Ondansetron is an effective antiemetic employed to prevent vomiting in children with gastroenteritis in high-income countries; data from low- and middle-income countries are sparse.
We conducted a randomized, double-blind, placebo-controlled superiority trial in 2 pediatric emergency departments in Pakistan. Dehydrated children aged 6 to 60 months with ≥1 diarrheal (ie, loose or liquid) stool and ≥1 vomiting episode within the preceding 4 hours were eligible to participate. Participants received a single weight-based dose of oral ondansetron (8–15 kg: 2 mg; >15 kg: 4 mg) or identical placebo. The primary outcome was intravenous administration of ≥20 mL/kg over 4 hours of an isotonic fluid within 72 hours of random assignment.
All 918 (100%) randomly assigned children completed follow-up. Intravenous rehydration was administered to 14.7% (68 of 462) and 19.5% (89 of 456) of those administered ondansetron and placebo, respectively (difference: −4.8%; 95% confidence interval [CI], −9.7% to 0.0%). In multivariable logistic regression analysis adjusted for other antiemetic agents, antibiotics, zinc, and the number of vomiting episodes in the preceding 24 hours, children administered ondansetron had lower odds of the primary outcome (odds ratio: 0.70; 95% CI, 0.49 to 1.00). Fewer children in the ondansetron, relative to the placebo group vomited during the observation period (difference: −12.9%; 95% CI, −18.0% to −7.8%). The median number of vomiting episodes (P < .001) was lower in the ondansetron group.
Among children with gastroenteritis-associated vomiting and dehydration, oral ondansetron administration reduced vomiting and intravenous rehydration use. Ondansetron use may be considered to promote oral rehydration therapy success among dehydrated children in low- and middle-income countries.
Ondansetron administration to dehydrated children with gastroenteritis-associated vomiting in emergency departments in high-income countries reduces vomiting and intravenous rehydration. Although it is ineffective among well-hydrated children, evidence of efficacy in dehydrated children in low- and middle-income countries is lacking.
Emergency department oral ondansetron administration to dehydrated children with gastroenteritis-associated vomiting in Pakistan safely reduces intravenous rehydration fluid administration and vomiting, and it should be considered to promote oral rehydration therapy in this population.
Globally, nearly 6 million children <5 years of age still die annually.1 Despite advances in oral rehydration therapy (ORT) and treatment of diarrhea, some 500 000 of these deaths are due to acute gastroenteritis (AGE).1 Important contributors to diarrhea-related mortality include limited access to services and the stagnated use of ORT,2,3 particularly in the presence of vomiting. In Pakistan, 80% of those who develop severe dehydration have persistent vomiting, with a high frequency in the first 6 hours of therapy.4 Although use of antiemetic agents such as domperidone or metoclopramide is commonplace,5,6 they are of limited benefit.7,8
A single oral dose of ondansetron reduces vomiting and intravenous rehydration use.9,10 Although administration in high-income countries is widespread,11–13 research on its use in low- and middle-income countries (LMICs) is limited but is necessary given the differences in etiology, clinical phenotypes,14 and complications.5 Consequently, there is a need to determine if ondansetron can enhance ORT success in a LMIC setting.
We conducted 2 separate but linked studies in Karachi Pakistan6 to answer 2 questions, and we planned a priori to publish 2 unique reports. In the first study, we reported that among children without dehydration,15 there were no benefits associated with ondansetron use. In the second study, we evaluated whether a single oral ondansetron dose administered to children with vomiting and dehydration secondary to AGE reduces the probability of intravenous rehydration fluid administration compared to the placebo.
Methods
Design and Setting
We performed a 2-center, randomized, double-blind, placebo-controlled superiority trial (Fig 1) in the emergency departments (EDs) of The Aga Khan University Hospital for Women and Children and The Aga Khan University Hospital, Karachi, Pakistan. Pediatric emergency medicine trained physicians treat ∼10 000 and 5000 patients annually in each of these institutions, respectively. The study was approved by the ethics committees of The Aga Khan University and University of Calgary.
Consolidated Standards of Reporting Trials flow diagram. a weight for height below -3z scores of the median WHO growth standards. b The one child that was lost to follow-up had complete data for 4 hour emergency department observation period. The child was included in the primary analysis as he received > 20 ml/kg of intravenous fluids and thus experienced the outcome of interest.
Consolidated Standards of Reporting Trials flow diagram. a weight for height below -3z scores of the median WHO growth standards. b The one child that was lost to follow-up had complete data for 4 hour emergency department observation period. The child was included in the primary analysis as he received > 20 ml/kg of intravenous fluids and thus experienced the outcome of interest.
Study Population
Potentially eligible children were consecutively screened by study-funded research officers 24 hours/day, 7 days/week. Eligible children were aged 0.5 to 5.0 years, weighed ≥8.0 kg, and had ≥1 episode of diarrhea (ie, a minimum of 1 loose or liquid stool) and ≥1 vomiting episode within the 4 hours preceding triage.13 As previously performed,13 we employed lower age and weight limits because infants <6 months of age are more likely to have alternative underlying etiologies (eg, urinary tract infection), and 8.0 kg corresponds to the weight at which the smallest study dose (2 mg) can be administered. Participants had “some” dehydration quantified by using the World Health Organization (WHO) dehydration tool,16,17 which requires the presence of ≥2 of the following: restlessness and/or irritability, sunken eyes, drinking eagerly and/or thirst, and skin pinch retracts slowly.16
Children with the following were excluded: severe dehydration, bloody or bilious vomiting, hypotension,18 vomiting or diarrhea for >7 days, previous abdominal surgery, hypersensitivity to ondansetron or any serotonin receptor antagonist, personal or family history of prolonged QT syndrome, taking a medication listed as causing torsades de pointes (https://crediblemeds.org/index.php/login/dlcheck:), previously enrolled in the study, and those for whom follow-up would not be possible. We excluded children whose weight for height was <−3 z scores of the median WHO growth standards because children with malnutrition are at greater risk of electrolyte abnormalities.19 Guardians of all participants provided written informed consent.
