Article

An international multicentre validation study of a pain classification system for cancer patients

University of Alberta, Edmonton, Canada.
European journal of cancer (Oxford, England: 1990) (Impact Factor: 5.42). 11/2010; 46(16):2896-904. DOI: 10.1016/j.ejca.2010.04.017
Source: PubMed

ABSTRACT

The study's primary objective was to assess predictive validity of the Edmonton Classification System for Cancer Pain (ECS-CP) in a diverse international sample of advanced cancer patients. We hypothesised that patients with problematic pain syndromes would require more time to achieve stable pain control, more complicated analgesic regimens and higher opioid doses than patients with less complex pain syndromes.
Patients with advanced cancer (n=1100) were recruited from 11 palliative care sites in Canada, USA, Ireland, Israel, Australia and New Zealand (100 per site). Palliative care specialists completed the ECS-CP for each patient. Daily patient pain ratings, number of breakthrough pain doses, types of pain adjuvants and opioid consumption were recorded until study end-point (i.e. stable pain control, discharge and death).
A pain syndrome was present in 944/1100 (86%). In univariate analysis, younger age, neuropathic pain, incident pain, psychological distress, addictive behaviour and initial pain intensity were significantly associated with more days to achieve stable pain control. In multivariate analysis, younger age, neuropathic pain, incident pain, psychological distress and pain intensity were independently associated with days to achieve stable pain control. Patients with neuropathic pain, incident pain, psychological distress or higher pain intensity required more adjuvants and higher final opioid doses; those with addictive behaviour required only higher final opioid doses. Cognitive deficit was associated with fewer days to stable pain control, lower final opioid doses and fewer pain adjuvants.
The replication of previous findings suggests that the ECS-CP can predict pain complexity in a range of practice settings and countries.

Full-text

Available from: Michael Fisch
An international multicentre validation study of a pain
classification system for cancer patients
5
Robin L. Fainsinger
a,
*
, Cheryl Nekolaichuk
a
, Peter Lawlor
b
, Neil Hagen
c
,
Michaela Bercovitch
d
, Michael Fisch
e
, Lyle Galloway
c
, Gina Kaye
f
, Willem Landman
g
,
Odette Spruyt
h
, Donna Zhukovsky
e
, Eduardo Bruera
e
, John Hanson
i
a
University of Alberta, Edmonton, Canada
b
University of Ottawa, Ottawa, Canada
c
University of Calgary, Calgary, Canada
d
Tel Hashomer Hospice, Tel Aviv, Israel
e
M.D. Anderson Cancer Center, Houston, United States
f
South Auckland Hospice, Auckland, New Zealand
g
Middlemore Hospital, Auckland, New Zealand
h
Peter MacCallum Cancer Centre, Melbourne, Australia
i
Cross Cancer Institute, Edmonton, Canada
ARTICLE INFO
Article history:
Received 4 March 2010
Accepted 19 April 2010
Available online 17 May 2010
Keywords:
Cancer pain
Pain assessment
Validation study
Edmonton Classification System for
Cancer Pain
ABSTRACT
Purpose: The study’s primary objective was to assess predictive validity of the Edmonton
Classification System for Cancer Pain (ECS-CP) in a diverse international sample of
advanced cancer patients. We hypothesised that patients with problematic pain syn-
dromes would require more time to achieve stable pain control, more complicated analge-
sic regimens and higher opioid doses than patients with less complex pain syndromes.
Methods: Patients with advanced cancer (n = 1100) were recruited from 11 palliative care
sites in Canada, USA, Ireland, Israel, Australia and New Zealand (100 per site). Palliative
care specialists completed the ECS-CP for each patient. Daily patient pain ratings, number
of breakthrough pain doses, types of pain adjuvants and opioid consumption were
recorded until study end-point (i.e. stable pain control, discharge and death).
Results: A pain syndrome was present in 944/1100 (86%). In univariate analysis, younger
age, neuropathic pain, incident pain, psychological distress, addictive behaviour and initial
pain intensity were significantly associated with more days to achieve stable pain control.
In multivariate analysis, younger age, neuropathic pain, incident pain, psychological dis-
tress and pain intensity were independently associated with days to achieve stable pain
control. Patients with neuropathic pain, incident pain, psychological distress or higher pain
intensity required more adjuvants and higher final opioid doses; those with addictive
behaviour required only higher final opioid doses. Cognitive deficit was associated with
fewer days to stable pain control, lower final opioid doses and fewer pain adjuvants.
0959-8049/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ejca.2010.04.017
5
Presentation list: Fainsinger R, Nekolaichuk C, Lawlor P, et al. An International Multicentre Validation Study of a Pain Classification
System for Advanced Cancer Patients. Oral Presentation at the 11th Congress of the European Association for Palliative Care, Vienna,
Austria, 10th May 2009.
* Corresponding author: Address: Division of Palliative Care Medicine, Grey Nuns Hospital, 217 Health Services Centre, 1090 Youville
Drive West, Edmonton, Alberta, Canada T6L 0A3. Tel.: +1 780 735 7727; fax: +1 780 735 7302.
E-mail address: Robin.Fainsinger@albertahealthservices.ca (R.L. Fainsinger).
EUROPEANJOURNALOFCANCER46 (2010) 2896 2904
available at www.sciencedirect.com
journal homepage: www.ejconline.com
Page 1
Conclusion: The replication of previous findings suggests that the ECS-CP can predict pain
complexity in a range of practice settings and countries.
Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction
Pain is one of the most prevalent and distressing symptoms in
patients with advanced cancer. Approximately 70% of these
patients will experience pain at some point during the pro-
gression of their disease.
