Which medicines work best to prevent people from being sick (vomiting) after an operation?

Why are people sick after an operation?

Feeling sick (nausea) or being sick (vomiting) is a common unwanted effect of general anaesthesia—medicine that makes people unconscious and unresponsive so they don't move or feel pain during an operation.

Most unwanted effects of general anaesthesia, including feeling or being sick, happen immediately and stop after a few hours, although some people may continue to feel sick for up to a day. If people carry on feeling or being sick, they might have to stay in hospital longer than expected and may experience other unwanted effects or complications.

Women are more likely to be sick after an operation, as are people taking opioid painkillers, those who have had motion sickness, and those who have been sick after previous operations. 

Medicines to prevent people from being sick

Medicines called antiemetics are given to prevent people from feeling or being sick. These medicines may be given before or during anaesthesia.

Antiemetic medicines are grouped into six main classes based on how they act. Combining medicines from different classes sometimes makes them work better.

Why we did this Cochrane Review

We wanted to find out which medicines work best to prevent people from being sick after an operation and cause the fewest unwanted effects. Some unwanted effects of antiemetic medicines include headache, constipation, movement disorders such as tremors, sleepiness, irregular heartbeat, and wound infection.

What did we do?

We searched for studies that looked at the use of antiemetic medicines in adults having general anaesthesia to prevent people from being sick afterwards.  

We looked for randomized controlled studies, in which the treatments people received were decided at random. This type of study usually gives the most reliable evidence about the effects of a treatment.

Search date

We included evidence published up to November 2017; in April 2020, we found another 39 studies, which are not yet included in the analysis.

What we found

We found 585 studies in 97,516 people who were given antiemetic medicines before or during general anaesthesia. People included in the studies were more likely to be sick after anaesthesia, as 83% were women and 88% were taking opioid painkillers. Most studies were conducted in Asia, Europe, or North America.

These studies either measured how many people were sick in the first 24 hours after their operation or how many unwanted effects were reported, or both of these outcomes. Most studies compared medicines (given alone or in combination) with a dummy (placebo) treatment.

We compared all antiemetic medicines with each other using a mathematical method called network meta-analysis.

What were our main results and how reliable are these results?

Compared with placebo treatment, 10 out of 28 single medicines and 29 out of 36 combinations of medicines prevented people from being sick in the first 24 hours after their operation (282 studies). Combinations of antiemetic medicines generally worked better than single medicines given alone. However, aprepitant, casopitant, and fosaprepitant worked as well alone as most combinations of antiemetics. The single medicine that worked best in the ranking of all medicines was fosaprepitant, followed by casopitant, aprepitant, ramosetron, granisetron, dexamethasone, tropisetron, ondansetron, dolasetron, and droperidol.

We are confident that aprepitant, ramosetron, granisetron, dexamethasone, and ondansetron prevent people from being sick. We are moderately confident about how well fosaprepitant and droperidol work, but this finding may change when further evidence becomes available. We are uncertain about how well casopitant, tropisetron, and dolasetron work.

Not all studies looked at serious, life-threatening unwanted effects. We are uncertain how many of these effects were reported when taking an antiemetic medicine and whether serious, life-threatening unwanted effects occur at a similar rate or are reduced compared to placebo (28 studies).

Of the best medicines with most reliable evidence for preventing being sick, granisetron and ondansetron probably made little to no difference in the occurrence of unwanted effects compared to placebo, whereas dexamethasone and droperidol may cause fewer unwanted effects than placebo. We are uncertain about unwanted effects with aprepitant and ramosetron (61 studies). We found no studies looking at unwanted effects for fosaprepitant.

We are less confident about unwanted effects of other antiemetic medicines because we found little reliable evidence about this. Our results for unwanted effects are likely to change when further evidence is available.


For people at higher risk, we found that some antiemetic medicines work well to prevent them from being sick after general anaesthesia. The best antiemetic medicines with reliable evidence were aprepitant, ramosetron, granisetron, dexamethasone, and ondansetron, followed by fosaprepitant and droperidol.

However, we did not find enough reliable evidence about potential unwanted effects to rank these medicines reliably according to how well they are tolerated.

Authors' conclusions: 

We found high-certainty evidence that five single drugs (aprepitant, ramosetron, granisetron, dexamethasone, and ondansetron) reduce vomiting, and moderate-certainty evidence that two other single drugs (fosaprepitant and droperidol) probably reduce vomiting, compared to placebo. Four of the six substance classes (5-HT₃ receptor antagonists, D₂ receptor antagonists, NK₁ receptor antagonists, and corticosteroids) were thus represented by at least one drug with important benefit for prevention of vomiting. Combinations of drugs were generally more effective than the corresponding single drugs in preventing vomiting. NK₁ receptor antagonists were the most effective drug class and had comparable efficacy to most of the drug combinations. 5-HT₃ receptor antagonists were the best studied substance class. For most of the single drugs of direct interest, we found only very low to low certainty evidence for safety outcomes such as occurrence of SAEs, any AE, and substance class-specific side effects.

