Singh, Sudha I. MD, FRCPC*; Rehou, Sarah BSc (Hons)†; Marmai, Kristine L. MD, FRCPC‡; Jones, and Philip M. MD, MSc, FRCPC§
Pain control after cesarean delivery presents unique demands compared with other surgeries, because women require a rapid recovery to ambulate and care for their babies. Pain relief must be rapid and effective, with minimal adverse effects for both the mother and her baby.
Epidural morphine is an acknowledged standard for postcesarean delivery pain relief.1,2 A single dose of epidural morphine provides superior analgesia compared with parenteral opioids.1 Nevertheless, the optimal dose of epidural morphine that maximizes analgesia while minimizing adverse effects is unknown. The side effects of epidural morphine include pruritus, nausea, sedation, and respiratory depression.1
A recent systematic review1 recommended 4 mg epidural morphine for good analgesia after cesarean delivery with an acceptable side effect profile. The studies included in this review differed from contemporary practice in that epidural morphine was not administered as part of a multimodal analgesia regimen, and most of the patients were having elective cesarean deliveries and had not labored. In current practice, multimodal analgesic therapy is recommended to provide effective pain control while reducing adverse effects. For pain relief after cesarean delivery, in addition to a neuraxial opioid, acetaminophen and nonsteroidal anti-inflammatory drugs (NSAIDs) are recommended.2 In addition, the majority of women in North America receiving epidural anesthesia for cesarean delivery have undergone a trial of labor for some period of time with epidural catheters in place for labor analgesia.3
At our institution, the standard postcesarean analgesic regimen for a patient with an epidural catheter in situ includes 3 mg epidural morphine.4 In addition, patients receive regularly scheduled doses of acetaminophen and ketorolac. Oral opioids are administered for breakthrough pain. Bothersome pruritus is a common side effect in up to 40% of our patients.
It is possible that with the regular use of systemic analgesics, a lower dose of epidural morphine may provide analgesic efficacy which is at least equal to that of 3 mg epidural morphine while also reducing the adverse effects associated with 3 mg epidural morphine. Therefore, we conducted a randomized, double-blind, noninferiority trial to test the hypothesis that 1.5 mg epidural morphine is not inferior to 3 mg for analgesia postcesarean delivery. We additionally hypothesized that adverse effects would be less frequent in the 1.5 mg group compared with the 3 mg group. The primary objective of this study was to evaluate the consumption of opioid analgesics in the first 24 hours in parturients receiving 2 different doses of epidural morphine in a setting reflective of current practice, using regular systemic analgesics in women undergoing intrapartum cesarean delivery.
The study protocol was approved by the Health Sciences Research Ethics Board at the University of Western Ontario. After giving written informed consent, eligible study participants were enrolled at a tertiary care obstetric academic hospital in London, Ontario, Canada. Parturients were eligible if they were at least 18 years old, ASA physical status I to III, at term gestation, were laboring with epidural analgesia, and required intrapartum cesarean delivery under epidural anesthesia. Patients were excluded if they had combined spinal–epidural anesthesia, if they were allergic to any of the study medications, or if they had a history of long-term opioid consumption, substance abuse, or chronic pain. Women were informed of the study when epidural labor analgesia was placed. If patients then subsequently required cesarean delivery, they were assessed for eligibility to be enrolled in the study by a member of the research group. Participant demographics were recorded after enrollment but before randomization. Group allocation was performed in the operating room, after surgery had commenced, by opening the next sequentially numbered, sealed, opaque envelope containing a computer-generated random code specifying the group assignment. The randomization sequence was generated using the ralloc program5 in Stata 11.0 for Mac OS X (StataCorp LP, College Station, TX) using randomly permuted blocks to ensure equal group sizes.
In the operating room, parturients were placed supine with left lateral tilt. Standard monitors included noninvasive arterial blood pressure, 3-lead electrocardiogram, and pulse oximeter oxygen saturation (SpO2); the fetal heart rate was checked. An anesthesiologist not involved in data collection extended epidural anesthesia using 2% lidocaine and epidural fentanyl 50 µg until a T4 sensory level to ice was attained. IV ondansetron 4 mg was given intraoperatively for nausea and vomiting prophylaxis. At skin closure, the traditional dose group received 3 mg epidural morphine, while the half-dose treatment group received 1.5 mg epidural morphine. Both doses were diluted with normal saline to a total volume of 5 mL. The epidural morphine doses were given by an anesthesiologist who was not involved in any subsequent outcome assessments. To ensure blinding, the patient and staff caring for the patient were not told the dose of epidural morphine given. Vasopressors, IV fluids, and anesthesia supplementation were given at the discretion of the attending anesthesiologist. Patients who had inadequate anesthesia intraoperatively and received supplementation other than IV ketorolac were excluded from data analysis. The epidural catheter was removed before transfer to postanesthesia care unit as per standard practice.
