Study reference

Study characteristics

Patient characteristics

Intervention (I)

Comparison / control (C)

Follow-up

Outcome measures and effect size

Comments

Fugetto, 2020

SR and meta-analysis of randomized controlled trials.

Literature search up to November 2019.

1. Cubeddu (1997)
2. Danewa (2016)
3. Freedman (2006)
4. Freedman (2018)
5. Golshekan (2013)
6. Hagbom (2017)
7. Marchetti (2016)
8. Ramsook (2002)
9. Rang (2019)
10. Rerksuppophol (2010)
11. Roslund (2008)
12. Stork (2006)
13. Yilmaz (2010)

Study design:

RCTs evaluating ondansetron administered orally or IV at any dosage versus placebo, prescribed to terminate or reduce vomiting in children and adolescents under the age of 18 with clinically suspected acute gastroenteritis.

Setting and Country:

Department of Medical and Surgical Sciences (DIMEC), Pediatric

Emergency Unit, St. Orsola Hospital, University of Bologna, via

Massarenti 9, 40138 Bologna, Italy.

Source of funding

There is no funding source.

Conflicts of interest:

The authors declare that they have no conflict of interest.

Inclusion criteria SR:

• RCTs written in English language;
• RCTs that included patients younger than 18 years of age.

Exclusion criteria SR:

• Not reported.

Seven studies included

Important patient characteristics at baseline:

N

1. 36 children
2. 170 children
3. 214 children
4. 626 children
5. 176 children
6. 81 children
7. 356 children
8. 145 children
9. 61 children
10. 74 children
11. 106 children
12. 137 children
13. 109 children

mean age

1. 6 months to 8 years
2. 3 months to 5 years
3. 6 months to 10 years
4. 6 months to 5 years
5. 1 to 10 years
6. 6 months to 16 years
7. 1 to 6 years
8. 6 months to 12 years
9. 11 to 60 months
10. 3 months to 15 years
11. 1 to 10 years
12. 5 months to 10 years
13. 5 months to 8 years

Sex, n/N M

1. 21/36 M
2. 99/170 M
3. 122/214 M
4. 372/626 M
5. 103/176 M
6. 37/81 M
7. 176/356 M
8. 82/145 M
9. 35/61 M
10. 38/74 M
11. 47/106 M
12. 74/137 M
13. 70/109 M

Groups comparable at baseline?

Yes.

1. Ondansetron;

Route: IV

Dose: 0.3 mg/kg single dose

2. Ondansetron;

Route: oral (syrup)

Dose: 0.2 mg/kg single dose

3. Ondansetron;

Route: oral (ODT)

Dose: 8-15 kg: 2 mg; 15-30 kg: 4 mg; > 30 kg: 8 mg (all single dose).

4. Ondansetron;

Route: oral (ODT)

Dose: 8-15 kg: 2 mg; 15-30 kg: 4 mg (single dose)

5. Ondansetron;

Route: oral (tablet)

Dose: < 15 kg: 2 mg; 15-30 kg: 4 mg; >30 kg: 6 mg (single dose)

6. Ondansetron;

Route: oral (syrup)

Dose: 0.15 mg/kg (single dose)

7. Ondansetron;

Route: Oral (syrup)

Dose: 1.6 mg every 8 h for patients aged 6 months to 1 year, 3.2 mg every 8 h for patients aged 1 to 3 years, and 4 mg every 8h for patients aged 4 to 12 years (six doses in total)

8. Ondansetron;

Route: Oral (syrup)

Dose: 0.15 mg/kg (single dose).

1. Ondansetron;

Route: IV

Dose: IV ondansetron at a dose of 0.2 mg/Kg up to maximal dose of 8 mg (single dose)

10. Ondansetron;

Route: IV

Dose: 0.15 mg/kg up to maximal dose of 8 mg (single dose)

11. Ondansetron;

Route: oral (ODT)

Dose: < 15 kg: 2 mg; 15-30 kg: 4 mg; >30 kg: 6 mg (single dose).

12. Ondansetron;

Route: IV

Dose: 0.15 mg/kg (single dose).

13. Ondansetron;

Route: Oral (ODT)

Dose: 0.2 mg/kg at 8h intervals (three doses in total).

All studies compared the intervention to placebo.

End-point of follow-up:

1. 24 hours
2. Not reported.
3. 8 hours
4. 8 hours
5. 8 hours
6. 8, 24 and 48 hours
7. 8 hours
8. 8 hours
9. 8 hours
10. 24 hours
11. 8 hours
12. 8 hours
13. 8 hours

For how many participants were no complete outcome data available?

Not reported.

