Presented as a poster at the 2011 Annual American Society of Clinical Oncology; June 1-5, 2011; Chicago, IL.
Cancer pain initiatives recommend using the personalized pain goal to tailor pain management. This study was conducted to examine the feasibility and stability of personalized pain goal, and how it compares to the clinical pain response criteria.
Records of 465 consecutive cancer patients seen in consultation at the Supportive Care Clinic were reviewed. Pain relief was assessed as clinical response (≥30% or ≥2 point pain reduction) and personalized pain goal response (pain ≤ personalized pain goal).
One hundred fifty-two (34%), 95 (21%), and 163 (37%) patients presented with mild (1-4), moderate (5-6), and severe (7-10) pain, respectively. Median age (59 years), males (52%), and advanced cancer status (84%) did not differ by pain category. Median personalized pain goal at initial clinic consult was 3 (interquartile range, 2-3), was similar across pain groups, and remained unchanged (P = .57) at follow-up (median, 14 days). Clinical response was higher among patients with severe pain (60%) as compared with moderate (40%) and mild pain (33%, P < .001). Personalized pain goal response was higher among patients with mild pain (63%) as compared with moderate (44%) and severe pain (27%, P < .001). By using personalized pain goal response as the gold standard for pain relief, the sensitivity of clinical response was highest (98%) among patients with severe pain, but it had low specificity (54%). In patients with mild pain, clinical response was most specific for pain relief (98%), but had low sensitivity (52%).
Pain is among the most frequent, feared, and burdensome of symptoms in cancer patients and negatively impacts quality of life.1-5 Pain intensity is recognized as the dominant factor that determines the effect of pain on the patient and the urgency of the treatment process,6, 7 and in most circumstances, its assessment relies primarily on the patient's self-report, which is considered the gold standard.
Pain control is a major outcome of quality improvement (QI) initiatives. An important part of current recommendations is the frequent assessment of pain intensity, prompt analgesic interventions for pain relief, and the reassessment of patient's response to such treatments.8-12 The frequent assessment for the presence and intensity of pain is now considered standard of care in US health care facilities. However, measures to capture the various multidimensional attributes of quality pain management and what constitutes meaningful improvement of pain to an individual patient have not been clearly defined. Studies suggest a low level of correlation between patient satisfaction with pain management and reported intensity of pain and/or pain relief,13-20 although higher satisfaction appears to be related to good communication between the health care team and the patient, and to patients being treated as informed partners and being involved in the management of their pain.13, 15 The most frequently reported outcome of cancer pain relief, used mainly in clinical trials of analgesic drugs, is the “clinically significant” reduction in pain intensity or clinical response; in acute cancer pain, it is suggested to be ≥2 points reduction on a 0 to 10 scale, or a 33% overall reduction in pain intensity.21 However, such reductions may not represent meaningful pain relief to the patient. As an example, in a patient reporting severe pain of 9-10, a pain reduction by 2 points would still leave the patient with a pain of 7, which would still be considered as severe pain, and although meaningful, the degree of pain relief may be insufficient for overall patient comfort.
One way of addressing how pain management goals can be set to individualize care is the personal pain goal, which relies on the patients' own criteria for meaningful pain relief. Similar to how the pain intensity is assessed, patients are asked to describe, on a 0 to 10 scale, the level/intensity of pain that will allow the patient to achieve comfort in physical, functional, and psychosocial domains. The personalized pain goal can convey to the medical provider the patients' desire and expectations of what his/her pain should be. The assessment of personalized pain goal along with the pain intensity is routinely done in the outpatient and inpatient settings at our institution.
The main objective of the present study was to examine the feasibility and stability of the self-reported personalized pain goal in cancer patients being treated at the outpatient Supportive Care Clinic. Our secondary objective was to compare pain response as determined by personalized pain goal versus traditional pain response criteria.
