A Single Prophylactic Dose of Ondansetron Given at Cessation of Postoperative Propofol Sedation Decreases Postoperative Nausea and Vomiting in Cardiac Surgery Patients: A Randomized Controlled Trial

Abstract

BACKGROUND: 

Postoperative nausea and vomiting (PONV) is a common occurrence after cardiac surgery. However, in contrast to other surgical populations, routine PONV prophylaxis is not a standard of care in cardiac surgery. We hypothesized that routine administration of a single prophylactic dose of ondansetron (4 mg) at the time of stopping postoperative propofol sedation before extubation in the cardiac surgery intensive care unit would decrease the incidence of PONV.

METHODS: 

With institutional human ethics board approval and written informed consent, we conducted a randomized controlled trial in patients ≥19 years of age with no history of PONV undergoing elective or urgent cardiac surgery procedures requiring cardiopulmonary bypass. The primary outcome was the incidence of PONV in the first 24 hours postextubation, compared by the χ² test. Secondary outcomes included the incidence and times to first dose of rescue antiemetic treatment administration, the incidence of headaches, and the incidence of ventricular arrhythmias.

RESULTS: 

PONV within the first 24 hours postextubation occurred in 33 of 77 patients (43%) in the ondansetron group versus 50 of 82 patients (61%) in the placebo group (relative risk, 0.70 [95% confidence interval {CI}, 0.51–0.95]; absolute risk difference, −18% [95% CI, −33 to −2]; number needed to treat, 5.5 [95% CI, 3.0–58.4]; χ² test, P = .022). Kaplan-Meier “survival” analysis of the times to first rescue antiemetic treatment administration over 24 hours indicated that patients in the ondansetron group fared better than those in the placebo group (log-rank [Mantel-Cox] test; P = .028). Overall, 32 of 77 patients (42%) in the ondansetron group received rescue antiemetic treatment over the first 24 hours postextubation versus 47 of 82 patients (57%) in the placebo group (relative risk, 0.73 [95% CI, 0.52–1.00]; absolute risk difference, −16% [95% CI, −31 to 1]); P = .047. There were no significant differences between the groups in the incidence of postoperative headache (ondansetron group, 5 of 77 patients [6%] versus placebo group, 4 of 82 patients [5%]; Fisher exact test; P = .740) or ventricular arrhythmias (ondansetron group, 2 of 77 patients [3%] versus placebo group, 4 of 82 patients [5%]; P = .68).

CONCLUSIONS: 

These findings support the routine administration of ondansetron prophylaxis at the time of discontinuation of postoperative propofol sedation before extubation in patients following cardiac surgery. Further research is warranted to optimize PONV prophylaxis in cardiac surgery patients.

KEY POINTS

• Question: Does ondansetron reduce the incidence of postoperative nausea and vomiting (PONV) in cardiac surgery patients when it is given at the time of discontinuing postoperative sedation?
• Findings: A single prophylactic dose of ondansetron given at the time of stopping postoperative propofol sedation resulted in a relative risk reduction of 30% and an absolute risk reduction of 18% of PONV in the first 24 hours postextubation after cardiac surgery; the number needed to treat was 5.5.
• Meaning: These findings highlight the topic of PONV prophylaxis in cardiac surgery patients and support the routine administration of ondansetron at the time of discontinuing postoperative propofol sedation in this population.

Postoperative nausea and vomiting (PONV) is a common occurrence after general anesthesia that affects a wide range of surgical populations, including patients undergoing cardiac surgery.^1–4 Numerous studies have reported sizeable incidences of PONV in cardiac surgery patients, with rates ranging as high as 46.5%–71%.^3,5–8 However, compared to other surgical populations, dedicated PONV prophylaxis in the cardiac surgery population has received comparatively limited study, although PONV is particularly distressing to patients following a median sternotomy and can be a significant source of morbidity^9–11 that is potentially preventable. Moreover, in contrast to noncardiac surgery patients, routine PONV prophylaxis does not appear to be a widespread standard of routine care in cardiac surgery,^5,12,13 including at our center.