1
Although most patients achieve
adequate pain control,
2,3
some particularly patients with
more complex pain syndromes fail to obtain satisfactory
analgesia. For these patients, clinicians may need to adopt a
more intense and complex programme of therapeutic inter-
vention, and as a result, more time is often required to
achieve adequate pain control.
4
Standardised approaches for assessing and classifying
cancer pain need to be developed to identify and treat pa-
tients with complex pain syndromes. However, the com-
plex, multidimensional nature of cancer pain presents
unique challenges for pain classification. A review of the
cancer pain literature has revealed the difficulty in compar-
ing research results of analgesic management for cancer
pain, due to the lack of a standardised approach.
5
Diverse
interpretations of the pain experience, as well as many fac-
tors that may contribute to it, have impeded the develop-
ment of a standardised classification system. Although
better characterisation and classification of pain syndromes
would allow for more valid clinical and research compari-
sons, there is no universally accepted pain classification
tool.
6,7
The development of a standardised classification system
that is comprehensive, predictive of difficulty in achieving
analgesia and simple to use could provide a common lan-
guage for the clinical management and research of cancer
pain. Bruera and colleagues recognised the need for such a
system, prompting the development of the Edmonton Staging
System (ESS).
8,9
The ESS has been used in a number of reports
where it was found useful in describing the underlying cancer
pain syndrome.
10–16
Interpretational difficulties with analge-
sic prognosis and feature definitions have limited the interna-
tional acceptance of the ESS. To overcome these limitations,
an expert panel, consisting of physicians and researchers in
the Edmonton Regional Palliative Care Program, developed
the revised Edmonton Staging System (rESS). We have subse-
quently conducted five validation studies: a pilot study, a re-
gional multicentre study,
17
secondary analysis looking at
pain intensity,
18
opioid escalation
19
and a construct validation
study for validating definitions using an expert panel.
20
Based
on feedback generated by the latter study and to reflect the in-
tended use as a classification system, the amended instru-
ment was renamed the Edmonton Classification System for
Cancer Pain (ECS-CP).
21
The ECS-CP includes five features
pain mechanism, incident pain, psychological distress, addic-
tive behaviour and cognitive function (Appendix A). These
features and the definitions and guidelines for use are the ba-
sis of the ECS-CP (Appendix B).
Using the revised definitions for the ECS-CP pain fea-
tures,
21
the primary objective of this study was to assess the
predictive validity of the ECS-CP as a tool for classifying can-
cer pain in a diverse international sample of patients, who
were referred to palliative care services. Our hypothesis was
that patients with more problematic pain features, as classi-
fied by the ECS-CP, would require a longer time to achieve sta-
ble pain control, require more complicated analgesic
regimens and use higher opioid doses than patients with less
complex pain syndromes.
2. Methods
A total of 1100 consecutive patients were recruited from 11
palliative care sites in Canada, the United States, Ireland, Is-
rael, Australia and New Zealand (100 patients per site). The
selection of these sites was purposeful, being limited to loca-
tions providing specialist palliative care services, such as a
palliative consult service (inpatient and outpatient), tertiary
palliative care unit or hospice setting. At study entry, a palli-
ative care specialist (physician or nurse consultant) com-
pleted the ECS-CP for each patient with cancer who was
referred to the service. The following additional information
was recorded until the study end-point: daily patient pain rat-
ings; daily number of breakthrough pain doses; initial and fi-
nal opioid consumption and types of adjuvant pain
treatments.
The primary inclusion criteria were patients with cancer,
18 years of age or older, who had been referred to a palliative
care service. Cancer patients who did not have a pain syn-
drome resulting from their cancer diagnosis were included
in the cohort description of patient demographics, but ex-
cluded from further analysis.
An initial ECS-CP assessment is recommended prior to
pain management by specialist palliative care services (e.g.
on admission to an inpatient unit or referral to a care consul-
tation service). Subsequent assessments may be conducted if
the patient’s condition changes or as additional information
regarding the five pain features is obtained. For the purposes
of this study, only an initial assessment was conducted. If the
patient did not have a cancer pain syndrome on initial referral
or admission, then no further data were collected for the
study.
Patients rated their current level of pain intensity on the
day of initial assessment and then daily until study termina-
tion, using the Pain-Numerical Rating Scale (NRS), ranging
from 0 (no pain) to 10 (worst possible pain). This assessment
is a routine clinical practice in the designated data collection
sites. A secure website was specially created to facilitate
investigator training at multiple study sites. All study materi-
als, including background information about the ECS-CP, the
ECS-CP Administration Manual,
21
other assessment tools
and password-protected data collection forms were posted
EUROPEANJOURNALOFCANCER46 (2010) 2896 2904 2897
Page 2
on this website. Once formal ethics board approval was ob-
tained at a designated study site, an individual teleconference
training meeting was organised between the principal inves-
tigators, the site-specific collaborator and research support
staff.
Patient demographics were recorded. The following infor-
mation was reviewed and recorded on initial assessment
and then as required until study termination (i.e. stable pain
control, discharge and death): patient rated daily pain inten-
sity (at the moment of rating), using a numerical rating scale
(0–10) in those who were cognitively intact, in the opinion of
the investigator; daily number of opioid doses given for
breakthrough pain; initial total morphine equivalent daily
dose (MEDD) of opioid on the initial assessment day, final
MEDD on the day of study termination; number and type
of adjuvant analgesics and/or other treatments used to man-
age pain; date of and reason for study termination (achieve-
ment of stable pain control, death or discharge resulting in
loss of follow-up).
For the purposes of this study, stable pain control was de-
fined as receiving less than three breakthrough analgesic
doses per day and a patient self-reported pain score of less
than or equal to 3/10 for three consecutive days.
17,18
If the pa-
tient was unable to self-report pain, then stable pain control
was defined solely as receiving less than three breakthrough
opioid analgesic doses per day for 3 consecutive days.