1. Success/failure of treatment

(Failure of oral rehydration therapy during the emergency department stay)

Effect measure: RR (95% CI:

A: 0.25 (0.03 to 1.92)

B: 0.50 (0.35 to 0.72

C: 0.41 (0.24 to 0.69)

E: 0.51 (0.27 to 0.94)

M: 0.33 (0.19 to 0.56)

Pooled effect (random effects model):

RR 0.43 (95% CI 0.34 to 0.55) favoring Ondansetron

Heterogeneity (I²): 0%

2. Time to rehydration

Not reported

3. Need for nasogastric tube/rehydration

Not reported

4. Need for hospitalization (incidence of hospitalization within eight hours after the administration of the treatment)

Effect measure: RR (95% CI:

C: 0.80 (0.22 to 2.90)

E: 0.56 (0.27 to 1.13)

G: 0.50 (0.24 to 1.01)

L: 0.21 (0.05 to 0.93)

K: 0.46 (0.13 to 1.69)

Pooled effect (random effects model / fixed effects model):

RR 0.49 (95% CI 0.32 to 0.75) favoring Ondansetron

Heterogeneity (I²): 0%

5. Length of hospital stay

Not reported

6. Intravenous rehydration during the ED stay

Effect measure: RR (95% CI:

B: 0.56 (0.30 to 1.07)

C: 0.45 (0.26 to 0.79)

D: 0.74 (0.34 to 1.58)

G: 0.41 (0.23 to 0.72)

H: 0.39 (0.16 to 0.94)

I: 0.26 (0.08 to 0.83)

K: 0.34 (0.18 to 0.65)

M: 0.20 (0.01 to 4.00)

Pooled effect (random effects model ):

RR 0.44 (95% CI 0.34 to 0.57) favoring Ondansetron

Heterogeneity (I²): 0%

7. Return visit to the ED

Effect measure: RR (95% CI:

C: 1.08 (0.59 to 1.95)

D: 1.87 (0.18 to 19.84)

G: 0.84 (0.39 to 1.80)

H: 8.64 (0.47 to 157.62)

K: 1.40 (0.25 to 7.97)

Pooled effect (random effects model / fixed effects model):

RR 1.14 (95% CI 0.74 to 1.76) favoring Ondansetron

Heterogeneity (I²): 0%

8. Weight loss

Not reported

9. Vomiting (Cessation of vomiting within 8 hours after the administration of the treatment)

Effect measure: RR (95% CI):

C: 1.31 (1.12-1.54)

D: 1.06 (0.97-1.15)

E: 1.05 (0.94-1.19)

G: 1.42 (1.20-1.69)

H: 1.33 (1.10-1.62)

I: 3.25 (1.63-6.45)

J: 3.33 (1.85-6.01)

K: 1.11 (0.84-1.47)

M: 2.35 (1.19-1.68)

Pooled effect (random effects model):

RR 1.41 (95% CI 1.19 to 1.68) favoring placebo

Heterogeneity (I²): 86%

9. Vomiting (Cessation of vomiting 24 hours after the administration of the treatment) – not meta-analyzed

Effect measure: RR (95% CI)

A: 0.50 (95% CI 0.24 to 1.02)

P=0.039

10. Diarrhea (number of diarrhea episodes 24 hours after the administration of the treatment) – not meta-analyzed

F (Hagbom, 2017)

I: 3.75 (SD 7.15)

C: 5.68 (SD 8.50)

Mann-Whitney U-test p=0.063

I (Rang, 2019)

I: median (IQR) 5 (2 to 18)

C: median 6 (1 to 18)

P=0.913

11. Complications

1. Cubeddu (1997)
2. Hagbom (2017)
3. Ramsook (2002)
4. Rerksuppophol (2010)
5. Roslund (2008)
6. Stork (2006)
7. Yilmaz (2010)

B Golshekan (2013): mentioned but not reported

Any

B (Danewa, 2016)

I 0/84

C 0/83

Cough

A (Cubeddu, 1997)

I: 3/12

C: 0/12 (0%)

Development of macular rash

H (Ramsook, 2002)

I: 1/74

C: 0/71 (0%)

Diarrhea, 24 hours

F (Rang, 2019)

I: median 5 (IQR: 2-18)

C: 6 (1-18)

H (Ramsook, 2002)

I: 64/74

C: 54/71

P=0.02

Diarrhea, 48 hours

H (Ramsook, 2002)

I: 62/74

C: 51/71

P=0.015

G (Marchetti 2016)

I: 49/118

C: 51/119

RR 1.10 (95% CI: 0.74 to 1.64)

Episodes of diarrhea during ORT

C (Freedman, 2006)

I: 1.4

C: 0.5

P<0.001

Abdominal distension

M (Yilmaz, 2010)

P=0.04 à not further specified.

