Volume 39, Issue 3 , Pages 548-554, March 2010
Lack of Benefit From Paracetamol (Acetaminophen) for Palliative Cancer Patients Requiring High-Dose Strong Opioids: A Randomized, Double-Blind, Placebo-Controlled, Crossover Trial
Article Outline
Abstract
Context
The adjunctive use of paracetamol (acetaminophen) with strong opioids has become entrenched practice in palliative care pain management, despite little evidence to support its use.
Objective
The study aim was to investigate potential analgesic benefits of 4
g of paracetamol daily for palliative cancer patients requiring high-dose opioids.
Methods
Thirty-one patients, using at least 200mg of oral morphine equivalent daily, were recruited to a prospective, double-blinded, randomized, crossover trial. Patients received usual medications plus 4g of paracetamol or placebo for five days each in random order. Primary outcome, effect on pain, was assessed using daily diaries, including a numerical rating scale (NRS) from zero (no pain) to 10 (unbearable) and recording numbers of breakthrough analgesics. Secondary outcomes—nausea, vomiting, cognitive impairment, constipation, and overall well-being—were assessed using the NRS. Data from the last four days of each treatment were analyzed. Patients also indicated in which part of the study their pain was better controlled.
Results
Twenty-two patients, requiring a median dose of 255mg of oral morphine equivalent daily, completed the trial. There were no significant order or treatment-by-order interaction effects for any variable; paired t-tests were conducted to investigate change in mean levels on outcome variables with placebo vs. paracetamol. For none of the variables was there a statistically significant difference when assessed with placebo compared with paracetamol. No change approached clinically significant levels, with a mean difference in rated pain of 0.16, and mean difference of 0.42 for a number of breakthrough medications. Fifteen patients were undecided whether paracetamol improved pain.
Conclusions
These data do not support the common practice of adding regular paracetamol daily as an adjunct to high-dose opioids for pain control in cancer patients receiving palliative care.
Key Words: Pain, paracetamol, acetaminophen, high-dose opioids, strong opioids, cancer, palliative care
Introduction
The management of pain in cancer patients receiving palliative care remains challenging.1 The World Health Organization (WHO) developed a three-step approach to pain that recommends nonopioid analgesics, such as paracetamol (acetaminophen) for mild pain, with the addition of a so-called weak opioid if pain persists or increases, and the substitution of the “weak” opioid with a so-called strong opioid, such as fentanyl, hydromorphone, methadone, morphine, or oxycodone, for severe pain.2 The use of paracetamol as an adjunct to strong opioids is optional at Step 3 of the ladder but has become entrenched in palliative care management.3 This is despite little evidence of any demonstrable additional analgesic benefits or the postulated opioid-sparing effects of this practice.4 Given the additional tablet burden and compliance issues associated with regular dosing of paracetamol, evidence is warranted either to support this practice or to trigger its cessation.
The rationale for using combinations of analgesic drugs with different pharmacologic properties is the potential for more effective analgesia coupled with an improved side-effect profile. There is a growing body of evidence confirming the usefulness of paracetamol as a coanalgesic to “weak” opioids, thus supporting Step 2 of the analgesic ladder.5, 6 The evidence at Step 3 of the ladder is scant and problematic. There have been two recent randomized controlled studies7, 8 investigating benefits from the addition of paracetamol to Step 3 in the cancer population receiving palliative care. Methodologically, both studies were well designed, taking into consideration many of the problems inherent in palliative care research, such as the need for short study times, easy assessments, and close patient follow-up.9 Results from the earlier study,7 based on measures of pain ratings and quality-of-life questions, did not support the routine use of paracetamol daily as an adjuvant to morphine therapy for pain control. Conversely, the later study8 found modest average effectiveness with the addition of paracetamol, and it was concluded that paracetamol improved pain and well-being without major side effects in patients with cancer and persistent pain despite a “strong” opioid regimen. The authors recommended its use in all such patients.
