Pain 81 (1999) 129–134

Breakthrough pain: characteristics and impact in patients with cancer pain

Russell K. Portenoy a ,*, David Payne b, Paul Jacobsen c

a
Department of Pain Medicine and Palliative Care, Beth Israel Medical Center, First Avenue at 16th Street, New York, NY, 10003, USA
b
Department of Psychiatry, Memorial Sloan–Kettering Cancer Center, 1275 York Avenue, New York, NY, 10021, USA
c
Department of Psychosocial Oncology, H. Lee Moffitt Cancer Center, 12902 Magnolia Drive, Tampa, FL, 33612-9497, USA

Received 9 July 1998; received in revised form 2 December 1998; accepted 24 December 1998

Abstract

Few surveys have been performed to define the characteristics and impact of breakthrough pain in the cancer population. In this cross-
sectional survey of inpatients with cancer, patients responded to a structured interview (the Breakthrough Pain Questionnaire) designed to
characterize breakthrough pain, and also completed measures of pain and mood (Memorial Pain Assessment Card (MPAC)), pain-related
interference in function (Brief Pain Inventory (BPI)), depressed mood (Beck Depression Inventory (BDI)), and anxiety (Beck Anxiety
Inventory (BAI)). Of 178 eligible patients, 164 (92.2%) met the criteria for controlled background pain. The median age was 50.6 years
(range 26 to 77 years), 52% were men, and 80.6% were Caucasian. Tumor diagnoses were mixed, 75% had metastatic disease, 65% had
pain caused directly by the neoplasm, and a majority had mixed nociceptive-neuropathic pain. The median Karnofsky Performance Status
score was 60 (range 40 to 90). Eighty-four (51.2%) patients had experienced breakthrough pain during the previous day. The median
number of episodes was six (range 1 to 60) and the median interval from onset to peak was 3 min (range 1 s to 30 min). Although almost
two-thirds (61.7%) could identify precipitants (movement 20.4%; end-of-dose failure 13.2%), pain was unpredictable in a large majority
(78.2%). Patients with breakthrough pain had more intense (P , 0.001) and more frequent (P , 0.01) background pain than patients
without breakthrough pain. Breakthrough pain was also associated with greater pain-related functional impairment (difference in mean BPI,
P , 0.001), worse mood (mood VAS, P , 0.05; BDI, P , 0.001), and more anxiety (BAI, P , 0.001). Multivariate analysis confirmed
that breakthrough pain independently contributed to impaired functioning and psychological distress. These data confirm that cancer-related
breakthrough pain is a prevalent and heterogeneous phenomenon. The presence of breakthrough pain is a marker of a generally more severe
pain syndrome, and is associated with both pain-related functional impairment and psychological distress. The findings suggest the need for
further studies of breakthrough pain and more effective therapeutic strategies.  1999 International Association for the Study of Pain.
Published by Elsevier Science B.V.

Keywords: Breakthrough pain; Incident pain; Functional impairment; Psychological distress; Cancer pain; Opioid therapy

1. Introduction some other voluntary action of the patient.

Patients with chronic cancer-related pain usually experi-
ence fluctuations in pain intensity (Portenoy, 1997). Often,
these fluctuations occur as discrete transitory flares of pain.
When these flares interrupt a tolerable background pain,
they are commonly described as ‘breakthrough pains’ or
‘incident pains.’ In the cancer setting, breakthrough pain
(BP) typically interrupts a background pain that is generally
well controlled by opioid therapy. Incident pain is usually
understood to be a subtype of BP induced by movement or
* Corresponding author. Tel.: +1-212-844-1505; fax: +1-212-844-1503;
e-mail rportenoy@bethisrael.org
The few existing studies of BP indicate that it is a pre-
valent phenomenon associated with adverse outcomes. A
survey of 90 inpatients with cancer pain observed that
64% reported breakthrough pain (Portenoy and Hagen,
1990) and a survey of 200 ambulatory cancer patients
noted that 90% experienced pain with movement (Banning
et al., 1991). Two other prospective surveys demonstrated
that the presence of BP reduces the likelihood of a satisfac-
tory response to opioid therapy (Mercadante et al., 1992;
Bruera et al., 1995).
Despite the clinical importance of BP, empirical informa-
tion about the phenomenon is very limited. In part, the lack
of data may relate to the difficulties inherent in defining and

