Vol. 15 No. 5 May 1998 Journal of Pain and Symptom Management 285

Original Article

The Prevalence and Severity of Cancer Pain:A Study of Newly-Diagnosed Cancer Patientsin TaiwanLuo-Ping Ger, RN, MPH, Shung-Tai Ho, MD, Jhi-Joung Wang, MD, DMSc,and Chen-Hwan Cherng, MD, DMScDepartment of Medical Education and Research (L-P.G.), Kaohsiung Veterans General Hospital,Kaohsiung, Taiwan, and Department of Anesthesiology (L-P.G., S-T.H., J-J.W., C-H.C.), NationalDefense Medical Center/Tri-Service General Hospital, Taipei, Taiwan, Republic of China

AbstractCancer pain is a relatively neglected public health issue in Taiwan. To characterize thenature of this problem, interviews were conducted on newly diagnosed cancer patientsadmitted to the Tri-Service General Hospital during a period of 18 months. Data werecollected on the prevalence and severity of cancer pain, its treatment, and impact onpatients in the week before the interview. Correlates of prevalence and severity of cancerpain were also examined. The final analysis included 296 patients who had no history ofrecent surgery. Most of the patients (69%) were interviewed within 14 days of theirdefinitive diagnosis of cancer. Thirty-eight percent (N 5 113) of the patients had cancer-related pain. Of these 113 patients, 65% had ‘‘significant worst pain’’ (worst pain levelat or above five on a ten-point scale) and 31% had ‘‘significant average pain’’ (averagepain level at or above five most of the time); 69% received no pain medication at all orinadequate medication (not ‘‘by the ladder’’), and 23% had pain medication that was notadministered at a fixed interval (not ‘‘by the clock’’). Multivariate analyses showed thatcancer pain was more prevalent in non-Mainlanders, those with a lower level ofinsurance, those with a history of excellent pain tolerance, those with poor EasternCooperative Oncology Group (ECOG) performance status, and those with distantmetastases. Patients who were at greater risk of ‘‘significant worst pain’’ were those withregional or distant metastases, those in whom an inadequate analgesic medication hadbeen prescribed (not ‘‘by the ladder’’), and those who had received an appropriate analgesicmedication but no fixed schedule dosing (‘‘by the ladder’’ but not ‘‘by the clock’’). Patientswho were at greater risk of ‘‘significant average pain’’ were those not undergoing anyresection of the tumor lesion and those who received an appropriate drug but no fixedschedule dosing (‘‘by the ladder’’ but not ‘‘by the clock’’). J Pain Symptom Manage1998;15:285–293. U.S. Cancer Relief Committee, 1998.

Key WordsCancer pain, prevalence, severity, correlates

Address reprint requests to : Shung-Tai Ho, MD, Department of Anesthesiology, Tri-Service General Hospital, Taipei,Taiwan, Republic of China.Accepted for publication: July 25, 1997.

U.S. Cancer Relief Committee, 1998Published by Elsevier, New York, New York

0885-3924/98/$19.00PII S0885-3924(98)00017-7

286 Ger et al. Vol. 15 No. 5 May 1998

AssessmentsIntroductionCancer has been the leading cause of death

Brief pain inventory (BPI). The BPI is a reliablein Taiwan since 1982.1 However, cancer pain re-

and valid tool for assessing the intensity of painlief, though important, is a neglected public

and the extent to which pain interferes with lifehealth issue because little information about

activities.3–6 A modified Taiwanese version ofcancer pain in Taiwan is available.

Brief Pain Inventory (TBPI) was used. PatientsIn 1991, we retrospectively reviewed 1523

used a scale with a horizontal row of equidistantcharts of new cancer patients treated at the Tri-

numbers from zero (no pain) to ten (extremeService General Hospital (TSGH), Taipei, Tai-

pain) to rate their worst pain, least pain, andwan.2 Pain was experienced by 30.9% of pa-

average pain during the week prior to the inter-tients at the time of admission. However, these

view and the pain they were having at the timedata were not well controlled because the ad-

of interview. The impact of cancer pain was alsomission notes were written by doctors in train-

examined on similar scales. Patients were askeding from different departments. Often, no data

to rate how much their pain interfered with var-were available in the admission notes regarding

ious aspects of life, including general activities,pain severity, the effect of pain relief, and the

mood, walking, work, relationships with otherimpact of cancer pain.

people, sleep, and enjoyment of life. ResponseTo obtain these data, a survey was carried out

options ranged from ‘‘not at all’’ as zero to ‘‘ex-on newly diagnosed cancer patients admitted

tremely disturbing’’ as ten.to TSGH between July 1992 and December1993. The information collected included the

Sociodemographic background. Age, gender,prevalence of cancer pain and its severity andmarital status, education, medical insuranceimpact during the week prior to the interview.coverage, religion, and ethnicity (patient’s eth-The correlates of the prevalence and severity ofnic identifications in Taiwan consist of Fu-cancer pain were also evaluated.kienese, Hakka, Aborigine, or Mainlander)were assessed.