There are a number of potential reasons why PONV prophylaxis may not be routinely included as part of standard care in cardiac anesthesia. First, individual anesthesia providers may not be aware of their patients’ PONV rates. Following cardiac surgery, most patients are sedated and ventilated for several hours after surgery in the cardiac surgery intensive care unit (CSICU) and they may be managed by a different team. Second, the literature regarding PONV prophylaxis in cardiac surgery has shown variable results and the optimal timing of when prophylaxis should be given is not clear.^{2,3,8,13–17} A majority of published trials have focused on interventions at the pre- or intraoperative stage, whereas PONV prophylaxis given in the postoperative phase has received limited study.³ However, the variable duration of cardiac surgical procedures and the potentially prolonged period of postoperative ventilation combined with the pharmacokinetic profile of the serotonin receptor type 3 (5-HT₃) antagonist ondansetron as the “gold standard” antiemetic^4,18 provide a rational basis for administering a prophylactic dose at a standardized postoperative time. To our knowledge, the routine postoperative administration of ondansetron as a single dose and with a standardized timing has not been studied in this patient population. On the basis of the above, we conducted a randomized controlled trial to test the hypothesis that a single prophylactic dose of ondansetron (4 mg) given at the time of discontinuation of postoperative propofol sedation reduces the incidence of PONV following cardiac surgery in the first 24 hours postextubation compared to placebo.

METHODS

With institutional human research ethics board approval (The University of British Columbia/Providence Health Care, Vancouver, BC, Canada) and participants’ written informed consent, we conducted a single-center, prospective, randomized, quadruple-blinded, parallel-group clinical trial in cardiac surgery patients at a quaternary care academic health sciences center (St Paul’s Hospital) in Vancouver, BC, Canada. The trial was registered at ClinicalTrials.gov under the identifier NCT02966041 (principal investigator: Dr Matthew Coley) on November 17, 2016, before the start of any participant enrollment. This article was prepared in accordance with the Consolidated Standards of Reporting Trials (CONSORT) statement for the reporting of randomized controlled trials.¹⁹

We included patients ≥19 years of age scheduled to undergo elective (ie, outpatient same-day admission) or urgent (ie, stable inpatient) cardiac surgery procedures requiring cardiopulmonary bypass (eg, coronary artery bypass grafting [CABG], valve replacement, or/and valve repair) and postoperative sedation with propofol. Excluded were patients undergoing off-pump CABG, heart transplantation, or ventricular assist device implantation, and those requiring extracorporeal membrane oxygenation. Patients were also excluded if they had a history of PONV, long-QT syndrome, migraine requiring treatment with triptans, depression requiring antidepressants, or previous allergy or intolerance to ondansetron. Finally, exclusion criteria also included inability to provide informed consent, intubation for >12 hours, or dexmedetomidine administration as part of anesthesia or postoperative sedation. Patients were screened for eligibility and provided written consent before surgery.

Patients were randomly assigned in blocks of 4 to receive either a single dose of ondansetron 4 mg intravenously (IV; 2 mL) or placebo (2 mL normal saline IV), administered before extubation at the time of discontinuation of the IV propofol infusion for postoperative sedation. The block randomization sequence was generated with the use of an online randomization service (https://www.sealedenvelope.com/simple-randomiser/v1/lists) by a pharmacist not otherwise involved in the study and concealed via sequentially numbered sealed opaque envelopes. The study drug syringes were batched by pharmacy technicians every morning after the names and particulars of the patients who were enrolled were faxed to pharmacy, where a pharmacist sequentially opened the sealed envelopes to determine the study drug allocated to each randomly assigned patient. The prepared syringes were labeled with each patient’s name and “Ondansetron 4 mg or normal saline, inject 2 mL IV × 1 dose at time of propofol discontinuation” and hand-delivered from the pharmacy to the CSICU by hospital porters. The study was quadruple-blinded; participants, health care providers, investigators/outcome assessors, and data analysts all were blinded to group allocation. If patients developed nausea or vomiting after study drug administration, the choice of rescue therapy was determined by the care provider and guided by the institution’s standard protocols of repeat dosing of ondansetron and/or dimenhydrinate. Additional therapy with such agents as haloperidol, metoclopramide, or dexamethasone was also at the discretion of the care provider.