MATERIALS AND METHODS
Study Design and Subjects
This study was approved by our institutional review board with waiver of informed consent. Consecutive patients with advanced cancer who were seen for the first time in consultation in the outpatient Supportive Care Clinic over a 15-month period (from June 1, 2008 to August 31, 2009) were identified. In this study, advanced cancer was defined as locally advanced, recurrent, or metastatic disease for solid tumors, and relapsed or refractory disease for hematologic malignancies. Patients who were unable or unwilling to receive curative-intent therapies for any reason or referred for phase 1 treatments were also considered to have advanced cancer. Only those patients who had their first follow-up visit in the clinic within 1 to 6 weeks of initial consultation were included in the analysis. The time period of 1 to 6 weeks was chosen in an effort to evaluate the response to the Supportive Care Clinic intervention with minimal impact from other factors such as increased pain from tumor growth.
The Supportive Care Clinic conducts assessment and management of cancer patients by a sophisticated team of board-certified palliative care physicians, nurses, and other psychosocial disciplines. All patients undergo comprehensive symptom assessments and management during each of their visits. All physicians follow a uniform set of standards for assessment and treatment interventions.
Patient's medical records were reviewed for information pertinent to patient demographics (age at the first consult visit, sex, ethnicity, and marital status); cancer diagnosis (type, location, and stage); dates of advanced cancer diagnosis, first consultation, and follow-up visit; symptoms such as pain, fatigue, nausea, and others (ascertained from the Edmonton Symptom Assessment Scale [ESAS]) and overall symptom burden; personalized pain goal; current use and dosage of opioids (calculation of morphine equivalent daily dose [MEDD]); screening test for history of alcohol dependence (cut down, annoy, guilt, eye-opener [CAGE]); Zubrod performance status; and cognitive status (Memorial Delirium Assessment Scale [MDAS]).
ESAS, CAGE score, Zubrod status, MDAS, and personalized pain goal are standard assessments that are used by our team at all consultations as well as all follow-up visits and are further described. The ESAS is a validated, self-report tool that assesses the intensity of the most common 10 individual symptoms (pain, fatigue, drowsiness, nausea, anxiety, depression, appetite, dyspnea, sense of well-being, and insomnia) on a 0 to 10 numerical scale (0 = best, 10 = worst) in patients with advanced incurable illness.22, 23 Pain and other symptoms on the ESAS were rated based on the average score in the previous 24-hour period. The ESAS has been found to be valid, reliable, and responsive when using this anchoring.24 Overall symptom burden was calculated as the total symptom distress score from summation of the 10 symptom item scores on the ESAS, giving values between 0 and 100.22 Personalized pain goal is defined as “the verbal or written goal stated by the patient describing the desired level/intensity of pain that will allow the patient to achieve comfort in physical, functional, and psychosocial domains.” The personalized pain goal was assessed by the outpatient palliative care nurse. Patients are asked to identify the maximal intensity of pain on a scale of 0 to 10 that would still be considered comfortable. Because of the nature of personalized pain goal, it was asked as a general question rather than an anchoring to any particular period of time.
The CAGE alcohol questionnaire is frequently used as a brief screening tool for alcohol abuse.25 A positive CAGE score of ≥2 has been shown to be of prognostic value in the opioid management of cancer pain.26 At each consultation visit, all patients complete the CAGE questionnaire. At each follow-up visit, the nurse confirms that a CAGE questionnaire has been previously completed. If not, patients are asked to complete questionnaire at the subsequent visits. The Zubrod score assesses patient functionality (performance status) and ranges from 0 to 5 (0 = fully active; 1 = restricted in strenuous activity; 2 = ambulatory, capable of self-care but not work; 3 = bedridden ≥50% of the time, limited self-care; 4 = totally bedridden and disabled, no self-care; and 5 = dead).27 Zubrod scores of 0 to 2 denote good functional status; scores of 3 and 4 denote poor functional status. The Zubrod score is calculated by the nurse, and confirmed by the physician at each visit. The MDAS is a validated tool to quantify the severity of delirium symptoms, and integrates objective cognitive testing and evaluation of behavioral symptoms.28 The MDAS yields a global score ranging from 0 to 30, with a suggested cutoff score of 7 for delirium. The MDAS is completed by the nurse, and confirmed by the physician at each visit.