MethodsPain tolerance. The past history of the ability totolerate toothache or stomachache (very good,Subjectsfair, poor, and extremely poor) and the pa-

This study was restricted to hospitalized con-tients’ reluctance (ever, never) to report their

scious patients, aged 10–80 years, who were re-pain to doctors or nurses in the past week were

cently diagnosed (cytologically or pathologi-assessed.

cally) with malignancies. Data were collected atTSGH between July 1992 and December 1993.

Performance StatusFrom a total of 978 eligible cases, 854 (87.3%)The Eastern Cooperative Oncology Groupwere available for evaluation. Among the 124

(ECOG) was used to assess performance status.excluded cases, 74 (7.6%) were discharged be-The scale is from zero to four, from fully activefore the interview, 32 (3.3%) were too ill to beto completely bedridden.7

interviewed, 13 (1.3%) refused to be inter-viewed, and the other 5 (0.5%) had incompletedata on the questionnaire. To rule out cases of Progress of Disease

Primary sites, stage of disease [The Nationalpostsurgical pain, the 558 patients who had sur-gery in the last 30 days were excluded from the Cancer Institute’s Surveillance, Epidemiology,

and End Results (SEER) program’s staging sys-data analysis. Thus, 296 patients were includedin the analysis of cancer pain prevalence and tem was adopted],8 remission status (active or

remission), and metastatic sites were the factorscorrelates of cancer pain prevalence during theweek prior to the interview. One hundred and considered in determining the progress of dis-

ease. After patients had been discharged, thethirteen patients who had pain attributed tocancer or to cancer treatment were included in medical records were reviewed by two experi-

enced cancer registrars to obtain the abovethe analysis of pain severity, correlates of painseverity, and impact of cancer pain. clinical information.

Vol. 15 No. 5 May 1998 Prevalence and Severity of Cancer Pain 287

Cancer Pain Management ProcedureThe interviewers screened cytological andPrevious anticancer treatment, analgesic

drug treatment during the week prior to the pathological reports every day at TSGH, thenvisited the qualified cancer patients during theinterview, and the degree (as a percentage) of

pain relief from the administered drugs following week, if possible. They interviewed allqualified patients in the wards, because manywere considered in cancer pain management.

Anticancer treatments include surgery, radio- patients could not complete the TBPI by them-selves. Fifty-six of the 296 patients (19%) weretherapy, chemotherapy, and hormone ther-

apy. The types of analgesic drugs were used illiterate and 86 patients (29%) had only gradu-ated from Chinese or Japanese elementaryin assessing ‘‘by the ladder.’’9 The interval

of prescription was used in assessing ‘‘by the school (before 1945, Taiwan was occupied byJapan). One hundred and fifty-seven of 296 pa-clock.’’9

tients (53%) were interviewed within 7 daysfrom the day of receiving the initial cytologicalPain Management Index (PMI)

The adequacy of analgesic drug therapy (‘‘by or pathological diagnosis. Sixty-nine percent ofthe patients were interviewed within 14 days,the ladder’’) for controlling cancer pain was as-

sessed by using Pain Management Index (PMI). and 84% were interviewed within 90 days of theinitial diagnosis. A random sample (N 5 19) ofThe PMI provides a conservative indicator of

analgesic drug management based on the most patients were re-interviewed by the supervisoron the next day to assess the test–retest reliabil-efficacious analgesic used relative to the level

of patient’s reported pain.10, 11 To construct the ity; the kappa coefficients were in the range be-tween 0.85 and 1.00, and the correlation coef-PMI, the most efficacious of the following four

levels of the analgesic drug treatment were ficients were in the range between 0.70 and0.94.used: zero, no analgesic drug; one, a nonopioid

(for example, nonsteroidal anti-inflammatorydrug or acetaminophen); two, a so-called