All patients received institutional standard of care with multimodal analgesia and rescue medications. Ketorolac 30 mg IV was given intraoperatively and every 6 hours after the first dose for the first 24 hours. Patients were given acetaminophen 975 mg by mouth on arrival in the postanesthesia care unit and every 6 hours after that for the first 24 hours. Patients were instructed to ask for rescue analgesia if they experienced inadequate pain relief. Rescue analgesia (oral oxycodone 5–10 mg) was administered every 4 hours as required. On request, patients were initially given oxycodone 5 mg. If their pain was not relieved after 30 minutes, then another 5 mg oxycodone was administered. For treatment of nausea or vomiting, ondansetron 4 mg IV was administered every 6 hours and metoclopramide 10 mg IV was administered every 8 hours as required. For moderate pruritus, diphenhydramine 25 to 50 mg IV was administered every 4 hours as needed. For severe pruritus, naloxone 0.1 mg was administered subcutaneously every 1 hour as required.
A research team member, unaware of group allocation, performed all data collection. Patient assessments were conducted during the usual time patients were required to stay in hospital for recovery. If a patient was discharged before the 48-hour study period was over, assessments occurred by telephone. Follow-up also occurred by telephone at 12 weeks to assess average pain scores.
The primary outcome was the total opioid consumption in IV morphine equivalents in the first 24 hours. Oral oxycodone was the only rescue analgesia administered, and these doses of oral oxycodone were converted to IV morphine equivalents in milligrams using a conversion ratio of 20 mg oral oxycodone being equivalent to 10 mg IV morphine.6 The time to the first request for opioid analgesia was also recorded.
Pain scores recorded the patient’s level of pain at rest and during movement (from lying to sitting) using an 11-point numerical rating scale (NRS; 0 = no pain to 10 = the worst pain imaginable) at 6, 12, 24, 36, and 48 hours.7 At 24 and 48 hours, the patient was asked about her overall pain score using the NRS, her overall pain relief using a 5-point ordinal verbal rating scale (0 = none, 1 = slight, 2 = moderate, 3 = good, 4 = complete),8 her satisfaction with pain management using a 5-point Likert scale (0 = very unsatisfied, 1 = unsatisfied, 2 = neutral, 3 = satisfied, 4 = very satisfied), and her quality of recovery (QOR) was assessed with a modified 9-item questionnaire.7,9
The incidence and severity of adverse effects were evaluated for pruritus, nausea, and sedation. Pruritus and nausea were rated using a 4-point ordinal verbal rating scale (0 = none, 1 = mild, 2 = moderate, 3 = severe).7 Sedation was assessed using a 5-point ordinal scale (0 = alert, 1 = occasionally drowsy, 2 = frequently drowsy, 3 = sleepy but easy to arouse, 4 = difficult to arose).10 All antiemetic (ondansetron and metoclopramide) and antipruritic (diphenhydramine and naloxone) consumption was recorded. Nursing notes that used the same scales were reviewed to fill in missing data when necessary. Nurses were blinded to treatment allocation.
The primary outcome of this trial was the 24-hour postoperative opioid consumption: we hypothesized that there would not be a clinically meaningful difference between the 2 groups. Therefore, we planned this trial with a noninferiority design. We considered that up to a one-third increase in opioid consumption postcesarean delivery would not be clinically important. Therefore, the noninferiority margin was chosen based on a one-third (33%) increase from values reported in a study,11 where it was shown that the 24-hour opioid consumption after epidural morphine in parturients after cesarean delivery was equal to 10 mg IV morphine equivalents (which is consistent with our clinical experience). We planned a priori to declare noninferiority of the 1.5 mg group with respect to the 3 mg group if the upper bound of the 1-sided 95% confidence interval (CI) of the difference in medians of the 24-hour opioid consumption between groups (1.5 mg group − the 3 mg group) was <3.33 mg. Because of the anticipated right-skewed distribution of the median 24-hour opioid consumption, CI construction was planned a priori to be done without distribution assumptions by using a bias-corrected bootstrapping technique (with 10,000 replications).12,13 These CIs may be approximate. A conventional 1-sided 95% CI was also constructed using normal distribution assumptions (based on the SEM) as a planned exploratory analysis. Because of the noninferiority design (where the null hypothesis is that the 2 groups being tested are different—opposite to that of a superiority trial), the primary analysis was planned to be per protocol, with an intention-to-treat analysis done as a sensitivity analysis if any study participants did not receive the intervention to which they were randomized.