The data collection team consisted of the principal investigator (S.D.) who is a palliative care specialist, a palliative care specialty fellow (L.N.), 2 research nurses, and the departmental data management team.
Assessment of Pain and Pain Relief
Pain score was obtained from the ESAS. Pain was considered to be mild if pain score was 1 to 4, moderate if 5 to 6, and severe if 7 to 10.29, 30 Pain response was determined at the first follow-up visit via 2 methods: clinical pain response and personalized pain goal response. Patients were considered to have achieved a clinical pain response if there was reduction in pain score by ≥2 points or a reduction by 30% from their baseline pain score.31 Patients were considered to have achieved a personalized pain goal response if their follow-up pain score was less than their personalized pain goal.
Descriptive statistics, including median and interquartile range (IQR), were used to summarize patient demographics, disease-related characteristics, symptom scores, and personalized pain goal. To evaluate the feasibility of personalized pain goal assessment, we assessed the proportion of charts containing a personalized pain goal score at the consultation visit. To assess the stability of the personalized pain goal score, we assessed the proportion of charts at follow-up that had a personalized pain goal score that was equal to or within 1 point of the baseline personalized pain goal score. Frequencies and proportions were used to summarize achievement of personalized pain goal and clinical pain response at follow-up visits. On the basis of the sample size of 300, the 95% confidence interval for the true rate of personalized pain goal achievement would be no wider than ±5.7%.
We used predefined cutpoints to classify patients into 3 pain categories (mild, moderate, or severe) based on their baseline pain scores on the ESAS. Differences between pain category groups were analyzed using the Kruskal-Wallis test for continuous variables and the chi-square test for categorical variables. A P value <.05 was considered statistically significant.
To determine the sensitivity and specificity of the clinical pain response criteria, we used the patient personalized pain goal as the gold standard. A 2 × 2 table was constructed with responses versus nonresponses of personalized pain goal and clinical responses.32 Sensitivity was defined as the number of patients who achieved a clinical response divided by all the patients who achieved a personalized pain goal response. Specificity was defined as the number of patients who did not achieve a clinical response divided by all the patients who did not achieve a personalized pain goal response. Statistical Package for the Social Sciences (version 17; SPSS Inc., Chicago, Ill) was used for statistical analyses.
Of the 944 cancer patients identified to have an outpatient Supportive Care consultation, 465 (49%) patients met the time to follow-up inclusion criteria (1-6 weeks from consultation visit) in the defined 15-month period of study (Fig. 1).
Twenty (4.5%) of the eligible 465 patients had no available personalized pain goal data and were excluded from analysis. The median age of 445 patients was 59 years (range, 16-89 years), and majorities were male (52%), Caucasian (69%), and married (66%), and had advanced cancers (84%). The most common malignancies were gastrointestinal, lung, and head and neck. At baseline, 35 (8%) patients reported no pain. Pain was in the mild category in 152 (34%) patients, moderate in 95 (21%) patient, and severe in 163 (37%) patients. Patient characteristics by pain category are presented in Table 1. There were no significant differences between pain groups with regard to ethnicity, marital status, or type of cancer (P > .5).
Table 1. Selected Baseline and Follow-up Characteristics for all Patients and by Baseline Pain Category
Baseline Pain Category
Abbreviations: CAGE, cut down, annoy, guilt, eye-opener; ESAS, Edmonton Symptom Assessment Scale; ESAS-SDS, sum of individual ESAS scores; IQR, interquartile range; MDAS, Memorial Delirium Assessment Scale; MEDD, morphine equivalent daily dose; PPG, personalized pain goal.