Results‘‘weak’’ opioid (for example, codeine); andthree, a so-called ‘‘strong’’ opioid (for exam-

Prevalence of Cancer Painple, morphine or meperidine). The patient’sA total of 296 cancer cases were analyzed, in-level of pain was determined by the worst pain

cluding 194 men (66%) and 102 womenscore on TBPI: zero, worst pain rating at zero;(34%). The mean age was 56.6 6 16.one, worst pain rating at one to three; two,

8 years (range, 12–80 years). Utilizing theworst pain rating at four to seven; and three,SEER program’s staging system, 71 patientsworst pain rating at eight to ten. The PMI was(24%) had localized disease, 103 patients (35%)computed by subtracting the pain level fromhad regional disease, 108 patients (36%) hadthe analgesic level. Negative PMI scores weredistant metastasis, and 14 patients (5%) had anconsidered to indicate analgesic drug treat-undetermined stage of disease. In terms of can-ment being not ‘‘by the ladder.’’ Scores of zerocer sites, 63 patients (21%) had lung cancer,or greater were considered to be a conservative58 patients (20%) had upper gastrointestinalindicator of ‘‘by the ladder.’’ The PMI mea-tract cancer, 36 patients (12%) had colon orsures the health-care provider’s response to arectal cancer, 34 patients (11%) had cervix can-patient’s pain. Note that pain management iscer, 29 patients (10%) had head or neck can-not considered inadequate if the patient is notcer, and 76 patients (26%) had other cancers.compliant in taking medication.11

Of the 296 patients, 113 patients (38%) hadpain. Ninety-two percent had cancer-related‘‘By the clock.’’ Analgesic drugs that were ad-pain, 5% had treatment-related pain, and 3%ministered on an ‘‘around-the-clock’’ basishad both cancer-related and treatment-relatedwere considered to indicate analgesic drugpains.treatment that was ‘‘by the clock.’’ If no analge-

sic drug was administered or the analgesic drugwas administered only on an ‘‘as-needed’’ ba- The Correlates of Cancer Pain

In univariate analysis, the chi-square test orsis, the regimen was considered to be not ‘‘bythe clock.’’ Mann–Whitney U test were used. Cancer pain

288 Ger et al. Vol. 15 No. 5 May 1998

Table 1Logistic Regression Analysis of Various Factors Associated with Cancer Pain During the

Week Prior to Interview (N 5 296)

Percentage Percentageof patients with cancer

Variables in category paina Coefficient SE Odds ratio (95% CI)

EthnicityMainlander 40 32Fukienese, Hakka, Aborigine 60 43 0.58 0.28 1.79 (1.03–3.11)*

Medical insurance coverageHigher levelb 42 30Lower levelc 58 44 0.67 0.28 1.96 (1.14–3.39)*

Past history of pain toleranceBad, not good, well 53 31Excellent 47 47 0.80 0.28 2.22 (1.29–3.82)**

ECOG StatusGoodd 91 34Poore 9 79 2.38 0.53 10.78 (3.85–30.32)**

Stage of diseaseLocalized 24 23Regional 35 32 0.29 0.38 1.34 (0.63–2.83)Distant f 41 53 1.22 0.36 3.38 (1.65–6.91)**

aPain attributed to cancer or cancer treatment.bCivilians, military, government employee, and relatives.cLabors, farmers, and military dependents.dFully active (without restriction) and ambulatory (restricted in physically strenuous activity).eCompletely disabled, capable of only limited self care (in bed . 50% day), and capable of self care (in bed , 50%day).fIncluding 14 (5%) patients with undetermined stage of disease.

*P , 0.05.**P , 0.01.

prevalence was significantly correlated with risk for cancer pain, compared with patientswith good ECOG status. Additionally, patientsmedical insurance coverage, past history of

pain tolerance, ECOG performance status, pri- with distant metastasis had a 3.4-times higherrisk for cancer pain compared with patients inmary sites, status of remission (active or remis-

sion), stage of disease, metastasis status (yes, a localized stage.no), bone metastasis (yes, no), and liver metas-tasis (yes, no). Severity of Cancer Pain

One hundred and thirteen patients with painIn a logistic regression model, medical insur-ance status, past history of pain tolerance, attributed to cancer and/or to cancer treat-

ment were included in the analysis of pain se-ECOG performance status, and stage of diseasewere statistically significant predictors of pain verity. There were 76 (67%) male and 37 (33%)

female patients, who had a mean age of 55.8 6(see Table 1). Because bone metastasis andliver metastasis also correlated with the stage of 16.7 years; range, 12–80 years. Sixteen patients

(14%) had local disease, 33 (29%) had regionaldisease, they were not included in the modelof logistic regression. Aborigines, Hakka, or Fu- disease, and 64 (57%) had distant metastasis.

