared
with results of the PD-Q. The ‘true’ diagnoses
(predominantly neuropathic pain: Yes/No) were used
as a target variable in logistic regression model. The
seven questions to characterise neuropathic pain, the
pain course pattern and the presence or absence of
radiating pain were included in the model. The ROC
(receiver operating characteristic) and area under the
curve were determined.
Our approach was driven by physicians’ needs
to be able to detect patients with predominantly
neuropathic pain, with high probability (> 80%);
the results of the logistic regression model were
approached by an empirical procedure using all the
parameters as described, easily combined to evaluate
patients’ answers in everyday consultation. This led to a
pencil-and-paper questionnaire version with an
expansion of the score range from (0, 35) to (–1, 38).
Internal consistency is reported as Cronbach’s alpha
(0–1.0).
Epidemiological survey
Five hundred PDAs (Palm Tungsten E, operating
system Palm OS 5.4) were issued to 472 general and
specialist practices and hospitals throughout Germany
(including 158 general practitioners, 45 orthopaedists,
67 neurologists and 202 pain specialists). A total
of 7772 patients with various forms of chronic low
back pain (lower back pain that lasts for longer than
3 months) were consecutively included in the study if
they fulfilled the inclusion and exclusion criteria. To
be included in the epidemiological survey low back
pain had to be the leading pain problem (e.g. with
the maximum pain localisation) but patients with
additional back pain problems (cervical, neck or upper
back problems, secondary chronic muscle tension, etc.)
were not excluded.
Apart from PD-Q, the following questionnaires
were used: the Medical Outcomes Study sleep
scale (MOS)
8
, the German-language Patient Health
Questionnaire (PHQ-9, short form)
9
, and the Hanover
Functional Ability Questionnaire for measuring back-
pain-related disability (FFbH-R)
10
. All of these are
validated and in common use. The programming
included automated plausibility checks that gave the
user a warning if impossible values were keyed in. At
intervals, PDAs were collected and data transfer and
processing were performed under secure conditions,
with anonymisation and encryption. Before the study,
investigators were introduced to the study procedures;
apart from this and PDA collection, further monitoring
was not considered appropriate. Physicians did not
receive a financial incentive for taking part in the
epidemiological study.
Study population
Patients presenting with chronic LBP, at least 18 years
old and after giving informed consent, were requested
to complete the PDA questionnaires, within either
the validation study or the epidemiological survey.
Recruiting took place from August 2004 to May 2005.
The study protocol and consent forms of the study
were submitted to and approved by the local ethics
committee, in accordance with the ethical principles
originating from the Declaration of Helsinki and in
compliance with Good Clinical Practice.
Data management and statistical analysis
Information from the paper forms was transferred to
the project data base by double data-entry, followed
by data review. Afterwards, the paper and PDA data
sources were merged for descriptive statistical analyses;
these were performed with the SAS package, version
8.2. Graphics were generated with SPSS version 11.5.
Patients were grouped into classes based on results of
the validation study regarding the painDETECT score.
Correlations between scores and continuous variables
© 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••) painDETECT: screening for neuropathic pain Freynhagen et al. P5
were calculated using Pearson correlation coefficient.
Relations between two dichotomous variables were
assessed by 2 × 2 contingency tables. An ANOVA
model was used to evaluate differences between groups
of patients.
Results
Validation study
The level of patients’ acceptance of the PDA
questionnaire was very high: 80% had no problems,
10% minor problems and only 10% required assistance.
In all, 411 patients were screened in the validation
study; of these 19 were excluded (ten because of
diagnosis of an unclear pain and nine because the raters
differed in assessment of the pain type). Results for
the remaining 392 patients (pain of predominantly
nociceptive origin, n = 164; pain of predominantly
neuropathic origin, n = 228) were evaluated by
logistical regression and are depicted in Figure 1(A),
which plots the cumulative frequency of scores on the
painDETECT scale up to 38. The corresponding curves
for the entire 392-patient population allow an equally
clear inference to be made [Figure 1(B)]. The set of
392 patients was used in a logistical regression analysis
to establish the sensitivity (which indicates how good
a test is at correctly identifying people who have the
disorder in question), specificity (which indicates how
good a test is at correctly identifying people who do
not have the disorder) and the probability of correct
assignment (positive predictive value refers to the
chance that a positive test result will be correct) of the
PD-Q. Results are displayed in Figure 2.