Recommended and high doses of granisetron, dexamethasone, ondansetron, and droperidol were more effective than low doses for prevention of vomiting. Dose dependency of side effects was rarely found due to the limited number of studies, except for the less sedating effect of recommended and high doses of ondansetron.

The results of the review are transferable mainly to patients at higher risk of nausea and vomiting (i.e. healthy women undergoing inhalational anaesthesia and receiving perioperative opioids). Overall study quality was limited, but certainty assessments of effect estimates consider this limitation. No further efficacy studies are needed as there is evidence of moderate to high certainty for seven single drugs with relevant benefit for prevention of vomiting. However, additional studies are needed to investigate potential side effects of these drugs and to examine higher-risk patient populations (e.g. individuals with diabetes and heart disease).

Read the full abstract...

Postoperative nausea and vomiting (PONV) is a common adverse effect of anaesthesia and surgery. Up to 80% of patients may be affected. These outcomes are a major cause of patient dissatisfaction and may lead to prolonged hospital stay and higher costs of care along with more severe complications. Many antiemetic drugs are available for prophylaxis. They have various mechanisms of action and side effects, but there is still uncertainty about which drugs are most effective with the fewest side effects.


• To compare the efficacy and safety of different prophylactic pharmacologic interventions (antiemetic drugs) against no treatment, against placebo, or against each other (as monotherapy or combination prophylaxis) for prevention of postoperative nausea and vomiting in adults undergoing any type of surgery under general anaesthesia

• To generate a clinically useful ranking of antiemetic drugs (monotherapy and combination prophylaxis) based on efficacy and safety

• To identify the best dose or dose range of antiemetic drugs in terms of efficacy and safety

Search strategy: 

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP), ClinicalTrials.gov, and reference lists of relevant systematic reviews. The first search was performed in November 2017 and was updated in April 2020. In the update of the search, 39 eligible studies were found that were not included in the analysis (listed as awaiting classification).

Selection criteria: 

Randomized controlled trials (RCTs) comparing effectiveness or side effects of single antiemetic drugs in any dose or combination against each other or against an inactive control in adults undergoing any type of surgery under general anaesthesia. All antiemetic drugs belonged to one of the following substance classes: 5-HT₃ receptor antagonists, D₂ receptor antagonists, NK₁ receptor antagonists, corticosteroids, antihistamines, and anticholinergics. No language restrictions were applied. Abstract publications were excluded.

Data collection and analysis: 

A review team of 11 authors independently assessed trials for inclusion and risk of bias and subsequently extracted data. We performed pair-wise meta-analyses for drugs of direct interest (amisulpride, aprepitant, casopitant, dexamethasone, dimenhydrinate, dolasetron, droperidol, fosaprepitant, granisetron, haloperidol, meclizine, methylprednisolone, metoclopramide, ondansetron, palonosetron, perphenazine, promethazine, ramosetron, rolapitant, scopolamine, and tropisetron) compared to placebo (inactive control). We performed network meta-analyses (NMAs) to estimate the relative effects and ranking (with placebo as reference) of all available single drugs and combinations. Primary outcomes were vomiting within 24 hours postoperatively, serious adverse events (SAEs), and any adverse event (AE). Secondary outcomes were drug class-specific side effects (e.g. headache), mortality, early and late vomiting, nausea, and complete response. We performed subgroup network meta-analysis with dose of drugs as a moderator variable using dose ranges based on previous consensus recommendations. We assessed certainty of evidence of NMA treatment effects for all primary outcomes and drug class-specific side effects according to GRADE (CINeMA, Confidence in Network Meta-Analysis). We restricted GRADE assessment to single drugs of direct interest compared to placebo.

Main results: 

We included 585 studies (97,516 randomized participants). Most of these studies were small (median sample size of 100); they were published between 1965 and 2017 and were primarily conducted in Asia (51%), Europe (25%), and North America (16%). Mean age of the overall population was 42 years. Most participants were women (83%), had American Society of Anesthesiologists (ASA) physical status I and II (70%), received perioperative opioids (88%), and underwent gynaecologic (32%) or gastrointestinal surgery (19%) under general anaesthesia using volatile anaesthetics (88%).

In this review, 44 single drugs and 51 drug combinations were compared. Most studies investigated only single drugs (72%) and included an inactive control arm (66%). The three most investigated single drugs in this review were ondansetron (246 studies), dexamethasone (120 studies), and droperidol (97 studies).

Almost all studies (89%) reported at least one efficacy outcome relevant for this review. However, only 56% reported at least one relevant safety outcome.

Altogether, 157 studies (27%) were assessed as having overall low risk of bias, 101 studies (17%) overall high risk of bias, and 327 studies (56%) overall unclear risk of bias.