The primary outcome was the incidence of PONV, defined as any nausea, retching, or vomiting²⁰ within the first 24 hours postextubation, or before discharge/transfer from the CSICU to the ward, whichever came first. Secondary outcomes included the incidence and times to the first dose of rescue antiemetic administration; the number of doses and different agents as well as the specific types of rescue antiemetic agents administered; the incidence and severity of postoperative nausea (PON) without vomiting; the incidence and severity of headaches; and the incidence of ventricular arrhythmias from the time of CSICU admission to discharge/ward transfer. Patients were asked about the severity of any PON as well as headache on a 10-point numeric rating scale where 0 reflects no symptoms and 10 the worst imaginable symptoms, administered every hour for 4 hours followed by every 4 hours for a total of 24 hours. In addition, we hand-searched the clinical charts of all patients who were given rescue antiemetics for any documented evidence of PONV. We also recorded any instances and times of patients indicating/exhibiting PONV before extubation. If a patient was discharged from the CSICU to the ward before 24 hours postoperatively, data collection was stopped.

Statistical Analysis

Our statistical analysis was based on a modified intention-to-treat approach whereby we sought to include all consenting randomly assigned patients who received a study intervention; a prespecified modification was that the a priori exclusion criterion “intubation for greater than 12 hours” inherently was to be applied after receipt of study intervention. We analyzed baseline demographic, surgical, intraoperative anesthetic, and postoperative data as well as the different types of rescue antiemetic agents administered with the use of descriptive statistics, supplemented by calculation of standardized mean differences. For comparison of the primary outcome as well as other dichotomous variables, we used the χ² test and computed relative risks as well as differences in proportions/absolute risk differences and the number needed to treat (NNT) with their respective 95% confidence intervals (CI). For dichotomous postrandomization data and secondary outcomes with expected sample sizes <5 in at least one of the cells, we used the Fisher exact test. For contingency tables with 2 columns and more than 2 rows arranged in an ordinal fashion, we used the Cochran-Armitage method (χ² test for trend) to test for associations. Continuous variables were analyzed using the Student t test or the Mann-Whitney U test for normal or nonnormal distributions (assessed with the use of the D’Agostino-Pearson omnibus K² test), respectively. We constructed Kaplan-Meier curves²¹ for patient “survival” free of PONV as well as the times to first administration of rescue antiemetic treatment and compared the underlying time-to-event data with the log-rank (Mantel-Cox) test. Comparisons were 2-tailed and declared significant at P < .05. We used Prism 8 (GraphPad Software, LLP, San Diego, CA) and Excel (Microsoft Corporation, Redmond, WA) software for the analyses; for computation of standardized mean differences (expressed as the Cohen d), we used the online calculator by Davis B. Wilson, PhD (George Mason University, Fairfax, VA; http://www.campbellcollaboration.org/escalc/html/EffectSizeCalculator-SMD-main.php).

Sample Size Projection

A clinical audit of our own CSICU patients at St Paul’s Hospital undertaken before this study yielded an estimated incidence of PONV of 40%, similar to PONV incidences among cardiac surgery patients previously reported in numerous studies.^2,6,8 We considered a reduction from 40% to 20% to be clinically important and meaningful. To be able to detect a decrease in proportion of 0.20 from 0.40 at an α = .05 (2-tailed) and β = .2 (ie, 80% power), 80 patients per group were required (StatMate 2.00, GraphPad Software, LLP, San Diego, CA). To account for data loss and patient dropout, we aimed to recruit n = 85 patients per group.

