Postoperative nausea and vomiting (PONV), the big “little problem”,1 is the most common postoperative medical problem. From the perspective of both patients and healthcare providers, prophylaxis of PONV is equally as important as postoperative analgesia.2,3 The incidence of PONV is estimated at 20-40% in adults,4 and for patients with multiple risk factors, the rate of PONV can be as high as 79%.5 Postoperative nausea and vomiting not only decreases patient satisfaction but also relates to rare but severe adverse consequences, including pulmonary aspiration, wound dehiscence, esophageal rupture, subcutaneous emphysema, and bilateral pneumothoraces.6

The main strategy for PONV prophylaxis is antiemetic drug therapy. The 5-hydroxytryptamine -3 receptor antagonists (5-HT3RAs) are the most popular pharmacologic class of antiemetics for PONV. The first-generation 5-HT3RAs have been shown to have similar efficacy and safety for PONV prophylaxis during the first 24 hr after surgery. The number needed to treat (NNT) with 5-HT3RAs in order to prevent one additional patient from experiencing nausea, vomiting, or PONV compared with placebo has been reported to be approximately 7, 6, and 3, respectively.711 Ondansetron was the first available 5-HT3RA, and its wide use in the prevention of PONV may be related to its relatively lower cost compared with other agents in the same class.7,8

Numerous studies of different prophylactic agents have been published over the past 50 years; however, a significant percentage of surgical patients still suffer from PONV.12 New antiemetic drugs continue to be introduced, including palonosetron, a 5-HT3RA approved by the United States Food and Drug Administration (FDA) in 2008 for PONV prophylaxis up to 24 hr. In 2012, palonosetron was recommended for authorization by Health Canada for the prevention of acute and delayed chemotherapy-induced nausea and vomiting. Nevertheless, palonosetron has not yet been approved for the prophylaxis or treatment of PONV in Canada.

Compared with the first-generation 5-HT3RAs, studies have shown that palonosetron has greater receptor binding affinity to 5-HT3R and a longer plasma half-life.13,14 In addition, Rojas et al.1517 showed features that differentiate palonosetron from other 5-HT3RAs: allosteric binding and positive cooperativity with 5-HT3R, triggering of 5-HT3R internalization and prolonged inhibition of receptor function, and inhibition of substance P (a mediator of emesis)-mediated responses.1517 Due to these unique characteristics, palonosetron may be a promising agent that could achieve long-lasting efficacy in the prevention of PONV. Importantly, preliminary studies have already shown that palonosetron is safer and more effective than other 5-HT3RAs in preventing chemotherapy-induced nausea and vomiting, although this requires more rigorous confirmation.18,19

In consideration of the above, we conducted this systematic review and meta- analysis to investigate whether the recommended palonosetron dose (0.075 mg iv)20 has better efficacy than ondansetron (≥ 4 mg iv), a widely used first-generation 5-HT3RA, for the prophylaxis of nausea and vomiting during the first 24 hr following surgery.


This systematic review and meta-analysis included randomized controlled trials (RCTs) that compared the efficacy of palonosetron with ondansetron in the prevention of PONV. We reported the study in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement.21,22

Systematic search and strategy

We searched the MEDLINE®, EMBASE® Cochrane Central Register of Controlled Trials (CENTRAL), and Web of Science™ databases up to January 6, 2015 without language limitations. We also searched the reference lists of included studies and grey literature using the System for Information on Grey Literature in Europe (SIGLE) database to identify potential RCTs.

The search strategy consisted of a combination of free text words and Medical Subject Headings (MeSH) terms as follows: “palonosetron”, “drug therapy”, “postoperative”, “postanaesthetic”, “postanesthetic”, “surgical”, “nausea”, “emesis”, “vomiting”, “retching”, “randomized controlled trial”, “controlled clinical trial”, “randomized”, “randomly”, and “trial”. Details of our search strategy are provided in the Appendix.

Eligibility criteria

Studies were included in the systematic review if they were RCTs that evaluated the effect of palonosetron 0.075 mg relative to that of ondansetron ≥ 4 mg for the prophylaxis of PONV. Participants were adults (18 yr or older) receiving any type of elective surgery involving general anesthesia and palonosetron or ondansetron administration by intravenous injection, regardless of the timing.

Exclusion criteria were studies that did not define nausea or vomiting; studies that did not report the primary outcome separately in the early and late postoperative periods; studies of treatments for established PONV; or data from meeting abstracts, reviews, posters, case reports, comments, letters to the editor, and animal studies.

