Management of vomiting in children and young people with gastroenteritis: ondansetron

Evidence in the NICE clinical guideline

The review of antiemetics in the full NICE guideline identified 5 relevant trials from a search to August 2008 (n=639, children aged 6 months to 12 years). Four trials were from the USA and 1 was from Venezuela. Three RCTs investigated oral ondansetron compared with placebo (Ramsook et al. 2002, Freedman et al. 2006 and Roslund et al. 2008) and 2 investigated intravenous ondansetron compared with either intravenous dexamethasone and placebo (Stork et al. 2006) or intravenous metoclopramide and placebo (Cubeddu et al. 1997). All of these trials are included in the Cochrane review.

Cochrane review

A Cochrane review of antiemetics for reducing vomiting related to acute gastroenteritis in children and young people was published in 2011 (Fedorowicz et al. 2011). This included 7 RCTs (n=1020) in children and young people aged 5 months to 12 years, 4 of which investigated oral ondansetron compared with placebo (Ramsook et al. 2002, Freedman et al. 2006, Roslund et al. 2008 and Yilmaz et al. 2010). Two RCTs investigated intravenous ondansetron (Stork et al. 2006 and Cubeddu et al. 1997) and 1 investigated a dimenhydrinate suppository.

An update to this Cochrane review was published in BMJ Open in 2012 (Carter and Fedorowicz 2012). The update included 10 RCTs (n=1479); 7 from the original Cochrane review plus 2 further RCTs investigating intravenous ondansetron compared with either intravenous metoclopramide (Al-Ansari et al. 2011) or placebo (Rerksuppaphol and Rerksuppaphol 2010), and 1 further RCT investigating oral granisetron. The 8 RCTs investigating ondansetron were conducted mostly in the emergency departments of children's hospitals in the USA, Canada, Turkey, Venezuela, Qatar and Thailand. Children and young people included in the trials were aged between 3 months and 13 years.

Dosing regimens varied between the studies but most used a weight‑dependent dosage. Of the 4 RCTs investigating oral ondansetron, 2 used a single dose of an orodispersible tablet at a dose of 2 mg for children under 15 kg, 4 mg for children weighing between 15 and 30 kg and 6 or 8 mg for children over 30 kg. The other 2 RCTs used multiple 8‑hourly doses of an ondansetron solution. In 1 study, 6 doses of 1.6 mg for children aged 6 months to 1 year, 3.2 mg for children aged 1 to 3 years and 4 mg for children aged 4 to 12 years were given. In the other study, 3 doses of 0.2 mg/kg were given. Of the 4 RCTs investigating intravenous ondansetron, 3 used a single dose of 0.15 mg/kg and 1 used a single dose of 0.3 mg/kg.

The primary outcome of the Cochrane review was the time taken from the first administration of the treatment until vomiting stopped, but none of the trials of oral ondansetron reported this. Pooled data from the 4 RCTs (n=574) which compared oral ondansetron with placebo (Ramsook et al. 2002, Freedman et al. 2006, Roslund et al. 2008 and Yilmaz et al. 2010) found ondansetron was more effective at stopping vomiting based on the proportion of children and young people who stopped vomiting (relative risk [RR] 1.44, 95% confidence interval [CI] 1.29 to 1.61, p<0.00001) with a number needed to treat (NNT) of 4 (95% CI 4 to 6). Substantial heterogeneity was noted and attributed to 1 study (Yilmaz et al. 2010), but when this was removed the result was similar (RR 1.33, 95% CI 1.19 to 1.49, p<0.00001, n=465).

Secondary outcomes included intravenous rehydration, hospitalisation and adverse events. In a pooled analysis of data from 3 of the RCTs (n=465) which compared oral ondansetron with placebo (Ramsook et al. 2002, Freedman et al. 2006 and Roslund et al. 2008), the proportion of children and young people needing intravenous fluid therapy was statistically significantly lower in the ondansetron group during the emergency department stay (RR 0.41, 95% CI 0.29 to 0.59, p<0.00001, NNT 5 [95% CI 4 to 8]) and up to 72 hours after discharge. Pooled data from these same studies found ondansetron reduced the immediate hospital admission rate during the emergency department stay (RR 0.40, 95% CI 0.19 to 0.83, p=0.01) but not hospitalisation up to 72 hours after discharge. There was no difference between oral ondansetron and placebo for emergency department revisit rates.