Facultative:

Author’s conclusion:

Mixed evidence suggests that ondansetron affects cessation of vomiting, oral rehydration therapy failures, need for intravenous rehydration, and hospitalization prevention in children with acute gastroenteritis. No differences could be examined in adverse events as the number of studies presenting these outcomes in a homogenous way was limited, and further studies are necessary.

Remark on Danewa: they reported no adverse events, and that they did “not observe any increase in diarrhea with odansetron use”

Fugetto, 2020

Yes

The objective of this meta-analysis was to provide up-to-date evidence regarding the efficacy and safety of a single dose of ondansetron versus placebo prescribed for vomiting due to acute gastroenteritis in children and adolescents.

Yes

Electronic databases were searched to identify relevant studies published up to November 2019, including MEDLINE (PubMed), Cochrane Central Register of Controlled Trials (CENTRAL), and Scopus (Elsevier). Reference lists of select- ed reviews, original articles, textbooks, organizations’ websites, and guideline clearinghouses were scanned in order to find additional articles. Investigators and corresponding authors were contacted to seek information about unpublished/incomplete studies. A snowball technique was applied to the search strategy

Yes

Thirteen placebo-controlled, double-blinded, peer-reviewed publications (2146 pediatric patients) were included. We re- stricted the focus to RCTs evaluating ondansetron adminis- tered orally or IV at any dosage versus placebo and prescribed to terminate or reduce vomiting in children and adolescents.

Yes

The trials were conducted in different countries and over different periods. Two RCTs included hospitalized patients,, while the remaining were carried out in emergency departments. The inclusion criteria for enrolment were similar for all studies, and the age of participants ranged from 3 months to 16 years. All studies included subjects with none to moderate dehydration, except four in which hydration status was not specified.

Not applicable

Yes

Substantial heterogeneity was found among the studies in- cluded in the primary outcome analysis. Using random- effects analysis, it was found that at least a part of the heterogeneity was due to differences in age and proportion of males included in the population in each RCT. There was some asymmetry observed in the funnel plot for the primary study outcome analysis, suggesting possible

publication bias or small study effects

Yes

Yes

There was some asymmetry observed in the funnel plot for the primary study outcome analysis, suggesting possible

publication bias or small study effects

Yes

The authors declare that they have no conflict of interest.

Study reference

(first author, publication year)

Was the allocation sequence adequately generated? ^a

Definitely yes

Probably yes

Probably no

Definitely no

Was the allocation adequately concealed?^b

Definitely yes

Probably yes

Probably no

Definitely no

Blinding: Was knowledge of the allocated

interventions adequately prevented?^c

Were patients blinded?

Were healthcare providers blinded?

Were data collectors blinded?

Were outcome assessors blinded?

Were data analysts blinded?

Definitely yes

Probably yes

Probably no

Definitely no

Was loss to follow-up (missing outcome data) infrequent?^d

Definitely yes

Probably yes

Probably no

Definitely no

Are reports of the study free of selective outcome reporting?^e

Definitely yes

Probably yes

Probably no

Definitely no

Was the study apparently free of other problems that could put it at a risk of bias?^f

Definitely yes

Probably yes

Probably no

Definitely no

Overall risk of bias

If applicable/necessary, per outcome measure^g

LOW

Some concerns

HIGH

Cubeddu, 1997

No information

Reason: “Patients were randomly assigned to receive either....”.

Defenitely yes;

Reason: “The study medication was prepared by a pharmacist not involved in patient care...".

No information

Reason: Measures used to blind study participants and personnel from knowledge of which intervention a participant received and information relating to whether the intended blinding was effective not provided.

Probably yes;

Reason: All data appear to have been reported in the study outcomes.

Defenitely yes;

Reason: No evidence of selective reporting. Outcomes listed in the methods section were comparable to those reported.

Definitely no;

Reason: The authors reported a baseline imbalance in age; weight; and height between the ondansetron and placebo treatment arm. This imbalance was not adjusted for in the analysis.

Fecal analysis performed: 4 samples resulted positive for bacteria and 1 for both bacteria and virus among the intervention group, while 2 samples result positive for bacteria among placebo group. All patients were included in final analysis.

The trialists reported that the study was supported by Glaxo Wellcome and although the level of support was unclear two of the investigators were in the employment of Glaxo Wellcome.

SOME CONCERNS

Danewa, 2016

Probably yes;

Reason: "Computer-generated block randomization with variable block sizes was used to assign patients...".