Allocation
Children were randomly assigned to receive a single ondansetron or placebo oral disintegrating tablet (ODT; both provided in-kind by GlaxoSmithKline, Inc, Philadelphia, PA) in a 1:1 ratio, stratified by age (<18 and ≥18 months) and study center by using variable block sizes of 4 and 6. Use of an Internet-based randomization service facilitated allocation concealment. The study team was unaware of block sizes.
As is commonly performed in clinical practice20 and trials,13 doses were weight based: 8 to 15 kg received a dose of 2 mg; >15 kg received a dose of 4 mg. Within the dose range of 0.13 to 0.26 mg/kg, higher doses of ondansetron are not superior to lower doses nor are they associated with increased side effects.21 The ODT was placed on the top of each child’s tongue, and the child was instructed to swallow 5 seconds later.13 Fifteen minutes after ODT administration, ORT was initiated. Children who vomited ≥1 times during that interval received a repeat dose.13 Blinded individuals included the ED physicians, research officers, families, patients, and on-site pharmacists. The placebo and active ODTs were of identical size, appearance, taste, and smell.13
A prespecified computer-generated randomization list with associated kit numbers was sent from www.randomize.net via password-protected files to the research pharmacist who prepared, packed, and shipped all drug kits. At enrollment, www.randomize.net randomly selected a kit number from the remaining kits containing the assigned treatment. Each kit contained two 4-mg ondansetron or placebo ODTs (cut in half if needed to provide a 2-mg dose) including 1 extra dose in case a repeat dose was required. If the extra dose was vomited, no additional medications were provided.
Study Interventions
Aside from study interventions, participants received therapy in keeping with WHO recommendations. Concomitant medication administration was at the discretion of the clinical team. The protocol emphasized a targeted weight-based ORT protocol during the 4-hour observation period and caregiver education regarding oral rehydration solution (ORS) administration. Breastfeeding continued ad lib in addition to giving WHO ORS. If the child vomited, caregivers waited 10 minutes and then resumed giving ORS more slowly. Children whose dehydration, assessed by using the WHO tool, had resolved were discharged; those with some dehydration after 4 hours had ORT treatment repeated for another 4 hours with food administration. Should children deteriorate and develop “severe” dehydration, rapid intravenous rehydration was administered.12,21 The need for hospitalization was determined by the treating physician. Postdischarge care was in keeping with WHO recommendations.16 Caregivers whose children were suitable for discharge were provided with a 2-day supply of ORS and were instructed to give as much fluid as the child desired to prevent dehydration. To reduce the likelihood of persistent diarrhea, participants were also provided with a 2-week supply of 20-mg zinc tablets to be administered daily for 14 days.22 Caregivers were instructed to initiate zinc therapy a minimum of 30 minutes after study drug administration.
Data
Data were collected by research officers. During the 4-hour study observation period, the following were recorded every 60 minutes: oral intake, intravenous fluids, and episodes of vomiting, diarrhea, and urination. We placed urine collection bags on children who could not urinate into measurement containers. Stool was quantified by weighing diapers for infants and toddlers and the use of collection devices for children who were toilet trained. Although the WHO tool was employed to assess dehydration regarding eligibility, we employed the clinical dehydration scale (CDS) score23,24 to perform repeat dehydration assessments during ED monitoring because the CDS allows for a better quantification of dehydration and thus is better suited to serve as a covariate during analysis.
Participants were reassessed 24 hours after discharge at their home or the enrolling institution. If there were no signs of dehydration and vomiting and diarrhea had resolved, 48- and 72-hour follow-up was done by telephone. For children with ongoing symptoms or signs of dehydration, a repeat in-person reassessment in 24 hours was required.
Study Outcomes
The primary outcome was intravenous rehydration defined as the administration of ≥20 mL/kg over 4 hours of an isotonic fluid for the purpose of rehydration within 72 hours of random assignment. This outcome was selected because we sought to include only those who had an intravenous line inserted for hydration purposes. It excludes those who received only maintenance fluids and those who had the intravenous line inserted for medication administration, while capturing those who received brief bolus fluid therapy or greater-than-maintenance fluids for several hours. The 72-hour time frame balances the potential benefits and side effects of ondansetron.