0304-3959/99/$20.00  1999 International Association for the Study of Pain. Published by Elsevier Science B.V.
PII: S03 04-3959(99)000 06-8
130 R.K. Portenoy et al. / Pain 81 (1999) 129–134
measuring a subtype of pain in the medically ill. Break-
through pains are heterogeneous and must be distinguished
from recurrent acute pains and other clinically insignificant
fluctuations in the intensity of chronic pain. Efforts to
improve the clinical management of BP will require a
clear definition of the phenomenon and valid data that illu-
minate its characteristics and impact.
To operationalize the definition of BP and acquire valid
epidemiologic data, the aforementioned survey of cancer
inpatients applied an assessment algorithm that required
patients to distinguish different types of pain according to
intensity and temporal characteristics (Portenoy and Hagen,
1990). This algorithm, which identifies a subgroup of
patients that can provide detailed information about BP,
has been applied to another sample of inpatients with cancer
pain. The information obtained from these patients provides
further detail about the characteristics and impact of BP.
2. Methods
The survey was approved by the Institutional Review
Board of Memorial Sloan–Kettering Cancer Center. All
patients gave informed consent prior to participation in
the study.
2.1. Procedure
Data were acquired through a cross-sectional survey of
cancer patients admitted to Memorial Sloan-Kettering Can-
cer Center. On each day of the survey, one of the inpatient
medical units of the hospital was randomly selected using a
random numbers table. All patients recruited for the survey
on that day were situated on the selected unit. To determine
initial eligibility for the study, an investigator reviewed the
nursing medication record for every patient on the unit.
Those patients who had a cancer diagnosis and were receiv-
ing a scheduled opioid regimen were considered eligible. A
list was made of the eligible patients and the names on this
list were randomly ordered. Patients were approached in this
order and asked to participate in the study. Patients were
excluded if not fluent in English or if the first contact with
the investigator revealed evidence of an organic brain syn-
drome or psychiatric disorder severe enough, in the judg-
ment of the investigator, to compromise data collection.
Patients who consented to the survey were first adminis-
tered a Breakthrough Pain Questionnaire in interview for-
mat. This interview also provided the information necessary
to determine usual performance status. Following the inter-
view, a questionnaire packet was administered. Additional
information about pain syndrome, medical status and
analgesic therapy was subsequently gleaned from the
patients’ medical records. For the purposes of data analysis,
opioid doses were converted into intramuscular morphine
equivalent milligrams using a standard relative potency
table (Agency for Health Care Policy and Research, 1994).
2.2. Measures
Data were collected using a combination of interview and
patient-rated instruments.
2.2.1. Breakthrough Pain Questionnaire
The Breakthrough Pain Questionnaire was administered
as a structured interview that comprised specific questions
designed to identify patients with breakthrough pain (the
assessment algorithm) and other inquiries about the phe-
nomenon when present.
The assessment algorithm categorized patients into three
groups; (1) those with uncontrolled background pain; (2)
those with controlled background pain and no BP, and (3)
those with controlled background pain and BP. Patients
were considered to have background pain if pain was
reported as experienced for more than half the waking
days during the previous week, or if a fixed schedule opioid
regimen had been used on more than half the days during
the previous week. Patients who did not meet these criteria
were considered to have transitory pains and were not
assessed further.
Patients with background pain were asked to describe the
usual intensity of this pain, which was defined as the inten-
sity experienced for more than half the time pain was
experienced. Patients were permitted to designate this
pain as usually absent if the analgesic regimen was highly
effective. Alternatively, the intensity of the background
pain could be described as usually mild, usually moderate,
or usually severe. Patients with background pain that was
described as usually severe were considered to have uncon-
trolled background pain, whereas patients who described
the pain as usually mild or moderate, or who described it
as usually absent, were considered to have controlled back-
ground pain.
Patients who met criteria for controlled background pain
were then asked whether they had experienced one or more
severe or excruciating episodes of pain during the past day.
If a patient responded affirmatively, he or she was classified
as having controlled background pain and BP. If none of
these transitory pains occurred, the patient was labeled as
having controlled background pain without BP.
Patients with BP described the characteristics of these
pains in response to a series of additional questions.
Although those with more than one type of BP were per-
mitted to describe each type separately, almost all of the
patients had only one distinct type of pain; the small number
of patients with more than one type had only the most pro-
minent type included in the analysis. The number of BP’s
during the previous day was recorded and patients were
asked to describe the typical pain in terms of the time
between onset and peak intensity; its location, quality and
