Habib, Ashraf S. MBBCh, FRCA; George, Ronald B. MD, FRCPC; McKeen, Dolores M. MD, MSc; White, William D. MPH; Ituk, Unyime S. MBBS, FCARCSI; Megalla, Sohair A. MBBCh; Allen, Terrence K. MBBS, FRCA
Intraoperative nausea and vomiting occur in 40–80% of women administered spinal anesthesia for cesarean delivery.1 Hypotension is a major risk factor, occurring in up to 80% of cases.1 Prophylactic phenylephrine infusions significantly reduce the incidence of intraoperative hypotension and the resultant intraoperative nausea and vomiting.2,3 The use of prophylactic antiemetics in conjunction with phenylephrine infusions has not been previously investigated.
Metoclopramide 10 mg was reported to be effective for both intraoperative and postoperative nausea and vomiting prophylaxis in obstetric patients.4–7 However, most of the studies investigating metoclopramide did not use neuraxial opioids and treated established hypotension with boluses of ephedrine. Therefore, the antiemetic efficacy of metoclopramide with modern techniques including neuraxial opioids and a prophylactic phenylephrine infusion is unknown.
Combination antiemetic therapy is more effective than monotherapy for postoperative nausea and vomiting prophylaxis in the general surgical population.8 However, there is paucity of data on the efficacy of a combination of two antiemetics for intraoperative and postoperative nausea and vomiting prophylaxis in women undergoing cesarean delivery with spinal anesthesia.
We therefore performed this study to estimate whether the addition of metoclopramide or the combination of ondansetron and metoclopramide provides improved intraoperative nausea and vomiting prophylaxis compared with a technique using a prophylactic phenylephrine infusion alone. We hypothesized that antiemetic prophylaxis will result in a lower incidence of intraoperative nausea and vomiting compared with using a prophylactic phenylephrine infusion alone.
MATERIALS AND METHODS
The study was performed at Duke University Medical Center, Durham, North Carolina (site 1), and at IWK Health Center, Halifax, Nova Scotia, Canada (site 2), between December 2008 and January 2011. This trial was registered with clinicaltrials.gov (NCT01216410). After Institutional Review Board approval (Duke University, Durham, NC, and IWK Health Center, Halifax, Nova Scotia, Canada) and written informed patient consent, English-speaking, American Society of Anesthesiologists class I or II, nonlaboring women with single gestation, 36 weeks of gestation or more, scheduled for cesarean delivery with spinal anesthesia were approached to participate in this randomized, double-blind study. The following were exclusion criteria: patients shorter than 152 cm or taller than 180 cm (because patient height might affect the spread of spinal anesthesia);9 laboring women needing an emergency cesarean delivery; receipt of a drug with antiemetic properties in the 24 hours before cesarean delivery; allergy to ondansetron, metoclopramide, or phenylephrine; severe hypertensive disease of pregnancy defined as systolic blood pressure more than 160 mm Hg, diastolic blood pressure more than 110 mm Hg requiring antihypertensive treatment, or associated with significant proteinuria; severe cardiac disease in pregnancy with marked functional limitations; insulin-dependent diabetes mellitus; patients using monoamine oxidase inhibitors or tricyclic antidepressants; morbid obesity (body mass index [calculated as weight (kg)/[height (m)]2] more than 45); and inclusion in another anesthetic study involving drug administration.
After obtaining informed consent, baseline noninvasive blood pressure was measured in the preoperative holding area with the patient supine with left uterine displacement. We determined baseline systolic blood pressure by calculating the mean of three consecutive systolic blood pressure measurements taken 5 minutes apart when the patient was left undisturbed. Patients were randomly assigned using sealed opaque envelopes to one of three groups: placebo group; metoclopramide group; and ondansetron with metoclopramide (combination) group. Randomization was performed in blocks of 20 using NQuery Advisor 6.0. An anesthesia care provider not involved in the study prepared the blinded study drugs in identical 3-mL syringes in a volume of 2 mL labeled study drug A and study drug B, respectively. Patients in the placebo group received 2 mL saline intravenously (IV) before placement of spinal anesthesia and after clamping of the cord. Patients in the metoclopramide group received metoclopramide 10 mg (2 mL) IV before placement of spinal anesthesia and a 2-mL saline placebo after clamping of the cord. Patients in the combination group received metoclopramide 10 mg (2 mL) IV before placement of spinal anesthesia and ondansetron 4 mg (2 mL) IV after clamping of the cord. The first study drug was administered as the patient was positioned sitting for the spinal anesthesia and the second was administered immediately after cord clamping. We administered metoclopramide before delivery because there are data supporting its safety for the neonate.7,10 However, although ondansetron is likely safe, its effects on the neonate have been less investigated.