Palliative care physicians at all sites were required to treat
cancer pain according to the National Cancer Institute of
Health Physician Data Query (PDQ) Pain Guidelines.
22
Adju-
vant analgesics and non-pharmacological treatments (e.g.
radiotherapy, chemotherapy, anaesthetic or surgical proce-
dures, acupuncture and transcutaneous nerve simulation)
were used as required.
Data were recorded on the ECS-CP Teleform. The Tele-
form
â
is an optical recognition-based technology that scans
and exports data from data collection forms directly to a com-
puter database, which is particularly useful with multiple
data collection sites. We have successfully implemented this
data collection process in previous research.
17,23
Pooled data from all 11 sites were analysed using
descriptive and inferential statistics. Kaplan–Meier survival
curves were constructed to estimate the probabilities of
achieving stable pain control over time for the eight explan-
atory variables identified in our previous validation studies
(i.e. age, gender, mechanism of pain, incident pain, psycho-
logical distress, addictive behaviour, cognitive function and
pain intensity).
17,18
Univariate and multivariate Cox regres-
sion analyses were performed, for identifying the associa-
tions between the eight explanatory variables and the
outcome variable, time to stable pain control.
24
A Chi-
square test was used to determine the differences in the
use of adjuvant analgesics and non-pharmacological treat-
ments for pain control. The non-parametric Kruskal–Wallis
One-Way ANOVA by ranks and Mann–Whitney tests were
used to examine differences in final median opioid doses,
as the sample distribution was not normal. Statistical sig-
nificance was set at p < 0.05 (2-tailed). The univariate and
multivariate Cox regression analyses were performed using
the SAS procedure PHREG (SAS 9.1 TSIM3; SAS Institute,
Cary, NC).
3. Results
Demographic and clinical characteristics of the 1100 patients
included in the study are listed in Table 1. Of these, 944 (86%)
had a pain syndrome. The patients with a pain syndrome
were significantly younger (p < .001), less likely to have lung
Table 1 Patient demographics and clinical characteristics
(n = 1100).
Characteristics Patients
with a pain
syndrome
n (%) (n = 944)
a
Patients
without
a pain
syndrome
n (%)
(n = 156)
p-Value
Chi-square
test
Age, years
Mean 61 69 <.001
Standard
deviation
15 13
Sex
Female 472 (50) 74 (47) .55
Diagnosis
Gastrointestinal 228 (24) 41 (26) .77
Lung 208 (22) 49 (31) .03
Genitourinary 146 (15) 12 (8) .01
Breast 125 (13) 17 (11) .33
Head and Neck 52 (6) 5 (3) .19
Other 59 (6) 16 (11) .09
Unknown origin 41 (4) 7 (4) .98
Haematological 47 (5) 9 (6) .76
Disposition
Stable pain 478 (51)
Death 160 (17)
Discharge 306 (32)
Mechanism of pain
Nociceptive (Nc) 636 (67)
Neuropathic
pain (Ne)
257 (27)
Unknown (Nx) 51 (5)
Incident pain
Absent (Io) 408 (43)
Present (Ii) 457 (48)
Unknown (Ix) 79 (8)
Psychological
distress
Absent (Po) 417 (44)
Present (Pp) 413 (44)
Unknown (Px) 114 (12)
Addictive
behaviour
Absent (Ao) 743 (79)
Present (Aa) 107 (11)
Unknown (Ax) 94 (10)
Cognition
Unimpaired (Co) 653 (69)
Impairment (Ci) 200 (21)
Unresponsive (Cu) 78 (8)
Unknown (Cx) 13 (1)
Note: Percentages do not always add up to 100% due to rounding.
a
Frequencies do not always add up to 944 due to missing values.
2898 EUROPEANJOURNALOFCANCER46 (2010) 2896 2904
Page 3
cancer (p = .03) and more likely to have genito-urinary cancer
(p = .01) than the patients with no pain syndromes. Fifty per-
cent of patients with a pain syndrome (n = 478) achieved sta-
ble pain control. The remaining patients had either died
(n = 160, 17%) or had been discharged and lost to follow-up
(n = 306, 32%). Most patients had nociceptive pain (67%), and
neuropathic pain was present in 27%; 48% had incident pain;
psychological distress was present and absent in similar
numbers (44%); addictive behaviour was present in 11% and
69% were cognitively intact. The ‘unknown’ classification op-
tion was used to varying degrees for all features, ranging from
1% (cognition) to 12% (psychological distress).
The associations between time to achieve stable pain con-
trol and age, gender, pain mechanism, incident pain, psycho-
logical distress, addictive behaviour, cognition and pain
intensity are summarised in Table 2. As shown in this table,
the median time to achieve stable pain control ranged from
3 d (unknown cognitive function and unknown pain mecha-
nism) to 16 d (neuropathic pain and unknown psychological
distress). Based on the univariate Cox regression analyses,
age less than 60 years (HR = 0.68, p < .0001), neuropathic pain
(HR = 0.55, p < .0001), incident pain (HR = 0.57, p < .0001), psy-
chological distress present (HR = 0.74, p = .002) and unknown
(HR = 0.63, p = .006), addictive behaviour (HR = 0.71, p = .029)
and pain intensity moderate (4–6) (HR = 0.65, p < .0001) and se-
vere (7–10) (HR = 0.52, p < .0001) were significantly associated
with longer time to achieve stable pain control. Unknown
pain mechanism (HR = 1.58, p = .031) and unresponsive cogni-
Table 2 Association between time to achieve stable pain control and age, gender, ECS-CP features and baseline pain
intensity ratings.
