Physician Attitudes and Practice in Cancer Pain Management: A Survey From the Eastern Cooperative Oncology Group

Pain is one of the most frequent and disturbing symptoms of cancer. The prevalence and severity of pain among cancer patients varies as a function of primary and metastatic sites of disease and disease stage [1-4]. It has been estimated that one of every three patients receiving active treatment for metastatic disease has significant cancer-related pain, with this percentage increasing to between 60% and 90% in patients with advanced disease [4-6]. In the specialized setting of a hospice or palliative care unit, as many as 90% of patients report pain relief [7, 8]. Yet, 1 analysis [9] of 11 published reports of cancer pain treatment covering nearly 2000 patients in nonhospice settings estimates that 50% to 80% of patients did not have adequate pain control. Experts estimate that 25% of all cancer patients die without adequate pain relief [7], despite the fact that the tools for adequate pain control are available. Several problems may account for the lack of good cancer pain management. Health care professionals may be overly concerned about addiction, the development of analgesic tolerance, side-effect management, and regulatory scrutiny of physicians who prescribe narcotics. They may not be aware of current knowledge concerning the mechanisms of pain and the modalities available for its control. Pain management has had a low priority in cancer care. Similarly, patients have also been concerned about addiction, side effects, and becoming tolerant to analgesics. Patients may also think that pain with cancer is inevitable. As a result, they have been reluctant to report pain or the lack of pain relief as well as to take adequate doses of analgesics [10-13]. In 1990, the Eastern Cooperative Oncology Group [ECOG], a National Cancer Institute supported cooperative group for the development and conduct of cancer-related clinical trials, assessed physicians' attitudes about and knowledge of cancer pain and its control as part of a larger project examining pain management within the group. The specific objectives of this project were 1) to determine the knowledge about cancer pain and its treatment among physicians practicing in ECOG-affiliated institutions; 2) to determine the methods of pain control being used by these physicians; and 3) to compare physician knowledge of and attitudes toward cancer pain and its treatment with the results of a study of cancer pain and its treatment in patients being treated by physicians at these same institutions. The latter study is currently underway throughout the ECOG; this report addresses the first two project objectives. Methods Subjects We surveyed all ECOG physicians with patient care responsibilities (medical oncologists, hematologists, surgeons, and radiation therapists), practicing in ECOG main institutions, Community Clinical Oncology Program (CCOP) institutions, and Cooperative Group Outreach Program (CGOP) institutions. Of the 1800 surveys sent out to institution data managers, 1177 were returned. Of these, 242 (21%) indicated that they did not wish to participate and 37 (3%) were ineligible (nonphysicians, left ECOG, or did not treat patients) leaving 897 respondents who form the basis of this report. Specialties included internal medicine (64%), surgery (10%), radiation therapy (14%), and other (10%). Subspecialty board certification included medical oncology (44%) and hematology (28%). Collectively, this group of physicians reported that they had treated more than 70 000 cancer patients in the 6 months before the survey. The primary treatment center in which these physicians practiced was best characterized as a comprehensive cancer center for 25% of the physicians, an outpatient treatment center for 27%, a community hospital for 19%, and a teaching hospital for 31%. The median age of the respondents was 42 years (range, 26 to 80 years) and 88% were men. Questionnaire We sent a physician cancer pain questionnaire developed by Cleeland and the Pain Research Group at the University of Wisconsin. A smaller physician study using this instrument has been previously reported [14]. The questionnaire was designed to assess physicians' estimates of the magnitude of pain as a specific problem for cancer patients, physicians' attitudes about the adequacy of pain management for cancer pain, and their report of how they manage pain in their own practice setting. As a way of describing more specific pain management practice questions, the questionnaire provided treatment recommendations for a hypothetical patient. Information was also gathered on the physicians' practice setting, training, experience with caring for patients with cancer pain, and personal experience with friends or family members with cancer, persistent pain, or substance abuse. The survey took about 45 minutes to complete. Statistical Analysis Descriptive statistics (frequencies, percentages, means, and ranges) for each response are reported. Percentages