Secondary outcomes identified a priori were the (1) presence and (2) frequency of vomiting during the 4-hour observation period, (3) hospitalization for >24 hours defined as the interval from ED arrival to hospital discharge, (4) volume of ORS consumed during the 4-hour observation period, (5) presence of some dehydration at any time after discharge up to the 72-hour follow-up assessment, and (6) number of diarrheal (ie, loose or liquid) stools during the 72 hours after random assignment. The composite outcome of treatment failure includes intravenous rehydration, nasogastric rehydration for >24 hours, or death within 72 hours. Nasogastric rehydration was included in our composite outcome measure because it is preferred to intravenous rehydration as second-line rehydration treatment, after ORT, in numerous guidelines.25
Statistical Analysis
The planned sample size of 868 patients was estimated to provide 90% power to detect an absolute between-group difference of 10% in the risk of receiving ≥20 mL/kg over 4 hours of an isotonic fluid for the purpose of rehydration within 72 hours of random assignment (risk ratio: 0.67) at a baseline risk of 30% (under the assumption of a 2-sided 5% level of significance) and a lack of primary outcome ascertainment of 5%. The consensus among investigators was that the probability of intravenous rehydration among children administered the placebo was higher than the 17% previously reported by the International Study Group on Reduced-Osmolarity ORS26 because all study participants have some dehydration. On the basis of local expert opinion, sample size calculations employed a minimally clinically important difference of 10%. Because of delays in data entry and concerns about completeness, an additional 50 patients were recruited. Full outcome data were not available until the final analysis. All children randomly assigned were included in the primary and secondary analyses.
Analyses were undertaken by intention-to-treat principles. The proportion of children receiving intravenous rehydration by 72 hours was analyzed by using a Mantel-Haenszel test, stratified by clinical center and age. Prespecified subgroup analyses based on subject age, duration of illness, and baseline diarrhea and vomiting frequency in the preceding 24 hours were conducted. Secondary analysis of the primary outcome employed a multivariable logistic regression model fitted with treatment group and baseline covariates (ie, antiemetic, antibiotics, zinc administration before random assignment, and the number of vomiting episodes in 24 hours before random assignment), which potentially predict intravenous rehydration and were associated with the outcome.
The Mantel-Haenszel test, stratified by clinical center, was used to analyze the secondary outcomes of vomiting (yes or no), hospitalization, presence of some dehydration recurring within 72 hours, and treatment failure. The van Elteren test, stratified by clinical center, was used for the continuous variables of vomiting frequency, volume of ORS consumed, and diarrheal stool frequency.
Because missing baseline values were present in <1% of cases, imputation was not required. A Bonferroni correction was used to correct for multiple comparisons, and adjusted P values are reported. The analysis plan was prespecified in the protocol and was performed with SPSS version 22.0 (IBM SPSS Statistics, IBM Corporation) and SAS version 9.4 (SAS Institute, Inc, Cary, NC).
Results
Among 918 randomly assigned children (median age, 18.0 [interquartile range (IQR), 12.0–30.0] months) recruited between June 5, 2014, and December 12, 2017 (Fig 1), 462 were assigned to ondansetron and 456 to placebo. Baseline characteristics (Table 1), laboratory parameters (Supplemental Table 5), and cointerventions (Table 2) were similar between groups. The intervention or placebo medication was vomited by 3.5% (16 of 462) and 3.7% (17/456) of those in the ondansetron and placebo groups, respectively. Primary outcome data were available for 100% (918 of 918) of study participants; 72-hour follow-up was completed for 99.9% (917 of 918; Supplemental Table 6). Overall, 20.9% (192 of 918) of children had an intravenous cannula inserted during the study period (placebo: 105 of 456 [23.0%]; ondansetron: 87 of 462 [18.8%]; OR: 0.77; 95% confidence interval [CI], 0.56 to 1.07).
Baseline Clinical Characteristics of Participants by Treatment Group