The primary outcome measures were the proportion of participants who experienced postoperative nausea (PON) or postoperative vomiting (POV) in the early or late period during the first 24 hr. If the 24-hr follow-up period was divided into two separate periods (e.g., 0-6 and 6-24 hr), we defined the first as the early period and the second as the late period. If the follow-up period was divided into three periods (e.g., 0-2, 2-8, and 8-24 hr), we defined the second as the early period and the third as the late period. The secondary outcome measures were the proportion of participants who experienced common adverse effects and the proportion of participants who received rescue antiemetics.

Nausea was defined as the subjectively unpleasant feeling associated with an awareness of the urge to vomit. Both vomiting (an actual physical phenomenon – the forceful expulsion of gastric contents from the mouth) and retching (laboured, spasmodic contractions of the respiratory muscle without expulsion of gastric contents) were defined as vomiting.23 We did not consider these definitions as mandatory inclusion criteria of an RCT.

Study selection and data collection

Two reviewers (C.X. and G.Y.L.) independently scanned the titles and abstracts of all identified citations to exclude duplicates and those that obviously failed to meet our inclusion criteria. Full articles of the selected citations were retrieved for further screening. The two reviewers resolved all discrepancies through discussion, and if necessary, they contacted another author (A.S.W.) for a decision. We used the PRISMA flow diagram to summarize the processes of study selection.

The following data were extracted from each study: name of first author; publication year; country; sample size; age of participants and their American Society of Anesthesiologists physical status; risk factors for PONV;5 anesthetic technique; type of surgery; dose and timing of target drug administration; dosage of administration. We also extracted the number of participants who experienced PON or POV in the early or late periods, those who received rescue antiemetics in the early or late periods, and those who experienced common adverse effects. All graphical data were converted into numerical data. Two reviewers (C.X. and G.Y.L.) were responsible for the extraction of all data mentioned above and another reviewer (J.X.X.) checked the extracted data.

Risk of bias in individual studies

We evaluated the overall risk of bias in individual studies with regard to the adequacy of randomization, concealment of allocation, blinding (of patients, healthcare providers, data collectors, and outcome assessors), incomplete outcome data, selective outcome reporting, and other sources of bias.

Summary measures and statistical analyses

Data of all primary and secondary outcomes were binary, and we used relative risks (RRs) and the NNTs with corresponding 95% confidence intervals (CIs) as the summary measures. If the 95% CI included the value 1.0, we assumed that the difference between the two groups was not statistically significant. The I2 test was used to measure heterogeneity. For I2 ≤ 40% and I2 > 40%, fixed effects and random effects models were used, respectively, (Mantel-Haenszel method) to compute the summary measures.24 The Egger regression asymmetry test was used to evaluate the existence of publication bias.

Substantial publication bias was considered to exist with P < 0.05 in the regression asymmetry test. If substantial heterogeneity (I2 > 40%) existed, we considered the possible reasons and performed a sensitivity analysis for the primary outcomes according to methodological quality, dose of ondansetron, type of anesthesia, and method for the combination of intervention effect estimates across studies. We used the software package STATA® (version 12.0) to conduct all statistical analyses.


Study selection

Initially, we identified 179 citations by searching the MEDLINE, EMBASE, CENTRAL, and Web of Science databases, and we identified three additional citations through other sources. Of these, 83 were duplications and were excluded. Next, we scanned the abstracts of the remaining 99 citations and found that 83 clearly did not meet our inclusion criteria. We then retrieved the full texts of the remaining 16 citations and excluded seven of these citations because five among them had no intended interventions and two had no outcomes of interest. Altogether, nine studies2533 fulfilled the criteria for systematic review and meta-analysis. The study selection processes are shown in Fig. 1.

Fig. 1
figure 1

Study selection flow diagram

Study characteristics

The studies in this review comprised a total of 741 eligible participants; 371 patients received palonosetron 0.075 mg iv and 370 patients received ondansetron ≥ 4 mg iv. Three studies involved ondansetron 4 mg and six studies involved ondansetron 8 mg. Five studies were conducted in Korea and four in India.

These studies scored their participants for risk factors for PONV using Apfel’s simplified risk factor scoring system.5 Among these studies, patients’ scores varied. Inhalation anesthesia was performed in eight studies and total intravenous anesthesia (TIVA) was used in one study. Palonosetron was administered before induction of anesthesia in eight studies and at the end of surgery in one study. Table 1 shows the characteristics of all included studies.