RESULTS

A total of 302 patients were screened for eligibility between March and July 2018, of which 186 met inclusion criteria and were randomly assigned. After randomization, 19 patients did not receive a study intervention due to meeting prespecified exclusion criteria, protocol violations, or withdrawal of informed consent. Eight patients who received a study intervention met the prespecified exclusion criterion “intubation >12 hours” or were lost to follow-up, leaving 159 patients for inclusion in the final modified intention-to-treat analysis, of which n = 77 were in the ondansetron group and n = 82 were in the placebo (control) group. Figure 1 provides a study flow chart with the details on screening, enrollment, allocation, and analysis.

Figure 1.: Study flow diagram. PONV indicates postoperative nausea and vomiting.

Baseline demographic and surgical data are shown in Table 1. Patients were overwhelmingly male and had an average age of 65 years; the majority had a medical history of coronary artery disease and/or hypertension. Approximately 10% were smokers and one-third were life-long nonsmokers. More than half of the surgical procedures performed were CABG, followed by valve surgery, combined procedures, and others. Table 2 shows the details on intraoperative anesthetic data with regard to induction agents, maintenance agents, and intraoperative opioids. For induction, the majority of patients received a combination of midazolam, sufentanil, and propofol; more than half also received ketamine. Nearly all patients received rocuronium for neuromuscular blockade, and all received sevoflurane and propofol as maintenance agents. The most common opioids used intraoperatively were sufentanil and hydromorphone; there was no evidence for significant differences in total intraoperative opioid dose administered. Table 3 summarizes postoperative data on the duration of propofol infusion, the interval between CSICU admission and extubation, the interval between study drug administration and extubation, the number of patients requiring resedation, and overall CSICU duration of stay/discharge.

Table 1. - Baseline Demographic and Surgical Data

	Ondansetron (n = 77)	Placebo (n = 82)	Standardized Mean Differencea
Age (y), mean ± SD	64 ± 12	67 ± 13	0.240
Male sex, n (%)	67 (87)	66 (80)	0.267
Weight (kg), mean ± SD	83 ± 15	82 ± 19	0.058
Medical history, n (%)
Coronary artery disease	57 (74)	53 (65)	0.245
Hypertension	55 (71)	45 (55)	0.397
Diabetes mellitus	25 (32)	23 (28)	0.116
COPD	6 (8)	7 (9)	0.055
Depression	2 (3)	4 (5)	0.361
Motion sickness	4 (5)	5 (6)	0.094
Smoker	7 (9)	9 (11)	0.115
Ex-smoker	33 (43)	24 (29)	0.328
Life-long nonsmoker	25 (32)	32 (34)	0.158
Surgical procedure, n (%)
CABG	44 (57)	44 (54)	0.078
Valve	20 (26)	19 (23)	0.084
CABG + valve	11 (14)	13 (16)	0.068
Other	2 (3)	5 (6)	0.491
CABG + valve + other	0 (0)	1 (1)	Infinity
CPB time (min), median (IQR)b	92 (76–123)	106 (82–128)	n/a
Surgical time (min), median (IQR)c	274 (226–324)	288 (225–313)	n/a

Abbreviations: CABG, coronary artery bypass grafting; COPD, chronic obstructive pulmonary disease; CPB, cardiopulmonary bypass; IQR, interquartile range; n/a, not applicable; SD, standard deviation.

^aCohen d, calculated for normally distributed continuous as well as dichotomous variables. Shown are absolute values without directionality.

^bMann-Whitney U test, P = .251.

^cMann-Whitney U test, P = .437.