Twenty-seven patients (24%) had lung cancer,kienese had a 1.8-times higher risk for cancerpain compared with Mainlanders. Patients with 23 patients (20%) had upper gastrointestinal

tract cancer, 13 patients (11%) had colon ormedical insurance of labors, farmers, or mili-tary dependents had a 2.0-times higher risk for rectal cancer, 11 patients (10%) had head or

neck cancer, and 39 patients (35%) had othercancer pain, compared with patients with insur-ance of civilians, military, government employ- cancers.

The severity of cancer pain during the weekees, or relatives of government employees. Pa-tients with a past history of excellent pain prior to the interview is shown in Table 2. The

median scores of worst pain, least pain, averagetolerance of toothache or stomachache had a2.2-times higher risk for cancer pain compared pain, and current pain (pain the patients were

suffering at the time of interview) were five,with patients with bad, not good, or good paintolerance. Patients with poor ECOG status zero, three, and one, respectively. The modal

scores of worst pain, least pain, average pain,prior to this admission had a 10.8-times higher

Vol. 15 No. 5 May 1998 Prevalence and Severity of Cancer Pain 289

Table 2Severity of Cancer Pain During the Week Prior to Interview (N 5 113)

Pain scorea percentiles

Pain during the week Lower Median Upperprior to interview quartile quartile quartile Mean (SD)

Worst pain 3 5 8 5.72 (3.19)Least pain 0 0 1 0.97 (1.75)Average pain 1 3 5 3.12 (2.34)Current pain 0 1 3 1.72 (2.30)

a Pain score was estimated by the Taiwan version of the Brief Pain Inventory (edited by Wisconsin Cancer Pain Initia-tive).

and current pain were ten (23%), zero (60%), those with local disease. Patients with distantmetastasis had an 8.2-times higher risk for sig-two (19%), and zero (49%), respectively. ‘‘Sig-

nificant’’ pain is defined as pain rated as five nificant worst pain compared to those with lo-cal disease. Patients who had received inade-or more on the zero-to-ten scale. Of the 113

patients, 65% had ‘‘significant’’ worst pain, and quate analgesic medication (not ‘‘by theladder’’) had a 27.4-times risk for significant31% had ‘‘significant’’ average pain; 69% re-

ceived inadequate analgesic medication (not worst pain compared with those with good anal-gesic management (‘‘by the ladder’’ and ‘‘by‘‘by the ladder’’), and 23% received pain medi-

cation not ‘‘by the clock.’’ the clock’’). Additionally, patients who re-ceived adequate analgesic medication (‘‘by theladder’’) but without a fixed interval (not ‘‘byThe Correlates of Significant Worst Pain

In the univariate analysis, significant worst the clock’’) had a 31.3-times higher risk for sig-nificant worst pain compared with those withpain was correlated with the stage of disease,

the patients’ reluctance to report their pain to good analgesic drug management (‘‘by the lad-der’’ and ‘‘by the clock’’).doctors or nurses, and the analgesic drug treat-

ment. In a logistic regression model, stage ofdisease and analgesic drug treatment were sta- The Correlates of Significant Average Pain

In the univariate analysis, the significant aver-tistically significant predictors of a worst painseverity being equal to or greater than five. The age pain was correlated with the patients’ reluc-

tance to report their pain to doctors or nursesresults of logistic regression analysis of signifi-cant worst pain are shown in Table 3. Patients and the analgesic drug treatment. In a logistic

model, previous surgery and analgesic drugwith regional metastasis had a 3.9-times higherrisk for significant worst pain compared with treatment were statistically significant pre-

Table 3Logistic Regression Analysis of Various Factors Associated with the Significant Worst Pain

During the Week Prior to Interview (N 5 113)

PercentagePercentage withof patients significant

Variables in category worst paina Coefficient SE Odds ratio (95% CI)

Stage of diseaseLocalized 14 38Regional 29 61 1.36 0.67 3.91 (1.05–14.57)*Distant 57 75 2.10 0.64 8.19 (2.33–28.83)**

Analgesic drug treatmentBy the ladder and 14 12

by the clockNot by the ladder 69 73 3.31 0.83 27.38 (5.42–138.24)**By the ladder but not 17 79 3.44 0.97 31.33 (4.65–211.05)**

by the clock

aThe worst pain score during the week being $ 5.*P , 0.05.