Preoperatively, we collected information about preoperative nausea, history of hyperemesis gravidarum, history of morning sickness in early pregnancy, history of postoperative nausea and vomiting, history of intraoperative nausea and vomiting if the patient had a previous cesarean delivery under neuraxial anesthesia, history of motion sickness, and smoking status.
The anesthetic technique was standardized. All women received antacid prophylaxis with 30 mL oral sodium citrate. In the operating room, spinal anesthesia was administered in the sitting position at the L2–L3 or L3–L4 interspace using 12 mg hyperbaric bupivacaine with 15 micrograms fentanyl and 150 micrograms preservative-free morphine. Patients were then immediately laid supine with left lateral tilt. A phenylephrine infusion was started immediately after placement of the spinal block at 50 micrograms/min and titrated according to a predefined algorithm to maintain systolic blood pressure within 20% of baseline or 90 mm Hg or more. The infusion rate was doubled to 100 micrograms/min if systolic blood pressure was less than 20% of baseline or less than 90 mm Hg and if heart rate was more than 50 beats per minute (bpm). A 100-microgram phenylephrine bolus was administered if systolic blood pressure was not restored after 1 minute. Glycopyrrolate 0.2 mg was administered if there was hypotension with bradycardia defined as a heart rate less than 50 bpm. If systolic blood pressure was more than 20% of baseline or if heart rate was less than 50 bpm in the absence of hypotension, the infusion rate was halved to 25 micrograms/min and then stopped if there was no response after 1 minute. A rapid fluid bolus was started immediately under pressure as the intrathecal drugs were administered (fluid coloading), with the aim of administering 2 L lactated Ringer solution before delivery followed by maintenance rate. The phenylephrine infusion was maintained until 10 minutes after delivery. Blood pressure was measured every minute for the first 10 minutes after spinal placement and then every 2.5 minutes for the remainder of the study. Oxytocin, 5 units IV, was administered after delivery, followed by an infusion of 25 units added to 1 L lactated Ringer solution, which was administered over the next 1–2 hours. Intraoperative nausea with a verbal rating scale score of 6 or more or vomiting (not related to a systolic blood pressure measurement less than 20% of baseline) was treated with ondansetron 4 mg IV. Diphenhydramine 6.25 mg IV was used if there was no complete resolution of symptoms within 5 minutes of ondansetron administration. Postoperative nausea and vomiting were treated with promethazine 6.25 mg IV(site 1) or dimenhydrinate 6.25 mg IV(site 2). Pruritus was treated with a naloxone infusion at .25 micrograms/kg/h (site 1) or nalbuphine 5 mg IV (site 2). Postoperative analgesia consisted of naproxen 500 mg orally every 12 hours and oxycodone 5 mg/acetaminophen 325 mg 1–2 tablets as needed for breakthrough pain in site 1, and regular oral acetaminophen 1,000 mg every 6 hours, oral naproxen 500 mg every 12 hours, and oxycodone 5–10 mg as needed for breakthrough pain in site 2.
Research personnel blinded to randomization assignment collected the data. In the operating room, we recorded blood pressure, heart rate, intraoperative nausea scores, vomiting, number of vomiting episodes, and the need for rescue antiemetics. Patients were asked about nausea every 5 minutes for the first 10 minutes and then every 10 minutes for the remainder of the procedure. At the end of surgery, patients were asked to rate their pruritus (0=no pruritus and 10=worst possible pruritus) and their satisfaction with intraoperative nausea and vomiting control (1=very satisfied, 2=somewhat satisfied, 3=neither satisfied nor dissatisfied, 4=somewhat dissatisfied, 5=very dissatisfied). We also recorded the duration of surgery, volume of IV fluids administered, estimated blood loss, total phenylephrine used, and exteriorization of the uterus.
Postoperatively, we collected nausea scores, number of vomiting episodes, need for rescue antiemetic therapy, pruritus scores, need for treatment for pruritus, side effects (headache and extrapyramidal side effects), and opioid consumption at 2, 6, and 24 hours after surgery. Patients were asked about their satisfaction with the management of postoperative nausea and vomiting at 24 hours postoperatively.