For purposes of sample size calculation, the expected standard deviation of the 24-hour postoperative opioid consumption was taken from a prospective randomized clinical trial of epidurally administered morphine after cesarean delivery.14 This standard deviation was 4.28 mg, which we rounded up to 5 mg to be conservative. With 90% power, the per-group sample size was calculated15 to be 39, but 45 patients per group were recruited to account for missing data due to loss to follow-up, patient dropouts, or protocol violations.
Descriptive statistics were calculated for demographic variables. Pain scores at rest and with movement had both a group component (1.5 mg vs 3 mg) and a time component (6, 12, 24, 36, and 48 hours after the intervention, clustered within each patient). Therefore, a linear mixed-effects regression model (maximum likelihood estimation) was used in which the dependent variable was the pain score and independent variables (fixed effects) were the group allocation and time period (main and interaction effects), while each patient was included in the random effects portion of the model. This controlled for the within-patient correlation of the repeated measures and allowed for a random intercept and slope for each patient in the regression model. An exchangeable correlation structure between pairs of within-patient measurements was assumed. Based on the statistical model, contrasts and 95% CIs for the contrasts were calculated for each measurement occasion both within groups and between groups. Contrasts of adjusted predictions were displayed graphically. Although it is conservative and forcibly increases the risk of a type II error, the Bonferroni adjustment for multiple comparisons was used for contrasts of all outcomes analyzed with the linear mixed-effects models because of the number of pairwise comparisons made.
Opioid consumption from 24 to 48 hours postoperatively, total opioid consumption, time to first request of any opioid, and QOR scores were analyzed using the Mann-Whitney test because these variables were anticipated to be right skewed (therefore, their CIs were estimated using bootstrapping with 10,000 replications). Overall mean pain scores at 24 hours, 48 hours, and 12 weeks were analyzed using Student t test. Satisfaction with pain control and adverse events (nausea, pruritus, and sedation) were analyzed using Fisher exact test (2-sided). To appreciate the time-dependent nature of adverse event data, an exploratory analysis was performed whereby the adverse event scales were assumed to be ratio data and modeled using a linear mixed-effect model (as described earlier for the pain scores, including Bonferroni adjustments).
Apart from the primary outcome, a 2-tailed P value of <0.05 was considered significant. Stata version 12.1 (StataCorp LP) for Mac OS X was used for all data management and analyses. The anonymized raw datasets (comma-separated values files containing the main dataset in wide and long format that can be imported into any statistical analysis package) are available as online data files S1 (see Supplementary Digital Content, http://links.lww.com/AA/A561) and S2 (see Supplementary Digital Content 2, http://links.lww.com/AA/A562) as well as instructions on how to use the raw datasets (online data file S3, see Supplementary Digital Content 3, http://links.lww.com/AA/A563) and approval from the Health Sciences Research Ethics Board of the University of Western Ontario, granting permission for the publication of the raw datasets (online date file S4, see Supplementary Digital Content 4, http://links.lww.com/AA/A564).
From January 2010 to January 2012, 90 participants received the planned intervention; however, 3 were excluded from the primary analysis (Fig. 1). One participant in the 1.5 mg EM group was lost to follow-up due to early discharge, and 2 participants in the 3.0 mg EM group were excluded due to protocol violations (Fig. 1). Data were analyzed from 44 subjects in the 1.5 mg EM group and from 43 patients in the 3.0 mg EM group. Participant demographics and baseline characteristics were similar in both groups (Table 1).