The PD-Q score is calculated by addition of the
entries in the right-hand column of the questionnaire
(maximum possible score = 38, minimum = –1; only
integral values are possible). For screening purposes
the following cut-off points have been found to be
the most appropriate: score ≤ 12, a neuropathic com-
ponent is unlikely (< 15%); score ≥ 19, a neuropathic
component is likely (> 90%). Between these, the result
is uncertain, i.e. a neuropathic pain component can be
present. The logistic regression indicated a sensitivity
of 84%, and a specificity and probability of correct
assignment, likewise, of 84%. When the score was
subjected to an empirical analysis for use as a pencil-
and-paper questionnaire, then the values found for
these were 85%, 80% and 83%, respectively. The area
under the curve for the logistical regression analysis was
found to be 0.91 (Figure 2); the theoretical maximum
is 1.00 for perfect discrimination). Repetition of the
0
20
(A) (B)
40
60
80
100
0 5 10 15 20 25 30 35 38
painDETECT score
Percentage
Percentage
0
20
40
60
80
100
0 5 10 15 20 25 30 35
38
painDETECT score
0
20
40
60
80
100
0 20 40 60 80 100
1 – Specifity (%)
Sensitivity (%)
Figure 1. (A) Cumulative frequency of scores on the painDETECT scale for various pain types. Key: red, nociceptive LBP
(n = 100); blue, other nociceptive pain (n = 64); green, neuropathic LBP (n = 52), mauve, other neuropathic pain (n = 132);
orange, DPN and PHN only (
n = 44). (B) The corresponding curves for the entire 392-patient population: red, all pain
diagnosed as nociceptive in origin (
n = 164); green, all pain diagnosed as neuropathic in origin (n = 228)
Figure 2. ROC curve of logistic regression model (blue) and ROC curve of empirical painDETECT score (orange). The area
under the curve for the logistic regression model is 0.91
P6 painDETECT: screening for neuropathic pain © 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••)
calculation for the LBP patients only led to similar
results, with slightly higher values.
According to the Kaiser criterion, four factors of
formal importance (eigenvalue of ≥ 1.0) could be
identified, of which two were found to be determin-
ative, as follows. The data structure of the pain course
pattern is dominated by the first principal component,
which consists of the seven questions about pain
quality. The second interpretable principal component
is dominantly described by the pain course pattern
‘Pain attacks with pain-free periods in between’ (one of
four, cf. Table 1) and ‘Radiating pain’. Considering the
first principal component, the bivariate correlations
between the seven items to grade pain were always
significant, with p < 0.001 (n = 392). These seven
items demonstrated adequate internal consistency
(Cronbach’s alpha = 0.83).
Epidemiological survey
A total of approximately 8000 patients with various
chronic back pain problems took part in the survey.
Of these, 7772 yielded complete data sets. The demo-
graphic profile of these patients is shown in Table 2.
Conspicuously, the percentage of patients with a PD-Q
score of ≥ 19 (synonymous with a probability of having
a neuropathic pain component > 90%) was high. The
prevalence of LBP patients with a NeP component
was found to be 37% (males, 36.9%; females, 37.1%).
Furthermore, our data demonstrate striking differences
between LBP patients with and without a neuropathic
pain component. Regarding the duration of pain
treatment, the number of physician visits for pain in
the past 4 weeks, the number of different therapists
(including on-going psychotherapy), the proportion
of intended or submitted pension applications and
the average pain intensity during the past 4 weeks,
LBP patients with a NeP component were found to
suffer longer and more severely (
p < 0.001; Table 2).
The breakdown of demographic and disease-related
variables by PD-Q score – and thus by the likely
involvement of different pain components – reveals
various trends, and these are summarised in Table 3.
Depression
On the PHQ-9 scale for depression, scores were
rated as follows: 0–4 (none), 5–9 (mild), 10–19
(moderate) and 20–27 (severe). As Table 3 shows,
the proportion of patients with severe depression is
greatest in the ‘neuropathic pain’ group (
p = 0.001).