Nausea was defined as a feeling of the urge to vomit, as solicited by the investigators during assessments. Verbal nausea scores were collected on an 11-point scale, with 0 indicating no nausea and 10 indicating as bad as it can possibly be. Any nausea score more than 0 was considered nausea. Vomiting was defined as expulsion of stomach contents through the mouth. Retching was defined as an attempt to vomit but not productive of stomach contents. Retching was analyzed as vomiting. A complete response was defined as no intraoperative or postoperative nausea or vomiting and no need for rescue antiemetics.
The primary endpoint was the incidence of intraoperative nausea and vomiting defined as any intraoperative nausea, vomiting, or retching. Secondary endpoints included the incidence of postoperative nausea and vomiting (defined as any postoperative nausea, vomiting, or retching), pruritus, and maternal satisfaction. Data from a previous study using a similar anesthetic technique in patients undergoing cesarean delivery using a prophylactic phenylephrine infusion reported an incidence of intraoperative nausea and vomiting of 40%.11 A sample size of 100 patients in each group was estimated to provide 80% power to detect an absolute reduction in the incidence of intraoperative nausea and vomiting to 20% and 15% in the metoclopramide and combination groups, respectively, with an overall significance level of α=.05 and accounting for multiple comparisons for a significance level of .05/3=.0167, as a Bonferroni adjustment would require.
Data were analyzed using Wilcoxon rank-sum test, Kruskal-Wallis test, and χ2 test as appropriate. Fisher exact test was used instead of a χ2, in tables which included any cell with an expected count of less than five. We adjusted for multiple pair-wise comparisons among treatment groups using Bonferroni correction for all dichotomous data and the Tukey-Kramer correction for all numeric comparisons such as intraoperative nausea scores. Potential risk factors were tested for univariable association with intraoperative nausea and vomiting. The initial list of candidate covariables included those with univariable association with intraoperative nausea and vomiting, or with previous clinical evidence of association with intraoperative nausea and vomiting, or that were significantly different among the treatment groups using a χ2 or rank-sum test. The following risk factors were evaluated: history of intraoperative or postoperative nausea and vomiting; history of motion sickness; smoking; history of morning sickness; history of hyperemesis gravidarum; exteriorization of the uterus; intraoperative hypotension defined as a decline in systolic blood pressure more than 20% of baseline; site; duration of surgery; age; baseline systolic blood pressure; maximum decline in systolic blood pressure compared with baseline; total fluids administered; and estimated blood loss. A forward stepwise multivariable logistic method was used to select covariables with significant effect on intraoperative nausea and vomiting for inclusion in a final model assessing treatment effect. All two-way interactions with treatment were tested. To assess for differences in treatment effect across the two sites, our analysis model included a test of the treatment by site interaction. We also performed a post hoc analysis for the primary endpoint by study site. All of the measured variables that differed between sites were tested for effect on intraoperative nausea and vomiting univariably, before the multivariable logistic model was framed, or in the selection of variables for the final multivariable logistic model. To further assure the similarity of treatment effect, the three-way interactions between treatment, site, and all of the variables, which differed between sites, also were tested. Data were analyzed using SAS 9.1. P<.05 for the overall three-group comparison and P<.017 for pair-wise comparisons were considered statistically significant.
Three-hundred six patients were enrolled in this study. Of those, 300 patients completed the study and were included in the analysis. The flow of patients in the study is shown in Figure 1 .
Flow of participants...Image Tools
There was no difference in patient demographics, risk factors for postoperative nausea and vomiting, duration of surgery, volume of fluids administered, estimated blood loss, or total phenylephrine administered among the groups. However, exteriorization of the uterus was significantly different among the groups (P=.046) (Table 1).
With unadjusted analysis, intraoperative nausea, intraoperative nausea and vomiting, need for rescue antiemetics, and intraoperative nausea scores were significantly different among the groups (Table 2). In pair-wise comparisons, intraoperative nausea and vomiting were significantly lower in the metoclopramide and combination groups compared with placebo (number needed to treat 6 [P=.001] and number needed to treat 4 [P<.001], respectively). Intraoperative nausea and vomiting were not different between the metoclopramide and combination groups. Need for rescue was significantly lower with the combination compared with placebo and metoclopramide groups (P<.001 and P<.017, respectively). The incidence of hypotension-associated intraoperative nausea and vomiting was not different among the groups. Of the 103 patients who had developed intraoperative nausea and vomiting, symptoms occurred at a systolic blood pressure 80% or less of baseline in 30 (29%) patients, more than 80% to 90% in 22 (21%), more than 90% to 100% in 22 (21%), and more than 100% in 29 (28%). In those with intraoperative nausea and vomiting, nausea occurred before delivery in 33 patients (32%), after delivery in 43 patients (42%), and occurred both before and after delivery in 26 patients (25%).
