554 Journal of Pain and Symptom Management Vol. 38 No. 4 October 2009

Original Article

Breakthrough Pain in Advanced CancerPatients Followed at Home:A Longitudinal StudySebastiano Mercadante, MD, Benedetta Veruska Costanzo, MD, Flavio Fusco, MD,Valeria Butta, MD, Valentina Vitrano, RS, and Alessandra Casuccio, BSPain Relief and Palliative Care Unit (S.M., V.V.), La Maddalena Cancer Center, Palermo; Palliative

Medicine (S.M.), Department of Anesthesia & Intensive Care, and Department of Clinical

Neuroscience (A.C.), University of Palermo, Palermo; Home Care Program (B.V.C., V.B.), SAMO,

Catania; and Home Care Program (F.F.), ASL-3, Genoa, Italy

Abstract

The aim of this study was to longitudinally assess breakthrough pain (BP) in advancedcancer patients who were admitted to home palliative care. One hundred and one consecutivepatients who were admitted to one of the two home care programs and were representative ofthe cancer population followed at home in Italy were included. Patients were excluded only ifat admission they were cognitively impaired or too unwell to provide reliable answers toquestions regarding data collection. At admission (T0), and one month later (T1), data wererecorded about the pharmacological treatment of background pain and its effectiveness; thepresence of BP and its intensity, duration, and number of episodes; the occurrence of BPinduced by movement; spontaneous relief after stopping activity and limitation of physicalactivity because of pain on movement; the analgesic drugs prescribed for BP; and theeffectiveness of BP medication. After the initial assessment, patients received analgesics forbackground pain and BP, as well as symptomatic treatment usually provided at home,according to local policies and protocols. At T0, 70.2% of patients were receiving analgesicdrugs, and 52% of them had uncontrolled pain. BP was reported by 49.2% of these patientsand had a mean duration and number of episodes per day of 35.1 minutes (�35.0) and 2.4(�1.5), respectively. Of patients with BP, 65.7% had pain on movement, which decreasedspontaneously after stopping activity in 73.9%. Physical activity was strongly limited in78.2% of these patients. Most patients did not have a BP medication prescription. At T1,a larger number of patients were receiving analgesics (P< 0.002), and the number ofpatients with BP medication significantly increased (P< 0.001). The incidence of BP andpain on movement decreased (33% and 38.5%, respectively). This longitudinal studysuggests that BP is a dynamic entity dependent on several factors, including analgesictreatment and the course of disease. This helps explain the different incidences reported inprevious studies performed in different settings and conditions. J Pain SymptomManage 2009;38:554e560. � 2009 U.S. Cancer Pain Relief Committee. Published byElsevier Inc. All rights reserved.

Address correspondence to: Sebastiano Mercadante,MD, Pain Relief and Palliative Care Unit, La Madda-lena Cancer Center, Via San Lorenzo 312, 90146

Palermo, Italy.E-mail: terapiadeldolore@la-maddalena.it

Accepted for publication: January 2, 2009.

� 2009 U.S. Cancer Pain Relief CommitteePublished by Elsevier Inc. All rights reserved.

0885-3924/09/$esee front matterdoi:10.1016/j.jpainsymman.2008.12.008

Vol. 38 No. 4 October 2009 555Breakthrough Pain in Home Care Cancer Patients

Key Words

Cancer pain, breakthrough pain, epidemiology

IntroductionIn the cancer population, breakthrough

pain (BP) is a transitory flare of pain superim-posed on an otherwise stable pain pattern inpatients treated with opioids.1 BP is still a diffi-cult entity to recognize. Several cross-sectionalstudies have reported a variable incidence,dependent on the setting, such as hospiceadmissions, oncology inpatients, pain clinicpatients, and patients followed at home, andthe type of evaluation.2 At admission to fourspecialist palliative care inpatient units in theUnited Kingdom, 93% of patients had BP. Ofthose with BP, movement-precipitated BPoccurred in 53%, and one-third reported BPas completely incapacitating.3 Similarly, 89%of patients admitted to a hospice reporteda mean of seven episodes of BP of differentintensities per day.4 In a home care setting,86% of patients with persistent pain reportedexperiencing BP, half of them in associationwith a specific activity.5 A lower prevalence ofBP was found in other studies. In a large surveyperformed on a predetermined day in differ-ent palliative care programs, including hos-pices and home care, in Catalonia, Spain, BPwas reported by 41% of patients, with a mean1.5 episodes per day.6 In patients admitted toinpatient units with a mean Karnofsky statusof 60, the incidence of BP was 51.2%.7 At ad-mission to a pain clinic, BP was reported by39% of patients,8 whereas oncology outpa-tients and inpatients admitted with uncon-trolled pain presented a global incidence of70%.9 Finally, in an international survey per-formed in a wide range of patients with painsevere enough to be treated with opioids,a BP incidence of about 65% was found, whichwas associated with poor pain control andfunctional impairment.10