**P , 0.01.

290 Ger et al. Vol. 15 No. 5 May 1998

dictors of having average pain of five or greater. sociated with cancer therapy, and 3% was unre-lated to cancer or cancer therapy (for example,The results of logistic regression analysis of

significant average pain are shown in Table 4. due to bed sores or headache).12 The rate oftreatment-related pain in this study was onlyPatients not undergoing any resection of the

tumor lesion had 11.3-times higher risk of sig- 8%. One possible explanation is that most ofthe cases in this study were newly diagnosed pa-nificant average pain than those undergoing re-

section of the tumor lesion. Patients who had tients who had just started receiving anticancertherapy. For example, 15% received initial sur-received adequate analgesic medication (‘‘by

the ladder’’) but without fixed interval dosing gery, 12% received initial radiotherapy, 13% re-ceived initial chemotherapy, and 0.3% received(not ‘‘by the clock’’) had 12.0-times higher risk

for significant average pain compared with hormone therapy.Ethnicity was correlated with the prevalencethose with good analgesic drug management

(‘‘by the ladder’’ and ‘‘by the clock’’). of cancer pain, but not correlated with the se-verity of cancer pain. This might reflect selec-

Impact of Cancer Pain tion bias because Mainlander patients may haveThe mean scores and standard deviations of health-care seeking behaviors that were more

pain-induced interference with life activities aggressive than other ethnic groups. Main-are as follows: general activity, 3.87 (64.25); lander patients had higher educational levelsmood, 3.93 (63.97); walking ability, 4.07 (55% with education years of 9 years or more)(64.19); normal work, 4.54 (64.50); relations than Fukienese, Hakka and Aborigine patientswith others, 2.86 (63.82); sleep, 5.08 (64.30); (26% with education of 9 years or more, χ 2 5and enjoyment of life, 4.17 (64.46). All of the 25.5, df 5 1, P , 0.001). When patients recog-cancer pain-induced interference with life ac- nized their initial symptoms and decided totivities among 113 patients was associated with seek care, ethnicity affected their health-carepain severity (P , 0.01) (see Table 5). How- seeking behaviors, especially the ‘‘type of initialever, interference with life activities, except clinic (or hospital) visit’’ and ‘‘health-caregeneral activity and walking ability, was not cor- seeking interval’’ in our study. For example,related with inadequate analgesic medication. more Mainlander patients (25%) visited ourFurthermore, patients with less interference in hospital directly than the Fukienese, Hakka, orgeneral activity and walking ability had a Aborigine (15%, χ 2 5 4.92, df 5 1, P 5 0.03)greater chance of having inadequate analgesic did, when they decided to seek care. TSGH is amedication. well-known medical center in southern Taipei.

The ‘‘type of initial clinic visit’’ was categorizedinto two groups as ‘‘original’’ for patients who

Discussion made their initial clinic visits at TSGH and ‘‘re-ferral’’ for patients whose initial clinic visitsFoley found that 77% of cancer pain was

caused by the disease process itself, 19% was as- were not at TSGH. In the ‘‘original’’ group, Fu-

Table 4Logistic Regression Analysis of Various Factors Associated with the Significant Average

Pain During the Week Prior to Interview (N 5 113)

PercentagePercentage withof patients significant

Variables in category average paina Coefficient SE Odds ratio (95% CI)

Previous operationYes 13 7No 87 35 2.42 1.09 11.3 (1.34–95.34)*

Analgesic drug managementBy the ladder and 14 12

by the clockNot by the ladder 69 29 1.27 0.80 3.55 (0.74–16.95)By the ladder but not 17 53 2.48 0.92 12.00 (1.99–72.41)**

by the clock

aThe average pain scoring during last week being $ 5.*P , 0.05.

**P , 0.01.

Vol. 15 No. 5 May 1998 Prevalence and Severity of Cancer Pain 291

Table 5Median Score of Pain-Induced Interference with Function among 113 Patients with

Cancer Pain According to Significant Worst Pain, Significant Average Pain, and Adequacyof Analgesic Medication

Significant Significant Adequacy of analgesicworst pain average pain medication

Types of function No Yes No Yes Adequate Inadequate

General activity 0 5*** 0 10*** 6 1**Mood 0 5*** 1 10*** 4 2Walking ability 0 5*** 1 10*** 6 2*Normal work 0 7*** 0 10*** 4 3Relations 0 2*** 0 6*** 0 0Sleep 0 9*** 2 10*** 5 5Enjoy of life 0 8*** 0 10*** 2 2

*P , 0.10.**P , 0.05.