Table 2. - Intraoperative Anesthetic Data

	Ondansetron (n = 77)	Placebo (n = 82)	P
Induction agents, n (%)a
Midazolam	74 (96)	73 (89)	.133b
Rocuronium	71 (92)	78 (95)	.525b
Propofol	61 (79)	73 (89)	.126b
Sufentanil	58 (75)	67 (82)	.327c
Ketamine	41 (53)	49 (60)	.408c
Fentanyl	19 (25)	15 (18)	.327c
Other	3 (4)	3 (4)	.937c
Maintenance agents, n (%)a
Sevoflurane	77 (100)	82 (100)	>.999b
Propofol	77 (100)	82 (100)	>.999b
Intraoperative opioids, n (%)a
Sufentanil	56 (73)	68 (83)	.121c
Hydromorphone	58 (75)	53 (65)	.142c
Fentanyl	19 (25)	15 (18)	.327c
Morphine	6 (8)	6 (7)	.910c
Total intraoperative opioid dose (mg), median (IQR)d
Morphine equivalents	175 (110–235)	180 (135–240)	.49e

Abbreviation: IQR, interquartile range.

^aShown is, in a descending manner, the number (percentage) of patients who received a particular agent.

^bFisher exact test.

^cχ² test.

^dMorphine equivalent doses were calculated on the basis of the following assumed equipotent intravenous doses: 10 mg morphine = 0.1 mg fentanyl = 0.01 mg sufentanil = 2 mg hydromorphone.

^eMann-Whitney U test.

Table 3. - Postoperative Data

	Ondansetron (n = 77)	Placebo (n = 82)	P
Duration of propofol infusion (min), median (IQR)a	305 (217–378)	338 (259–445)	.05b
Interval between CSICU admission and extubation (min), median (IQR)	265 (218–329)	290 (225–371)	.15b
Interval between study drug administration and extubation (min), median (IQR)c	80 (38–131)	80 (47–131)	.61b
Patients requiring resedation with propofol infusion, n (%)	1 (1)	4 (5)	.368d
Duration of CSICU stay (h), median (IQR)	24.5 (22.1–28.2)	23.4 (20.2–26.8)	.037b
Patients discharged from CSICU <24 h postadmission, n (%)	39 (51)	46 (56)	.491e

Abbreviations: CSICU, cardiac surgery intensive care unit; IQR, interquartile range.

^aTimes reflect intervals between commencement of infusions in the operating room and discontinuation in the cardiac surgery intensive care unit postoperatively.

^bMann-Whitney U test.

^cTwo patients in the placebo group and 4 patients in the ondansetron group received the study drug after extubation; their times were entered as negative values.

^dFisher exact test.

^eχ² test.

The primary outcome, PONV in the first 24 hours postextubation, occurred in 33 of 77 patients (43%) in the ondansetron group compared to 50 of 82 patients (61%) in the placebo group (relative risk, 0.70 [95% CI, 0.51–0.95]; absolute risk difference, −18% [95% CI, −33 to −2]; NNT, 5.5 [95% CI, 3.0–58.4]; χ² test; P = .022). Figure 2A depicts the comparative incidence in both groups of PONV within the first 24 hours postextubation in the form of Kaplan-Meier “survival” curves.

Figure 2.: A, Kaplan-Meier “survival” curves depicting the percentage over time of patients free of the primary outcome, PONV within the first 24 h postextubation (*log-rank [Mantel-Cox] test). Four (5%) patients in the ondansetron group and 6 (7%) patients in the placebo group met the primary outcome before or at the time of extubation (Fisher exact test; P = .747); these patients were considered “at risk” and arbitrarily assigned an event time of 1 s. B, Kaplan-Meier curves depicting the percentage over time of patients receiving rescue antiemetic treatment for PONV within the first 24 h (*log-rank [Mantel-Cox] test). Two patients in the ondansetron group and 2 in the placebo group received a rescue agent before extubation; these patients were considered “at risk” and assigned an event time of 1 s. PONV indicates postoperative nausea and vomiting.