predictability; and the factors that precipitated it or provided
relief once it occurred. BP duration was not queried since
the availability of analgesic therapy would complicate inter-
pretation of this recollection. The use of supplemental
 R.K. Portenoy et al. / Pain 81s (1999) 129–134
analgesic medication and the typical relationship between
the occurrence of BP and the timing of analgesic therapy
were specifically queried.
Using information obtained from the patient and medical
records, the structural lesion etiologically related to the BP
was identified and a pathophysiology for the pain was
inferred. The inferred pathophysiology was categorized as
nociceptive if the pain was related to injured somatic or
visceral structures and was described as usually aching,
sharp, or throbbing. The pathophysiology was categorized
as neuropathic if the pain was related to an injured neural
structure, conformed to a known neuropathic pattern (e.g.
radicular pain), and/or was described as having aberrant
characteristics (dysesthetic pain). Pains that had both
types of characteristics were labeled as mixed.
Similar questions were used to elicit information about
the background pain. An additional question assessed the
frequency of the background pain on a five-point catego-
rical scale, and a final question determined whether the
patient perceived that the BP represented an acute worsen-
ing of the more chronic pain or was an entirely distinct
phenomenon.
2.2.2. Memorial Pain Assessment Card
The Memorial Pain Assessment Card (MPAC) measures
pain and global mood using three 100 mm visual analogue
scales (VAS) – pain intensity, pain relief and mood, respec-
tively – and an eight-item verbal rating scale. The instru-
ment has been validated in the cancer population (Fishman
et al., 1987). The score on the mood VAS reflects global
psychological distress. For this study, patients were asked to
complete the mood VAS and the pain intensity VAS for the
background pain ‘on average,’ and the time frame for both
scores was focused on the ‘past day.’
2.2.3. Brief Pain Inventory
A seven-item subscale of the Brief Pain Inventory (BPI)
was used to evaluate the degree to which pain interfered
with function and quality of life. This subscale, which has
been extensively validated in the cancer population (Daut et
al., 1983; Serlin et al., 1995), measures pain interference
with activity, mood, social relations, walking, work, sleep
and overall enjoyment of life. Each item is assessed on a 0–
10 numeric scale (0, does not interfere; 10, completely inter-
feres). The sum of these scores provides a global measure of
pain-related interference (Portenoy et al., 1992).
2.2.4. Beck Depression Inventory
The Beck Depression Inventory (BDI) is a 21-item self-
report instrument designed to measure the symptoms asso-
ciated with depression (Beck et al., 1961). The instrument is
reliable and has two main factors; a cognitive factor, which
accounts for most of the variance, and a vegetative factor.
2.2.5. Beck Anxiety Inventory
The Beck Anxiety Inventory (BAI) is a 21-item self-
131
report instrument designed to measure the symptoms asso-
ciated with anxiety (Beck et al., 1988). The instrument is
highly reliable and encompasses two main factors: somatic
symptoms and subjective anxiety or panic symptoms.
2.2.6. Karnofsky Performance Status Score
The Karnofsky Performance Status (KPS) score is a
widely used observer-rated measure of performance status.
The score indicates the ability to function physically (Yates
et al., 1980).
2.3. Data analysis
Demographic data, tumor-related information, and the
characteristics of both the background pain and BP were
tabulated. Univariate comparisons between patients with
and without BP were evaluated using Chi-square for cate-
gorical variables and t-tests for continuous variables. To
assess the degree to which the occurrence of BP contributed
independently to pain-related interference with function and
quality of life, or to affective disturbances, multivariate
models were created using the BPI score, the BDI score
and the BAI score as the dependent variable, respectively.
Each of these regression models evaluated, as independent
variables, the presence versus absence of BP, demographic
variables, usual intensity of the background pain, number of
supplemental analgesic doses and the performance status.
3. Results
There were 178 patients who had evidence of chronic
pain, were otherwise eligible for study, and agreed to parti-
cipate. Fourteen (7.8%) of these patients had uncontrolled
background pain (intensity worse than moderate for more
than half the time experienced) and were not assessed
further. The remaining 164 patients (92.2%) had controlled
background pain (Table 1). The median age of these patients
was 50.6 years (range, 26 to 77 years). Fifty-two percent
were men and 80.6% were Caucasian. Tumor diagnoses
were mixed and 75% of the group had metastatic disease.
The median KPS score was 60 (range, 40 to 90).
Of the 164 patients with controlled background pain, 84
(51.2%) had experienced one or more BP’s during the pre-
ceding day. Almost all patients with these transient episodes
of severe or excruciating pain reported that they experi-
enced only one distinct type.
3.1. Pain characteristics
Sixty-five percent (N = 107) of the patients with con-
trolled background pain had pain caused directly by the
neoplasm. The remainder had pain related to antineoplastic
treatment. The pathophysiology of the background pain had
both nociceptive and neuropathic features in 52% of the
patients; the pain was labeled nociceptive in 38% and neu-
132 R.K. Portenoy et al. / Pain 81 (1999) 129–134