Kaplan–Meier Cox regression
b
Univariate (n = 944) Multivariate (n = 860)
No. of patients
a
Median time
(days) to stable
pain control
(95% confidence interval (CI))
Chi-square p HR (95% CI) p HR (95% CI)
Age
P60 518 7 (6–8) 1 1
<60 426 11 (9–14) 17.34 <.0001 0.68 (0.56–0.81) .049 0.82 (0.67–1.00)
Sex
Female 472 8 (7–9) 1 1
Male 472 9 (7–11) 0.76 .382 0.92 (0.77–1.11) .866 0.98 (0.81–1.20)
Pain mechanism
Nociceptive 636 7 (6–8) 1 1
Neuropathic 257 16 (11–25) 27.54 <.0001 0.55 (0.45–0.69) <.0001 0.62 (0.49–0.79)
Unknown 51 3 (3–6) 4.68 .031 1.58 (1.04–2.39) .837 0.93 (0.48–1.80)
Incident pain
Absent 408 7 (5–7) 1 1
Present 457 11 (10–15) 33.20 <.0001 0.57 (0.48–0.69) .001 0.72 (0.59–0.88)
Unknown 79 6 (4–9) 0.05 .828 1.04 (0.73–1.49) .886 1.04 (0.60–1.81)
Psych distress
Absent 417 7 (6–8) 1 1
Present 413 9 (8–11) 10.01 .002 0.74 (0.61–0.89) .018 0.79 (0.64–0.96)
Unknown 114 16 (9–28) 7.65 .006 0.63 (0.46–0.88) <.0001 0.37 (0.22–0.60)
Addict behaviour
Absent 743 8 (7–9) 1 1
Present 107 14 (8–26) 4.75 .029 0.71 (0.52–0.97) .199 0.81 (0.58–1.12)
Unknown 94 10 (5–16) 0.12 .728 0.94 (0.67–1.32) .671 0.90 (0.55–1.47)
Cognition
Normal 653 9 (7–10) 1 1
Impaired 200 9 (7–12) 0.09 .760 0.97 (0.77–1.21) .614 0.94 (0.73–1.20)
Unresponsive 78 5 (3–8) 6.71 .010 1.55 (1.11–2.16) .366 1.25 (0.77–2.01)
Unknown 13 3 (3 to –) 2.06 .152 1.81 (0.81–4.06) .218 1.92 (0.68–5.39)
Pain intensity
Mild (0–3) 310 5 (4–6) 1 1
Mod (4–6) 267 10 (8–11) 13.91 <.0001 0.65 (0.52–0.82) .005 0.72 (0.57–0.91)
Severe (7–10) 283 13 (9–17) 30.64 <.0001 0.52 (0.41–0.66) .001 0.65 (0.51–0.83)
Missing 84
Abbreviations: HR hazard ratio; n.s. non-significant; PD psychological distress; AB addictive behaviour.
a
Total number of patients may vary due to missing values.
b
Forced entry model.
EUROPEANJOURNALOFCANCER46 (2010) 2896 2904 2899
Page 4
tive status (HR = 1.55, p = .01) were significantly associated
with shorter time to stable pain control. In the multivariate
Cox regression analysis, only younger age (HR = 0.82,
p = .049), neuropathic pain (HR = 0.62, p < .0001), incident pain
(HR = 0.72, p = .001), psychological distress present (HR = 0.79,
p = .018) and unknown (HR = 0.37, p < .0001) and pain intensity
(moderate, HR = 0.72, p = .005; severe, HR = 0.65, p = .001), were
significantly associated with longer time to achieve stable
pain control.
The number of adjuvant treatments used for pain control
was calculated for each individual. These were further classi-
fied into three groups, according to the number required to
achieve stable pain control: no adjuvant treatments required;
one adjuvant treatment required and two or more adjuvant
treatments required (Table 3). Patients with neuropathic pain,
incident pain and moderate to severe pain intensity used sig-
nificantly more modalities to achieve stable pain control
(p < .001). Patients with psychological distress also required
a greater variety of management options ( p = .041). Patients
reported as having an unknown pain mechanism (p = .019),
unknown addictive behaviour status (p < .001) or a cognitively
unresponsive status (p < .001) required significantly less treat-
ment options.
Analysis of the final morphine equivalent daily dose
(MEDD) demonstrated significantly higher opioid doses for
patients with neuropathic pain (p < .001); incident pain
(p < .001); psychological distress (p = .002); addictive behaviour
(p = .038) and moderate to severe pain intensity (p < .001).
Unknown cognition or unresponsive patients required a sig-
nificantly lower final MEDD (p = .01 and p < .001, respectively),
as did patients with unknown pain mechanism (p = .002) and
unknown addictive behaviour (p < .001) (Table 4).
4. Discussion
The results of this international multicentre study confirm
the findings of our previous research: neuropathic pain, inci-
dent pain, psychological distress, addictive behaviour and
moderate to severe pain intensity are significant predictors
of complexity of pain management as measured by the out-
comes of longer duration (days) to achieve stable pain control,
the use of more adjuvant treatments and the use of higher
opioid doses. As noted previously
17
these findings reflect clin-
ical practice, in which patients with more complex pain syn-
dromes, such as neuropathic and incident pain, psychological
distress, addictive behaviour and moderate or severe pain
intensity, can still achieve stable pain control, but may require
more time and more complex management strategies involv-
ing more adjuvant approaches and higher opioid doses, in
comparison with patients with less complex pain syndromes.
Conversely, declining cognition and the proxy measure of the
‘unknown’ option for the other features (perhaps in some
cases due to difficulty obtaining a history in cognitively im-
paired patients) were associated with less complexity in pain
Table 3 Use of adjuvant analgesics and non-pharmacological management for pain control (n = 944).