With regard to secondary outcomes, Kaplan-Meier “survival” analysis of the times to first rescue antiemetic treatment administration over the first 24 hours indicated that patients in the ondansetron group fared better than those in the placebo group (log-rank [Mantel-Cox] test; P = .028) (Figure 2B). Overall, 32 of 77 patients (42%) in the ondansetron group received rescue antiemetic treatment over the first 24 hours postextubation versus 47 of 82 patients (57%) in the placebo group (relative risk, 0.73 [95% CI, 0.52–1.00]; absolute risk difference, −16% [95% CI, −31 to 1]; χ² test; P = .047). Table 4 summarizes details on rescue antiemetic treatment: shown are the number of patients in each group who received 1, 2, or 3 or more different rescue antiemetic agents, respectively; the number of total rescue antiemetic doses administered irrespective of the type of agent; and the frequency with which particular types of antiemetic agents were administered. The most common ones were ondansetron and dimenhydrinate.

Table 4. - Rescue Antiemetic Treatment: Number of Total Doses, Number of Different Agents, and Types of Agents Administered

	Ondansetron (n = 77)	Placebo (n = 82)	P
Number of total doses, n (%)a			.252b
None	45 (58)	35 (43)
1	11 (14)	23 (28)
2	10 (13)	8 (10)
3 or more	11 (14)	15 (18)
Number of different agents, n (%)c			.054b
None	45 (58)	35 (43)
1	14 (18)	24 (29)
2	17 (22)	15 (18)
3 or more	1 (1)	8 (10)
Type of agent, n (%)d			.012e
None	45 (58)	35 (43)
Ondansetron	30 (39)	47 (57)
Dimenhydrinate	19 (25)	23 (28)
Dexamethasone	0 (0)	5 (6)
Haloperidol	1 (1)	5 (6)

^aShown is the number (percentage) of patients per category (number of rescue antiemetic doses administered within the first 24 h postextubation, regardless of the type of agent).

^bCochran-Armitage method (χ² test for trend).

^cShown is the number (percentage) of patients per category (number of different rescue antiemetic agent types administered within the first 24 h postextubation, regardless of the number of doses).

^dShown is the number (percentage) of patients who received a listed type of rescue antiemetic agent within the first 24 h postextubation, regardless of the number of doses.

^eχ² test (after combining the dexamethasone and haloperidol rows to meet sample size conditions for valid calculation; P value before = .024).

PON only (without vomiting) within the first 24 hours postextubation occurred in 22 of 77 patients (29%) in the ondansetron group versus 33 of 82 patients (40%) in the placebo group (relative risk, 0.79 [95% CI, 0.59–1.07]; absolute risk difference −12% [95% CI, −26 to 4]; χ² test; P = .122). We found no significant difference between the groups in the highest reported PON severity on a 0–10 numeric rating scale (ondansetron group: median, 0 [interquartile range {IQR}, 0–0; range, 0–10; n = 74 patients with data] versus placebo group: median, 0 [IQR, 0–3; range, 0–10; n = 82]; Mann-Whitney U test, P = .165). Among patients with PON only (ie, those with “zero” values [with no PON] excluded from the comparison), we similarly observed no significant difference (ondansetron group: median, 6 [IQR, 4–9; n = 14 patients with data] versus placebo group: median, 7 [IQR, 4–10; n = 23 patients with data]; Mann-Whitney U test, P = .65). There were no patients who experienced vomiting and/or retching that was not accompanied by nausea.

The number of patients who reported headache over the first 24 hours postextubation generally was low and there was no evidence of a difference between the groups (ondansetron group, 5 of 77 [6%] versus placebo group, 4 of 82 [5%]; Fisher exact test, P = .740). Among those patients who reported headache, we observed no difference between the groups in the highest reported headache severity (data not shown). There also was no significant difference between the ondansetron and placebo groups with regard to the incidence of postoperative ventricular arrhythmias (2 of 77 [3%] vs 4 of 82 [5%]; Fisher exact test; P = .68). There were no instances of ventricular arrhythmias lasting >30 seconds or requiring treatment other than magnesium.

DISCUSSION

The present randomized controlled trial showed that a single prophylactic dose of ondansetron given at the time of cessation of postoperative propofol sedation significantly reduced the rate of PONV in the first 24 hours postextubation after cardiac surgery. Somewhat remarkably, this is the first trial to our knowledge to study the routine postoperative administration of ondansetron in this patient population, as a single dose and with a standardized timing.