Disparity in the findings may have arisen because neither study stratified outcomes according to oral morphine equivalent dose; the dose range for the two studies was 20–2100mg of oral morphine equivalents daily. The benefits of paracetamol at Step 3 of the pain ladder may be dependent on the dose of “strong” opioid. A first step in clarifying this issue is to determine whether paracetamol is of benefit to those requiring higher doses of “strong” opioids. If so, this would add support to the current WHO analgesic recommendations. If not, it could trigger an investigation into the paracetamol or opioid dose relationship and perhaps lead to appropriate modification of the recommendations of the ladder.
This trial further investigates the potential benefits of using paracetamol as an adjuvant for pain control at Step 3 of the analgesic ladder, specifically focusing on cancer patients receiving palliative care for pain and requiring high-dose strong opioids. The primary objective was to assess whether, in patients receiving the equivalent of at least 200mg of oral morphine daily, the addition of 4g of oral paracetamol daily improves pain, as measured by subjective pain ratings and the objective number of treatments for breakthrough pain required.
Patients and Methods
Approval for the study was obtained from two institutional Human Research Ethics Committees and all participants signed informed consent. Enrollment spanned an 18-month period.
The study was a prospective, double-blind, randomized, crossover trial targeting cancer patients receiving palliative care and requiring high-dose “strong” opioids, defined as the equivalent of at least 200mg of oral morphine daily. In addition to their usual medications, all patients received five days of paracetamol, two 500mg tablets per oral four times daily, and five days of an identical-appearing placebo. Randomization of the order of active and placebo agents was controlled by a computer-generated randomization schedule determined by a research pharmacist at a distant site.
The primary outcome, effect on pain, was assessed subjectively by asking patients, “Please rate your overall level of pain today.” Patients completed a daily diary of numerical ratings of pain on an 11-point scale, with values ranging from zero (none) to 10 (unbearable), for each of the five days of both treatments. In addition, the number of breakthrough analgesic medications used each day was recorded. Secondary outcomes of subjective ratings of nausea and vomiting, cognitive impairment (drowsiness and unclear thinking), constipation, and overall feeling of well-being were also assessed using an 11-point numerical rating scale (NRS). Only data from the last four days of each treatment were analyzed to exclude the influence of possible carryover effects. At the completion of the study, patients were asked whether, overall, they felt their pain was better controlled in the first or second part of the study or whether they were undecided.
Ambulatory patients, both inpatient and community-based, were recruited from the Brisbane South Palliative Care Service and Mt. Olivet Palliative Care Service in Brisbane, Australia.
Specific inclusion criteria included the following:g of oral paracetamol daily
Specific exclusion criteria included the following:
After informed consent was obtained, baseline data, including pain ratings, demographics, medical and surgical history, pain history, analgesic therapy history, current medication, any known allergies, and MMSE scores, were documented. All medications containing paracetamol were ceased. Patients were then randomly allocated to receive either five days of oral paracetamol followed by five days of an identical appearing placebo or vice versa.
Patients were asked to complete a daily diary for 10 days that included subjective measures to monitor pain, nausea and vomiting, cognitive impairment, and feeling of overall well-being. Each breakthrough pain medication taken was recorded.
Community-based patients were visited by a research nurse, uninvolved with their clinical care, at least twice during the study and telephoned every second day. Any changes in condition or medications were recorded. Inpatients were visited daily. Any patient who became febrile was withdrawn from the study.
The outcome variable of primary interest was change in rated pain. To provide a more stable estimate of pain in each treatment condition, mean pain ratings over the last four days of each treatment were computed for each participant. A priori power analysis was carried out on the assumption that a difference of one point between the means with paracetamol and placebo on the NRS could be regarded as a clinically significant difference. Although there is no objective criterion for what constitutes a clinically significant difference in rated pain, empirical evidence with cancer patients10 and emergency department patients11 has suggested that changes of 2 or 1.39, respectively on an NRS may be clinically important differences. For the present study, we required sufficient power to detect a change of at least one unit, assuming a standard deviation of 1.5 in change scores. This yields a standardized mean difference, d, of 0.67, which represents a medium-large effect on the scale-free measure of effect size,12 which is the basis of power analysis.13 At the 0.05 significance level in a nondirectional test, a sample of 20 participants completing the study would result in power of at least 80% to detect an effect of this size using a paired t-test. Given the fact that attrition rates tend to be very high in research within the palliative care population,14 power was calculated for the number of participants completing the study rather than the number actually recruited.