Vomiting within 24 hours postoperatively

Relative effects from NMA for vomiting within 24 hours (282 RCTs, 50,812 participants, 28 single drugs, and 36 drug combinations) suggest that 29 out of 36 drug combinations and 10 out of 28 single drugs showed a clinically important benefit (defined as the upper end of the 95% confidence interval (CI) below a risk ratio (RR) of 0.8) compared to placebo. Combinations of drugs were generally more effective than single drugs in preventing vomiting. However, single NK₁ receptor antagonists showed treatment effects similar to most of the drug combinations. High-certainty evidence suggests that the following single drugs reduce vomiting (ordered by decreasing efficacy): aprepitant (RR 0.26, 95% CI 0.18 to 0.38, high certainty, rank 3/28 of single drugs); ramosetron (RR 0.44, 95% CI 0.32 to 0.59, high certainty, rank 5/28); granisetron (RR 0.45, 95% CI 0.38 to 0.54, high certainty, rank 6/28); dexamethasone (RR 0.51, 95% CI 0.44 to 0.57, high certainty, rank 8/28); and ondansetron (RR 0.55, 95% CI 0.51 to 0.60, high certainty, rank 13/28). Moderate-certainty evidence suggests that the following single drugs probably reduce vomiting: fosaprepitant (RR 0.06, 95% CI 0.02 to 0.21, moderate certainty, rank 1/28) and droperidol (RR 0.61, 95% CI 0.54 to 0.69, moderate certainty, rank 20/28).

Recommended and high doses of granisetron, dexamethasone, ondansetron, and droperidol showed clinically important benefit, but low doses showed no clinically important benefit. Aprepitant was used mainly at high doses, ramosetron at recommended doses, and fosaprepitant at doses of 150 mg (with no dose recommendation available).

Frequency of SAEs

Twenty-eight RCTs were included in the NMA for SAEs (10,766 participants, 13 single drugs, and eight drug combinations). The certainty of evidence for SAEs when using one of the best and most reliable anti-vomiting drugs (aprepitant, ramosetron, granisetron, dexamethasone, ondansetron, and droperidol compared to placebo) ranged from very low to low. Droperidol (RR 0.88, 95% CI 0.08 to 9.71, low certainty, rank 6/13) may reduce SAEs. We are uncertain about the effects of aprepitant (RR 1.39, 95% CI 0.26 to 7.36, very low certainty, rank 11/13), ramosetron (RR 0.89, 95% CI 0.05 to 15.74, very low certainty, rank 7/13), granisetron (RR 1.21, 95% CI 0.11 to 13.15, very low certainty, rank 10/13), dexamethasone (RR 1.16, 95% CI 0.28 to 4.85, very low certainty, rank 9/13), and ondansetron (RR 1.62, 95% CI 0.32 to 8.10, very low certainty, rank 12/13). No studies reporting SAEs were available for fosaprepitant.

Frequency of any AE

Sixty-one RCTs were included in the NMA for any AE (19,423 participants, 15 single drugs, and 11 drug combinations). The certainty of evidence for any AE when using one of the best and most reliable anti-vomiting drugs (aprepitant, ramosetron, granisetron, dexamethasone, ondansetron, and droperidol compared to placebo) ranged from very low to moderate. Granisetron (RR 0.92, 95% CI 0.80 to 1.05, moderate certainty, rank 7/15) probably has no or little effect on any AE. Dexamethasone (RR 0.77, 95% CI 0.55 to 1.08, low certainty, rank 2/15) and droperidol (RR 0.89, 95% CI 0.81 to 0.98, low certainty, rank 6/15) may reduce any AE. Ondansetron (RR 0.95, 95% CI 0.88 to 1.01, low certainty, rank 9/15) may have little or no effect on any AE. We are uncertain about the effects of aprepitant (RR 0.87, 95% CI 0.78 to 0.97, very low certainty, rank 3/15) and ramosetron (RR 1.00, 95% CI 0.65 to 1.54, very low certainty, rank 11/15) on any AE. No studies reporting any AE were available for fosaprepitant.

Class-specific side effects

For class-specific side effects (headache, constipation, wound infection, extrapyramidal symptoms, sedation, arrhythmia, and QT prolongation) of relevant substances, the certainty of evidence for the best and most reliable anti-vomiting drugs mostly ranged from very low to low. Exceptions were that ondansetron probably increases headache (RR 1.16, 95% CI 1.06 to 1.28, moderate certainty, rank 18/23) and probably reduces sedation (RR 0.87, 95% CI 0.79 to 0.96, moderate certainty, rank 5/24) compared to placebo. The latter effect is limited to recommended and high doses of ondansetron. Droperidol probably reduces headache (RR 0.76, 95% CI 0.67 to 0.86, moderate certainty, rank 5/23) compared to placebo. We have high-certainty evidence that dexamethasone (RR 1.00, 95% CI 0.91 to 1.09, high certainty, rank 16/24) has no effect on sedation compared to placebo. No studies assessed substance class-specific side effects for fosaprepitant.

Direction and magnitude of network effect estimates together with level of evidence certainty are graphically summarized for all pre-defined GRADE-relevant outcomes and all drugs of direct interest compared to placebo in http://doi.org/10.5281/zenodo.4066353.