In the 3 RCTs which compared intravenous ondansetron with placebo, only 1 (Rerksuppaphol and Rerksuppaphol 2010, n=74) reported data for the primary outcome. This found that the time taken from the first administration of the treatment until cessation of vomiting was statistically significantly reduced with ondansetron (last vomiting after drug administration: mean 4.2±11.3 hours in the ondansetron group and 13.5±13.6 hours in the placebo group, p<0.01). Pooled data from the 3 RCTs (Cubeddu et al. 1997, Stork et al. 2006 and Rerksuppaphol and Rerksuppaphol 2010, total n=186) found intravenous ondansetron was more effective than placebo at stopping vomiting based on the proportion of children and young people who stopped vomiting (RR 2.01, 95% CI 1.49 to 2.71, p<0.00001, NNT 3 [95% CI 3 to 5]). However, there was heterogeneity between the studies, possibly because of their small sample size.

Two RCTs (Al-Ansari et al. 2011, n=186 and Cubeddu et al. 1997, n=36) compared intravenous ondansetron with intravenous metoclopramide (0.3 mg/kg). These trials found no statistically significant differences between the drugs in various outcomes. However, the updated Cochrane review authors stated that both RCTs were underpowered superiority studies and could only demonstrate no evidence of a difference between the drugs.

The authors of the updated Cochrane review (Carter and Fedorowicz 2012) carried out a mixed treatment comparison analysis using both direct and indirect evidence to estimate which was the most likely treatment to stop children and young people from vomiting. This has limitations, but found that oral or intravenous ondansetron was the treatment most likely to be effective compared with intravenous metoclopramide, oral granisetron, intravenous dexamethasone or placebo. No evidence on the use of cyclizine or domperidone was included in this systematic review.

Since this review was published, 2 further RCTs have been published comparing oral ondansetron with placebo in the emergency department of an Iranian children's hospital (Golshekan et al. 2013, n=176) and oral ondansetron with oral domperidone in the outpatient department of a hospital in Thailand (Rerksuppaphol and Rerksuppaphol 2013, n=76). These small studies have not been reviewed further in this evidence summary. There is also an Italian RCT (n=356) comparing oral ondansetron with oral domperidone and placebo which has been completed but not yet published (ClinicalTrials.gov Identifier: NCT01257672).

Observational study

In 2014, a retrospective cohort study of ondansetron use in the emergency departments of children's hospitals in the USA was published (Freedman et al. 2014). The objective of this study was to determine whether increasing emergency department use of oral ondansetron had resulted in a reduction in intravenous rehydration rates.

The cohort included 804,000 eligible patient visits across 18 children's hospital emergency departments in the USA between January 2002 and December 2011. Patients were children and young people aged less than 18 years who had a diagnosis of acute gastroenteritis (mean age 3.1 years [standard deviation 3.9 years]). Administration of oral ondansetron was identified for each patient through database review (no information on the dose or formulation of oral ondansetron administered was given). Visits were then categorised based on institutional use of ondansetron as low use (<5% of patients with acute gastritis received ondansetron), medium use (5 to 25% of patients with acute gastritis received ondansetron) or high use (more than 25% of patients with acute gastritis received ondansetron), and hospital‑level analyses of the association between ondansetron use and various outcomes was conducted. The primary outcome was the proportion of children or young people given intravenous rehydration. Secondary outcomes included the proportion of children or young people who were hospitalised and the proportion who revisited the emergency department within 3 days.