are reported as whole numbers. Not all respondents answered each question completely; therefore, the numbers that constituted the basis for the analysis of each item are included with the reported responses. For the categorical variables, the Fisher exact test [15] was used to determine candidate variables that were significantly associated with time-to-start, maximum-tolerated opioid analgesic therapy (outcome). The association between continuous predictors and outcome was tested for significance by examining the log-likelihood ratio, chi-square statistic [16]. Differences in mean rankings for barriers to pain control were tested by means of the Mann-Whitney U test [15]. Univariate analyses (two-way associations) were used to initially screen the predictors significantly associated with the outcome variable prognosis (< 6 months versus more than 6 months of the start of maximum opioid analgesic therapy in the treatment of severe pain). Several passes through the data were made to screen variables to obtain the most parsimonious model and at the same time to use as much of the data as possible. The prognostic variables were considered in a multiple logistic regression analysis using stepwise selection. A predictor was considered a candidate if it exhibited marginal association (P 0.25). A F-to-enter probability of 0.25 was therefore used initially and was decreased to 0.05 in the final model. Using this approach, we are confident that strong predictors were not inadvertently excluded from the model. This resulted in eliminating 29 of the 35 possible predictors. Results Concerning the use of analgesics for cancer pain in the United States (n = 864), 86% of the respondents thought that the majority of patients with pain are undermedicated, although 13% thought that most patients receive adequate treatment for pain. Most of the sample (67%) thought that at least 50% of the cancer patients they treat had pain at some point during their illness. Physicians estimated that almost one half of cancer patients (48%) had pain for more than 1 month. Evaluation of Physician Pain Management Practices To evaluate individual physician pain management practice, the following hypothetical case scenario was presented: A 40-year-old man is hospitalized with severe untreated back pain of more than 1-month duration, attributable to bone metastases without vertebral collapse. He weighs 70 kg; he has no cardiovascular or respiratory problems, and his prognosis is more than 24 months. He has no history of medication allergies and is opioid naive. What would be your recommendation for an initial analgesic regimen for this patient? Drug? Dosage? Route? Most ECOG physicians (93%, n = 882) stated that they would prescribe an opioid analgesic, with 41% indicating a strong opioid (morphine or a similar drug), although 51% of physicians chose a weak opioid (codeine or an equivalent). Eleven percent prescribed a regimen that also included acetaminophen or an anti-inflammatory agent. Eighty-eight percent chose an oral route of analgesic administration. In a continuation of the scenario, the patient requests additional pain medication 2 hours before his next dose is due. Physicians were asked to choose an immediate and long-term response to the patient's request. As an immediate response, 68% (n = 876) chose to give the next dose of analgesia and 11% gave a less potent non-narcotic analgesic. No respondent chose to refuse or ignore the patients' request. As a long-term response, 54% (n = 834) chose to change the dosage or interval between doses or both, 23% of physicians chose to change to another analgesic medication, 7% chose to supplement the analgesic prescription with another less potent (non-narcotic) analgesic, and 8% chose to supplement the analgesic prescription with another potent opioid analgesic. Fewer than 1% chose to leave the prescription unchanged. Other choices accounted for the remaining 8% of responses. The scenario continues, The patient's pain persists despite palliative radiation therapy to treat the pain. The patient's disease is stable. There are no signs of complications, and he is having no side effects from the medication. What is the most aggressive analgesic drug regimen that you would recommend? Drug? Dose? Route? Radiotherapy failure was a trigger for more aggressive pain management. At this stage of the scenario, 86% of physicians included a potent opioid (morphine or equivalent) in their pain treatment recommendations, although 11% of physicians recommended a weak opioid. A small number (4% of the sample) continued to recommend an anti-inflammatory agent as their most aggressive analgesic. The oral route of analgesic administration was chosen by 86% (n = 797) of physicians. Prophylactic drugs to manage typical opioid analgesic side effects, such as antiemetics or laxatives, were prescribed by fewer than 1% of the sample. Respondents were asked: At what disease stage (in terms of prognosis) would you recommend maximum-tolerated narcotic analgesic therapy for tre