***P , 0.01.

kienese, Hakka, or Aborigines had a 2.7-times good pain tolerance (52.2%) than patients withhigher education levels (38.2%,χ 2 5 4.92, df 5higher risk for cancer pain compared with

Mainlander (χ 2 5 2.94, P 5 0.08). In the ‘‘re- 1, P 5 0.03). This result is consistent with ourclinical experience. However, the relationshipferral’’ group, the risk of cancer pain for Fu-

kienese, Hakka, or Aborigines was comparable between a past history of pain tolerance andcancer pain prevalence needs further study.with Mainlander’s (odds ratio, 1.3; P 5 0.24).

Additionally, Mainlander patients had shorter Peteet et al. found that severity of cancerpain was not correlated with patients’ perfor-‘‘health-care seeking interval,’’ which was de-

fined as the number of days between symptom mance status (measured by ECOG scale).14 Thisstudy demonstrated that cancer pain was moreonset and the initial clinic visit or hospital visit

(median, 9; 95% confidence interval, 4.46– prevalent in patients with poor performancestatus; however, the severity of cancer pain was13.54), than Fukienese, Hakka, or Aborigine

patients (median, 15; 95% confidence interval, not correlated with performance status. Addi-tionally, cancer pain patients with poor perfor-11.73–18.27, P 5 0.03 by log-rank test). There-

fore, we put ethnicity in the model of logistic mance status had a statistically much higherrate of adequate analgesic medication (54.6%)regression to control its selection effect of can-

cer pain. than patients with good performance status(25.3%) (Yates corrected χ 2 5 5.82, P 5 0.016).The study by Greenwald et al. indicated that

patients with low pain thresholds reported high This result is in agreement with a previousfinding that patients being rated as more ill arelevels of pain.13 However, this study showed that

a past history of pain tolerance was not associ- more likely to have adequate analgesia.11 Thisresult showed the different effects of patients’ated with severity of cancer pain. Nevertheless,

cancer pain was more prevalent in those with performance status on the prevalence and theseverity of cancer pain.a past history of good pain tolerance. Recall

bias may have occurred despite the use of a The prevalence of cancer pain is far higherin patients with widespread disease than instandardized questionnaire combined with in-

tensive interviewer training in evenhanded those with local or regional disease.15–17 Fur-thermore, the degree of cancer pain is also gen-probing techniques for patients with or without

cancer pain. Although a random sample (N 5 erally higher in patients with advanced diseasethan in those at an earlier stage.16 These find-19) of the patients were re-interviewed to assess

the test–retest reliability, the kappa coefficient ings are consistent with results of the presentstudy. However, Greenwald et al. found that thewas 0.85 for past history of pain tolerance. The

relationship between patients’ education and pain level was not correlated with the stage ofdisease.13the history of pain tolerance was evaluated to

check the validity of pain tolerance. A higher The management of cancer pain revolvesaround two principal approaches. If the tumorpercentage of patients with lower education lev-

els (education years # 9 years) had history of is the source of pain, anticancer treatment (sur-

292 Ger et al. Vol. 15 No. 5 May 1998

gery, radiotherapy, chemotherapy, or hormone were undermedicated. A possible factor forsuch discrepancy was that most patients in thistherapy) can lead to significant pain relief.

More frequently, however, pain symptoms are study had just received the diagnosis of cancerwere waiting for anticancer treatment. How-relieved with the use of analgesic drugs. Other

pain management procedures can range from ever, it is evident that inadequate prescriptionof analgesics by physicians is a significant con-the application of psychosocial interventions to

the use of psychopharmacologic agents, physi- tributing factor to patients’ experiencing mod-erate or severe cancer pain. Hence, Taiwanesecal modalities, nerve blocks, or neurosurgical

procedures.9,18,19 The results of this study con- physicians need to be encouraged to prescribeanalgesic drugs to control cancer pain. In partic-firmed that previous surgery and the use of an-

algesic drugs were the most important methods ular, Taiwanese physicians should alter currentapproaches and adopt the following methods:to reduce pain.