Overall, oral ondansetron use increased from a median institutional rate of 0.11% (interquartile range 0.04% to 0.44%) of patient visits in 2002 to 42.2% (interquartile range 37.5% to 49.1%) in 2011 (p<0.001). However, this increased use of oral ondansetron was not associated with reductions in the rates of intravenous rehydration or hospitalisation. Intravenous rehydration was given to 18.7% of children or young people during the low ondansetron use period and 17.8% during the high ondansetron use period. The hospitalisation rate was 6.0% during the low ondansetron use period and 6.7% during the high ondansetron use period. Increased use of oral ondansetron was, however, associated with a decrease in emergency department revisits within 3 days from 5.7% during the low‑use period to 5.0% during the high‑use period (p<0.05).

Most children and young people receiving intravenous rehydration did not receive oral ondansetron and the authors suggest that although oral ondansetron use has increased, it may not be being targeted to those most likely to fail oral rehydration therapy (no information on the use of oral rehydration therapy was given). Alternatively, they suggest that intravenous rehydration may be being used excessively. Of the children and young people given intravenous rehydration, 13.5% were given oral ondansetron, but 54.1% received intravenous ondansetron. The authors suggest that these patients may not be being given the opportunity to benefit from oral ondansetron in conjunction with oral rehydration therapy.

Safety and tolerability

During the review of antiemetics for the full NICE guideline on diarrhoea and vomiting caused by gastroenteritis in children younger than 5 years, the guideline development group expressed concern that ondansetron might have adverse effects such as worsening diarrhoea. They stated that if ondansetron does worsen diarrhoea it would be crucially important to determine the clinical significance of this effect, for example in relation to the risk of dehydration recurring or re‑admission to hospital.

The updated Cochrane review (Carter and Fedorowicz 2012) found that in 3 of the 4 RCTs that compared oral ondansetron with placebo (Ramsook et al. 2002, Freedman et al. 2006 and Yilmaz et al. 2010), there was an increase in the number of episodes of diarrhoea (p<0.05). Other side effects included a single episode of macular rash in the ondansetron group and a single episode of urticaria with placebo. However, the authors point out the limitation that none of the studies were powered to detect rare but serious adverse effects.

In Ramsook et al. 2002, there was no statistically significant difference between groups in the numbers of episodes of diarrhoea while undergoing rehydration in the emergency department. However, over the next 48 hours, children receiving ondansetron had statistically significantly more diarrhoea than those receiving placebo. In the first 24‑hour period, the mean number of diarrhoeal episodes in the ondansetron group (n=64) was 4.70 compared with 1.37 in the placebo group (n=54; p=0.002) and in the second 24 hours was 2.98 episodes (n=62) compared with 0.96 episodes (n=51; p=0.015).

In Freedman et al. 2006, the mean number of diarrhoeal episodes while undergoing rehydration was statistically significantly higher in children who had received ondansetron (mean 1.40, n=107) compared with the placebo group (mean 0.50, n=107; p<0.001) even after adjustment for number of episodes prior to admission.

In Yilmaz et al. 2010, at 8 hours there was no statistically significant difference in diarrhoeal episodes between the groups (mean 2.14 with ondansetron [n=55] compared with 2.0 with placebo [n=54]). However, children who received ondansetron had more episodes of diarrhoea while undergoing rehydration than those who received placebo at 24 hours (mean 5.04 with ondansetron [n=55] compared with 4.3 with placebo [n=54]; p=0.04).

In the 3 RCTs that compared intravenous ondansetron with placebo, 2 trials did not report the presence of any significant side effects. In 1 RCT (Cubeddu et al. 1997), more episodes of diarrhoea were reported in the ondansetron group in the first 24 hours compared with placebo (p=0.013).

The authors of the updated Cochrane review concluded that there is an increased incidence of diarrhoea when using ondansetron, but this is likely to vary according to the dosage. It may also be that the increase in diarrhoea is the result of retention of fluids which would otherwise been removed by vomiting.

The retrospective cohort study of ondansetron use in the emergency departments of 18 children's hospitals in the USA (Freedman et al. 2014) provided no information on safety or tolerability.