Other than examining different settings andpopulations, all these surveys had a cross-sectional design. No study has assessed changesover time in reported BP, either in relation tothe course of disease or the analgesic treatment.For example, a type of BPdincident pain be-cause of movement in patients with bone

metastasesdmay be influenced by physical ac-tivity, which is dependent on the stage of dis-ease. Moreover, patients often receive ananalgesic treatment, which frequently can be ef-fective or not sufficient, the outcome stronglyinfluencing the feature of BP.

Patients with cancer pain can be placed in twoextremes, with a large variability between thesepoints: patients receiving oncologic treatmentare quite different from advanced cancerpatients with a short survival followed at home.Because of the dynamic course of cancer diseaseand analgesic regimen, it is likely that changesof these components influence BP.

The aim of this study was to longitudinallyassess BP in different settings according tothe different stages of disease and treatments.The present study focuses on advanced cancerpatients who were followed at home in two rep-resentative palliative care programs existing inItaly. Data regarding patients assessed in oncol-ogy units longitudinally followed over a periodof six months during the course of disease willbe the subject of a concomitant study witha similar design.

Patients and MethodsThis survey was a prospective observational

cohort study. A consecutive sample of patientsadmitted to one of two home care programs inhealth care districts existing in Genoa and Cat-ania, two large cities based in northern andsouthern Italy, respectively, who were represen-tative of cancer populations followed at homein Italy, was surveyed over a period of threemonths. Patients were eligible regardless oftheir expected survival. Patients were excludedonly if cognitively impaired or too unwell toprovide reliable answers to questions duringdata collection. Informed consent and ethicalcommittee approval were obtained.

At admission to the program, in addition tocommon demographic and disease-relateddata, data about the background pain, includ-ing pharmacological treatment and its effective-ness (if pain was moderate-severe in intensity,

556 Vol. 38 No. 4 October 2009Mercadante et al.

i.e., four out of 10 or more on a numerical scalefrom 0 to 10), were collected. The presence ofBP was noted, and if present, information wasobtained about its intensity, duration, numberof episodes, presence induced by movement,spontaneous relief after stopping activity, andlimitation of physical activity because of painon movement were recorded. Analgesic drugsprescribed for BP and the effectiveness of thesemedications also were noted. For the purposesof this study, the definition of BP was a transitoryflare of pain superimposed on an otherwise sta-ble pain pattern in patients treated with opioids,with only pain of greater than moderate inten-sity being considered a BP. Parameters relativeto background pain and BP in the previous 24hours were measured using the patient’s self-report. Data were recorded by home physicianson the first visit. After the initial assessment,patients received analgesics for backgroundpain and BP, as well as symptomatic treatmentusually provided at home, according to localpolicies and protocols, without changing rou-tine activity. At a one-month interval, the samedata collected at admission were recorded,again by interview conducted by the homecare physician.

StatisticsData were collected and analyzed using SPSS

Software 14.0 version (SPSS, Inc., Chicago, IL).Statistical analysis of quantitative data, whichincluded descriptive statistics, was performedfor all the items. The paired Wilcoxon signed-rank test and the paired samples Student’st-test were used to compare the differences inpain intensity scores and other parameters inthe time intervals. All P values were two-sidedand P values less than 0.05 were considered toindicate statistical significance.