Analysis of variance with order as a between-subjects variable and treatment (placebo vs. paracetamol) as a within-subjects variable was conducted to test for order effects and treatment by order interactions. If no order or treatment by order interaction effects were found, the data were collapsed over order and paired t-tests were conducted to test for mean differences between treatment conditions. This procedure was followed for pain and breakthrough pain medications as well as the other outcomes of interest.
Results
Figure 1 illustrates patient progress through the study. Thirty-one patients, 26 of whom were prescribed regular paracetamol before commencement of the study, were randomized and 22 completed the study; reasons for study withdrawal included fever (4), protocol violation (3), development of unstable pain (1), and unexpected rapid deterioration in condition so that carers considered it burdensome for the patient to continue (1). Eight of the nine withdrawals had been prescribed regular paracetamol and seven withdrew in the placebo phase. Patients who withdrew did not differ significantly from those who completed in terms of age, sex, or daily oral morphine equivalent dose. Withdrawal was not systematically related to treatment or order of treatments. The lowest MMSE score achieved by participants who completed the trial was 26 out of 30.
Fig. 1
Flow of participants through the study.
Table 1 summarizes the characteristics of those who completed the study. All patients had a primary diagnosis of cancer, the average age was 56.3 years and 12 were male. All patients, except for one, were community based. Before commencing the trial, 18 were prescribed 4g of oral paracetamol daily. The median baseline daily opioid dose was equivalent to 255mg of oral morphine, with a range of 200–9000mg. The extreme value of 9000mg of oral morphine equivalent was for a patient who had been fitted with a Medtronic intraspinal infusion device; the next highest dose was 900mg.
Table 1. Characteristics of Patients (n=22) Who Completed the Trial
| Characteristic | n |
|---|---|
| Sex (male/female) | 12/10 |
| Average age in years (range) | 56.3 (28–79) |
| Environment of care (inpatient/home) | 1/21 |
| Primary diagnosis | |
| Colorectal cancer | 4 |
| Pancreatic cancer | 3 |
| Prostate cancer | 3 |
| Breast cancer | 2 |
| Unknown primary | 2 |
| Melanoma | 2 |
| Esophageal cancer | 1 |
| Ewing's sarcoma | 1 |
| Hodgkin's lymphoma | 1 |
| Lung cancer | 1 |
| Malignant fibrous histiocytoma | 1 |
| Renal cell carcinoma | 1 |
| Prescribed regular paracetamol (4g daily) before commencement of trial (yes/no) | 18/4 |
| Baseline daily opioid dose expressed as oral morphine equivalents in milligrams—median (range) | 255 (200–9000) |
Means for each outcome in the placebo and the paracetamol conditions, with relevant information on changes across treatments, are shown in Table 2. Analysis of variance indicated no significant order or treatment-by-order interaction effects for any of the outcome variables, so paired t-tests were conducted to investigate evidence of change in mean levels on the outcome variables with placebo vs. paracetamol. For none of the variables was there a statistically significant difference when the variable was assessed with placebo compared with paracetamol. Additionally, none of the changes approached the level deemed to be clinically significant, with the mean difference in rated pain, for example being 0.16, and the mean difference in number of breakthrough medications required being 0.42. Figure 2 illustrates the uniformly small mean differences (and 95% confidence intervals) for the outcome variables when the mean in the placebo condition is compared with the mean in the paracetamol condition.