ECS-CP features # of adjuvants used n (%) p value
0 1 2+ Chi-square test
Pain mechanism <.001
Nociceptive (n = 636) 247 (39) 246 (39) 143 (22)
Neuropathic (n = 257) 28 (11) 87 (34) 142 (55) <.001
Unknown (n = 51) 30 (59) 14 (27) 7 (14) .019
Incident pain <.001
Absent (n = 408) 177 (43) 139 (34) 92 (23)
Present (n = 457) 89 (19) 182 (40) 186 (41) <.001
Unknown (n = 79) 39 (49) 26 (33) 14 (18) 0.53
Psychological distress 0.11
Absent (n = 417) 149 (36) 147 (35) 121 (29)
Present (n = 413) 114 (28) 161 (39) 138 (33) .041
Unknown (n = 114) 42 (37) 39 (34) 33 (29) 0.97
Addictive behaviour .004
Absent (n = 743) 224 (30) 279 (38) 240 (32)
Present (n = 107) 34 (32) 40 (37) 33 (31) 0.93
Unknown (n = 94) 47 (50) 28 (30) 19 (20) <.001
Cognition <.001
Normal (n = 653) 188 (29) 238 (36) 227 (35)
Impaired (n = 200) 68 (34) 79 (40) 53 (27) 0.083
Unresponsive (n = 78) 43 (55) 26 (33) 9 (12) <.001
Unknown (n = 13) 6 (46) 4 (31) 3 (23) 0.38
Pain intensity <.001
0–3 (n = 310) 136 (44) 101 (33) 73 (24)
4–6 (n = 267) 64 (24) 111 (42) 92 (34) <.001
7+ (n = 283) 54 (19) 114 (40) 115 (41) <.001
Missing (n = 84)
n.s. non-significant (p value > 0.05).
2900 EUROPEANJOURNALOFCANCER46 (2010) 2896 2904
Page 5
management. The surprise finding of increased time to
achieve stable pain control in those with unknown psycholog-
ical distress status requires further explanation. This may be
explained by the investigator’s difficulty in applying the defi-
nition for this feature on initial assessment or difficulties in
relation to patients’ cognitive status. We cannot exclude the
possibility that many of these patients could have met the
definition on reassessment.
The magnitude of the association between pain intensity
and time to achieve stable pain control in the multivariate
analysis is not as strong as in our previous study (i.e.
p = .001 versus p < .0001).
18
However, the robust multivariate
findings suggest that the association of initial pain intensity
with time to stable pain control is truly independent, and
needs to be incorporated into the classification system. A
simple and practical suggestion proposed at a recent work-
shop on ‘Pain classification and assessment’ at the Mari Negri
Institute in Milan was to add the pain intensity number to the
other ECS-CP features. For example, a patient with neuro-
pathic pain (Ne), incident pain (Ii), psychological distress
(Pp) and addictive behaviour (Aa), with normal cognition
(Co) and a pain intensity of 7, could be classified as 7-NeIiPpA-
aCo. There are a number of items captured by the ECS-CP, but
there are likely other confounding issues for some individuals
such as age, chronic pain, true analgesic tolerance or genetic
factors. Further research and mathematical modelling may
provide a mechanism to attribute a numerical value (e.g.
weighting system) to the items included in the ECS-CP as
we work towards developing a comprehensive, prognostic,
standardised classification system for cancer pain.
A standardised, comprehensive and simple classification
of cancer pain would support physicians to better manage pa-
tients’ cancer pain and inform resource allocation decisions
within cancer programmes, through earlier identification of
patients with more difficult to manage pain. This type of clas-
sification would also enable researchers to compare results of
outcome surveys and clinical trials in cancer pain manage-
ment. Currently, it is possible that large discrepancies in the
efficacy of a given treatment between groups can simply
result from different characteristics in the population under
study. It is our hope that, in the future, the ECS-CP could play
a significant role in treatment planning, evaluating and
reporting research results in the assessment and manage-
ment of cancer pain. One positive development arising from
site participation in this study is the potential establishment
of an international working group for further development
and validation of this pain classification system. This study
involving diverse palliative populations strengthens the po-
tential use of the ECS-CP as the base for ongoing development
and evolution of an internationally recognised classification
system.
At a European Palliative Care Research Collaborative
(EPCRC) Research forum in the Lofoten Islands prior to the
European Association of Palliative Care Conference in Trond-
heim, Norway in May 2008, the discussion of a Classification
system for Cancer Pain was informed and guided by a
Systematic Literature review on the topic.
6,25
As there is no
single accepted framework on how to classify cancer pain,
one of the aims of the EPCRC is to develop a classification sys-
tem for advanced cancer patients with pain, based on inter-
national consensus. A pivotal issue of the discussion was
whether to go ahead and gain experience with introducing
and applying the items included in the ECS-CP and continue
to develop this system or to develop a new consensus tool.
There was agreement that the ECS-CP offered the best start-
ing point for evolution of an international classification sys-
tem for cancer pain and would be used in multi-site
research initiatives being planned by the EPCRC.
25
The current study has several limitations. The patient pop-
ulation was heterogeneous by design due to the multiple sites
and centres. Further research on more homogeneous patient
populations may improve understanding of how the different
features of the ECS-CP influence the achievement of stable
pain control and complexity of management in different set-
tings. We cannot exclude a selection bias due to referral of pa-
tients with more difficult pain problems to the respective
palliative care services at the multiple study sites. Despite
the intent to have patients report current pain intensity, the
possibility that some patients reported worst or average pain
intensity cannot be excluded. We did not systematically re-
cord the degree of chronicity of the cancer pain presentation.
It is possible that some patients had pain for a longer period
before referral to the different sites, and neurophysiological
changes due to chronicity of pain could have resulted in a
longer time to achieve stable pain control.