Table 1 vary significantly between those with and without break-

Sample characteristics (N = 164) through pain (mean (SD) morphine equivalent milligrams
for those with and without BP, 58.0 (86.7) and 53.6 (99.7),
Age
Median 53 years
respectively; P = ns). As would be expected, patients with
Range 26–77 BP did use more supplemental opioid doses than those with-
out breakthrough pain (mean (SD), 5.1 (5.0) vs. 2.3 (2.4);
Sex P , 0.001).
Men 87 (52%) Although the use of an analgesic drug was the typical
Women 79 (48%)
therapy for BP, many other palliative approaches were
Tumor type also used (Table 2). Overall, the mean (SD) effectiveness
Leuk/Lym 35 (21%) of all approaches to manage BP was 6.9 (1.9) on a 0–10
Breast 29 (17%) scale (0, not effective; 10, very effective).
GI 26 (16%)
Lung 24 (14%)
GU 22 (13%)
3.2. Impact of breakthrough pain
Sarcoma 9 (5%)
Other 21 (13%) On the pain interference scale of the BPI, patients with
BP had scores indicative of greater functional impairment
Race on all items than patients with controlled background pain
Caucasian 144 (81%)
African American 23 (13%)
and no breakthrough episodes (Table 3). The mean (SD)
Hispanic 7 (4%) total score was 24.8 (6.3) in the group with BP and 16.7
Asian 3 (2%) (6.7) in the group without BP (P , 0.001).
Other 1 (1%) Similarly, BP was associated with relatively worse scores
on the measures of affective disturbance. On the mood
Extent of disease
Remission 14 (8%)
VAS, mean (SD) score was 64.4 (24.9) for those without
Local Extension 32 (17%) and 56.2 (25.6) for those with BP (P , 0.05). On the BDI,
Metastatic 134 (75%) the mean (SD) scores for patients with and without BP were
18.2 (9.5) and 12.8 (6.3), respectively (P , 0.001), and the
Karnofsky Performance Scale mean (SD) scores on the BAI for those with and without BP
Median 60 (range 40–90)
were 17.9 (11.1) and 9.9 (7.9), respectively (P , 0.001).
GI, gastrointestinal; GU, genitourinary.
Table 2
Characteristics of Breakthrough Pain
ropathic in 10%. The mean (SD) VAS score for background Number of types per patient Number per 24 h
pain intensity was 28.4 (26.1) mm and the mean (SD) opioid
One 83.1% Median 6
consumption during the 1-day period was 54.0 (95.0) mor- Two 14.5% Mean (SD) 5.8 (8.7)
phine equivalent milligrams. Three 2.4% Range 1–60
The characteristics of the BP’s were diverse (Table 2).
The median number during the 1-day period was six (range Duration till peak Predictability
1 to 60). The patients recalled that the usual interval from Mean (SD) min 3.2 (19.5) Never 48.2%
onset to peak pain was brief (median, 3 min; mean (SD), 3.2 Range 1 s to 30 min Sometimes 18.8%
(19.5) min; range, 1 s to 30 min). Although 61.7% of the Often 7.1%
patients could identify precipitants (Table 2), almost half Almost always 10.6%
Always 15.3%
(48.2%) also stated that BP was never predictable. About
one-fifth of the patients noted that movement could preci- Precipitants Palliatives
pitate a BP and 13.2% reported that BP could be related to
Movement 20.4% Medication 61.0%
the end of a dosing interval for an analgesic drug (end-of- End of dose 13.2% Othersb 26.0%
dose failure). Uniformly, BP represented a flare at a site of Othersa 28.1% Unsure 13.0%
chronic pain and was recognized as being directly related to Unsure 38.3%
the background pain.
Patients with BP experienced more intense background Effectiveness of palliatives (0–10 scale)
pain than patients without BP (mean (SD), 34.2 (24.1) vs. Mean 6.9
16.7 (18.1); P , 0.001). On the scale of background pain Range 3–10
frequency, 81.4% of patients with BP noted that pain was a
Walking, 7.4%; defecation, 5.7%; urination, 3.8%; sit, 3.7%; cough,
experienced ‘frequently’ or ‘constantly,’ compared with 3.7%; breathe, 1.9%; eat, 1.9%.
56.2% of those with background pain alone (P , 0.01). b
Sit/lie, 11.8%; move/turn, 7.9%; touch, 2.6%; breathe, 1.3%; defecation,
In contrast to these variables, opioid consumption did not 1.3%.
 R.K. Portenoy et al. / Pain 81s (1999) 129–134 133