Consistent with previous studies,^3,6–8,12 the rate of PONV was considerable (albeit higher than expected following our prior audit). In the placebo group, 61% of patients experienced PONV within the first 24 hours, despite the use of a propofol infusion intra- and postoperatively, a population that overwhelmingly comprised older men, and having excluded patients with a history of PONV.²² Contributing factors might include the PONV definition used,²⁰ the proportion of nonsmokers, the high opioid doses during cardiac surgery, the duration of sevoflurane exposure, opioid analgesia in the CSICU, and other postoperative events.²³ Of note, 2 studies on fast-track cardiac anesthesia^5,13 and a recent retrospective registry data analysis²⁴ have reported lower PONV rates; although differences in definitions, cohorts, and management may be factors, the incidences in the latter of 0.38%–0.41% appear implausibly low. Nonetheless, our findings combined with those in the literature collectively support a call for future research aimed to elucidate if cardiac surgery represents an independent risk factor for PONV. A recent systematic review of risk factors that examined 13 types of surgery did not include cardiac surgery.²³

Despite the comparatively low-risk population in our study, a single postoperative dose of ondansetron produced a sizeable relative risk reduction of 30% and an absolute risk reduction (ARR) of 18% that translated to an NNT of 5.5, which is remarkably consistent with the findings of a systematic review of noncardiac surgery trials published in 1997.²⁵ Our results also are similar to those of another placebo-controlled trial in which ondansetron was given both intra- and postoperatively (4 mg at the end of cardiac surgery followed by 12 mg added to postoperative patient-controlled analgesia [PCA])³; the PONV incidence at 48 hours was 46% in the ondansetron group versus 71% in the placebo group (ARR, 25%). However, the PCA use limits widespread applicability and direct comparisons to our findings because the majority of cardiac surgery patients at our institution are not prescribed PCA. Furthermore, the total ondansetron dose inherently was variable due to individual PCA use differences, and the relative contribution of the 2 dosing components remains uncertain.

Two other active-control trials have studied ondansetron in cardiac surgery; one investigated preoperative administration compared to metoclopramide⁸ and another intraoperative dosing compared to midazolam.¹⁴ We opted for postoperative administration because we felt that this best aligns with IV ondansetron’s pharmacokinetics.¹⁸ We selected the timing of the cessation of propofol infusion because of the inherent variability in extubation times (cf Table 3). Our results support this approach and are consistent with the body of evidence from the realm of ambulatory/noncardiac surgery showing that ondansetron is most effective when administered at the end of surgery.²⁶ We chose 4 mg as the most commonly studied and lowest effective dose to minimize the risks of adverse effects including QTc prolongation.^27–32 We observed no differences in adverse effects as very few patients experienced headache or ventricular arrhythmias.

Given the body of evidence on ondansetron prophylaxis in noncardiac surgery,⁴ it may be argued that it is ethically questionable to conduct a placebo-controlled trial. However, it has not been standard practice at our institution to administer antiemetics prophylactically as part of routine cardiac anesthesia care, similar to other North American and international centers surveyed. The most recent iteration of the Consensus Guidelines for the Management of Postoperative Nausea and Vomiting does not mention cardiac surgery.⁴ We are unaware of cardiac anesthesia practice guidelines recommending the routine use of prophylactic antiemetics. In contrast, 2 studies have concluded that prophylactic antiemetic administration is unnecessary in fast-track cardiac anesthesia.^5,13 The present results would indicate the opposite.

