Page 121 - 15Diarrhoeaandvomiting

Diarrhoea and vomiting caused by gastroenteritis in children under 5 years

four children in the placebo group died, three from persistent diarrhoea with septicaemia and
one who additionally had congestive heart failure (re-feeding syndrome). This difference was
statistically significant (P = 0.041).

A comparative trial 150 conducted in Egypt was identified that examined the effect of nitazoxanide
and co-trimoxazole compared with placebo for clinical and microbiological ‘cure’ of
cryptosporidium infection. This trial was poorly reported, with no details regarding the methods
used or the baseline comparability of the treatment groups. Consequently, it was considered to
be potentially highly biased. [EL = 1−]
Of 1087 patients with diarrhoea examined in the outpatient clinic, 150 were found to have
cryptosporidiosis. This was confirmed by two stool diagnostic tests (Ziehl–Neelsen stain and direct
immunofluorescent technique). Of these 150 patients, 73 were children. Adults and children
were divided into three treatment groups (nitazoxanide, co-trimoxazole or placebo) and were
followed for up to 10 days. Clinical cure was not defined in the study. Microbiological cure was
defined as two consecutive negative stool samples. Results of numbers of patients ‘cured’ were
presented, although it was not clear which ‘cure’ was measured and when measurements were
taken – microbiological cure by the 10th day was presumed. There was a statistically significant
difference in the number of children cured (21/24) following administration of nitazoxanide
compared with placebo (9/25) (RR 2.43; 95% CI 1.41 to 4.19; P = 0.001). A statistically significant
difference was not demonstrated for the comparison of co-trimoxazole (8/24) versus placebo
(RR 0.93; 95% CI 0.43 to 2.00; P = 0.84).

Evidence summary
A good-quality RCT conducted in a population of malnourished children showed that nitazoxanide
was effective in achieving a clinical and parasitological response to treatment, and in preventing
deaths. Results from the other, potentially biased, controlled trial suggested that nitazoxanide but
not co-trimoxazole was effective in achieving a microbiological cure in children younger than
12 years with diarrhoea of cryptosporidium infection.

7.7 Treatment without prior identification of a pathogen

Evidence overview

Four RCTs were identified as relevant: two studies were conducted in South Africa 151,152 and two
in Mexico. 153,154 Data from 378 children (aged 2 months to 7 years) suffering from gastroenteritis
were collected across the four studies, Three trials had two treatment arms 151–153 and one trial had
three treatment arms. 154 The proportion of patients randomised but lost to follow-up was reported
in all the studies (less than 20%). None of the trials included a sample size power calculation.
A trial conducted in South Africa 151 compared the effect of trimethoprim/sulphonamide with
placebo for the treatment of gastroenteritis. Children aged 5–30 months admitted to hospital for
gastroenteritis (n = 34) were randomised into two treatment arms. There were 18 participants in
one group and 26 in the other, although the authors did not specify which group received which
treatment. No details of the randomisation process were given, but the investigators, patients and
outcome assessors were blinded to treatment allocation. At baseline, the treatment groups were
similar for age. No statistically significant differences were found between the two groups for the
mean duration (in days) of diarrhoea, vomiting, pyrexia or hospital stay. [EL = 1−]
The second trial from South Africa 152 examined the effect of erythromycin compared with
placebo for the treatment of non-specific gastroenteritis. Included in the study were children
aged 1 month to 2 years who had been admitted to a hospital with a history of diarrhoea not
exceeding 96 hours and who had received no antimicrobial therapy for the illness (n = 78).
Participants were randomised into two treatment groups and received either erythromycin oral
suspension 40 mg/kg per day in divided doses for 5 days (n = 39) or placebo oral suspension
(n = 39). Follow-up was by daily examination for 7 days. Treatments for the trial were supplied
by the manufacturer as granules for reconstitution in pre-randomised and coded containers.
The patients and the outcome assessors were blind to treatment allocation. The comparability
of the groups at study entry was adequate. The distribution of pathogens was similar between

96