At baseline, patients with severe pain had higher Zubrod scores (median 2 vs 1, P = .001) and were on higher opioid dosage (median MEDD, 60 mg vs 0; P < .001) than all other pain categories. There were more CAGE-positive patients among patients reporting no pain versus those patients with any type of pain (P < .001).
At follow-up (median, 14 days; IQR, 11-25 days), we observed a significant decrease in pain and overall symptom burden. As shown in Table 2, pain decreased from a median of 5 (IQR, 3-8) to 4 (IQR, 2-7; P < .001), and ESAS symptom distress score from a median of 39 (IQR, 29-50) to 34 (IQR, 23-48; P < .001).
Table 2. Epidemiological Performance of Clinical Response When Using the Achievement of Personalized Pain Goal as the Gold Standard for Pain Relief
Baseline Pain Category
Sensitivity/ False-Negative Rate
Specificity/ False-Positive Rate
Pain ≥ 1 (n=410)
Mild pain (n=152)
Moderate pain (n=95)
Severe pain (n=163)
We observed no significant difference in baseline demographics, such as sex (P = .6), age, ethnicity, and advanced cancer status (P > .1), between patients who were not eligible for the current study (Fig. 1, 479 patients) and patients who met the eligibility criteria (data not shown). However, with regard to baseline symptoms, patients who returned for follow-up in <1 week had higher pain (median, 6 vs 5; P < .001) and overall symptom burden (median, 46 vs 38; P < .001), whereas those who came later than 6 weeks for follow-up had generally lower overall symptom distress (P < .01; data not shown).
Personalized Pain Goal
The median baseline personalized pain goal was 3 (IQR, 2-3). Despite significant differences in pain intensity among the 3 baseline pain categories, median personalized pain goal remained consistent at 3 (Table 2). However, patients with no baseline pain reported a small, albeit significantly different baseline personalized pain goal (median personalized pain goal, 2; IQR, 0-3; P < .001) as compared with patients with any pain. Baseline personalized pain goal value was 0 in only 8 of 410 patients (2%) with pain ≥1, and in 16 of 35 (46%) patients with no baseline pain (P < .001). Baseline personalized pain goal was 1 in 23 (5%), 2 in 111 (25 %), 3 in 201 (45%), and ≥4 in 66 (15%) patients overall.
At follow-up visit, the overall median personalized pain goal remained unchanged (P = .57). Figure 2 illustrates that for the vast majority of patients, personalized pain goal at follow-up was quite stable, either showing no variation (52% patients) or being within ±1 baseline personalized pain goal value in 85% patients. The Spearman correlation between baseline and follow-up personalized pain goal was 0.469 (P < .001).
Assessment of Pain Response
Pain response was assessed as the proportion of patients who achieved a personalized pain goal response and clinical pain response. Figure 3 illustrates the breakdown of pain responses by baseline pain category and for all patients. Overall, personalized pain goal response and clinical response were achieved in 207 (47%) and 184 (41 %) patients, respectively. Figure 3 clearly shows that clinical response was significantly higher among patients with baseline severe pain (60%), as compared with patients who had moderate (40%) and mild (33%) pain (P < .001). Conversely, personalized pain goal response showed the opposite results, and was achieved more often among patients with mild pain (63%) as compared with patients with moderate (44%) and severe pain (27%) (P < .001). One area where clinical response and personalized pain goal response appeared to behave similarly was among those patients with moderate pain (40% and 44%, respectively). Table 2 demonstrates the epidemiological performance of clinical response, when using the achievement of personalized pain goal as the gold standard for pain relief. The sensitivity of clinical response was highest (98%) among patients with severe baseline pain, but in these patients its specificity was lowest (54%), resulting in a large number of false-positive responses (46%). In patients with mild pain, the opposite was true. Among these patients, clinical response was quite specific for pain relief (98%), but had low sensitivity (52%), resulting in a large number of false negatives (48%). Among patients with moderate pain, sensitivity and specificity were 84% and 94%, respectively.