Table 1 Characteristics of included studies

Risk of bias within studies

The overall risk of bias within studies was low in two studies and unclear in seven (Fig. 2)

Fig. 2
figure 2

Summary of the risk of bias of the included studies

Syntheses of results

Fig. 3 shows that palonosetron 0.075 mg iv was significantly more effective than ondansetron for the prevention of early PON (RR, 0.51; 95% CI, 0.37 to 0.71; I2 = 36%) and late PON (RR, 0.53; 95% CI, 0.36 to 0.77; I2 = 56%). Fig. 4 indicates that palonosetron was more effective than ondansetron in preventing late POV (RR, 0.41; 95% CI, 0.28 to 0.62; I2 = 0%) but was similarly effective in the prevention of early POV (RR, 0.77; 95% CI, 0.45 to 1.34; I2 = 8%). As both the relative treatment effect and the prevalence of disease determine the clinical effectiveness of a given treatment, we estimated the NNT. The numbers needed to treat for early PON (NNT, 9.3; 95% CI, 6.4 to 31.7), late PON (NNT, 5.4; 95% CI, 4.0 to 11.1), and late POV (NNT, 9.7; 95% CI, 7.6 to 17.0) characterized the benefit of palonosetron when compared with ondansetron.

Fig. 3
figure 3

Forest plot of meta-analysis for palonosetron vs ondansetron: (A) Early postoperative nausea (PON); (B) late PON. CI = confidence interval; RR = relative risk

Fig. 4
figure 4

Forest plot of meta-analysis for palonosetron vs ondansetron: (C) Early postoperative vomiting (POV); (D) late POV. CI = confidence interval; RR = relative risk

Six studies compared the need for rescue antiemetics between the treatment groups. The combined results of these studies failed to identify a statistically significant difference between palonosetron and ondansetron in the need of rescue antiemetics, either in the early period (RR, 0.67; 95% CI, 0.40 to 1.13; I2 = 0%) or in the late period (RR, 0.59; 95% CI, 0.30 to 1.20; I2 =50 %). The meta-analysis showed no difference between palonosetron and ondansetron with regard to the rates of common adverse effects (Table 2), such as headache (RR, 0.73; 95% CI, 0.52 to 1.03; I2 = 0%), dizziness (RR, 0.84; 95% CI, 0.56 to 1.25; I2 = 0%), or drowsiness (RR, 1.00; 95% CI, 0.53 to 1.87; I2 = 0%).

Table 2 Comparison of incidences of adverse effects between palonosetron and ondansetron

Publication bias and sensitivity analysis

None of the results of the Egger regression asymmetry tests for early PON, late PON, early POV, or late POV achieved statistical significance (P = 0.50, 0.14, 0.16, 0.34, respectively).

We performed four separate sensitivity analyses (Table 3) as follows: (1) excluding studies with an unclear or high risk of bias, (2) studies in which more than 4 mg of ondansetron were administered, (3) studies in which TIVA was used, and (4) a comparison of the fixed and random-effects estimates of the efficacy of palonosetron relative to ondansetron. In the subgroup analysis comparing different doses of ondansetron with palonosetron, each primary outcome showed that the RR was lower in the 4-mg group than in the 8-mg group, and the I2 value was lower in each of the 4-mg and 8-mg groups than in the two groups combined.

Table 3 Sensitivity analysis


With nine included RCTs, our study investigated the efficacy and safety profile of palonosetron relative to ondansetron for PONV prophylaxis during the first 24 hr. Our meta-analysis shows that palonosetron 0.075 mg iv compared with ondansetron ≥ 4 mg significantly reduces the incidence of early PON (RR, 0.51; 95% CI, 0.37 to 0.71), late PON (RR, 0.53; 95% CI, 0.36 to 0.77), and late POV (RR, 0.41; 95% CI 0.28 to 0.62) but shows no difference between groups for early POV (RR, 0.77; 95% CI, 0.45 to 1.34). Furthermore, palonosetron did not reduce the need for rescue antiemetics in either the early or the late postoperative periods. The NNTs of palonosetron are large, and the corresponding 95% CIs are fairly wide, raising a question regarding the clinical impact of these findings. Among all the included trials, the overall baseline risk for PONV (based on the 2014 consensus guidelines)20 was variable. As the NNT is likely affected by the baseline incidence of PONV, the same intervention might achieve a smaller NNT if it included more studies with high-risk participants.

Our findings may be explained partially by the longer half-life and receptor binding characteristics of palonosetron.1317 Palonosetron may also have other advantages for perioperative care. For example, a recent study reported by Cho et al.34 found that palonosetron 0.075 mg iv administered before induction of anesthesia reduced injection pain associated with rocuronium administration; the NNT was 2.94. Ondansetron was also found to be effective in reducing injection pain, with an NNT (2.78) similar to that of palonosetron,35 but it might be inappropriate to administer ondansetron before induction due to its short duration of action.36

The recommended intravenous dose of ondansetron is 4 mg,20 and the combined results of the dose-ranging studies11 suggest no significant difference between 4 mg and 8 mg. Thus, we included studies in which at least 4 mg of ondansetron were administered in this meta-analysis. Nevertheless, in our subgroup analysis comparing palonosetron and ondansetron 4 mg, the difference between the two agents may be even bigger. This may explain some of the heterogeneity in this meta-analysis when combining studies of all doses of ondansetron.