| Ondansetron, n = 462 | Placebo, n = 456 | |
|---|---|---|
| Age, mo | 18 (12–30) | 18 (12–29) |
| Male | 271 (58.7) | 279 (61.2) |
| Wt, kg | 10.0 (8.6–12.0) | 10.0 (8.6–12.0) |
| Chronic medical conditions | 0 | 5 (1.1) |
| Time interval, last vomit to medication administration, h | 1.5 (0.8–2.5) | 1.6 (0.8–2.7) |
| Maximal vomit episodes per 24-h period | 5 (3–6) | 5 (3–7) |
| Vomit episodes in past 24 h | 4 (3–6) | 4 (3–6) |
| Vomiting duration, d | 1 (1–2) | 1 (1–2) |
| Maximal diarrheal episodes per 24-h period | 4 (2–6) | 3 (2–6) |
| Diarrheal episodes in past 24 h | 3 (2–5) | 3 (2–5) |
| Diarrhea duration, d | 1 (1–2) | 1 (1–2) |
| Fevera | 122 (26.4) | 104 (22.8) |
| Previous ED visit, current illness | 60 (13.0) | 62 (13.6) |
| Previous intravenous rehydration, current illness | 20 (4.3) | 25 (5.5) |
| Previous hospitalization, current illness | 4 (0.9) | 6 (1.3) |
| Medications administered, past 24 hb | ||
| Antacids | ||
| Omeprazole and/or ranitidine | 4 (0.9) | 4 (0.9) |
| Antipyretics | 51 (11.0) | 46 (10.1) |
| Acetaminophen | 45 (9.7) | 44 (9.6) |
| Ibuprofen | 6 (1.3) | 6 (1.3) |
| Antibiotics and/or antihelminthics | 69 (14.9) | 60 (13.2) |
| Azithromycin and/or clarithromycin | 1 (0.2) | 0 (0) |
| Amoxicillin and/or ampicillin | 1 (0.2) | 1 (0.2) |
| Cefixime and/or cefotaxime/ceftriaxone | 31 (6.7) | 25 (5.5) |
| Diloxanide and/or mebendazole | 6 (1.3) | 10 (2.2) |
| Metronidazole | 41 (8.9) | 36 (7.9) |
| Other | 8 (1.7) | 3 (0.7) |
| Antiemetics | 97 (21.0) | 86 (18.9) |
| Dimenhydrinate | 32 (6.9) | 31 (6.8) |
| Domperidone | 76 (16.5) | 63 (13.8) |
| Metoclopramide | 1 (0.2) | 4 (0.9) |
| Antihistamines and/or anticholinergics | ||
| Cetirizine, clemastine, cyclizine, and/or diphenhydramine | 9 (1.9) | 10 (2.2) |
| Nutrition | ||
| Zinc | 10 (2.2) | 14 (3.1) |
| Probiotics | ||
| Saccharomyces boulardii and/or Lactobacilus aicdophilus | 13 (2.8) | 14 (3.1) |
| Rotavirus vaccine received | 201 (43.5) | 193 (42.3) |
| Exclusively breastfed | 14 (3.0) | 9 (2.0) |
| CDS score23 | 2 (2–3) | 2 (2–3) |
| Ondansetron, n = 462 | Placebo, n = 456 | |
|---|---|---|
| Age, mo | 18 (12–30) | 18 (12–29) |
| Male | 271 (58.7) | 279 (61.2) |
| Wt, kg | 10.0 (8.6–12.0) | 10.0 (8.6–12.0) |
| Chronic medical conditions | 0 | 5 (1.1) |
| Time interval, last vomit to medication administration, h | 1.5 (0.8–2.5) | 1.6 (0.8–2.7) |
| Maximal vomit episodes per 24-h period | 5 (3–6) | 5 (3–7) |
| Vomit episodes in past 24 h | 4 (3–6) | 4 (3–6) |
| Vomiting duration, d | 1 (1–2) | 1 (1–2) |
| Maximal diarrheal episodes per 24-h period | 4 (2–6) | 3 (2–6) |
| Diarrheal episodes in past 24 h | 3 (2–5) | 3 (2–5) |
| Diarrhea duration, d | 1 (1–2) | 1 (1–2) |
| Fevera | 122 (26.4) | 104 (22.8) |
| Previous ED visit, current illness | 60 (13.0) | 62 (13.6) |
| Previous intravenous rehydration, current illness | 20 (4.3) | 25 (5.5) |
| Previous hospitalization, current illness | 4 (0.9) | 6 (1.3) |
| Medications administered, past 24 hb | ||
| Antacids | ||
| Omeprazole and/or ranitidine | 4 (0.9) | 4 (0.9) |
| Antipyretics | 51 (11.0) | 46 (10.1) |
| Acetaminophen | 45 (9.7) | 44 (9.6) |
| Ibuprofen | 6 (1.3) | 6 (1.3) |
| Antibiotics and/or antihelminthics | 69 (14.9) | 60 (13.2) |
| Azithromycin and/or clarithromycin | 1 (0.2) | 0 (0) |
| Amoxicillin and/or ampicillin | 1 (0.2) | 1 (0.2) |
| Cefixime and/or cefotaxime/ceftriaxone | 31 (6.7) | 25 (5.5) |
| Diloxanide and/or mebendazole | 6 (1.3) | 10 (2.2) |
| Metronidazole | 41 (8.9) | 36 (7.9) |
| Other | 8 (1.7) | 3 (0.7) |
| Antiemetics | 97 (21.0) | 86 (18.9) |
| Dimenhydrinate | 32 (6.9) | 31 (6.8) |
| Domperidone | 76 (16.5) | 63 (13.8) |
| Metoclopramide | 1 (0.2) | 4 (0.9) |
| Antihistamines and/or anticholinergics | ||
| Cetirizine, clemastine, cyclizine, and/or diphenhydramine | 9 (1.9) | 10 (2.2) |
| Nutrition | ||
| Zinc | 10 (2.2) | 14 (3.1) |
| Probiotics | ||
| Saccharomyces boulardii and/or Lactobacilus aicdophilus | 13 (2.8) | 14 (3.1) |
| Rotavirus vaccine received | 201 (43.5) | 193 (42.3) |
| Exclusively breastfed | 14 (3.0) | 9 (2.0) |
| CDS score23 | 2 (2–3) | 2 (2–3) |
Data are n (%) or median (IQR).
Fever was defined as an adjusted rectal temperature of ≥38.0°C. Axillary and oral temperatures were adjusted to rectal temperatures by adding 1.1°C and 0.6°C, respectively.27
Some children received >1 medication in the past 24 h.