There was a significant difference in the incidence of intraoperative nausea and vomiting between the two sites (47% compared with 20%; P<.001). There was no difference between patients at the two sites in age, body mass index, history of intraoperative or postoperative nausea and vomiting, history of motion sickness, previous cesarean delivery, hyperemesis, or smoking. However, there were significant differences between the two sites in surgical duration, total fluids administered, estimated blood loss, total phenylephrine administered, exteriorization of the uterus, and incidence of hypotension. The incidence of hypotension-associated intraoperative nausea and vomiting was not different between the two sites (28% compared with 35%; P=.47). When analyzed by site, intraoperative nausea and vomiting were still significantly different among the groups, with the combination being more effective than placebo at both sites (Table 3).
The variables selected for adjusting the treatment effect in the multivariable model were site (P<.001), exteriorization of the uterus (P<.001), decline in systolic blood pressure more than 20% of baseline (P<.001), and age (P=.04). All two-way and three-way interactions with treatment were found to be nonsignificant and were dropped from the model. The nonsignificant interaction between treatment and site (P=.63) indicates that the effect of treatment on intraoperative nausea and vomiting was not significantly different between sites. In the final model (Table 4), intraoperative nausea and vomiting were still significantly lower in the metoclopramide and combination groups compared with placebo (P=.001 and P<.001, respectively), whereas exteriorization became nonsignificant in the presence of site (P=.65). To estimate whether exteriorization is a significant predictor of intraoperative nausea and vomiting irrespective of site, we evaluated another model with treatment, exteriorization, and decline in blood pressure more than 20% from baseline as predictors. Without site in the model, exteriorization was a significant predictor of intraoperative nausea and vomiting (P=.002), and the treatment effect remained significant (P<.001).
No significant two-way or three-way interactions were found between treatment, site, and the effects that differed between sites. The nonsignificant interactions indicate that the effect of treatment seen is consistent regardless of the other measures, for example, that the treatment effect is similar irrespective of whether the uterus was or was not exteriorized.
Postoperative results are shown in Table 5. The incidences of nausea, vomiting, postoperative nausea and vomiting, number of vomiting episodes, and need for rescue antiemetics were significantly different among the groups in the first 2 hours postoperatively. Need for a rescue antiemetic also was significantly different among the groups at 2–6 hours. Pair-wise comparisons showed a lower incidence of nausea, vomiting, postoperative nausea and vomiting, number of vomiting episodes, and need for rescue in the combination group compared with placebo at 2 hours, and showed less need for rescue in the combination group compared with the placebo group at 2–6 hours after surgery.
Complete response (no intraoperative or postoperative nausea and vomiting or need for rescue) also was significantly lower in the combination group compared with the placebo group (odds ratio 3.2, 95% confidence interval 1.7--6; P<.001), with the other pair-wise comparisons being nonstatistically significant. Unlike intraoperative nausea and vomiting, there was no difference in the incidence of postoperative nausea and vomiting between the two sites (59% compared with 52%; P=.21).
There was no difference in the incidence, severity, or need for treatment of pruritus among the groups. There was also no difference in patient satisfaction with intraoperative or postoperative nausea and vomiting control. The incidence of headache was not different among the groups (11% in placebo and metoclopramide groups, 19% in combination group). No extrapyramidal side effects were observed in any of the patients. Consumption of acetaminophen and oxycodone was not different among the groups (data not reported).
Coadministration of metoclopramide or the combination of metoclopramide and ondansetron with a prophylactic phenylephrine infusion was significantly more effective than phenylephrine infusion alone for intraoperative nausea and vomiting prophylaxis in women undergoing cesarean delivery with spinal anesthesia. The combination was also associated with a lower incidence of early postoperative nausea and vomiting compared with placebo.