Noninferiority was demonstrated because the difference in median 24-hour opioid consumption between the 1.5 mg EM and 3 mg EM groups was 0 mg (1-sided 95% CI, 2.5 mg; Fig. 2). This CI was below the prespecified noninferiority margin of 3.33 mg. The planned exploratory analysis showed a difference in mean 24-hour opioid consumption between the 1.5 mg EM and 3 mg EM groups of −0.1 mg (1-sided 95% CI, 1.7; Table 2 and Fig. 2). There were no significant differences observed between groups in the median 24- to 48-hour opioid consumption or the median total opioid consumption within 48 hours (Table 2). The proportion of patients not consuming opioids during the first 48 hours after cesarean delivery in the 1.5 mg EM and 3 mg EM group was 25% (11 of 44) and 42% (18 of 43) participants, respectively (relative risk of not consuming opioids during the first 48 hours after cesarean delivery 0.60, 95% CI, 0.3–1.1, P = 0.12).
There were no statistically significant differences between groups in NRS pain scores at rest or with movement during the 48-hour study period (Fig. 3). The overall mean pain scores were not significantly different between groups at 24 hours, 48 hours, or 12 weeks (Table 3).
At 24 and 48 hours, overall pain relief and maternal satisfaction (Table 4) were not significantly different between groups. The 1.5 mg EM group had a statistically significantly higher QOR score at 24 hours, but there was no difference between groups in the QOR score at 48 hours (Table 5).
There was a decrease in the incidence of pruritus in patients receiving the lower dose of epidural morphine when compared with the patients who received the traditional dose at 6 and 12 hours (Table 6). The planned exploratory analysis using a linear mixed-effects model demonstrated decreasing pruritus severity over time with significant differences between groups at 6, 12, and 24 hours (Fig. 4). The lower dose of epidural morphine also resulted in a less frequent incidence of nausea and vomiting for patients at 6 hours (Table 6). The planned exploratory analysis demonstrated less severe nausea and vomiting in the 1.5 mg EM group at 6 hours (Fig. 5). Fewer patients in the 1.5 mg EM group consumed ondansetron than the 3.0 mg EM group in the first 24 hours (P = 0.002), but no other significant differences were found at other time points or with other antiemetics and antipruritics (Table 7).
There were no significant differences found in the incidence of sedation between groups (Table 8) or in sedation severity over time using the exploratory analysis (Fig. 6).
In this randomized, double-blinded, noninferiority study, we found that when using multimodal analgesia consisting of systemic ketorolac and acetaminophen half our traditional dose of epidural morphine, 1.5 mg, compared with our traditional 3 mg epidural morphine provided noninferior (<3.33 mg difference between groups) analgesia after cesarean delivery. We found no difference in opioid consumption between the 2 groups. No differences in pain scores, at rest or with movement, were observed between groups. Participants expressed similar levels of satisfaction in both groups. Importantly, less frequent pruritus and nausea were observed in the lower dose group. In our study, there were reductions in the severity of pruritus at 6 and 12 hours and nausea at 6 hours with the lower dose of epidural morphine.
It is interesting to note that fewer patients in the 1.5 mg group consumed ondansetron than the 3 mg group in the first 24 hours, but the 1.5 mg group had less severe nausea and vomiting only at the 6-hour assessment point.
The modified QOR scores were high in both groups although there was a statistically significantly improved QOR for the lower dose EM study group at 24 hours. Effective pain control can improve maternal function and lessen side effects from opioid consumption.2 The QOR includes questions about side effects and impact of pain on daily function after a cesarean delivery. Given that side effects can lower the QOR score and that studies have shown the side effects increase in a dose-dependent manner, it is possible a lower dose of epidural morphine resulted in a higher QOR score on the first day.16 However, the clinical significance of this very small difference is unclear. Nonetheless, it is reassuring that the lower dose of epidural morphine did not result in a lower QOR score.
The optimal dose of epidural morphine that provides effective analgesic with minimal side effects is controversial. In a retrospective chart review of 4880 patients in 1990, Fuller et al.4 examined the analgesic effects of epidural morphine for cesarean delivery. They found that the incidence of side effects, especially respiratory depression, increased with the dose used. They recommended that 3 mg epidural morphine provided effective analgesia with limited side effects.4 A recent systematic review1 of 10 prospective randomized studies evaluating the analgesic efficacy of epidural morphine for postcesarean pain relief recommended 4 mg as the optimal dose. Of note, the most recent randomized clinical trial in this review was published in 2000 by Palmer et al.17 In this dose-response study, the authors examined different doses of epidural morphine (0, 1.25, 2.5, 3.75, or 5 mg) and the quality of analgesia and adverse effects in 60 parturients.17 There was no improvement in analgesia above 3.75 mg, and pruritus increased with the dose of epidural morphine. A recent randomized controlled trial by Carvalho et al.6 that compared extended release epidural morphine with conventional epidural morphine used 4 mg epidural morphine as the conventional dose based on the study by Palmer et al.17
Neither the Palmer et al. study17 nor any of the others included in the systematic review1 used regular systemic NSAIDs which has become standard in contemporary practice. The use of a combination of analgesics with different mechanisms of action results in improved analgesia after cesarean delivery and a reduction in the incidence of adverse effects.2,6 NSAIDs, in particular, are effective in alleviating uterine cramping pain and are opioid sparing, thus resulting in a lower incidence of opioid-related side effects.
