Page 115 - 15Diarrhoeaandvomiting

7 Antibiotic therapy

Introduction
Gastroenteritis is most often caused by a viral enteric pathogen, and even in those with
bacterial or protozoal infections the disease is generally self-limiting. Treatment has usually
been recommended for dysenteric shigellosis and amoebiasis, cholera, pseudomembranous
colitis, and for some other organisms in particular circumstances. 140 The evidence relating to
the use of antibiotics in young children with gastroenteritis was reviewed with a view to making
recommendations on good practice.

Clinical question
What is the role of antibiotic therapy in children with gastroenteritis?
A search for randomised trials or systematic reviews comparing antibiotic treatment with placebo
or no treatment of gastroenteritis in children was performed. Of 130 citations identified, 25
were retrieved in full copy. Of these, 11 were excluded, but 14 compared use of antibiotics with
placebo and were included in this review.

7.1 Salmonella

Evidence overview
Four randomised trials 141–144 evaluated the effectiveness of antibiotic treatment compared with
placebo or no treatment for children presenting to hospital with acute diarrhoea. These trials were
conducted in the USA, Taiwan, Canada and Colombia. Three trials had three treatment
141
144
142
143
arms 141–143 and one trial had two treatment arms but according to protocol changed the route
144
of antibiotic administration from intramuscular (IM) to oral in the second year of the study.
The first RCT, from the USA, 141 included children aged up to 8 years (n = 45) presenting at
hospital with acute diarrhoea and with salmonella species subsequently isolated in rectal swab
cultures. Children with a history of adverse drug reactions to penicillins, with another focus of
infection or who were under 6 weeks of age were excluded. Participants were randomised to
one of three treatment arms to compare the effects of ampicillin (100 mg/kg per day) (n = 15),
amoxicillin (100 mg/kg per day) (n = 15) and placebo (n = 14) given in four equal doses daily
for 5 days for the treatment of salmonella gastroenteritis. Computer-generated random number
lists were used to assign the children to pre-coded drugs. Separate randomisation lists were used
for children under and over 1 year of age. The main outcomes assessed were the mean days
until diarrhoea stopped (diarrhoea cessation defined as the day of the first formed stool without
mucus), the mean days until diarrhoea improved (defined as improved stool consistency and
a decrease in the number of stools) , the mean days until the first negative culture (defined as
the first of at least two consecutive negative cultures), the mean days until last positive culture,
bacteriological relapse and diarrhoea relapse. Outcome assessors were blind to the treatment the
children received. [EL = 1−]

The groups were broadly comparable at baseline, except that children receiving amoxicillin were
younger (mean age 7.7 ± 1.7 months) than those in the ampicillin (mean age 15.7 ± 5.7 months)
and placebo groups (mean age 19.8 ± 7.4 months). There were no statistically significant differences
between the ampicillin, amoxicillin and placebo groups for the mean number of days until
diarrhoea stopped or the mean number of days until the first negative culture. Participants receiving
ampicillin and amoxicillin did have a significantly reduced mean number of days until diarrhoea
improved compared individually with placebo (WMD −1.20 days; 95% CI −1.65 to −0.75 days
and WMD −1.00 days; 95% CI −1.45 to −0.55 days, respectively), but the difference across all
three groups was not significant (Kruskal–Wallis nonparametric ANOVA P > 0.2). Excretion of

90