Table 3 1990). Most patients experienced multiple pains during the

day (median, six pains; range, 1 to 60) and noted that the
Impact of breakthrough pain
onset of pain was rapid each time it occurred. Precipitating
No BP (N = 80) BP (N = 84) P-value factors were often identified, but pains occurred without
mean (SD) mean (SD) warning about half the time.
Brief Pain Inventory In general, the occurrence of BP was a marker of a more
Activity 2.7 (1.4) 4.0 (1.1) ,0.001 severe pain syndrome. Background pain was both more
Mood 2.5 (1.2) 3.6 (1.2) ,0.001 intense and more frequent among those with BP. This rela-
Walk 2.3 (1.4) 3.5 (1.4) ,0.001
Work 2.5 (1.3) 3.9 (1.1) ,0.001
tively greater pain intensity was not matched by higher
Social relations 1.9 (1.0) 2.9 (1.4) ,0.001 doses of a regularly scheduled opioid drug. Patients with
Sleep 2.2 (1.1) 3.2 (1.2) ,0.001 BP consumed a larger proportion of their opioid regimen
Enjoy life 2.7 (1.4) 3.7 (1.2) ,0.001 in the form of supplemental ‘as needed’ doses.
Total score 16.7 (6.7) 24.8 (6.3) ,0.001 BP was also associated with relatively more severe pain-
Mood VAS 75.1 (25.4) 60.2 (28.4) ,0.001
related functional impairment and psychological distress.
Beck depression 12.8 (6.3) 18.2 (9.4) ,0.001 The multivariate analyses suggest that this association is
Beck anxiety 9.9 (7.9) 17.9 (11.1) ,0.001 not explained by the more intense background pain experi-
enced by those with BP, or by greater physical debilitation
in this group.
These scores reflect significantly higher levels of depression The clinical relevance of BP in the management of cancer
and anxiety in the group with BP. pain is suggested by the widespread use of supplemental ‘as
Separate multivariate models were constructed to evalu- needed’ opioid administration during long-term therapy
ate the degree to which BP independently contributed to with a fixed schedule opioid regimen. This strategy, which
pain-related functional interference (BPI), depression is sometimes called ‘rescue’ dosing, was prescribed to most
(BDI), and anxiety (BAI). Demographic variables, usual of the patients in this survey. Although the higher levels of
intensity of the background pain, number of supplemental pain, functional impairment and distress experienced by the
doses of analgesic drugs, and performance status were also patients with BP in this survey suggest deficiencies in this
entered in each of these models. The presence of BP con- approach, the data cannot distinguish problems with the
tributed significantly to pain-related interference (P , nature of the treatment from the more general problem of
0.001), accounting for 13% of the shared variance. Simi- undertreatment. The finding that patients with BP had more
larly, the presence of BP contributed significantly to depres- background pain but did not receive a higher scheduled
sion (P , 0.01) and anxiety (P , 0.01), and accounted for opioid dose suggests the possibility of treatment inadequa-
5 and 6% of the shared variances in these variables, respec- cies, which, if present, would complicate efforts to assess
tively. the value of optimally administered conventional treatment
approaches for BP.
It must be acknowledged, however, that optimally admi-
4. Discussion nistered conventional treatment for breakthrough pain, spe-
cifically the oral ‘rescue’ dose, is likely to be inadequate for
This survey of inpatients with cancer pain applied an a substantial proportion of patients. The onset time of break-