Table 4 Final MEDD
*
by ECS-CP features (n = 944).
ECS-CP features Median final
MEDD (25–75%
quartile range)
p value
Kruskal–
Wallis
p value
Mann–
Whitney
Pain mechanism <.001
Nociceptive pain 32 (12–100)
Neuropathic pain 100 (36–235) <.001
Unknown pain
syndrome
15 (4–60) .002
Incident pain
Absent 32 (10–110) <.001
Present 60 (20–170) <.001
Unknown 27 (10–60) 0.17
Psychological distress .009
Absent 37 (12–104)
Present 50 (16–150) .002
Unknown 49 (12–137) 0.22
Addictive behaviour <.001
Absent 48 (14–135)
Present 65 (18–192) .038
Unknown 24 (10–68) <.001
Cognition <.001
Normal 50 (15–138)
Impaired 45 (15–154) 0.91
Unresponsive 20 (10–60) <.001
Unknown 12 (4–50) .012
Pain intensity <.001
0–3 25 (8–80)
4–6 64 (16–169) <.001
7+ 68 (20–200) <.001
* Morphine equivalent daily dose; n.s. non-significant (p value >
0.05).
EUROPEANJOURNALOFCANCER46 (2010) 2896 2904 2901
Page 6
5. Conclusion
The ECS-CP is a simple, comprehensive categorical classifica-
tion system for meaningfully assessing cancer pain. While
many factors have been proposed as prognostic for pain con-
trol, the ECS-CP is the first pain classification system to simul-
taneously integrate these factors within a cohesive
framework. The items included in the ECS-CP represent only
initial efforts to define a standard core of variables, and addi-
tional items such as analgesic tolerance, genetic variations
and age would be worthy of further research. This interna-
tional validation study confirms its predictive validity, repro-
ducibility and generalisability in diverse palliative care
settings and advances the process of development towards
an internationally recognised pain classification system.
The ECS-CP and future modifications could play a significant
role in routine clinical assessment and management of can-
cer pain, as well as providing a standard for describing the pa-
tient population in reporting research results.
6. Conflict of interest statement
None declared.
Acknowledgements
We thank colleagues who assisted with patient assessments
and/or data collection: Carla Stiles, Audra Arlain, Laureen
Johnson, Donna deMoissac, Lorelei Sawchuk, Viki Muller,
Hue Quan, Pablo Amigo, Doreen Oneschuk, Bei Pei, Gayle
Jones, Tonya Edwards, B.J. Clayton, Jenny Thurston, Vina Ngu-
yen, Larry Hasson, Kate McLoughlin.
Robin Fainsinger, Neil Hagen and Cheryl Nekolaichuk are
supported by the Canadian Institutes of Health Research
through grant support for the CIHR New Emerging Team in
Difficult Pain Problems (CIHR PET69772) and an international
multicentre validation study of a pain classification system
for advanced cancer patients (Grant No. 79367).
Appendix A
Edmonton Classification System for Cancer Pain
For each of the following features, circle the response that
is most appropriate, based on your clinical assessment of the
patient.
1. Mechanism of Pain
No No pain syndrome
Nc Any nociceptive combination of visceral and/or
bone or soft tissue pain
Ne Neuropathic pain syndrome with or without any
combination of nociceptive pain
Nx Insufficient information to classify
2. Incident Pain
Io No incident pain
Ii Incident pain present
Ix Insufficient information to classify
3. Psychological Distress
Po No psychological distress
Pp Psychological distress present
Px Insufficient information to classify
4. Addictive Behaviour
Ao No addictive behaviour
Aa Addictive behaviour present
Ax Insufficient information to classify
5. Cognitive function
Co No impairment. Patient able to provide accurate
present and past pain history unimpaired
Ci Partial impairment. Sufficient impairment to affect
patient’s ability to provide accurate present and/or past
pain history
Cu Total impairment. Patient unresponsive, delirious
or demented to the stage of being unable to provide
any present and past pain history
Cx Insufficient information to classify.
ECS-CP profile: (combination of the five circled responses,
one for each category)
Assessed by: __________________ Date:
______________________
Appendix B
Definitions of Terms
Incident Pain
Pain can be defined as incident pain when a patient has
background pain of no more than moderate intensity with
intermittent episodes of moderate to severe pain, usually
having a rapid onset and often a known trigger.
Io No incident pain
Ii Incident pain present
Ix Insufficient information to classify
j
Guidelines for use
There are six key characteristics of incident pain, as defined
in the ECS:
Relationship with background pain: The intensity of incident
pain is significantly greater than background pain.
Severity: The intensity of incident pain is moderate to
severe.
Predictability: The trigger is often known such as move-
ment, defecation, urination, swallowing and dressing
change. However, clinically significant episodic pain (i.e.
no predictable trigger) can be included (e.g. bladder or
bowel spasm).
Onset: Its onset is rapid, with intensity often peaking
within 5 minutes.
Transiency: Incident pain is transient, and may return to
baseline shortly after the trigger is stopped or removed.
j
(Insufficient information to classify due to factors such as
questionable/unknown diagnosis, patient’s unwillingness to par-
ticipate or physical impairments (e.g. aphasia)).
2902 EUROPEANJOURNALOFCANCER46 (2010) 2896 2904
Page 7
Recurrence: It is intermittent, recurring when the trigger is
reinitiated or reapplied.
Psychological Distress
Po No psychological distress present
Pp Psychological distress present
Px Insufficient information to classify
j
Psychological distress, within the context of the pain expe-
rience, is defined as a patient’s inner state of suffering result-
ing from physical, psychological, social, spiritual and/or
practical factors that may compromise the patient’s coping
ability and complicate the expression of pain and/or other
symptoms.