Table 1Epidemiological Data of Patients Admitted

to Home Care

Age (years) 71.9 (12.2)Gender (M/F) 54/47Mean survival (days) 22

Primary tumorLung 21Breast 10Urogenital 9Colon-rectum 8Stomach 7Pleura 6Others 40

ResultsOne hundred and one consecutive patients

were admitted to home care programs inGenoa and Catania. The characteristics of thepatients are described in Table 1. Mean (stan-dard deviation [SD]) Eastern CooperativeOncology Group performance status score atstudy entry was 2.7 (1.1). The median survivaltime was 22 days (range: 3e91).

At admission (T0), 73 patients (70.2%) werereceiving analgesic drugs for their pain. Morethan half of the patients had uncontrolledpain, as 39 patients reported moderate-severepain (exactly 39% of the entire sample, 52%of patients receiving analgesics). Of the 73patients receiving analgesics, analgesic therapywas considered very effective, acceptable,poorly effective, and ineffective by 27, 26, 15,and 5 patients, respectively. Five patients werereceiving nonopioid drugs, including paraceta-mol (acetaminophen) or ketorolac, 33 pa-tients were receiving so-called ‘‘weak’’ opioidsor low doses of ‘‘strong’’ opioids (less thanthat equivalent to 60 mg of oral morphine/day), and 33 patients were receiving strongopioids in doses equal to or greater than60 mg of oral morphine or its equivalent perday. In two patients, data regarding analgesicconsumption for basal pain were not obtained(Table 2).

Of 73 patients receiving analgesics for theirpain at admission, 35 patients (49.2%) had BP,with a mean (SD) pain intensity of 8.0 (�1.2)on a 0e10 scale. The mean (SD) duration ofBP and the mean (SD) number of episodes perday were 35.1 minutes (35.0) and 2.4 (1.5),respectively. Of the 35 patients with BP, 23(65.7%) had pain on movement, with a mean(SD) pain intensity of 7.0 (2.2). In 17 of thesepatients (73.9%), pain decreased spontaneouslyafter stopping activity. The mean duration of thiskind of episode was less than 30 minutes in sixpatients, 30e60 minutes in five patients, and un-defined or dependent on the type of movementin the remaining patients. Physical activity wasstrongly limited in 18 of these 23 patients(78.2%).

Data regarding the treatment of BP painwere available in 93 patients. Of these, 63

Table 2Parameters Recorded at Admission (T0) and One Month Later (T1)

Parameters T0 (n¼ 101) T1 (n¼ 41) P

Uncontrolled pain 39/101 (39%) 9/41 (22%) 0.057Background pain medication 73/101 (72%) 39/41 (95%) 0.002Nonopioids 5/73 (7%) 3/39 (8%) 0.868Low-dose opioids 33/73 (45%) 10/39 (26%) 0.042Strong opioids 33/73 (45%) 13/39 (33%) 0.223Unavailable data 2/73 (3%) 13/39 (33%) 0.001

Breakthrough pain 35/73 (48%) 13/39 (33%) 0.136Pain on movement 23/35 (66%) 5/13 (38%) 0.088

BP medicationAvailable data 93/101 (92%) 30/41 (73%) 0.002No prescription 63/93 (68%) 10/30 (33%) 0.001BP medication 30/93 (32%) 20/30 (67%) 0.001OTFC 7/30 (23%) 7/20 (35%) 0.368Oral morphine 8/30 (27%) 7/20 (35%) 0.528Subcutaneous morphine 0/30 (0%) 1/20 (5%) 0.400Tramadol-codeine 6/30 (20%) 0/20 (0%) 0.037Oxycodone 0/30 (0%) 4/20 (20%) 0.021Nonopioids 9/30 (30%) 1/20 (5%) 0.030

OTFC ¼ oral transmucosal fentanyl citrate.

Vol. 38 No. 4 October 2009 557Breakthrough Pain in Home Care Cancer Patients

patients had no prescription of BP medication.Seven patients were prescribed transmucosalfentanyl, eight patients oral morphine, sixpatients tramadol or codeine, and the remain-ing, nonopioid analgesics (Table 2). The treat-ment was considered very effective, acceptable,poorly effective, and ineffective, by 14, seven,eight, and one patients, respectively.