Table 2. Mean Scores Across Four Days of Intervention with Placebo and Paracetamol and Information About Changes (Placebo—Paracetamol) Across Interventions
| Treatment | Placebo Mean (SD) | Paracetamol Mean (SD) | Difference Mean (SD) | 95% CI | da | Pb |
|---|---|---|---|---|---|---|
| Rated pain | 3.59 (1.58) | 3.43 (1.44) | 0.16 (1.42) | −0.47 to 0.79 | 0.11 | 0.60 |
| Breakthroughs | 1.41 (1.45) | 0.99 (1.26) | 0.42 (1.02) | −0.03 to 0.87 | 0.41 | 0.07 |
| Well-being | 5.26 (1.40) | 5.46 (1.63) | −0.21 (1.42) | −0.83 to 0.42 | 0.15 | 0.50 |
| Nausea | 0.86 (1.46) | 0.83 (1.22) | 0.03 (1.48) | −0.62 to 0.69 | 0.02 | 0.92 |
| Drowsiness | 2.31 (2.19) | 1.88 (1.93) | 0.43 (1.30) | −0.15 to 1.01 | 0.33 | 0.14 |
| Unclear thinking | 1.06 (1.38) | 0.94 (1.09) | 0.12 (0.85) | −0.26 to 0.50 | 0.14 | 0.52 |
| Constipation | 1.05 (1.68) | 0.81 (1.23) | 0.24 (1.29) | −0.34 to 0.81 | 0.17 | 0.40 |
ad |
bP-value from paired t-test. |
Fig. 2
Mean change (placebo—paracetamol) and 95% confidence interval for outcomes.
Given the observed, albeit small, trends toward higher pain ratings combined with more breakthrough medications in placebo compared with paracetamol, a multivariate repeated measures analysis of variance was conducted, but the observed trends across these two outcome variables did not approach statistical significance (Pillai's trace=0.15, F=1.81, degrees of freedom=2, 20, P=0.19). Clinically, an increase in the background dose of an opioid is unlikely to be considered necessary unless the number of daily breakthrough doses is at least two to three for a few days, and the observed means do not approach this level.
At the completion of the trial, when patients were asked whether they felt their pain was better controlled in the first or second part of the study, 15 patients reported that they were unable to decide, six chose the part of the study when they were using paracetamol, and one chose the placebo part. Of the four patients who were not prescribed 4g of oral paracetamol daily before commencing the trial, two chose paracetamol and two were undecided.
Discussion
The data presented in this article do not support the common palliative care practice of using 4g of oral paracetamol daily as an adjunct to high-dose “strong” opioids for pain control in cancer patients receiving palliative care. A limitation of the study was the small sample size; nonetheless, it was adequately powered and multiple outcome measures were considered. None of the measures, including subjective pain ratings, objective scores of numbers of analgesic breakthroughs required, or outcomes related to possible opioid sparing effects, such as drowsiness, nausea, constipation, and unclear thinking, favored the use of paracetamol. Overall well-being was also not significantly affected by the use of paracetamol. A strength of these findings was that the measures were averaged over four days to enhance stability.
Additionally, most of the patients were undecided as to whether their pain was better controlled with or without paracetamol. This is not surprising, as there was only a mean difference of 0.16 in rated pain between the placebo and paracetamol phases, well below the difference of one point conservatively chosen to be the clinically important difference in pain ratings. The mean difference of 0.42 in the number of breakthrough medications favors the use of paracetamol, but the result is neither statistically nor clinically significant, and there were no differences in opioid side-effect measures.
Patient withdrawals from the study require discussion, as seven of the nine occurred during the placebo phase. Four of these were consequent to patients becoming febrile, necessitating unblinding and subsequent administration of paracetamol. Interestingly, all four of these patients had previously been prescribed paracetamol; perhaps the use of regular paracetamol in cancer patients serves to decrease the emergence of fever. Two patients were withdrawn during the placebo phase because they failed to take adequate study medications as directed, in spite of close monitoring by the research nurse. The final withdrawal in the placebo phase was precipitated by family insistence when the patient's condition deteriorated unexpectedly. Six placebo withdrawals (five of whom were previously prescribed regular paracetamol) took place when placebo was assigned first; the authors are unable to explain this occurrence. On initial consideration, it may suggest an effect because of withdrawal of paracetamol but this seems unlikely given that the effect was not replicated when placebo was assigned second. In the paracetamol phase, one patient was withdrawn because the pain became unstable and the background opioid needed considerable adjustment, and the other because the patient was hospitalized and left the study medications at home.