Previous studies of epidural morphine have evaluated women having elective cesarean delivery. The findings from these studies may not be applicable to women having intrapartum cesarean delivery. It is this latter group who most commonly receive epidural morphine in contemporary practice rather than women having elective cesarean delivery.
There are some limitations to our study. First, the use of opioid consumption as our primary outcome to measure analgesic efficacy may be modulated by the mother if she is concerned that opioids may affect her or her baby. If the difference detected was biased toward the null because women systematically did not take their medications, and in fact, the 3 mg EM group had better pain relief than the 1.5 mg EM group, one would expect lower pain scores in the 3 mg EM group. This was not observed. The pain scores were essentially the same, which implies that the degree of analgesia was the same in both groups. Also, the predisposition toward not taking opioids because of a maternal concern about neonatal outcomes should be equalized by randomization.
It is also possible that small differences in opioid requirements for breakthrough pain may have been missed by nurse-administered oxycodone pills. Patient-controlled IV analgesia may have more easily detected small differences. However, at our institution, we do not routinely use patient-controlled IV analgesia in cesarean delivery patients given epidural morphine. Our protocol was designed to allow the study to be conducted using our current standard of care.
We also recognize that the use of Bonferroni correction that biases toward the null may potentially exaggerate bias in a noninferiority study. However, with the exception of the pain scores over time figure, the adjustment was only used for the exploratory analyses and therefore would not affect the overall conclusions of the study. When the study data were reanalyzed without Bonferroni correction, all changes from statistically nonsignificant to significant (nausea at 6 and 12 hours, and sedation at 12 hours) favored the 1.5 mg EM group.
Another limitation of this study was the dose selection. We chose 3 mg epidural morphine as the traditional dose because this is what is used most commonly at our institution based on a retrospective chart review.4 We chose to compare our traditional dose with half that amount, rather than to a variety of doses, because we believed 1.5 mg epidural morphine and multimodal analgesia would provide effective pain relief and to maintain high statistical power. Therefore, we cannot make any conclusions regarding other doses.
We did not collect information regarding use of epidural fentanyl boluses for inadequate labor analgesia. It is possible that during labor, parturients may have received fentanyl boluses that may have affected the efficacy and side effects of epidural morphine. However, any concern regarding use of fentanyl boluses should be equal between groups due to randomization. Finally, another potential limitation of this study was the number of patients lost to follow-up at the 12-week assessment. Only 21 patients from the original 87 included in the primary outcome analysis could be assessed. Therefore, the ability of this study to make any inferences regarding pain at 12 weeks is limited to hypothesis generation.
In conclusion, when using multimodal analgesia after cesarean delivery in term parturients, 1.5 mg epidural morphine demonstrated noninferior analgesia (<3.33 mg difference between groups), similar maternal satisfaction, and less pruritus than 3 mg epidural morphine. For effective control of postoperative pain after cesarean delivery, 1.5 mg epidural morphine should be used with regular systemic NSAID therapy.
Name: Sudha I. Singh, MD, FRCPC.
Contribution: This author helped design and conduct the study, collect the data, and prepare the manuscript.
Attestation: This author reviewed the original study data and data analysis, approved the final manuscript, and is the archival author.
Name: Sarah Rehou, BSc (Hons).
Contribution: This author helped design and conduct the study, collect the data, and prepare the manuscript.
Name: Kristine L. Marmai, MD, FRCPC.
Contribution: This author helped design and conduct the study, collect the data, and prepare the manuscript.
Name: Philip M. Jones, MD, MSc.
Contribution: This author helped design the study, performed the statistical analysis, and helped in manuscript preparation.
Attestation: This author also reviewed the original study data and data analysis.
This manuscript was handled by: Cynthia A. Wong, MD.
We would like to thank all the anesthesiologists, obstetricians, and obstetrical nursing staff at St. Joseph’s and Victoria Hospital in London, Ontario, Canada.