One month later (T1), 41 patients were stillalive. The number of patients receiving analge-sic drugs for their pain significantly increased(39 patients [95.1%], P< 0.002). Only ninepatients (22%) reported moderate-severe pain(P¼ 0.057 compared with T0 [Table 2]). Whenspecifically considering the 41 patients whofollowed up at T1, the number of patients withuncontrolled background pain decreased from13 patients (at T0) to nine patients (at T1). Thedifference was not significant, possibly becauseof the low number of patients (P< 0.318). Forthe same reasons, the intensity of backgroundpain decreased from 5.8� 1.9 to 4.6� 1.9, butwas not significant (P< 0.172).

Three patients were receiving nonopioiddrugs; 10 patients were receiving tramadol,codeine, or oral morphine or equivalent ofless than 60 mg per day; and 13 patients werereceiving oral morphine or equivalent equalto or more than 60 mg per day. Data regardinganalgesic consumption for background painfor the remaining patients were not obtained.Data regarding the effectiveness of analgesictherapy were available in 30 of 39 patients

receiving analgesics. Analgesic treatment wasconsidered highly effective in 26 patients,acceptable in three patients, and poorly effec-tive in one patient.

In the 41 patients who had a follow-up, BPwas present in 15 (36.5%) and 13 patients(31.7%) at T0 and T1, respectively, and themean (SD) BP intensity did not change(8.2� 1.3 and 8.2� 1.4, at T0 and T1, respec-tively). The number of patients with pain onmovement decreased from eight to five(P< 0.364). Mean (SD) intensity of pain onmovement was 6.9 (2.3) and 7.8 (1.6) at T0

and T1, respectively (P< 0.391). Thesechanges were not significant, possibly becauseof the low number of patients.

Of 13 patients who were receiving analgesicdrugs and had episodes of BP (33%), fivepatients (38.5%) had pain on movement,with a mean (SD) pain intensity of 7.8 (1.64),and four patients (30.8%) had pain whichdecreased spontaneously after stopping activ-ity. The mean duration of this kind of episodewas less than 30 minutes in 11 patients, andmostly, only one episode per day was recorded(nine patients). Physical activity was stronglylimited in three patients.

Data regarding the treatment of BP at T1

were available in 30 patients. Of these, only10 patients had no prescription of BP medica-tion (33%, P< 0.001). Seven patients wereprescribed transmucosal fentanyl, seven pa-tients were prescribed oral morphine, four

558 Vol. 38 No. 4 October 2009Mercadante et al.

patients oxycodone, and one patient paren-teral morphine. The treatment was consideredsatisfactory in most patients, very effectiveand acceptable in 10 and three patients,respectively.

DiscussionThis survey was performed in two standard

home care programs existing in Italy, whichfollow advanced cancer patients with a limitedsurvival time of about three weeks. About two-thirds of patients were receiving analgesics forbackground pain, which was uncontrolled inabout half of the cases, possibly becausepatients were undertreated before admissionto the home care program in terms of typeand doses of analgesics, as confirmed by therelatively high number of patients who weredissatisfied. It is likely that dissatisfied patientshaving uncontrolled background pain havesignificantly more occurrences of BP.1,4,9 Inother studies, a high intensity of backgroundpain has been associated with BP.1 The inten-sity of baseline pain also was higher in outpa-tients and inpatients without BP referred toa pain clinic, where neuropathic pain pre-vailed as a cause of BP.8 These data confirmthat different types of referral, setting, andassessment may provide different information.

About half of these patients had episodes ofBP clearly distinguishable from backgroundpain, usually pain because of movement, whichstrongly influenced physical activity. Thisapproximates a previous observation whereonly pain of greater than moderate intensityon an otherwise controlled background painwas considered as a BP, with an incidence of51.2%.1 It is similar to the findings in a hospiceand home care population in Spain.6 This con-trasts with previous experiences in other set-tings reporting higher rates of BP.3,4 Thesedifferences can be explained by differentpain intensities across studies or the lack ofbackground pain in a large number ofpatients. In another study reporting an 86%incidence of BP, the timing of the interviewis unclear; 5 data recorded by the same groupreport large differences in prevalence, rangingfrom 39% to 68%.8 In patients admitted withuncontrolled background pain, BP incidencewas 70%.9