Of the subjects without a NeP component, 63.7%
(n = 1727) met the criteria for depression. Their
median PHQ-9 score (possible maximum 27) was 6.0
(mean 7.1; SD 5.0). In contrast, among the patients
with a NeP component, 91.6% (n = 2605) of subjects
met the criteria for depression, and their PHQ-9 score
was doubled (median 12.0; mean 12.3; SD 5.8). We
found a strong association between depression severity
and the PD-Q score (Pearson correlation coefficient =
0.43; p < 0.001). The corresponding odds ratios for
mild, moderate and severe depression were found to be
3.2, 8.9 and 25.9, respectively.
Panic/anxiety disorder
According to the PHQ-9 screening, 6747 patients were
evaluable for panic/anxiety disorders; 707 (10.5%)
revealed the presence of these (Table 3). Here again a
correlation was seen, with approximately four times as
many patients in the ‘neuropathic’ group than in the
‘nociceptive’ group (
p < 0.001). The corresponding
odds ratio was 4.5 (95% confidence limits [3.6, 5.6];
Table 3).
Sleep disorders
The MOS sleep questionnaire was evaluated such as to
record for each patient ‘optimal sleep’, ‘sleep disturb-
ance’, ‘somnolence’, ‘sleep quality’ and ‘sleep adequacy’.
The correlation seen is unusually clear: in each case a
clear and substantial worsening was observed with rising
PD-Q score for all reported items (
p < 0.001; Table 3).
Intensity of back pain (VAS)
This also increased markedly with rising PD-Q score
(
p < 0.001), and neuropathic patients had more severe
pain with significantly higher current, average and
worst pain scores on the VAS than the ‘nociceptive’
group (Table 2).
Functionality
The FFbH-R scale provides the percentage of remain-
ing functionality of the patient as compared with a
healthy subject under everyday conditions. There was
a clear decrease in functionality with increasing PD-Q
Score (
p < 0.001). Table 3 illustrates clearly how a NeP
component in the majority of the sufferers severely
affects functionality.
Discussion
Our results are noteworthy because of three major
findings.
First of all, we established a reliable, simple and
validated screening tool to predict the likelihood of a
neuropathic pain component in chronic pain disorders,
© 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••) painDETECT: screening for neuropathic pain Freynhagen et al. P7
for use either in hand-held computers (e.g. in clinical
trials) or as a pencil-and-paper questionnaire (in routine
practice). In the validation, the PD-Q showed a slightly
higher sensitivity and specificity in comparison with
other screening questionnaires for neuropathic pain
such as DN4
11
, LANSS
12
, the NPQ
13
or the NPS
14
. It
can easily be applied fully by the patient, without any
prior physical examination by medical personnel. It is
the first tool to use unique pain patterns as a principal
component and, while it incorporates radiation, it is
particularly suitable for initial screening on LBP, for
example in the waiting-room or (if necessary, blinded)
in the context of a clinical trial, in order to distinguish
between pathophysiological subgroups that are
assumed to respond differently to a drug under study.
The level of patients’ acceptance of the PDA version
was very high.
Our validation study has a number of strengths.