ED and Discharge Cointerventions
| Ondansetron (n = 462), n (%) | Placebo (n = 456), n (%) | |
|---|---|---|
| Antacid in the ED | 10 (2.2) | 26 (5.7) |
| Omeprazole | 10 (2.2) | 25 (5.5) |
| Ranitidine | 0 (0) | 1 (0.2) |
| Antibiotic in the ED | 114 (24.7) | 99 (21.7) |
| Amoxicillin | 5 (1.1) | 1 (0.2) |
| Azithromycin | 5 (1.1) | 12 (2.6) |
| Cefixime | 28 (6.1) | 18 (3.9) |
| Ceftriaxone | 38 (8.2) | 39 (8.6) |
| Ciprofloxacin | 20 (4.3) | 10 (2.2) |
| Metronidazole | 22 (4.8) | 24 (5.3) |
| Antibiotic recommended at discharge or given after discharge | 112 (24.2) | 98 (21.5) |
| Any antibiotics during the whole study period | 125 (27.1) | 109 (23.9) |
| Antiemetic in the ED | 168 (36.4) | 180 (39.5) |
| Dimenhydrinate | 1 (0.2) | 2 (0.4) |
| Domperidone | 154 (33.3) | 157 (34.4) |
| Ondansetron | 18 (3.9) | 28 (6.1) |
| Metoclopramide | 1 (0.2) | 1 (0.2) |
| Antihistamine in the ED | ||
| Cetirizine | 5 (1.1) | 3 (0.7) |
| Antipyretic in the ED | 82 (17.7) | 69 (15.1) |
| Acetaminophen | 59 (12.8) | 52 (11.4) |
| Ibuprofen | 25 (5.4) | 18 (3.9) |
| Other in the ED | ||
| Saccharomyces boulardii | 119 (25.8) | 117 (25.7) |
| Zinc | 84 (18.2) | 73 (16.0) |
| Ondansetron (n = 462), n (%) | Placebo (n = 456), n (%) | |
|---|---|---|
| Antacid in the ED | 10 (2.2) | 26 (5.7) |
| Omeprazole | 10 (2.2) | 25 (5.5) |
| Ranitidine | 0 (0) | 1 (0.2) |
| Antibiotic in the ED | 114 (24.7) | 99 (21.7) |
| Amoxicillin | 5 (1.1) | 1 (0.2) |
| Azithromycin | 5 (1.1) | 12 (2.6) |
| Cefixime | 28 (6.1) | 18 (3.9) |
| Ceftriaxone | 38 (8.2) | 39 (8.6) |
| Ciprofloxacin | 20 (4.3) | 10 (2.2) |
| Metronidazole | 22 (4.8) | 24 (5.3) |
| Antibiotic recommended at discharge or given after discharge | 112 (24.2) | 98 (21.5) |
| Any antibiotics during the whole study period | 125 (27.1) | 109 (23.9) |
| Antiemetic in the ED | 168 (36.4) | 180 (39.5) |
| Dimenhydrinate | 1 (0.2) | 2 (0.4) |
| Domperidone | 154 (33.3) | 157 (34.4) |
| Ondansetron | 18 (3.9) | 28 (6.1) |
| Metoclopramide | 1 (0.2) | 1 (0.2) |
| Antihistamine in the ED | ||
| Cetirizine | 5 (1.1) | 3 (0.7) |
| Antipyretic in the ED | 82 (17.7) | 69 (15.1) |
| Acetaminophen | 59 (12.8) | 52 (11.4) |
| Ibuprofen | 25 (5.4) | 18 (3.9) |
| Other in the ED | ||
| Saccharomyces boulardii | 119 (25.8) | 117 (25.7) |
| Zinc | 84 (18.2) | 73 (16.0) |
Primary Outcome
The administration of ≥20 mL/kg over 4 hours of an intravenous rehydration solution within 72 hours of random assignment occurred in 14.7% (68 of 462) vs 19.5% (89 of 456) of those in the ondansetron and placebo groups, respectively (odds ratio [OR]: 0.71; 95% CI, 0.50 to 1.00; difference: 4.8%; 95% CI, 0.0% to 9.7%; Table 3). Employing a multivariable logistic regression model fitted with the treatment group and adjusted for the administration of other antiemetics, antibiotics, and zinc before random assignment (Supplemental Tables 7 and 8) and the number of vomiting episodes in the preceding 24 hours yielded an OR of 0.70 (95% CI, 0.49 to 1.00) in favor of the ondansetron treatment arm. Antibiotic administration (OR: 1.75; 95% CI, 1.08 to 2.84) and the number of vomit episodes in the preceding 24 hours (OR: 1.12; 95% CI, 1.06 to 1.19 per episode) were also associated with intravenous rehydration (Supplemental Tables 7 and 8). There was no evidence of interaction between treatment group and age (Fig 2), presence of ≥3 diarrheal stools in the preceding 24 hours, or presence of ≥3 vomits in the preceding 24 hours (Table 4, Supplemental Tables 9 and 10).
Participant Clinical Outcomes by Treatment Group
 |
| |
Data are n (%) or median (IQR) unless otherwise stated. N/A, not applicable.
P values presented for secondary outcomes are adjusted by using the Bonferroni correction procedure; for secondary outcomes, adjustment was performed for 7 comparisons; for other outcomes, adjustment was performed for 11 comparisons. Statistical tests performed were either the van Elteren test stratified by enrollment center and age (<18 and ≥18 mo) (continuous variables) or the Cochran-Mantel-Haenszel test stratified by enrollment center and age (<18 and ≥18 mo) (categorical variables).
Hospital length of stay was defined as a total length of stay from the ED arrival until discharge.
Dehydration status was assessed employing the WHO dehydration assessment approach.
Diarrhea was defined as loose or liquid stools.
Treatment failure is a composite outcome measure that includes children who experienced any of the following: intravenous rehydration (≥20 mL/kg per 4 h), nasogastric rehydration for >24 h, death within 72 h from any cause, in or out of hospital. No children experienced the outcome of death or nasogastric rehydration.