Probably yes;

Reason: "Bottles were coded with this randomization scheme by a person not directly involved with the study".

Probably yes;

Reason: "Study subjects, their caregivers, the person assigning intervention, doctors managing the patient, and outcome assessors were blinded ...".

Probably yes

Reason: The authors indicated that they followed the intention to treat principle.

Definitely yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

Probably yes;

Reason: Apparently free of other bias.

LOW

Freedman, 2006

Probably yes;

Reason: “The patients were randomly assigned in blocks of six to receive Ondansetron or placebo and were stratified according to the dose of medication”. “An independent statistician provided the code to the pharmacy”.

Probably yes;

Reason: “An independent statistician provided the code to the pharmacy, which dispensed in an opaque bag a weight-appropriate dose of active drug or placebo”.

Probably yes;

Reason: “active drug or placebo of similar taste and appearance".

Probably yes;

Reason: The authors indicated that they followed the intention to treat principle.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

No information;

Reason: Although no potential conflicts of interest were reported, this trial was supported partly by a grant from GlaxoSmithKline but the level of support was not declared. The risk of ’other bias’ was therefore judged unclear.

LOW

Freedman, 2018

Probably yes;

Reason: "... Randomization was stratified by study center and age (<18 and >18 months), using variable block sizes of 4 to 6...".

Probably yes;

Reason: " Allocation was concealed through the use of an Internet-based randomization service...".

Probably yes;

Reason: "A prespecified computer-generated randomization list with associated kit numbers was sent directly to the University of Calgary research pharmacist from http://www.randomize.net through password-protected files. At patient enrollment, http://www.randomize.net randomly assigned treatment and then randomly selected a kit number containing..." "Patients, treating physicians, investigators, and data assessors were masked...".

"Patients, treating physicians, investigators, and data assessors were masked...".

Probably yes;

Reason: The Authors indicated no lost on follow-up or excluded from the final analysis.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

Probably no;

Reason: Only children without evidence of dehydration assessed with the WHO dehydration tool were included in this study, The Authors state that this population was important to study because they believed they were at lower risk of potential complications associated with ondansetron administration compared with children who are dehydrated because the latter group is more likely to have electrolyte abnormalities and perhaps more likely to receive intravenous rehydration without a proper trial of oral rehydration therapy.

Many study participants were coadministered with other medications, including antibiotic and antiemetic agents including domperidone.

SOME CONCERNS

Golshekan, 2013

Probably yes;

Reason: "The patients were randomized in two groups by computer randomization using a block of two...".

Probably yes;

Reason: "The patients were randomized in two groups by computer randomization using a block of two...".

Probably yes;

Reason: "The patients were randomized in two groups by computer randomization using a block of two..." "Also investigators were blinded to group assignment until after complete statistical analysis".

"Also investigators were blinded to group assignment until after complete statistical analysis".

Probably yes;

Reason: The authors indicated that they followed the intention to treat principle.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

Probably no;

Reason: Hydration status of included children not known.

SOME CONCERNS

Hagbom, 2017

Probably yes;

Reason: " Randomization was done in blocks (n = 8) and the code key was sealed...".

Probably yes;

Reason: "Drug and placebo were labeled blinded with A or B and a randomization list was...".

Probably yes;

Reason: "Children who met the inclusion criteria and whose parents or guardians gave written consent to participate in the study were examined by a physician and randomized by a nurse into ondansetron or placebo treatment, blinded as A or B solution".

"A blinded member of the study team, located at Linko¨ping University in Sweden, phoned the guardian 24–72 hour".

Probably no;

Reason: In the absence of an intention-to-treat analysis, losses to follow-up resulted in incomplete data for several of the outcomes which were pre-specified.

Probably no;

Reason: The addition of a secondary endpoint after treatment was introduced from the 21st participant and forward could have had serious and invaluable effects on final analyses.

Probably no;

Reason: The study randomized patients before their full eligibility had been ascertained which resulted in 4 patients dropping out.

HIGH

Marchetti, 2016

Probably yes;

Reason: "Patients were randomly assigned in fixed blocks of nine".

Probably yes;

Reason: "The randomization procedure was centralized. The randomization sequence was transmitted to the pharmaceutical development service (Monteresearch S.r.l.), that prepared an...".

Probably yes;

Reason: "The randomization procedure was centralized. The randomization sequence was transmitted to the pharmaceutical development service (Monteresearch S.r.l.), which prepared and sent directly to participating hospitals, active drugs and placebo in closed, opaque and consecutively numbered bags. Drug preparations were indistinguishable...".