Whereas hypotension is a major risk factor for intraoperative nausea and vomiting, other factors such as fundal pressure at delivery, exteriorization of the uterus, and visceral manipulation also result in intraoperative nausea and vomiting.3 The use of prophylactic phenylephrine infusions significantly decreases the incidence of hypotension,11,12 which was associated with a significant reduction in the incidence of intraoperative nausea and vomiting in some studies,12 whereas others failed to show such a reduction.11 This is likely related to the effect of other emetogenic stimuli at and after delivery. It is important to note that studies reporting a reduction in intraoperative nausea and vomiting with prophylactic phenylephrine infusions did not report data after delivery when factors other than hypotension led to intraoperative nausea and vomiting.3 For instance, in this current study, the incidence of postdelivery intraoperative nausea and vomiting was numerically higher than predelivery intraoperative nausea and vomiting (42% compared with 32%). We therefore investigated whether the addition of prophylactic antiemetics to a prophylactic phenylephrine infusion reduces the incidence of intraoperative nausea and vomiting attributable to factors other than hypotension, and we collected data for the entire duration of the cesarean delivery.
In our study the incidence of hypotension was 16–19%. Of all the patients who had developed intraoperative nausea and vomiting, symptoms occurred below our defined threshold of hypotension in 29% of cases, and at systolic blood pressure 80–100% of baseline in 42% of cases. Titrating the phenylephrine infusion to maintain systolic blood pressure at baseline might have resulted in a lower incidence of intraoperative nausea and vomiting.2 In 28% of cases, nausea occurred at a systolic blood pressure higher than baseline, supporting the fact that other emetogenic stimuli contribute to intraoperative nausea and vomiting.
Exteriorization of the uterus is a known risk factor for intraoperative nausea and vomiting. Siddiqui et al13 reported an incidence of intraoperative nausea and vomiting of 38% with exteriorization compared with 18% with in situ repair. Other factors, such as intra-abdominal irrigation, also increase the risk of intraoperative nausea and vomiting. Viney et al14 reported an increase in intraoperative nausea and vomiting from 28% to 46% with irrigation, with no reduction in postoperative infectious complications. Whereas hypotension is the major cause of predelivery intraoperative nausea and vomiting, those factors together with visceral manipulation result in postdelivery intraoperative nausea and vomiting. Therefore, the surgical technique might result in differences in intraoperative nausea and vomiting between different centers. In fact, we found more than two-fold difference between the two sites involved in this study in the incidence of intraoperative nausea and vomiting. Risk factors for intraoperative nausea and vomiting were significantly different between the two sites. In particular, exteriorization of the uterus was performed more frequently in the site with the higher incidence of intraoperative nausea and vomiting. In addition, intra-abdominal irrigation is performed routinely at site 1, but is not performed at all at site 2. Interestingly, however, postoperative nausea and vomiting were not different between the two sites. This underscores the fact that risk factors for intraoperative nausea and vomiting are likely different than those for postoperative nausea and vomiting.
Our study confirmed the efficacy of metoclopramide for intraoperative nausea and vomiting prophylaxis. However, metoclopramide did not result in any reduction in the incidence of postoperative nausea and vomiting. This agrees with data from the general surgical population suggesting the lack of efficacy of the 10-mg dose of metoclopramide.15 The reduction of postoperative nausea and vomiting in the combination group was limited to the early postoperative period. This likely reflects the short duration of action of ondansetron. Longer-acting antiemetics such as transdermal scopolamine and dexamethasone might achieve better control of postoperative nausea and vomiting in the later postoperative period.
This study has limitations. We did not include an ondansetron-only group, so we cannot comment on whether the combination of ondansetron and metoclopramide would be more effective than ondansetron alone, or on the comparison of metoclopramide with ondansetron. We titrated the phenylephrine infusion to maintain systolic blood pressure within 20% of baseline. Stricter control of blood pressure, aiming for a target of baseline, might have achieved better control of intraoperative nausea and vomiting.
In conclusion, metoclopramide and a combination of metoclopramide and ondansetron added to a prophylactic phenylephrine infusion provided improved prophylaxis against intraoperative nausea and vomiting compared with placebo in women undergoing cesarean delivery with spinal anesthesia. Surgical factors, especially exteriorization of the uterus and irrigation, contributed to significantly different intraoperative nausea and vomiting rates between the two sites. Early postoperative nausea and vomiting were also reduced with the combination. The overall postoperative nausea and vomiting rates, however, remain high, and more studies are needed to optimize antiemetic prophylaxis in this patient population.