The summaries of product characteristics for ondansetron state that hypersensitivity reactions have been reported in people who have exhibited hypersensitivity to other selective 5HT3 receptor antagonists. Respiratory events should be treated symptomatically and clinicians should pay particular attention to them as precursors of hypersensitivity reactions.

Ondansetron prolongs the QT interval in a dose‑dependent manner and cases of Torsade de Pointes have been reported in people using ondansetron. Ondansetron should be avoided in people with congenital long QT syndrome and should be used with caution in people who have or may develop prolongation of QTc, such as people with electrolyte abnormalities, congestive heart failure, bradyarrhythmias or people taking other medicinal products that lead to QT prolongation or electrolyte abnormalities. Hypokalaemia and hypomagnesaemia should be corrected prior to ondansetron administration (see ondansetron summaries of product characteristics for details). Dose‑dependent QT interval prolongation with ondansetron is discussed in more detail in the July 2013 Drug Safety Update from the MHRA.

The summaries of product characteristics also state that there have been reports of people experiencing serotonin syndrome (including altered mental status, autonomic instability and neuromuscular abnormalities) following the concomitant use of ondansetron and other serotonergic drugs, such as selective serotonin reuptake inhibitors (SSRIs) and serotonin noradrenaline reuptake inhibitors (SNRIs). Ondansetron is known to increase large bowel transit time and people with signs of subacute intestinal obstruction should be monitored following administration.

Very common or common side effects (reported in at least 1 in 100 people) listed in the summaries of product characteristics for ondansetron are headache, a sensation of warmth or flushing and constipation. Uncommon side effects (reported in between 1 in 100 and 1 in 1000 people) are hiccups, seizures, movement disorders (including extrapyramidal reactions), arrhythmias, chest pain with or without ST segment depression, bradycardia, hypotension and asymptomatic increases in liver function tests.

Evidence strengths and limitations

The updated Cochrane review (Carter and Fedorowicz 2012) included 10 RCTs (n=1479). Whilst the review was well‑conducted, it is limited by the quality of the trials it included. The authors suggest that study design in the included studies was adequate overall but there was an unclear or high risk of bias in some of the domains (for example random sequence generation, allocation concealment, blinding, incomplete outcome data and selective reporting). They suggest this reflects the challenges faced in screening and following up children and young people presenting to emergency departments.

The outcomes reported in the included studies were generally clinician centred outcomes rather than patient or parent preferred outcomes. For example, none of the RCTs of oral or intravenous ondansetron reported parenteral satisfaction and only 1 RCT of intravenous ondansetron reported the time taken from the first administration of the treatment until vomiting stopped.

The 8 RCTs included in the updated Cochrane review that investigated ondansetron were conducted in the emergency departments of children's hospitals in the USA, Canada, Turkey, Venezuela, Qatar and Thailand. The findings of these may not be generalisable to UK practice. It is also difficult to define a usual dose of oral or intravenous ondansetron for the management of nausea and vomiting in children and young people with gastroenteritis because the doses used in the studies were weight dependent and the dose, dosing schedule and formulation varied considerably between the studies.

The authors of the updated Cochrane review (Carter and Fedorowicz 2012) carried out a mixed treatment comparison analysis using both direct and indirect evidence. Although these analyses are useful to assess treatments that have not been compared in direct head‑to‑head clinical trials, they have limitations and are no substitute for large, well‑designed RCTs. They are based on many assumptions and are only as good as the data that are included.

The retrospective cohort study of ondansetron use (Freedman et al. 2014) is limited by its observational nature; in that bias and confounding is an issue. The authors did adjust the findings for age, sex, race, season, admission time and prior visit for acute gastroenteritis, laboratory testing, diagnostic imaging and rotavirus diagnosis; however some residual confounding may remain. The authors state that misclassification of patient visits is also a possibility and the increasing use of diagnostic testing and treatments in emergency departments over time may have affected the findings. The results of this study may not be generalisable to non‑children's hospital settings or to UK practice, because the study was conducted in the emergency departments of children's hospitals in the USA. However, it provides useful data to determine the clinical effectiveness of ondansetron in 'real‑world' conditions.