The findings are based on a large, multicentre sample
(n = 392), and the pain classification used in developing
and validating the PD-Q was based upon the current
gold standard – i.e. diagnoses by expert pain physicians
based upon all available criteria. Unlike other screening
questionnaires, the PD-Q validation also included
LBP diagnosed as predominantly neuropathic or
nociceptive in origin. In addition, the validation
Table 2. Socio-demographic and clinical characteristics of the low back pain population
painDETECT score
12
13–18
19
Total
Patients, n (%) 2743 (35.3%) 2153 (27.7%) 2876 (37.0%) 7772 (100%)
male, n (%)
1010 (36.8%) 823 (38.2%) 1070 (37.2%) 2903 (37.4%)
female, n (%) 1733 (63.2%) 1330 (61.8%) 1806 (62.8%) 4869 (62.6%)
Age (years)* 53.4 ± 15.1 52.6 ± 13.1 51.5 ± 11.9 52.5 ± 13.4
P25–P75 41.0–65.0 42.0–63.0 42.0–60.0 42.0–63.0
Height (cm)*
males 178 ± 7 178 ± 7 178 ± 7 178 ± 7
females 165 ± 6 165 ± 6 166 ± 6 165 ± 6
Weight (kg)*
males 85 ± 13 86 ± 16 87 ± 14 86 ± 14
females 71 ± 13 72 ± 14 73 ± 15 72 ± 14
BMI (kg/m
2
)*
males 26.9 ± 3.9 27.0 ± 4.2 27.4 ± 4.0 27.1 ± 4.0
females 26.0 ± 4.8 26.4 ± 4.7 26.8 ± 5.3 26.4 ± 5.0
Pain treatment (months)* 56 ± 85 67 ± 88 73 ± 90 65 ± 88
P25–P75 25.0–67.0 10.0–82.0 14.0–93.0 9.0–80.0
Number of visits to a physician for pain in the past 4 weeks
no visit, n (%)
553 (20.2%) 253 (11.8%) 197 (6.8%) 1003 (12.9%)
1 visit, n (%)
830 (30.3%) 607 (28.2%) 663 (23.1%) 2100 (27.0%)
2 visits, n (%)
657 (24.0%) 582 (27.0%) 735 (25.6%) 1974 (25.4%)
3 or more, n (%) 703 (25.6%) 711 (33.0%) 1281 (44.5%) 2695 (34.7%)
Number of different therapists* 1.7 ± 1.3 2.0 ± 1.4 2.5 ± 1.6 2.1 ± 1.5
P25–P75 1.0–3.0 1.0–3.0 1.0–3.0 1.0–3.0
Ongoing psychotherapy, n (%) 152 (5.5%) 183 (8.5%) 356 (12.4%) 691 (8.9%)
Intended or submitted pension application, n (%) 403 (14.7%) 551 (25.6%) 1235 (42.9%) 2189 (28.2%)
Average pain intensity recorded on VAS during the past 4 weeks, n (%)
mild 651 (23.7%) 303 (14.1%) 177 (6.2%) 1131 (14.6%)
moderate 1532 (55.9%) 1258 (58.4%) 1462 (50.8%) 4252 (54.7%)
severe 560 (20.4%) 592 (27.5%) 1237 (43.0%) 2389 (30.7%)
VAS worst pain* 7.0 ± 2.2 7.6 ± 1.9 8.1 ± 1.6 7.6 ± 2.0
VAS average pain* 5.2 ± 2.0 5.8 ± 1.9 6.6 ± 1.8 5.9 ± 2.0
VAS current pain* 4.6 ± 2.4 5.5 ± 2.2 6.5 ± 2.1 5.5 ± 2.4
*Mean ± standard deviation
BMI, body mass index; VAS, visual analogue scale from 0 to 10; P25/P75, 25% and 75% percentiles
Percentages are given by painDETECT score group, except in the top row, where the basis is the row total
Exploratory comparisons between patients with painDETECT score ≤ 12 versus patients with painDETECT score ≥ 19 were carried out
(ANOVA,
2
cross-tabulation for categorical data). p values were < 0.001 for all comparisons except the rate of patients with two
physician’s visits in the past 4 weeks. Regarding demographic characteristics painDETECT score groups were comparable
P8 painDETECT: screening for neuropathic pain © 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••)
process fulfilled the criteria for validation of pain
questionnaires
15
. To ensure validity of content, a large
patient database from the DFNS was reviewed, expert
clinical opinion was obtained and the relevant pain
literature was consulted; this validity was borne out by
the principal-components analysis as applied to the set
of items presented in Table 1 and the use of Kaiser’s
criterion (see Methods). Although traditionally the
diagnosis of NeP relies largely on medical history and
sensory examination, there is still no absolute test for
neuropathic pain available. However, our gold standard-
based approach, involving two experts working
independently of each other and using all available
diagnostic methods, represents the most objective
classification into neuropathic and nociceptive pain
that can currently be attained. Furthermore, the very
small number of inter-rater discrepancies (9/411, or
≈2%) confirms the reliability of the reference standard.