Despite the strengths of our trial, which include its pragmatic design, detection of a clinically meaningful effect size, and real-world applicability, it has several limitations. First, it was a single-center study that was not powered for a robust comparison of adverse events or other hard outcomes reflecting morbidity or mortality; the number needed to harm remains unknown. Our study also was not powered to identify differences between men and women. The detected ARR was just below our prespecified goal of 20%. Second, we selected a 24-hour time frame for the primary outcome to reflect typical length of CSICU stay and account for the variability of extubation times. While the median CSICU stay approximated 24 hours in both groups, slightly more than half of the patients were transferred before 24 hours. It is possible that patients developed PONV afterward; the incidence at 48 hours is unknown. However, it is our cardiac surgery pathway’s standard of care that PONV symptoms must be controlled before ward transfer. Further, Figure 2 indicates that while ondansetron’s effect persisted throughout the 24-hour time frame, the separation of the Kaplan-Meier curves mainly took place within the first ~10 hours, which would be consistent with ondansetron’s expected duration of action. Third, the results may have been affected by the use of propofol infusions,³³ postoperative drugs and/or events, as well as rescue medication administration. However, the treatment protocol was the same for both groups and followed our institutional standard; the present trial was not intended or designed to identify postoperative risk factors in the CSICU. Fourth, a history of motion sickness was not an exclusion criterion; nonetheless, the number of patients who reported this was small and balanced between the groups. Finally, we cannot rule out that a higher dose, different timing, or repeated administration of ondansetron would have produced a larger effect size.²⁵ Similarly, it remains unknown whether alternative agents such as aprepitant or palonosetron would have been more effective.^34,35 Noteworthy in this regard is a recent report that identified intraoperative steroid administration as a risk factor for PONV after cardiac surgery.²² Overall, the 43% residual incidence in the treatment group also raises the possibility that combination therapy may further reduce the risk of PONV in this population.³⁶ Clearly, room exists for future studies to further optimize PONV prophylaxis in cardiac surgery patients.

In conclusion, a single prophylactic dose of ondansetron given at the time of cessation of postoperative propofol sedation reduced the rate of PONV after cardiac surgery without apparent adverse effects. Compared to the relatively scarce previous research on ondansetron prophylaxis in cardiac surgery, our study was the first to use it postoperatively, as a single dose and with a standardized timing. The results reflect a context of real-world practice and demonstrate that this effective approach is feasible, pragmatic, and easily implementable into standardized cardiac surgery pathways. Despite the observed benefit of ondansetron, the routine use of propofol, and a relatively low-risk cohort, the residual PONV incidence was still high, warranting further research dedicated to the cardiac surgery population.

ACKNOWLEDGMENTS

The authors are grateful for the invaluable assistance of the St Paul’s Hospital Pharmacy Department staff in randomization, study drug preparation, and dispensing; and the CSICU nursing staff as well as the members of the Department of Anesthesia for their tireless support of the conduct of this trial. Finally, the authors thank Dr Gregory L. Bryson (Department of Anesthesiology and Pain Medicine, University of Ottawa, Ottawa, ON, Canada) for helpful comments.

DISCLOSURES

Name: Erica H. Z. Wang, PharmD.

Contribution: This author helped design the study; acquire and interpret the data; write, critically revise, and finalize the manuscript.

Name: Sarah Sunderland, MD, FRCPC.

Contribution: This author helped conceive and design the study, critically revise, and finalize the manuscript.

Name: Nicola Y. Edwards, BHSc (Hons).

Contribution: This author helped acquire, analyze, and interpret the data; critically revise; and finalize the manuscript.

Name: Navraj S. Chima, MD, FRCPC.

Contribution: This author helped conceive and design the study, critically revise, and finalize the manuscript.

Name: Cynthia H. Yarnold, MD, FRCPC.

Contribution: This author helped conceive and design the study, interpret the data, critically revise, and finalize the manuscript.

Name: Stephan K. W. Schwarz, MD, PhD, FRCPC.

Contribution: This author helped conceive and design the study; analyze and interpret the data; write, critically revise, and finalize the manuscript.

Name: Matthew A. Coley, MD, FRCPC.

Contribution: This author helped conceive and design the study; acquire and interpret the data; write, critically revise, and finalize the manuscript.

This manuscript was handled by: Tong J. Gan, MD.