assessment algorithm designed to identify a subgroup able through pain for most patients is measured in minutes. The
to distinguish the occurrence of transitory severe pains from time required to reach peak concentration after oral admin-
relatively controlled background pain. A previous survey istration of a supplemental analgesic dose is far longer. For
used the same methodology (Portenoy and Hagen, 1990) many patients, particularly those whose pains are both
and suggested that such a systematic approach could effec- unpredictable and rapid in onset, the utility of conventional
tively yield specific information about the phenomenon of therapy may be limited.
BP. The present survey evaluated a larger sample, confirms Studies are needed to clarify the true effectiveness of the
the high prevalence and heterogeneity of BP in this popula- conventional therapeutic approaches to BP. These studies
tion, and provides new information about the impact of should address titration of the baseline opioid regimen and
these pains. the supplemental ‘as needed’ drug. It may be valuable to
Slightly more than half of the patients with controlled compare escalation of the baseline dose with optimal use of
background pain (51.2%) experienced BP during the day a rescue dose as alternative interventions for breakthrough
prior to the interview. Although this prevalence is less pain. Although the dose of the medication used for BP tra-
than the previous survey, which observed the phenomenon ditionally has been selected as a fixed proportion of the total
in almost two-thirds, it nonetheless confirms the clinical daily opioid consumption (Portenoy, 1997), recent con-
impression that BP is common in the clinical setting. trolled trials of oral transmucosal fentanyl citrate, a new
The characteristics of BP varied, mirroring the heteroge- opioid formulation currently undergoing investigation as a
neity observed in the previous survey (Portenoy and Hagen, treatment for BP, failed to confirm the necessity of this
134 R.K. Portenoy et al. / Pain 81 (1999) 129–134
approach and, instead, suggested the value of titration to
identify the optimal dose for a breakthrough pain medica-
tion (Portenoy et al., 1999). Additional studies that evaluate
the size and timing of the rescue dose are needed.
Similarly, further studies are needed to confirm the extent
to which the phenomenon of BP is associated with adverse
functional and psychological outcomes, and clarify the
direction of causation. Although the multivariate analyses
in this study suggest a link between BP and both functional
impairment and distress that cannot be explained by the
overall background pain or physical debilitation, they can-
not clarify whether episodic pain contributes directly to
these adverse outcomes. Intuitively, such a direct effect
may be suggested by the severity and unpredictability of
BP, and the rapidity with which it usually occurs. These
characteristics could compromise function and undermine
the patient’s sense of control. Longitudinal studies and treat-
ment trials that incorporate assessment of functional and
psychosocial variables, as well as pain, would be very useful
to illuminate these relationships.
Future studies can address some, but not all, of the poten-
tial limitations of the present survey. The assessment of BP
is difficult, given its heterogeneity and lack of predictability.
The assessment algorithm developed for this survey has not
been validated, and validity would be difficult to confirm
empirically, in the absence of a gold standard for the assess-