The number of episodes per day of BP ishighly variable in the literature, and may berelated to the adequacy of pharmacologicaltherapy, or, given the origin often because ofmovement-induced pain, may be dependenton the balancing of physical activity and inci-dent pain. Mobilization may be the most fre-quent trigger for BP.6 Physical activity wasstrongly limited by incident pain in a largenumber of patients, possibly explaining thelow number of BP episodes recorded in thispopulation. In a hospice study, over a thirdof patients were reported as completely inca-pacitated, but this finding was not specificallyassociated with movement-induced pain.3 Incontrast to this low rate of BP, undertreatedneuropathic pain may produce even hundredsof episodes of BP per day,8 again demonstrat-ing that the number of episodes of BP perday cannot be easily predicted.

The mean duration of BP at admission wasabout 30 minutes, in accordance with previous fig-ures reported in different settings.3e6,8 It is unclearwhether the duration of BP reported in most epi-demiological studies of BP is spontaneous, becauseof stopping activity, or the consequence of a medi-cal intervention. This issue should be better de-fined and appropriately debated.

Only a smaller number of patients (32%)were prescribed medications for BP at admis-sion. Similar figures were reported in previousstudies at hospice admission, where 43% ofpatients were prescribed no rescue medica-tion.4 In contrast, the rate of BP medication,mainly oral morphine, was very high inpatients surveyed in Catalonia.6

After one month, only about 40% of patientswere still living. Most patients had controlledpain and were satisfied with the analgesic treat-ment, possibly because the pattern and theefficacy of background pain medication andBP medication changed. In particular, thenumber of patients with a BP prescriptionincreased. Moreover, the pattern of BP medica-tion changed, with oral transmucosal fentanyl,oral morphine, and oral oxycodone prevailingover nonopioids, or opioids for moderatepain, which were significantly less frequentlyused. Of interest, the incidence of BPdecreased, although it did not attain signifi-cance because of the low number of patientsliving at the time of the second assessment per-formed one month later. This finding can be

Vol. 38 No. 4 October 2009 559Breakthrough Pain in Home Care Cancer Patients

explained by the better analgesic regimen, thereduced physical activity in patients close todeath, or both.

Data regarding BP remains quite controver-sial in the literature, mainly because of theinfluence of different settings, treatments, andthe lack of a longitudinal perspective inassessing cancer patients. This study showedthat BP characteristics may change over time.For example, one can assume that the incidenceof BP, particularly movement-related pain, maydecrease with the worsening of disease, as a con-sequence of a reduced level of activity, or be-cause patients stop even minimal activity afterfeeling an increase in pain intensity. Alterna-tively, a better analgesic approach may limitthe occurrence of BP episodes. The durationof BP episodes is evidently limited by a betterBP medication prescribed after admission.Thus, the changes in physical activity as well asanalgesic regimen may modify conclusionsabout BP when it is longitudinally examined.

This underscores the need for a longitudinalassessment of a phenomenon that is invariablydependent on stage of disease, patient, andtherapeutic factors. In other words, this phe-nomenon is impossible to assess as a pure event,because it is strongly influenced by activity or an-algesic regimen. This approach better reflectswhat happens in clinical practice, rather thanproviding just a photograph of a phenomenonthat is dynamic and influenced by several fac-tors. As any photograph taken when a subjectis moving, the development of such a photo-graph could be out of focus.

The findings of this study contradict some ofthe information drawn from previouscross-sectional studies, which have suggestedthat a large number of patients may experienceBP.4,5 These data are probably dictated by previ-ous undertreatment, rather than by real inci-dence of the phenomenon. Paradoxically, the‘‘pure’’ entity of BP should be studied in patientsreceiving no analgesic treatment, which is ethi-cally unacceptable and unrealistic, and wouldcontradict the original definition of BP, or in dif-ferent conditions where patients receive analge-sics with no, partial, or adequate pain relief.