Guidelines for use
There are five key characteristics of psychological distress,as
defined in the ECS:
Relationship with pain: The definition of psychological dis-
tress is limited to patients who are experiencing psycho-
logical distress within the context of the pain experience
and who appear to express their suffering through physi-
cal symptoms.
Relationship with suffering: It is an expression of suffering,
often referred to as total pain.
Multidimensional: It is multidimensional in nature, influ-
encing many spheres of a patient’s experience, including
but not necessarily limited to physical, psychological,
social and spiritual factors.
Relationship with coping: It may impair a patient’s ability to
cope with his/her illness.
Physical symptom expression: It is often expressed as an
exacerbation of pain and/or other symptoms, which may
be conceptualised as a form of somatisation.
Assessment
Assessment of psychological distress may include, but is
not necessarily limited to, the following:
Assessment of patient’s experience in multidimensional
domains
Patient’s behavioural presentation and symptom reporting
profile
Collateral history from primary caregivers
Addictive behaviour
Ao Addictive behaviour not present
Aa Addictive behaviour present
Ax Insufficient information to classify
j
Addiction is a primary, chronic, neurobiologic disease,
with genetic, psychosocial, and environmental factors influ-
encing its development and manifestations. It is character-
ised by behaviours that include one or more of the
following: impaired control over drug use, compulsive use,
continued use despite harm and craving.
Guidelines for use
There are five key characteristics of addictive behaviour,as
defined in the ECS:
chronicity: It is a chronic disorder, which may have periods
of relapse and remission.
Multidimensional: It is multidimensional in its development
and expression, including genetic, psychosocial and envi-
ronmental factors.
Compulsivity
persistent use despite harm
craving
This definition is limited to the following:
A remote history of prior alcohol/substance use may not
be considered relevant as a complicating factor in ongoing
pain assessment and management.
Substances of abuse include alcohol, prescription/non pre-
scription medications, and illicit drugs.
It does not include chronic tobacco use.
Assessment
Assessment of addictive behaviour may include, but is not
necessarily limited to, the following:
Use of CAGE as screening tool for possible alcohol abuse
Patient’s behavioural presentation over a series of visits
A strong clinical history of substance abuse provided by
patient
Collateral history from primary caregivers
Reference: Nekolaichuk C, Fainsinger R, Lawlor P. A valida-
tion study of a pain classification system for advanced cancer
patients using content experts: The Edmonton classification
system for cancer pain. Palliat Med 2005; 19(6):466–476.
REFERENCES
1. Portenoy RK, Lesage P. Management of cancer pain. Lancet
1999;353:1695–700.
2. Ventafridda V, Tamburini M, Caraceni A, et al. A validation
study of the WHO method for cancer pain relief. Cancer
1987;59:850–6.
3. Zech DF, Grond S, Lynch J, et al. Validation of World Health
Organization guidelines for cancer pain relief: a 10-year
prospective study. Pain 1995;63:65–76.
4. Fainsinger RL, Nekolaichuk CL. Cancer pain assessment can
we predict the need for specialist input? Eur J Cancer
2008;44(8):1072–7.
5. Boisvert M, Cohen SR. Opioid use in advanced malignant
disease: Why do different centres use vastly different doses?
A plea for standardized reporting. J Pain Symptom Manage
1995;10:632–8.
6. Knudsen AK, Aass N, Fainsinger R, et al. Classification of pain
in cancer patients a systematic review. Palliat Med
2009;23:295–308.
7. Hjermstad MJ, Kaasa S, Fainsinger RL. Assessment and
classification of cancer pain. Curr Opin Support Palliat Care
2009;3:24–30.
EUROPEANJOURNALOFCANCER46 (2010) 2896 2904 2903
Page 8
8. Bruera E, Macmillan K, Hanson J, et al. The Edmonton staging
system for cancer pain: preliminary report. Pain
1989;37:203–9.
9. Bruera E, Schoeller T, Wenk R, et al. A prospective multicenter
assessment of the Edmonton staging system for cancer pain. J
Pain Symptom Manage 1995;10:348–55.
10. Bruera E, Fainsinger R, Spachynski, et al. Clinical efficacy and
safety of a novel controlled-release morphine suppository
and subcutaneous morphine in cancer pain: a randomized
evaluation. J Clin Oncol 1995;13:1520–7.
11. Bruera E, Sloan P, Mount B, et al. A randomized, double-blind,
double-dummy, crossover trial comparing the safety and
efficacy of oral sustained-release hydromorphone with
immediate-release hydromorphone in patients with cancer
pain. Canadian Palliative Care Clinical Trials Group. J Clin
Oncol 1996;14:1713–7.
12. Bruera E, Watanabe S, Fainsinger RL, et al. Custom-made
capsules and suppositories of methadone for patients on
high-dose opioids for cancer pain. Pain 1995;62:141–6.
13. De Stoutz ND, Bruera E, Suarez-Almazor M. Opioid rotation
for toxicity reduction in terminal cancer patients. J Pain
Symptom Manage 1995;10:378–84.
14. Ernst DS, Brasher P, Hagen N, et al. A randomized, controlled
trial of intravenous clodronate in patients with metastatic
bone disease and pain. J Pain Symptom Manage 1997;13:319–26.
15. Fainsinger RL, Toro R. Opioids, confusion and opioid rotation.
Palliat Med 1998;12:463–4.
16. Obiols M, Lossignol D. A staging system for cancer pain.
Support Care Cancer 1995;5(3):357 [abstract].
17. Fainsinger RL, Nekolaichuk CL, Lawlor PG, et al. A
multicenter study of the Revised Edmonton Staging System
for classifying cancer pain in advanced cancer patients. J Pain
Symptom Manage 2005;29:224–37.