The use of analgesic drugs is the mainstay ofcancer pain management. It has been shown

1. Prescribe the adequate analgesic drugs (‘‘bythat the World Health Organization (WHO)the ladder’’) and not to undermedicate;analgesic ladder is effective in relieving pain in

2. Adopt fixed dosing intervals (‘‘by theapproximately 90% of patients with cancer 20

clock’’) for patients with persistent pain;and over 75% of cancer patients who are termi-nally ill.21 Two key concepts underlying the use 3. Use morphine (especially oral morphine),of analgesic drugs are ‘‘by the ladder’’ and ‘‘by not meperidine, as a ‘‘strong’’ opioid;the clock.’’9,18,22 In this study, for both signifi- 4. Initiate the analgesic ladder following thecant worst pain and significant average pain, initial pain assessment, and do not let pa-correlates were found not only with the right tients wait for anticancer treatment and suf-drug (‘‘by the ladder’’) but also with the right fer unnecessary pain.drug given at a fixed interval (‘‘by the clock’’).In the relief of average pain, the analgesic Cleeland et al. found that patients with ade-

quate analgesic medication (‘‘by the ladder’’)drugs being given ‘‘by the clock’’ was more im-portant than ‘‘by the ladder’’ (shown in Table reported less pain-related impairment.11 This

study showed that cancer pain-induced inter-4). This indicated that both ‘‘by ladder’’ and‘‘by clock’’ should be considered as tools for ference with different aspects of function was

not correlated with adequate analgesic medica-cancer pain management, especially in Taiwan.Additionally, this study showed that 69% tion, with the exception of general activity and

walking ability. Furthermore, patients with(78/113) of patients were undermedicated(not ‘‘by the ladder,’’ according to the WHO more interference in general activity and walk-

ing ability had a greater likelihood of receivingladder, as measured by PMI). Forty-four per-cent (17/39) of patients with moderate pain adequate analgesic medication (see Table 5).

This discrepancy might be due to the fact thatdid not receive any analgesic drug, and 31%(12/39) of patients with severe pain did not re- our patients were all newly diagnosed, in con-

trast to those patients from other countries whoceive any analgesic drugs. Moreover, both 83%(15/18) of the patients receiving ‘‘strong’’ opi- were at an advanced stage. Additionally, physi-

cians in Taiwan are inclined to prescribed anal-oids and 30% (3/10) of the patients receiving‘‘weak’’ opioids were given these drugs as gesic drugs to those patients with more pain-

induced activity interference. However, most ofneeded. Furthermore, 83% (15/18) of the pa-tients who received a ‘‘strong’’ opioid were them did not prescribe the adequate analgesic

drugs (both ‘‘by the ladder’’ and ‘‘by thegiven intramuscular meperidine, rather thanoral morphine. All patients (ten/ten) who re- clock’’) to patients so that the patients’ pain-

induced interference could not be reduced ef-ceived ‘‘weak’’ opioids were given oral codeineor tramadol. fectively.

In conclusion, cancer pain, significant worstPrevious studies reported that the rate of un-dermedication for cancer pain (not ‘‘by the pain, significant average pain, and undermedi-

cation (not ‘‘by the ladder’’ and not ‘‘by theladder,’’ as measured by the PMI) was only30%–42%.10,11,23 However, a significantly higher clock’’) are very common among newly diag-

nosed cancer patients in Taiwan. The patients’proportion (69%) of the patients in this study

Vol. 15 No. 5 May 1998 Prevalence and Severity of Cancer Pain 293

8. National Institutes of Health (NIH). Self-sociodemographic backgrounds (ethnicity,medi-instructional Manual for Tumor Registrars, Book 6–cal insurance coverage), past history of pain tol-Classification for Extent of Disease, 1st ed. U.S. De-

erance, ECOG performance status and stage of partment of Health and Human Service: National In-disease were the most important correlates of stitutes of Health, 1981:59–194.cancer pain prevalence. The patient’s stage of 9. World Health Organization (WHO). Cancerdisease and cancer pain management [previous pain relief, 1st ed. Geneva: WHO, 1986:45–70.operation and analgesic drug treatment (‘‘by 10. Zelman D, Cleeland C, Howland E. Factors inthe ladder’’ and ‘‘by the clock’’)] were the appropriate pharmacologic management of cancer

pain: a cross-institutional investigation. Pain 1987;most important correlates of their pain severity.(suppl), S136.