The common adverse effects of palonosetron include headache, dizziness, drowsiness, constipation, and bradycardia.37 Our study showed that palonosetron was no different than ondansetron in the rates of headache, dizziness, or drowsiness. Previous meta-analyses showed that other 5-HT3RAs, including ondansetron, granisetron, and ramosetron, had similar rates of adverse effects when used for the prevention of PONV,10,38 and use of palonosetron did not result in a higher incidence of adverse effects compared with other 5-HT3RAs for the prevention of chemotherapy-induced nausea and vomiting.18,19 Nevertheless, we cannot easily draw a conclusion regarding the safety profile of palonosetron because the studies in this meta-analysis did not report the incidence of all common adverse effects or any rare but severe adverse events.

Reports of recent clinical studies30,3941 showed that granisetron, ramosetron, and palonosetron had similar rates of associated adverse effects when used to prevent PONV, and in general, the incidence of common adverse events seemed to be similar among different 5-HT3RAs. With regard to serious adverse events, however, significant differences between palonosetron and ondansetron may exist. For example, in 2011, the FDA announced that the use of ondansetron may bring about abnormal heart rhythms42 by prolonging the QT interval. This may lead to the potentially fatal abnormal heart rhythm, Torsade de Pointes, reported in some patients receiving ondansetron.43 In contrast, palonosetron thus far appears to be safe in terms of QT intervals. Results of a number of studies4446 revealed that palonosetron had no effect on the QT interval during surgery.

Serotonin syndrome is another serious adverse event associated with the use of 5-HT3RAs.47 Serotonin syndrome is a potentially life-threatening condition, even in young patients, and should be avoided in the perioperative settings. Importantly, according to the FDA review,47 29 cases of serotonin syndrome were reported with the use of ondansetron, but no such events have been associated with the use of palonosetron. Thus, palonosetron may be safer than ondansetron with regard to serious adverse events, but this idea requires further investigation.

Our study has several limitations. Both the number of RCTs included in our meta-analysis and the sample size were small, and all patients were Asian (Indian or Korean). Furthermore, the risk of bias within studies was low in only two studies, and we did not investigate the efficacy of prophylaxis with palonosetron combined with other antiemetics for PONV. The latter warrants investigation because patients at high risk often require more aggressive intervention.20

Baseline risks for PONV were not identical among these included trials, and the definitions for the postoperative early and late periods varied. Our study did not consider patient satisfaction or intensity of nausea as primary or secondary outcomes because the scale used to measure these outcomes differed among studies and we could not find an appropriate way to combine the data. We also did not include economic data as an outcome, although it is important to consider since palonosetron is more costly than other drugs in its class.

We based our study on the presumption that the optimal intravenous dose of palonosetron was 0.075 mg; equipotent doses of palonosetron and ondansetron are unknown. In 2008, the FDA approved a single intravenous dose of palonosetron for PONV prophylaxis for up to 24 hours based on the results of two earlier dose-ranging studies.48,49 Subsequently, most studies2533 chose palonosetron 0.075 mg as the dose to administer in their trials, which makes it difficult to re-evaluate the optimal dose of palonosetron.

A final limitation of the present meta-analysis is that, although palonosetron holds promise for application during the postoperative 24-72 hr, we did not investigate its prophylactic efficacy for that period. In previous studies reported by Candiotti et al.45 and Kovac et al.,46 there was no statistical difference between palonosetron 0.075 mg and placebo in the reduction of PONV during the postoperative 24-72 hr. Compared with ondansetron, the combined results of four studies indicated that palonosetron can significantly reduce the incidence of PON (RR, 0.53; 95% CI, 0.33 to 0.86) but not POV within 24 to 48 or 72 hr after surgery.25,28,30,31 Given that the overall rate of POV (0-10%) is low at the 24-48 hr or 24-72 hr, these studies comparing palonosetron with placebo or ondansetron suffered from small sample sizes, and it was difficult to show a statistically significant difference. Thus, it is inappropriate to make conclusions regarding palonosetron’s efficacy during the postoperative 24-72 hr; it calls for more RCTs to address this question.

In conclusion, compared with ondansetron, palonosetron 0.075 mg iv showed statistically better efficacy for the prophylaxis of early PON, late PON, and late POV, but there may be questions regarding the clinical significance of these results due to the large NNTs. Future studies should focus on evaluating the safety profile of palonosetron and investigate its prophylactic efficacy for PONV in the postoperative 24-72 hr.