Our findings support those reported by Axelsson and Christensen,7 using a crossover design with 30 patients. These authors questioned whether the routine use of paracetamol with concurrent strong opioids is clinically sound and suggested that because the daily intake of eight large paracetamol tablets inflicts a substantial burden to some patients with advanced cancer, a less dogmatic attitude should prevail. These two studies are at odds with the findings of Stockler et al.,8 also using a crossover design with 30 patients, who reported modest effectiveness of paracetamol in conjunction with strong opioids. In that trial, the significant result in favor of adding paracetamol was a mean difference between two single assessment days (paracetamol plus opioid vs. placebo plus opioid) of 0.4 on an 11-point verbal numeric scale of pain (P = 0.03). The clinical importance of this result is unclear, particularly as no other pain endpoint (visual analog scale for pain, patient preference to continue or cease paracetamol, and the number of breakthrough medications) differed significantly across paracetamol and placebo conditions. Stockler et al. concluded that people can try paracetamol and decide whether the degree of analgesia justifies the number of tablets.
Our study is the first to restrict the patient population to those requiring high-dose “strong opioids” and finds no evidence to support the use of adjuvant paracetamol. We would suggest that for this group of patients, clinicians consider ceasing paracetamol, thus decreasing the burdensome tablet load for this vulnerable population. This view is reiterated in a recent publication.3
Although we have no evidence to support the common palliative care practice of using 4g of oral paracetamol daily as an adjunct to high-dose “strong” opioids for pain control in cancer patients receiving palliative care, patients requiring lower doses of opioids may well benefit. It would be useful to repeat this study with similar patients requiring lower doses.
Acknowledgments
The authors thank Prof. Janet Hardy for her useful comments during preliminary discussions related to this research. Particular thanks to Ben Sankey for his professional and personal input during recruitment and data collection.
References
- . Cancer pain relief. Geneva, Switzerland: World Health Organization; 1986;
- . Palliative cancer pain: Pain relief and management of other symptoms. Geneva, Switzerland: World Health Organization Publications Centre; 1992;
- . Analgesic effect of paracetamol on cancer related pain in concurrent strong opioid therapy. A prospective clinical study. Acta Oncol. 2008;47:891–895
- . NSAIDS or paracetamol, alone or combined with opioids, for cancer pain. Cochrane Database Syst Rev. 2005;(2):
- Supra-additive effects of tramadol and acetaminophen in a human pain model. Pain. 2008;136:262–270
- . A double-blind placebo-controlled comparison of tramadol/acetaminophen and tramadol in patients with postoperative dental pain. Pain. 2004;109:250–257
- . Is there an additive analgesic effect of paracetamol at step 3? A double-blind randomized controlled study. J Palliat Med. 2003;18:724–725
- . Acetaminophen (paracetamol) improves pain and well-being in people with advanced cancer already receiving a strong opioid regimen: a randomised, double-blind, placebo-controlled cross-over trial. J Clin Oncol. 2004;22:3389–3394
- . Acetaminophen in cancer pain. J Clin Oncol. 2005;23:1586
- . Clinically important changes in acute pain outcome measures: a validation study. J Pain Symptom Manage. 2003;25:406–411
- . The minimum clinically significant difference in patient-assigned numeric scores for pain. Am J Emerg Med. 2005;23:828–832
- . A visitor's guide to effect sizes—Statistical significance versus practical (clinical) importance of research findings. Adv Health Sci Educ Theory Pract. 2004;9:241–249
- . Practical statistics for medical research. Boca Raton, FL: Chapman & Hall/CRC; 1991;
- . Principles of designing clinical trials in palliative care. In: Addington-Hall JM, Bruera E, Higginson I, Payne S editor. Research methods in palliative care. Oxford, UK: Oxford University Press; 2007;p. 13–26
This research was supported by the Brisbane South Palliative Care Collaborative, a joint partnership between Queensland Health, Metro South Health Service District, Griffith University School of Medicine, and Mt. Olivet Community Services.
PII: S0885-3924(09)01138-5
doi:10.1016/j.jpainsymman.2009.07.008
Crown Copyright © 2010. Published by Elsevier Inc. All rights reserved.
Volume 39, Issue 3 , Pages 548-554, March 2010