11. Cleeland CS, Gonin R, Hatfield AK, et al. PainAcknowledgment and its treatment in outpatients with metastatic can-

cer. N Engl J Med 1994;330:592–596.This work was supported by grant no.12. Foley KM. Pain syndromes in patients with can-DOH82-HR-C03 and DOH83-HR-203 fromcer. In: Bonica JJ, Ventafridda V, eds. Advances inthe Department of Health, ROC. The authorspain research and therapy, vol 2. New York: Raven,

wish to express the highest appreciation to Dr. 1979:59–75.Wen-Hsien Wu for his initiative ideas and en-

13. Greenwald HP, Bonica JJ, Bergner M. The preva-couragement to us to implement the study, to lence of pain in four cancers. Cancer 1987;60:2563–Dr. Charles S. Cleeland for his generous provi- 2569.sion of the Brief Pain Inventory, and to Dr. 14. Peteet J, Tay V, Cohen G, et al. Pain characteris-Kathleen M. Foley for her comments and sug- tics and treatment in an outpatient cancer popula-gestions on the draft of this manuscript. We tion. Cancer 1986;57:1259–1265.also wish to express our gratitude to Drs. Chih- 15. Ahles TA, Ruckdeschel JC, Blanchard EB.Shung Wong, Ching-Yue Yang, and Yuan-Ji Cancer-related pain. I. Prevalence in an outpatient

setting as a function of stage of disease and type ofDay, Jeng-Tieng Shen, and Misses Joy Fan,cancer. J Psychosom Res 1984;28:115–119.Woei-Wei Chen, and Mei-Lan Chu for their16. Daut RL, Cleeland CS. The prevalence and se-constant support to this project.verity of pain in cancer. Cancer 1982;50:1913–1918.

17. Portenoy RK. Cancer pain epidemiology andReferences syndrome. Cancer 1989;63:2298–2307.

18. U.S. Department of Health Services. Manage-1. Department of Health (DOH), Executive Yuan,Republic of China. Health statistics, Vol II Vital statis- ment of cancer pain. Clinical practice guideline, no.tics 1983–1993 Taipei: Department of Health, 1983– 9. U.S. Department of Health Services, Public Health1993. Service, Agency for Health Care Policy and Re-

search, 1994:7–21.2. Cherng CH, Ho ST, Kao SJ, et al. The study ofcancer pain and it’s correlates. Acta Anaesth Sinica 19. Dorrepaal KL, Aaronson NK, Van Dam FSAM.(ROC) 1991;33:653–657. Pain experience and pain management among hos-

pitalized cancer patients. Cancer 1989;63:593–598.3. Daut RL, Cleeland CS, Flanery RC. Develop-ment of the Wisconsin brief pain questionnaire to 20. Ventafridda V, Caraceni A, Gamba A. Field-assess pain in cancer and other diseases. Pain 1983; testing of the WHO guidelines for cancer pain relief:17:197–210. summary report of demonstration projects. In: Foley

KM, Bonica JJ, Ventafridda V, eds. Advances in pain4. Cleeland CS, Ladinsky JL, Serlin RC, et al. Multi-research and therapy, vol 16. New York: Raven, 1990:dimensional measurement of cancer pain: compari-451-464.sons of US and Vietnamese patients. J Pain Symptom

Manage 1988;3:23–27. 21. Grond S, Zech D, Schug SA, et al. Validation ofWorld Health Organization guidelines for cancer5. Caraceni A, Mendoza TR, Mencaglia E, et al. Apain relief during the last days and hours of life. Jvalidation study of an Italian version of the Brief PainPain Symptom Manage 1991;6:411–422.Inventory. Pain 1996;65:87–92.22. Bonica JJ. Treatment of cancer pain: current sta-6. Wang XS, Mendoza TR, Gao SZ, et al. The Chi-tus and future needs. In: Fields HL, Dubner R, Cer-nese version of the Brief Pain Inventory: its develop-vero R, eds. Advances in pain research and therapy,ment and use in a study of cancer pain. Pain 1996;vol 9. New York: Raven, 1985:589–616.67:407–416.

7. Oken MM, Creech RH, Tormey DC, et al. Toxic- 23. Ward SE, Goldberg N, Miller-McCauley V, et al.Patient-related barriers to management of cancerity and response criteria of the Eastern Cooperative

Oncology Group. Am J Clin Oncol 1982;5:649–655. pain. Pain 1993;52:319–324.