18. Fainsinger RL, Fairchild A, Nekolaichuk C, et al. Is pain
intensity a predictor of the complexity of cancer pain
management? J Clin Oncol 2009;27(4):585–90.
19. Lowe SS, Nekolaichuk CL, Fainsinger RL, et al. Should the rate
of opioid dose escalation be included as a feature in a cancer
pain classification system? J Pain Symptom Manage
2008;35(1):51–7.
20. Nekolaichuk C, Fainsinger R, Lawlor P. A validation study of a
pain classification system for advanced cancer patients using
content experts: the Edmonton classification system for
cancer pain. Palliat Med 2005;19(6):466–76.
21. Fainsinger RL, Nekolaichuk CL, Lawlor PG, et al. Edmonton
Classification System for Cancer Pain (ECS-CP).
<www.palliative.org/PC/ClinicalInfo/AssessmentTools/ECS-
CP%20Manual.pdf>.
22. National Cancer Institute. Pain (PDQ
â
). <http://
www.cancer.gov/cancertopics/pdq/supportivecare/pain/
HealthProfessional>.
23. Quan KH, Vigano A, Fainsinger RL. Evaluation of a data
collection tool (TELEformÓ) for palliative care research. J
Palliat Med 2003;6:401–8.
24. Parmar MKB, Machin D. Survival analysis: a practical
approach. New York, NY: John Wiley and Sons; 1995.
25. Hagen NA, Klepstad P, Hjermstad MJ, et al. Lofoten seminar:
the pain sessions. Palliat Med 2008;22:891–4.
2904 EUROPEANJOURNALOFCANCER46 (2010) 2896 2904
Page 9
  • Source
    • "Neuropathic pain (NP) is a core domain in the classification of cancer patients with pain [8] [15] [22] [23]. It is associated with a worse response to conventional analgesic treatment [12,22–24] and can be relieved by specific adjuvant drugs enhancing the efficacy of opioid analgesia [5] [9]. "
    [Show abstract] [Hide abstract] ABSTRACT: Neuropathic pain (NP) in cancer patients lacks standards for diagnosis. This study is aimed at reaching consensus on the application of the NeuPSIG criteria to the diagnosis of NP in cancer patients and on the relevance of patient reported outcome (PRO) descriptors for the screening of NP in this population. An international group of 42 experts was invited to participate in a consensus process through a modified two-round internet-based Delphi survey. Relevant topics investigated were: peculiarities of NP in patients with cancer, IASP NeuPSIG diagnostic criteria adaptation and assessment, standardized PRO assessment for NP screening. Median consensus scores (MED) and inter-quartile ranges (IQR), were calculated to measure expert consensus after both rounds. 29 experts answered and good agreement was found on the statement "the pathophysiology of NP due to cancer can be different from non-cancer NP" (MED=9, IQR=2). Satisfactory consensus was reached for the first three NeuPSIG criteria (pain distribution, history and sensory findings) (MEDs>=8, IQRs<=3), but not for the fourth one (diagnostic test/imaging) (MED=6, IQR=3). Agreement was also reached on clinical examination by soft brush or pin stimulation (MEDs>=7 and IQRs<=3) and on the use of PRO descriptors for NP screening (MED=8, IQR=3). Based on the study results a clinical algorithm for NP diagnostic criteria in cancer patients with pain was proposed. Clinical research on PRO in the screening phase and on the application of the algorithm will be needed to examine their effectiveness in classifying NP in cancer patients.
    Full-text · Article · Oct 2014 · Pain
  • Source
    • "The original version categorised cancer pain as ''nociceptive ,'' ''neuropathic,'' ''mixed,'' and ''unclassified'' [11]. The updated version has reduced the classification to a dichotomous yes/no response to denote the presence or absence of NP based on the clinician opinion [10]. There is clearly uncertainty surrounding the classification and diagnosis of NP in cancer patients. "
    Full-text · Article · Nov 2013 · Pain
  • Source
    • "In another 10 studies, the primary aim was to evaluate the aetiology of pain in cancer patients [1,9,11,17,25,30,31,33–35]. Assessment of verbal pain description or the performance of screening tools to discriminate between types of pain was the primary aim in another 5 studies [13] [26] [27] [29] [36]. Finally, 1 paper was a controlled , single-dose study that assessed response to opioids on the basis of pain type [10] "
    [Show abstract] [Hide abstract] ABSTRACT: Pain in cancer patients remains common and is often associated with insufficient prescribing of targeted analgesia. An explanation for undertreatment could be the failure to identify neuropathic pain mechanisms, which require additional prescribing strategies. We wanted to identify the prevalence of neuropathic mechanisms in patients with cancer pain to highlight the need for detailed assessment and to support the development of an international classification system for cancer pain. We searched for studies that included adult and teenage patients (age above 12 years), with active cancer and who reported pain, and in which a clinical assessment of their pain had been made. We found 22 eligible studies that reported on 13,683 patients. Clinical assessment methods varied, and only 14 studies reported confirmatory testing for either sensory abnormality or diagnostic lesion to corroborate a diagnosis of neuropathic pain. We calculated that the prevalence of patients with neuropathic pain (95% confidence interval) varied from a conservative estimate of 19% (9.4% to 28.4%) to a liberal estimate of 39.1% (28.9% to 49.5%) when patients with mixed pain were included. The prevalence of pain with a neuropathic mechanism (95% confidence interval) ranged from a conservative estimate of 18.7% (15.3% to 22.1%) to a liberal estimate of 21.4% (15.2% to 27.6%) of all recorded cancer pains. The proportion of pain caused by cancer treatment was higher in neuropathic pain compared with all types of cancer pain. A standardised approach or taxonomy used for assessing neuropathic pain in patients with cancer is needed to improve treatment outcomes.
    Full-text · Article · Nov 2011 · Pain
Show more