Impact of age in 6-month increments on the primary outcome. IVF, intravenous fluid.
Impact of age in 6-month increments on the primary outcome. IVF, intravenous fluid.
A Priori Specified Subgroup Analysis of the Primary Outcome
| n | Ondansetron, n (%) | Placebo, n (%) | OR (95% CI) | Pa | |
|---|---|---|---|---|---|
| Baseline diarrhea episodes in a 24-h period | |||||
| ≥3 episodes | 556 | 50 (17.4) | 60 (22.3) | 0.75 (0.49 to 1.15) | >.99 |
| <3 episodes | 362 | 18 (10.3) | 29 (15.5) | 0.58 (0.31 to 1.10) | .57 |
| Age | |||||
| <18 mo | 449 | 34 (15.2) | 49 (21.7) | 0.62 (0.38 to 1.02) | .35 |
| ≥18 mo | 469 | 34 (14.2) | 40 (17.4) | 0.81 (0.49 to 1.34) | >.99 |
| Baseline duration of illness | |||||
| <48 h | 475 | 38 (15.5) | 34 (14.8) | 1.07 (0.65 to 1.78) | >.99 |
| ≥48 h | 443 | 30 (13.8) | 55 (24.3) | 0.46 (0.28 to 0.71) | .02 |
| n | Ondansetron, n (%) | Placebo, n (%) | OR (95% CI) | Pa | |
|---|---|---|---|---|---|
| Baseline diarrhea episodes in a 24-h period | |||||
| ≥3 episodes | 556 | 50 (17.4) | 60 (22.3) | 0.75 (0.49 to 1.15) | >.99 |
| <3 episodes | 362 | 18 (10.3) | 29 (15.5) | 0.58 (0.31 to 1.10) | .57 |
| Age | |||||
| <18 mo | 449 | 34 (15.2) | 49 (21.7) | 0.62 (0.38 to 1.02) | .35 |
| ≥18 mo | 469 | 34 (14.2) | 40 (17.4) | 0.81 (0.49 to 1.34) | >.99 |
| Baseline duration of illness | |||||
| <48 h | 475 | 38 (15.5) | 34 (14.8) | 1.07 (0.65 to 1.78) | >.99 |
| ≥48 h | 443 | 30 (13.8) | 55 (24.3) | 0.46 (0.28 to 0.71) | .02 |
The statistical test performed was the Cochran-Mantel-Haenszel test stratified by the enrollment center.
P values presented are adjusted by using the Bonferroni correction procedure for 6 comparisons.
Secondary Outcomes
Overall, 13.2% (61 of 462) of children in the ondansetron group vomited during the 4-hour observation period compared with 26.1% (119 of 456) in the placebo group (OR: 0.43; 95% CI, 0.31 to 0.61; difference: 12.9%; 95% CI: 7.8% to 18.0%; Table 3, Fig 3). There were fewer vomiting episodes in the ondansetron group (P < .001; Table 3) but no difference in the volume of oral fluids consumed during the observation period. The proportion of children hospitalized >24 hours and that had some dehydration develop at any time up to 72 hours after discharge did not differ between groups. The number of diarrheal stools during the 72-hour follow-up period and the median volume of diarrhea during the 4-hour observation period were similar between groups. Primary and secondary outcomes were similar at both study sites (Supplemental Tables 9–11).
Adverse Events
No serious adverse events or admissions to the step-down units or ICUs were reported. Reported adverse events were similar between groups (Supplemental Table 11).
Discussion
In this 2-center trial, young children with some dehydration were less likely to receive intravenous rehydration if they received ondansetron compared with children who received the placebo. This effect stems from the reduction in vomiting associated with ondansetron administration. These results are important because >500 000 children continue to die each year from AGE,1 and most deaths in LMICs could be prevented by the use of known and cost-effective interventions.28 The evidence from this study has the potential to lead to further evaluations in more rural contexts where a disproportionate number of children die.29
It is important to consider our results in the context of the companion study that included 626 children without dehydration in which the authors identified no benefits associated with ondansetron administration.15 Participants in the current study were older, had more frequent vomiting, and higher CDS scores. They were thus more likely to benefit from an effective antiemetic. Although the absolute reduction in intravenous rehydration use was lower than anticipated, the reduction was significant, and the number needed to treat is 21. The benefits are symptomatically meaningful with the number needed to treat to prevent vomiting being 8. Because these benefits are in keeping with findings from high-income countries where the absolute reductions in vomiting and intravenous rehydration are 25% and 19%, respectively,30 ondansetron use may be considered to promote ORT success in children similar to those enrolled in our study.
The lower than anticipated intravenous rehydration rate likely relates to the baseline frequency of vomiting, which was lower than anticipated. The median frequency of vomiting in the preceding 24 hours was only 4; other reports have exceeded 9.13,31 The connection between ondansetron benefits and vomiting frequency is highlighted by our multivariable regression model that retained vomiting frequency as an independent predictor of treatment failure. Although it is also possible that concomitant antiemetic administration (ie, domperidone) may have played a role, authors of most studies have found it to be ineffective.7,8 Additionally, it may be that in this academic tertiary care center, there was greater adherence to guidelines with an emphasis on ORT, and the use of higher thresholds for intravenous rehydration may be in routine use than in earlier reports.