Also, the score distributions of patients with different
disorders of similar algogenetic nature are comparable
[Figure 1(A)]. Hence, these ‘true’ diagnoses were then
used as reference variable in the logistic regression
model. Moreover, the PD-Q questions require graded
answers, rather than binary yes/no responses or a
visual analogue scale. This fine division may enhance
the questionnaire’s value, not only as a research tool
for monitoring clinical progress or helping to discover
pathogenic mechanisms in clinical studies, but also in
daily clinical practice.
The validation study has some potential limitations.
While it took account of a number of likely important
variables based on a large base of patient data, an
analysis of the current literature and experts’ clinical
experience, the possibility cannot be excluded that
some other factor or set of factors associated with
neuropathic, nociceptive, or both types of pain may
have been omitted. The validation did not include a
‘test–retest’; this we considered ethically and scientif-
ically unjustifiable, the former because it would have
required withholding or interrupting treatment of
the patients’ pain, and the latter because pain does
not necessarily remain stable from one day to the next,
so that ‘test–retest’ stability coefficients have only
limited utility as estimates of the reliability of pain
measures
15
.
The classification of NeP based on mechanisms,
aetiology or location has considerable shortcomings
16
;
while several studies have shown that the presence of a
cluster of symptoms allows separation of NeP from non-
neuropathic pain
11–14
other workers, who failed to find
a single symptom or descriptor, contradict this idea
17
.
However, patients ultimately report symptoms rather
than pain mechanisms and, therefore, the development
and validation of screening tools for identifying NeP on
the basis of symptoms and signs seems to be the most
rational way to improve the use of questionnaires in
daily clinical routine
18
.
Our second finding is the demonstration, in a large
representative survey of approximately 8000 patients
with LBP, that 37.0% (n = 2876) had a predominantly
neuropathic type of pain (PD-Q scores ≥ 19) and
35.3% (n = 2743) a predominantly nociceptive type
of pain (PD-Q scores ≤ 12). To estimate how many
patients are affected in the whole of Germany, we
Table 3. Summary of the epidemiological results for various co-morbidities
painDETECT score:
≤ 12 13–18 ≥ 19
p*
Odds ratio†
PHQ-9 score, depression
None (0–4) 985 (36.3%) 383 (18.0%) 239 (8.4%) < 0.001 –
Mild (5–9) 992 (36.6%) 834 (39.1%) 775 (27.3%) < 0.001 3.2
Moderate (10–19) 676 (24.9%) 821 (38.5%) 1460 (51.3%) < 0.001 8.9
Severe (20–27) 59 (2.2%) 94 (4.4%) 370 (13.0%) < 0.001 25.9
Panic/anxiety disorder present 105 (3.9%) 163 (7.6%) 439 (15.4%) < 0.001 4.5
MOS sleep scale
Optimal sleep 1310 (47.8%) 825 (38.4%) 703 (24.5%) < 0.001 0.4
Sleep disturbance 38.5 ± 24.4, 36 48.0 ± 22.7, 46 59.0 ± 61.0, 61 < 0.001 –
Somnolence 36.1 ± 21.9, 33 42.7 ± 21.2, 40 50.9 ± 21.3, 53 < 0.001 –
Sleep quality 6.5 ± 1.4, 7 6.1 ± 1.5, 6 5.6 ± 1.6, 6 < 0.001 –
Sleep adequacy 52.8 ± 27.7, 50 44.2 ± 26.5, 40 35.2 ± 25.2, 30 < 0.001 –
FFbH-R functionality (0–100%) 62.3 ± 24.0, 62 52.8 ± 22.4, 50 41.4 ± 21.2, 41 < 0.001 –
For PHQ-9 and ‘optimal sleep’, n and % are given; percentages are given by group; for the remaining variables, mean ± standard deviation
and median are given
*p values for exploratory comparisons between patients with painDETECT score ≤ 12 versus patients with painDETECT score ≥ 19
(ANOVA,
2
cross-tabulation for categorical data)
†Odds ratios for the various PHQ-9 scores describe patients with the respective grading of depression compared with patients without
depression
© 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••) painDETECT: screening for neuropathic pain Freynhagen et al. P9
took into account the latest data from the Robert
Koch Institute on chronic back pain
19
. On the basis
of given prevalence of LBP in the general population,
we calculated that 14.5% of all female and 11.4% of
all male Germans suffer from LBP with a predominant
neuropathic pain component. Hence, NeP is a major
contributor to chronic LBP, and a correct differential
diagnosis to detect the most likely operating
pain type is essential. The distinction between
neuropathic and nociceptive pain types and the
estimation of its relevance are crucial, since the
different pain types require different therapeutic
approaches. Although non-steroidal anti-inflammatory
drugs are still the first-line drugs for chronic
pain
7
, sufficient evidence for their use in chronic
LBP
20
or NeP
21,22
is lacking. However, there is strong
evidence that opioids, cyclic antidepressants and
anticonvulsants are equally effective for the treatment
of NeP and, in addition, the realisation that symptoms
other than pain are sometimes more important and/
or easier to overcome can increase the benefits of
therapy
21
.