Thus, it is difficult to have a clear idea abouta complex phenomenon, and a prospectiveevaluation may help understand what happensin individual patients. For example, in a surveyperformed in selected patients seen in an

oncologic setting with predetermined uncon-trolled background pain, data were reviewedone week after a visit. Of 70% of patients ini-tially reporting BP, only half (36%) still hadBP after pain management began, suggestingthat medical intervention may decrease the oc-currence of BP, which is unmasked by a poorefficacy of background medication. Of inter-est, the duration of BP, a median of 15 min-utes, did not change despite prescription ofrescue medication, presumably based on oralmorphine, which was otherwise consideredeffective by 83% of patients;9 however, theselection criteria applied in this latter study,where patients with uncontrolled pain wereadmitted, cannot allow any comparison withdata from the present study.

Limitations of this study include the rela-tively high number of missing data, mainlybecause of the inability to answer specific ques-tions in severely ill patients, and the low num-ber of patients living after one month of homecare. However, this finding exactly reflects theclinical setting of home care programs in Italy,with a survival time of about three weeks onaverage because of a regrettable bias towardlate referral to home care programs, oftenjust close to death.11 The same reasonsstrongly influence collecting data in patientswith physical exhaustion, with large attritionrates typical of any longitudinal palliativecare study with different collection points.12

The fact that BP is highly prevalent among pa-tients with cancer pain and predicts more severepain7,10 may reflect that patients with uncon-trolled pain are more likely to unmask BP epi-sodes.13 If patients with BP have scoresindicative of greater functional impairment,6

paradoxically, changes in physical activitybecause of progression of disease in severely illpatients and an improvement of an analgesicregimen may reduce the occurrence of BP. Asurvey of the course of BP over a six-month ob-servation in patients admitted to an oncologyward, concomitantly planned with the presentstudy, should provide further data about a com-plex phenomenon such as BP.

References1. Portenoy RK, Hagen NA. Breakthrough pain:

definition, prevalence, and characteristics. Pain1990;41:273e281.

560 Vol. 38 No. 4 October 2009Mercadante et al.

2. Mercadante S, Radbruch L, Caraceni A, et al. Epi-sodic (breakthrough) pain. Cancer 2002;94:832e839.

3. Swanwick M, Haworth M, Lennard RF. The apre-valence of episodic pain in cancer: a survey of hospicepatients on admission. Palliat Med 2001;15:9e18.

4. Zeppetella G, O’Doherty CA, Collins S. Preva-lence and characteristics of breakthrough pain incancer patients admitted to a hospice. J Pain Symp-tom Manage 2000;20:87e92.

5. Fine PG, Busch MA. Characterization of break-through pain by hospice patients and their care-givers. J Pain Symptom Manage 1998;16:179e183.

6. Gomez-Batiste X, Madrid F, Moreno F, et al.Breakthrough cancer pain: prevalence and charac-teristics in patients in Catalonia, Spain. J Pain Symp-tom Manage 2002;24(1):45e52.

7. Portenoy RK, Payne D, Jacobsen P. Break-through pain: characteristics and impact in patientswith cancer pain. Pain 1999;81:129e134.

8. Petzke F, Radbruch L, Zech D, Loick G,Grond S. Temporal presentation of chronic cancerpain: transitory pains on admission to a multidisci-plinary pain clinic. J Pain Symptom Manage 1999;17:391e401.

9. Hwang SS, Chang VT, Kasimis B. Cancerbreakthrough pain characteristics and responsesto treatment at a VA medical center. Pain 2003;101:55e64.

10. Caraceni A, Martini C, Zecca E, et al. WorkingGroup of an IASP Task Force on Cancer Pain.Breakthrough pain characteristics and syndromesin patients with cancer pain. An international sur-vey. Palliat Med 2004;18:177e183.

11. Costantini M, Toscani F, Gallucci M, et al. Ter-minal cancer patients and timing of referral to pal-liative care: a multicenter prospective cohort study.Italian Cooperative Research Group on PalliativeMedicine. J Pain Symptom Manage 1999;18:243e252.

12. Jordhoy MS, Kaasa S, Fayers P, et al. Challengein palliative care research; recruitment, attritionand complications: experience from a randomizedcontrolled trial. Palliat Med 1999;13:299e310.

13. Mercadante S, Villari P, Ferrera P, Casuccio A.Optimization of opioid therapy for preventing inci-dent pain associated with bone metastases. J PainSymptom Manage 2004;28:505e510.