In Delhi, India,17 25% of children with some dehydration who were administered the placebo received intravenous fluids compared with 14% of children who were administered ondansetron (relative risk: 0.56). Benefits attributed to ondansetron administration included expedited resolution of dehydration, reduced vomiting, and greater satisfaction.17 Thus, our findings, supported by previous LMIC work17 and evidence from high-income countries,10 lead to the conclusion that despite being of borderline statistical significance,32 it is highly likely that ondansetron administration to children with dehydration is beneficial in resource-poor settings. Identification of an effective antiemetic in this setting is important because although domperidone has been revealed to be ineffective at treating gastroenteritis-associated vomiting,7,8 it is routinely employed in LMICs. This is likely because of the propensity for self-medication in LMICs,33 the desire to treat vomiting in children with dehydration, and the widespread34 belief that domperidone is effective.8,35 Because both ondansetron and domperidone are readily available in Pakistan, educational efforts disseminating recent evidence are needed to improve care.
Antibiotic use was common in our study. It is indicated in recent reports in the region that antimicrobial agents are prescribed to nearly 40% of children with acute watery diarrhea due to viral pathogens and 60% of unknown etiology.36 The excessive use of antimicrobial agents in Southeast Asia has led to a resistance crisis.37 Our findings further these concerns with use also being associated with increased intravenous rehydration usage (OR: 1.75; 95% CI, 1.08 to 2.84), which may reflect the propensity of antibiotics to cause diarrhea in exposed children.38
Although we had intended to conduct stool microbial analyses, because of an insufficient number of specimens submitted, this objective was not completed. Although not different between groups, the extensive coadministration of antiemetics such as domperidone was not anticipated. Although a more restrictive approach to concomitant medication use could have been employed, we focused on conducting a pragmatic real-world trial.39 Although, in theory, concomitant antiemetic use could have influenced the outcomes of the study because this was a randomized clinical trial, it is unlikely to have altered the effect of the intervention. Moreover, we incorporated this covariate in regression models to further minimize any impact it may have had. It should be noted that dehydration assessment using clinical scores is suboptimal.40 Although concerns have been raised regarding use of the WHO dehydration tool,41 in keeping with local standards of care, we used it to determine eligibility. The CDS score23,24 was used to assess dehydration as an outcome because, unlike the WHO tool, it can be employed as a quantitative tool. Future studies investigating ondansetron use barriers in LMIC settings are needed.
Conclusions
Among children with gastroenteritis-associated vomiting and dehydration, oral ondansetron administration reduces vomiting and intravenous rehydration use. These findings should be replicated in a larger multicenter trial, and if successful, ondansetron use should be considered to promote ORT success among dehydrated children in LMICs.
Acknowledgments
We thank the extended study team based at the Alberta Children’s Hospital, Calgary, Alberta, Canada, and the team based in Karachi, Pakistan. We also thank our funders: (1) the Thrasher Research Fund (award 10025), (2) Bill and Melinda Gates Foundation (grant OPP1058793), and (3) Alberta Children’s Hospital Foundation. We also thank GlaxoSmithKline, Inc, for supplying the study drug and placebo.
This work was presented at the annual meeting of the Pediatric Academic Societies; April 24, 2019, to May 1, 2019; Baltimore, MD.
The authors of this trial commit to making requested, deidentified participant data (including data dictionaries) and study protocols, the statistical analysis plan, and the informed consent form available after reasonable request after publication of the manuscript up until 5 years after publication. Requests for access to data require evidence of ethics approval of a methodologically sound proposal for use, and data sharing agreements must be in place. Requests should be addressed to the corresponding author at [email protected].
The lead author (S.B.F.) affirms that the manuscript is an honest, accurate, and transparent account of the study being reported, no important aspects of the study have been omitted, and any discrepancies from the study as planned (and, if relevant, registered) have been explained.
Dr Freedman designed, conceived, and developed the trial, secured funding, oversaw all aspects of data collection analysis, wrote the manuscript, had full access to all the data in the study, and takes responsibility for the integrity of the data and the accuracy of the data analysis; Dr Soofi executed the trial, recruited patients, acquired data, was a treating clinician, and critically revised the manuscript; Dr Willan wrote the statistical plan, analyzed the data, and critically revised the manuscript; Ms Williamson-Urquhart developed and executed the trial, oversaw all aspects of data collection and analysis, and critically revised the manuscript; Dr Siddiqui executed the trial, performed data extraction (entering and monitoring), oversaw local data collection, and critically revised the manuscript; Dr Xie analyzed data and critically revised the manuscript; Dr Dawood monitored data entry, performed query management, reviewed data for accuracy, and prepared the data for analysis; Dr Bhutta designed, conceived, and developed the trial, secured funding, and critically revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
This trial has been registered at www.clinicaltrials.gov (identifier NCT01870648).
FUNDING: Funded by the Bill and Melinda Gates Foundation, the Thrasher Research Fund, and the Alberta Children’s Hospital Foundation. Dr Freedman is the Alberta Children’s Hospital Foundation Professor in Child Health and Wellness. None of the study funders played any role in the study design, data collection, data analysis, data interpretation, or writing of the article. The corresponding author had full access to all the data in the study, takes responsibility for the integrity of the data and the accuracy of the data analysis, and had final responsibility for the decision to submit for publication. The researchers conducted the work independently of the funders.