ment of transitory pains of this type in cancer patients.
Techniques to validate the approach, such as caregiver
observations and a prospective method of data collection
(which would reduce reliance on more problematic patient
recall), should be explored. Whatever method is used, the
generalizability of the data to an outpatient population must
be established.
Notwithstanding the limitations of this survey, the asso-
ciations among BP, pain-related interference in function,
and psychological distress underscore the need for aggres-
sive pain management and suggest the utility of a multi-
dimensional understanding of cancer pain. BP should be
recognized as a common problem, which is often associated
with other aversive experiences. Interventions should
attempt to provide comfort, support the patient’s sense of
control over the pain, and address the concurrent physical
and psychosocial burdens that contribute to impaired quality
of life.
References
Agency for Health Care Policy and Research, U.S. Dept. of Health and
Human Services: Clinical Practice Guideline Number 9: Management of
Cancer Pain. Washington, DC, 1994.
Banning, A., Sjogren, P. and Henriksen, H., Treatment outcome in a multi-
disciplinary cancer pain clinic, Pain, 47 (1991) 129–134.
Beck, A., Ward, C.H., Mendelson, M. et al., An inventory for measuring
depression, Arch. Gen. Psychiatry, 4 (1961) 561–671.
Beck, A., Epstein, N., Brown, G. and Steer, R., An inventory for measur-
ing anxiety: psychometric properties, J. Consult. Clin. Psychol., 56
(1988) 893–897.
Bruera, E., Schoeller, T., Wenk, R., MacEachern, T., Marcelino, S., Han-
son, J. and Suarez-Almazor, M., A prospective multi-center assessment
of the Edmonton staging system for cancer pain, J. Pain Symptom
Manage., 10 (1995) 348–355.
Daut, R.L., Cleeland, C.S. and Flanery, R.C., Development of the Wis-
consin Brief Pain Questionnaire to assess pain in cancer and other
diseases, Pain, 17 (2) (1983) 197–210.
Fishman, B., Pasternak, S., Wallenstein, S., Houde, R., Holland, J.C. and
Foley, K.M., The Memorial Pain Assessment Card: a valid instrument
for the assessment of cancer pain, Cancer, 60 (1987) 1151–1157.
Mercadante, S., Maddaloni, S., Roccella, S. and Salvaggio, L., Predictive
factors in advanced cancer pain treated only by analgesics, Pain, 50
(1992) 151–155.
Portenoy, R.K., Treatment of temporal variations in chronic cancer pain,
Semin. Oncol., 24 (5) (1997) S16–12.
Portenoy, R.K. and Hagen, N.A., Breakthrough pain: definition, preva-
lence and characteristics, Pain, 41 (1990) 273–281.
Portenoy, R.K., Miransky, J., Thaler, H.T., Hornung, J., Bianchi, C.,
Cibas-Kong, I., Feldhamer, E., Lewis, F., Matamores, I., Sugar, M.Z.,
Olivieri, A.P., Kemeny, N.E. and Foley, K.M., Pain in ambulatory
patients with lung or colon cancer: prevalence, characteristics and
impact, Cancer, 70 (1992) 1616–1624.
Portenoy, R.K., Payne, R., Coluzzi, P., Raschko, W., Lyss, A., Busch,
M.A., Frigerio, V., Ingham, J., Loseth, D.B., Nordbrook, E. and Rhiner,
M., Oral transmucosal fentanyl citrate (OTFC) for the treatment of
breakthrough pain in cancer patients: a controlled dose titration study,
Pain, 79 (1999) 303–312.
Serlin, R.C., Mendoza, T.R., Nakamura, Y., Edwards, K.R. and Cleeland,
C.S., When is cancer pain mild, moderate or severe? Grading pain
severity by its interference with function, Pain, 61 (2) (1995) 277–284.
Yates, J.W., Chalmer, B. and McKegney, F.P., Evaluation of patients with
advanced cancer using the Karnofsky Performance Status, Cancer, 40
(1980) 2220–2224.