It is important to note that the search of the specific
diagnosis of the disease needs to be performed without
delay as in some cases the identification of a neuro-
pathic component may reflect the need of disc surgery
or in rare cases even malignancy. However, although
screening tools can not replace clinical judgment in the
assessment of an individual patient’s pain they can at
least alert clinicians to undertake prompt further diag-
nostic evaluation which may subsequently influence
pain management.
Thirdly, the present study reveals fundamental
differences, in respect of perceived pain and of various
co-morbidities, between LBP patients with neuropathic
and those with nociceptive components. The additional
effect on peripheral nerves in low back pain obviously
leads to considerable changes in the entire central
processing of nociceptive information in the brain,
as well as to more (and more distinct) co-morbidities
such as depression, panic/anxiety disorder and sleep
disturbance, seriously affecting the quality of patients’
lives. NeP patients appeared more likely to have a lower
residual functionality under everyday conditions and
to suffer longer and with a more severe pain intensity.
On the basis of average pain intensity reported for the
past 4 weeks 43% were suffering severe pain (pain
score > 7–10 on an 11-point scale
23,24
), compared
with 24% in the nociceptive group. NeP patients
had more frequent visits to physicians and more
different therapists, more psychotherapy, a longer
duration of pain treatment and made pension
applications earlier. These findings illustrate the impact
of the NeP component on health-care resource util-
isation. As we demonstrated, the detrimental effects
of NeP ultimately result in a decreased quality of life,
so that our results imply that separating predominantly
neuropathic from non-neuropathic pain components
should be made a central goal in the management
of LBP.
Growing evidence exists to support a multidisciplin-
ary therapeutic approach, and physicians should follow
clinical practice guidelines in their daily management
of LBP
2,3,20
. The integral role played by psychological
and social factors in LBP is still largely ignored;
attention to it is reserved for patients who do not
respond to cheaper treatment options. However,
our finding that a NeP component has such an
immense impact on the burden of the individual
sufferers was an unexpected result of our study and,
therefore, at least a cheap and simple screening on the
presumed pain type should be performed to sensitise
the physician towards potential co-morbidity problems.
This study demonstrates that the PD-Q is a valuable
tool that may cover an unmet need. It is currently
being developed for application in different linguistic
and cultural contexts.
Conclusion
The PD-Q developed, validated and applied to 8000
LBP patients constitutes a new, reliable and simple
screening tool to predict the likelihood of a neuropathic
pain component being present in individual patients
and in inhomogeneous cohorts of LBP patients. The
present findings demonstrate a high prevalence of LBP
patients with a predominant NeP component (37%)
and provide important information on the association
between NeP and the occurrence and severity of
different co-morbidities. Patients with a neuropathic
pain component suffer more (and more severe) than
those without. The future use of the PD-Q may be
helpful both in clinical research and daily clinical
routine.
Acknowledgements
Declaration of interest: This research was supported by
Pfizer Pharma GmbH, Germany, without restriction
on publication. UG is an employee of Pfizer Germany.
RF, RB and TRT received research support, consult-
ancy and lecture fees from a number of pharmaceutical
companies but have no direct stock holding in any of
them. The authors would like to thank all participating
patients, colleagues and the staff of the institutions for
their contributions to data collection. The research
benefited from the technical and statistical support of
StatConsult GmbH, Germany.