Comments
RE: Oral Ondansetron Administration to Dehydrated Children in Pakistan: A Randomized Clinical Trial
We thank Drs. Kuniyoshi, Banno, and Watanabe for providing feedback regarding our manuscript.1 We do indeed concur with their comment that we have not demonstrated that ondansetron administration reduces mortality from acute gastroenteritis in low- and middle-income countries (LMICs) and that additional research building on the evidence we have presented is needed to better address this important, but different question. While we concur that mortality in young children should be a primary focus of research in this field, we excluded young and very small children, as has been done in studies in high-income countries,2 because of concerns related to alternative diagnoses (e.g. bowel obstruction) and dosing (i.e. inability to split the 4 mg oral disintegrating tablet into small enough a dose). Nonetheless, given the importance of promoting oral rehydration therapy and reducing the unnecessary use of intravenous rehydration, our findings still demonstrate that ondansetron administration improves clinical outcomes when administered to dehydrated, but not to non-dehydrated,3 children seeking emergency department care in a LMIC.
In keeping with current approaches to the conduct of pragmatic clinical trials,4 we did not direct physicians when to administer intravenous rehydration. By leaving it to the discretion of the responsible physician we were able to evaluate the real world effectiveness of single dose oral ondansetron in this clinical setting. As requested, we have analyzed our primary outcome data by study site and found that the primary outcome occurred more frequently at The Aga Khan University Hospital (23.3%) compared with The Women and Children Hospital (9.9%). Nonetheless, in the primary outcome analyses, which adjusted for study site and age, the administration of ≥20 mL/kg over 4 hours of an intravenous rehydration solution within 72 hours of randomization occurred in 14.7% vs 19.5% of those in the ondansetron and placebo groups, respectively (odds ratio [OR]: 0.71; 95% CI, 0.50 to 1.00; difference: 4.8%; 95% CI, 0.0% to 9.7%). Nonetheless, we do believe it would be important for additional research to evaluate ondansetron use in other settings, and perhaps more importantly in remote communities where access to intravenous or nasogastric rehydration may be limited.
1. Freedman SB, Soofi SB, Willan AR, et al. Oral Ondansetron Administration to Dehydrated Children in Pakistan: A Randomized Clinical Trial. Pediatrics. 2019;144(6).
2. Freedman SB, Adler M, Seshadri R, Powell EC. Oral ondansetron for gastroenteritis in a pediatric emergency department. N Engl J Med. 2006;354(16):1698-1705.
3. Freedman SB, Soofi SB, Willan AR, et al. Oral Ondansetron Administration to Nondehydrated Children With Diarrhea and Associated Vomiting in Emergency Departments in Pakistan: A Randomized Controlled Trial. Ann Emerg Med. 2019;73(3):255-265.
4. Ford I, Norrie J. Pragmatic Trials. N Engl J Med. 2016;375(5):454-463.
RE: Oral Ondansetron Administration to Dehydrated Children in Pakistan: A Randomized Clinical Trial
Freedman et al. (1) reported that oral administration of ondansetron to dehydrated children in low- and middle-income countries (LMICs) reduced vomiting and use of intravenous rehydration. We appreciate the author’s efforts to provide evidence of the efficacy of oral ondansetron therapy. However, we have two main concerns regarding the applicability of the findings to clinical practice.
First, ondansetron was shown to reduce the incidence and number of vomiting episodes within 4 hours after administration compared with placebo. A systematic review by Tomasik et al (2) also found that ondansetron may reduce the incidence of vomiting within 1 hour after administration. However, the incidence of vomiting after 4 hours, the WHO dehydration severity score at 4 hours, and the risk of hospitalization were almost similar in the ondansetron and placebo groups. There were no significant between-group differences with respect to the clinical course, expect for the rate of use of intravenous rehydration. Moreover, in LMICs, many children aged 1–11 months died from dehydration due to acute watery diarrhea (3); however, Freedman et al excluded children aged <6 months and those who weighed <8 kg. Therefore, whether ondansetron administration reduces mortality from acute gastroenteritis in LMICs is unclear from this study alone. Further studies with different outcome measures are required to reflect more severe states such as “treatment failure”; this would help confirm whether this intervention improves the clinical course in LMICs.
Second, the criteria for intravenous rehydration use (the primary outcome) were not provided in advance and were left to the discretion of the clinician. Furthermore, the timing of intravenous fluid therapy and the associated clinical information were not reported; in addition, the differences in intravenous rehydration rates between the two centers were not mentioned. The authors should provide this information. The magnitude of efficacy may be influenced by the implementation rate of intravenous rehydration in the study setting (2). Therefore, the results may not be extrapolated to other settings. Further studies are required to determine whether this intervention has similar efficacy in other settings in LMICs.
Reference
1. Freedman SB, Soofi SB, Willan AR, et al. Oral Ondansetron Administration to Dehydrated Children in Pakistan: A Randomized Clinical Trial. Pediatrics. 2019; 144(6): e20192161.
2. Tomasik E, Ziółkowska E, Kołodziej M, Szajewska H. Systematic review with meta-analysis: ondansetron for vomiting in children with acute gastroenteritis. Aliment Pharmacol Ther. 2016; 44(5): 438–446.
3. Rahman AE, Moinuddin M, Molla M, et al. Childhood diarrhoeal deaths in seven low- and middle-income countries. Bull World Health Organ. 2014; 92(9): 664–671.