Pain control is vital to the quality of life of cancer patients, and is a major outcome of QI initiatives. In the past few decades, several professional organizations have advocated for cancer pain relief by calling for improvements in pain education, health care policies, and institutional commitment to pain-related QI initiatives.9, 12, 33-43 An important part of current recommendations is the frequent assessment of pain intensity, prompt interventions for pain relief, and reassessment of the patient's response to such treatments.9, 44-46 QI initiatives thus far have predominantly been focused on documentation of pain assessment at frequent intervals, with much less emphasis on individual pain outcomes such as effectiveness of prescribed medications. This may be in part because as of yet methods to best determine the effectiveness of cancer pain management are still lacking, and what constitutes meaningful pain relief to an individual patient is not clear. As pain is inherently subjective and personal, greater patient involvement in decision making about pain management options is now increasingly advocated, with recommendations that pain treatments and goals be tailored to the needs, desires, and circumstances of individual patients.9, 11 One method adopted by our institution recently is to use the patients' own criteria for meaningful pain relief, the personalized pain goal, which is included as part of the comprehensive pain assessment in the outpatient and inpatient settings. A pain score that is equal to or less than the personalized pain goal score could immediately relay to the provider that pain is adequately controlled to the patient's satisfaction. The present study was conducted to examine: 1) the feasibility and stability of personalized pain goal; and 2) to compare personalized pain goal response to the traditional clinical pain response criteria.
We found that the vast majority (445 of 465 patients; 95%) of eligible cancer patients seen in our outpatient Supportive Care Clinic had documentation of personalized pain goal scores. This suggests that this simple assessment is feasible in a busy outpatient setting, and that patients are capable of stating their desired value (personalized pain goal) for pain relief on a 0 to 10 scale, and that the majority considered a value of 3 to be very reasonable, which is consistent with literature suggesting that pain in the mild category is an achievable patient outcome.6, 7 The median personalized pain goal score of 3 was consistent among the 3 pain groups, suggesting that patients do not decide their personalized pain goal based on their pain intensity at a given time. Furthermore, at the follow-up visit, patient's personalized pain goal score either remained the same or was within 1 point in 52% and 85% of patients, respectively, thereby suggesting that individual patients are able to be consistent in their desired goal for pain relief. Because the use of the scale is subjective, there were some patients who had extreme variation depending on how they interpreted their personalized pain goal. For instance, 30% of patients had personalized pain goal scores of ≤2, and 15% of patients had scores of ≥4. This suggests that rather than using 3 as a gold standard for all patients, an individualized pain goal (the personalized pain goal) is more appropriate for the treatment plan and for quality assurance. Of surprise to us was that only a small minority (2%) of patients with pain reported a personalized pain goal value of 0. This may suggest that cancer patients with pain are realistic regarding what can be achieved in the management of their pain, and that complete elimination of their pain is very unlikely.