"The randomization procedure was centralized. The randomization sequence was transmitted to the pharmaceutical development service (Monteresearch S.r.l.), which prepared and sent directly to participating hospitals, active drugs and placebo in closed, opaque and consecutively numbered bags. Drug preparations were indistinguishable...".

Probably yes;

Reason: The authors indicated that they followed the intention to treat principle.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

Probably yes;

Reason: Apparently free of other bias.

LOW

Ramsook, 2002

Probably yes;

Reason: "random allocation procedure was designed using standard random number allocation tables".

Probably yes;

Reason: "The pharmacy research section assigned treatment or placebo according to

this individual randomization”. “The pharmacy team was not privy to the enrolled patients or the outcome measures. This code remained locked within the pharmacy research section and was broken and revealed to the investigators only at the close of the study".

Probably yes;

Reason: "the pharmacy provided the drug or a color, taste, and odor-matched placebo in identical packaging.".

"This code remained locked within the pharmacy research section and was broken and revealed to the investigators only at the close of the study".

Probably yes;

Reason: The authors indicated that they followed the intention to treat principle.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

No information

Reason: This trial was supported partly by a grant from GlaxoSmithKline but the level of support was not declared. The risk of ’other bias’ was therefore judged unclear.

LOW

Rang, 2019

Probably yes;

Reason: "The random number sequence was generated in Excel".

No information

Reason: Not specified.

Probably yes;

Reason: "The treatment assignments were concealed in opaque, sealed envelopes".

" ... both drug and placebo were provided in the same volume (10 mL) in syringes, with code numbers, by a clinical pharmacist".

Probably yes;

Reason: No attrition bias detected.

No information

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

No information

Reason: The treating physicians prescribed anti-diarrheal medications such as smectite or racecadotril for some patients, which could have affected the duration of emesis and diarrhea.

The sample size may have been inadequate to detect small differences in some outcomes, especially adverse effects.

SOME CONCERNS

Rerksuppaphol, 2010

Probably yes;

Reason: "Participants were randomized in two groups by a computerized program using a block of two".

No information

Reason: Not explained in the text.

No information

Reason: Not explained in the text.

Probably yes;

Reason: The authors indicated that they followed the intention to treat principle.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

No information

Reason: It is not clear if the two randomized groups are similar in initial dehydration grade.

SOME CONCERNS

Roslund, 2008

Probably yes;

Reason: " patients were randomized to receive oral Ondansetron or placebo”, “block randomization of 10".

No information

Reason: "Each subject was assigned a packet with a corresponding number. Each packet contained a tracking form used for documenting the subject’s course in the ED".

Probably yes;

Reason: "[placebo] looked smelled and tasted like ondansetron..." "The packets were prefilled (oral ondansetron or placebo)..." "The markings on the blister pack were obscured...".

"healthcare providers were the assessors during the study and the research nurse, who was blinded to the treatment allocation, completed the follow-up for the study.".

Probably no;

Reason: Trialists considered different denominators (n= 51 and n=48) in the ondansetron group when presenting the outcomes.

Post-discharge informations not available.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

Probably yes;

Reason: Apparently free of other bias.

SOME CONCERNS

Stork, 2006

Probably yes;

Reason: "Randomization was performed in the pharmacy using specific instructions to access a provided table of random numbers to identify the patient’s study group".

Probably yes;

Reason: "Randomization was performed in the pharmacy".

Probably yes;

Reason: "all study drug dispensed was uniform in design and color".

The healthcare providers were the assessors but blinding was ensured, and unlikely that the blinding could have been broken.

Probably no;

Reason: Losses to follow-up and missing data for several outcomes.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

Probably no;

Reason: The study randomized patients before their full eligibility had been ascertained which resulted in 19 patients dropping out and a further 10 being discharged shortly after randomization.

SOME CONCERNS

Yilmaz, 2010

Probably yes;

Reason: "computerized randomization codes .. in blocks of six".

Probably yes;

Reason: "...by a statistician who was not one of the investigators of this study..." "Code numbers were written on the injectors and...drawn up...by a pharmacist who was not one of the investigators in this study.".

Probably yes;

Reason: "orally disintegrating Ondansetron tablets dissolved.. and identical-looking placebo liquid”. “The subjects, parents, study personnel and other medical staff were blinded to which study drug was given".

" identical-looking"...."study personnel and other medical staff were blinded to which study drug was given".

Probably no;

Reason: Two participants in the placebo group were removed from the study after receiving IV rehydration and were admitted to the hospital but were not included in the analysis.

Probably yes;

Reason: Outcomes listed in the methods section comparable to the reported results. No evidence of selective choice of data for outcomes.

Probably yes;

Reason: Apparently free of other bias.

SOME CONCERNS