P10 painDETECT: screening for neuropathic pain © 2006 LIBRAPHARM LTD – Curr Med Res Opin 2006; 22(••)
References
1. Deyo RA, Weinstein JN. Low back pain. New Engl J Med
2001;344:363–70
2. Feuerstein M, Hartzell M, Rogers HL, et al. Evidence-based
practice for acute low back pain in primary care: patient
outcomes and cost of care. Pain 2006;124:140-9
3. van Tulder MW, Tuut M, Pennick V, et al. Quality of primary care
guidelines for acute low back pain. Spine 2004;29:E357-E362
4. O’Sullivan P. Diagnosis and classification of chronic low back pain
disorders: maladaptive movement and motor control impairments
as underlying mechanism. Man Ther 2005;10:242-55
5. Cruccu G, Anand P, Attal N, et al. EFNS guidelines on neuro-
pathic pain assessment. Eur J Neurol 2004;11:153-62
6. Baron R, Binder A. How neuropathic is sciatica? The mixed
pain concept. Orthopade 2004;33:568-75
7. Breivik H, Collett B, Ventafridda V, et al. Survey of chronic
pain in Europe: prevalence, impact on daily life, and treatment.
Eur J Pain 2006;10:287-333
8. Hays RD, Martin SA, Sesti AM, et al. Psychometric properties
of the Medical Outcomes Study Sleep measure. Sleep Med
2005;6:41-4
9. Lowe B, Unutzer J, Callahan CM, et al. Monitoring depression
treatment outcomes with the patient health questionnaire-9.
Med Care 2004;42:1194-201
10. Kohlmann TRH. Hannover Functional Questionnaire in
ambulatory diagnosis of functional disability caused by backache.
Rehabilitation 1996 Feb;35:1-8
11. Bouhassira D, Attal N, Alchaar H, et al. Comparison of pain
syndromes associated with nervous or somatic lesions and
development of a new neuropathic pain diagnostic question-
naire (DN4). Pain 2005;114:29-36
12. Bennett M. The LANSS Pain Scale: the Leeds assessment of
neuropathic symptoms and signs. Pain 2001;92:147-57
13.
Krause SJ, Backonja MM. Development of a neuropathic pain
questionnaire. Clin J Pain 2003;19:306-14
14. Galer BS, Jensen MP. Development and preliminary validation
of a pain measure specific to neuropathic pain: the Neuropathic
Pain Scale. Neurology 1997;48:332-8
15. Jensen MP. Questionnaire validation: a brief guide for readers of
the research literature. Clin J Pain 2003;19:345-52
16. Baron R. Disease mechanisms in neuropathic pain: a clinical
perspective. Nature Clin Pract Neurol 2006;2:95-106
17. Rasmussen PV, Sindrup SH, Jensen TS, et al. Symptoms
and signs in patients with suspected neuropathic pain. Pain
2004;110:461-9
18. Finnerup N, Jensen TS. Mechanisms of disease: mechanism-
based classification of neuropathic pain – a critical analysis.
Nature Clin Pract Neurol 2006;2:107-15
19. Neuhauser H, Ellert U, Ziese T. Chronic back pain in the general
population in Germany 2002/2003: prevalence and highly
affected population groups. Gesundheitswesen 2005;67:685-
93
20. Airaksinen O, Brox JI, Cedraschi C, et al. European guidelines
for the management of chronic nonspecific low back pain. Eur
Spine J 2006;15(Suppl 2):192-300
21. Finnerup NB, Otto M, McQuay HJ, et al. Algorithm for
neuropathic pain treatment: an evidence based proposal. Pain
2005;118:289-305
22. Dworkin RH, Backonja M, Rowbotham MC, et al. Advances
in neuropathic pain: diagnosis, mechanisms, and treatment
recommendations. Arch Neurol 2003;60:1524-34
23. Breivik EK, Bjornsson GA, Skovlund E. A comparison of pain
rating scales by sampling from clinical trial data. Clin J Pain
2000;16:22-8
24. Paul SM, Zelman DC, Smith M, et al. Categorizing the severity
of cancer pain: further exploration of the establishment of
cutpoints. Pain 2005;113:37-44
CrossRef links are available in the online published version of this paper:
http://www.cmrojournal.com
Paper CMRO-3586_3, Accepted for publication: 16 August 2006
Published Online:
00 September 2006
doi:10.1185/030079906X132488