When we compared the 2 pain responses, we found that the overall rate of personalized pain goal response and clinical response criteria were not different. However, as demonstrated in Figure 3, we observed significant differences between the 3 pain categories. Patients with severe pain at baseline were more likely to achieve a clinical response (60% of patients), as compared with patients with moderate (40%) or mild (33%) pain. Conversely, patients with severe pain were less likely to achieve a personalized pain goal response (27%), as compared with patients with moderate (44%) or mild pain (63%). Our findings with personalized pain goal response by pain category are also much more consistent with what would be expected in the clinical care of these patients. Patients with mild to moderate pain would be expected to achieve a more acceptable level of analgesia than those patients who complain of moderate to severe pain, and in this population the use of clinical response criteria resulted in appropriate results. When using achievement of personalized pain goal as the gold standard for pain relief, the sensitivity of clinical response was highest (98%) among patients with severe baseline pain, but in these patients its specificity was lowest (54%), resulting in a large number of false-positive responses (46%). Because patient's overall impression of the estimation of their pain control as compared with their expectations is the proposed goal for clinical care and quality assurance, our results suggest that using clinical response criteria for pain relief has variable sensitivity and specificity, and comparing pain intensity over time with the patient's personalized pain goal may be the more useful and simple outcome. Thus, the personalized pain goal may be the way to operationalize the recommended outcome suggested by the American Pain Society and the National Comprehensive Cancer Network for cancer pain management. Studies have shown that higher satisfaction in pain management appeared to be related to good communication between the health care team and the patient, and to patients being treated as informed partners and involved in the management of their pain.13, 15
Although clinical response might be useful in clinical trials comparing different analgesic treatments, the inclusion of personalized pain goal response might be more useful in determining which patients have achieved satisfactory pain control. Our results reflect the importance of obtaining a personalized pain goal, because a drop in pain intensity by 2 points is much easier to achieve in patients with severe pain, and that can falsely suggest that these patients achieved acceptable analgesia. For example, in our cohort, follow-up pain improved clinically and statistically (from median pain score of 8 to 6, P < .001); however, according to the personalized pain goal response criteria, the majority (53%) of our patients did not achieve their personalized pain goal. These findings suggest that although the interventions at our clinic successfully reduced pain by the clinical response criteria in the first follow-up visit, they were not lowered enough to the personalized pain goal level for a majority of patients. This suggests that perhaps other interventions such as earlier follow-up intervals, phone calls, or aggressive pharmacological treatment at the first consult might be more appropriate for some patients. The presence of several factors such as neuropathic pain, breakthrough pain, pain intensity, history of chemical coping or addiction, psychological distress, and cognitive failure have been shown to be associated with the complexity of cancer pain management.6, 26, 47-50 Most of these risk factors have been incorporated into the recently developed and validated cancer pain classification system, the Edmonton Classification System for Cancer Pain,51 and they also bring to light that cancer patients may have diverse interpretations of their pain experience, and an individualized approach to pain management and assessment of pain relief should be implemented. We plan to conduct further research into identifying factors associated with nonachievement of personalized pain goal response. In our study, the overall frequency of CAGE positivity was 11%. Previous studies by our group and others have shown frequency of CAGE positivity in cancer patients referred to a palliative care service to range from 11% to 38%.26, 52-55 Most of these studies were conducted in patients who did report cancer pain, and CAGE positivity was associated with higher levels of pain expression, opioid use, or longer time to achieve stable pain control. Our results regarding higher frequency of CAGE positivity in patients with no reported pain are surprising and may be explained by the very small number of patients in this category. More research is necessary to better characterize the role of CAGE status on both personalized pain goal and pain intensity reporting.
Our study has several limitations, including issues related to the retrospective design such as missing data. Our eligibility criteria restricted our analysis to only those patients who had their first follow-up visit in the clinic within 1 to 6 weeks of initial consultation, thereby excluding approximately half the patients who were initially seen in outpatient consultation. However, this time frame was chosen in an effort to evaluate the response to the Supportive Care Clinic interventions. Patients returning <1 week after initial visit have minimal time to allow the recommended interventions to be fully effective. Conversely, patients returning after >6 weeks had a higher risk of complications related to tumor progression that may not have been present during the initial visit. Furthermore, our study was conducted in a highly specialized comprehensive cancer center, and the type of care delivered in the supportive care clinic may have affected the response (eg, pharmacological management, psychosocial management of other symptoms such as depression, anxiety, and fatigue, or family care). Further studies are required to better characterize the multiple different interventions on both the personalized pain goal and pain intensity reporting, and to determine whether our findings are generalizable to other settings.
Our preliminary study suggests that the vast majority of cancer patients with pain were capable of stating their desired level for pain relief on a 0 to 10 scale. The median personalized pain goal was 3, and was highly stable over our follow-up period. Personalized pain goal represents a novel target for pain response. Regular documentation of personalized pain goal can potentially facilitate personalized pain management.
E.B. is supported in part by National Institutes of Health grants RO1NR010162-01A1, RO1CA122292-01, and RO1CA124481-01.