Page 120 - 15Diarrhoeaandvomiting

Antibiotic therapy

7.5 Escherichia coli

Evidence overview
The Colombian trial described above also investigated the effects of ampicillin compared
144
with placebo on serology-confirmed enteropathogenic E. coli infection, as well as shigella and
salmonella infections (total n = 282). In total, 35 of 282 infants and children younger than 2 years
admitted to hospital with diarrhoea as a major symptom had E. coli isolated by stool culture. Of
these, 18 received either IM or oral ampicillin (100 mg/kg per day in equally divided doses every
6 hours for 5 days) and 17 received either IM fructose or oral placebo in every 6 hours for 5 days.
There were no statistically significant differences between either ampicillin groups, or between the
ampicillin and placebo groups in the mean number of days until diarrhoea improved or ceased or
in the mean number of days until the patient became afebrile or stool culture negative. [EL = 1+]

Evidence summary

Results from one trial showed no statistically significant differences between either ampicillin
groups or between the ampicillin and placebo groups in the mean number of days until diarrhoea
improved or ceased or in the mean number of days until the patient became afebrile or culture
negative. Although this trial was of good quality, there were only 35 children evaluated to inform
this question.

7.6 Cryptosporidium

Evidence overview
Two studies were identified for inclusion. One was a good-quality trial from Zambia but it
included HIV-seropositive and malnourished children, while the other trial had methodological
limitations and included both children and adults as the study population.
An RCT 149 conducted in Zambia recruited children (n = 100) who were admitted to hospital
with diarrhoea, who had Cryptosporidium parvum oocytes identified from a pre-enrolment stool
sample and whose parents consented to the child having an HIV test. All children were stabilised
with fluid therapy, antibiotics and mineral supplementation as required throughout the study.
Exclusion criteria were age under 1 year and receipt of a drug with antiprotozoal activity within
2 weeks of enrolment to the study. Following stratification by HIV status (n = 50 in each group),
children were randomly assigned using a computer-generated random number list to treatment
with 20 g/l nitazoxanide (n = 25) or placebo (n = 25) oral suspension (5 ml twice daily for
three consecutive days). The patients and outcome assessors were not aware of the treatment
allocation. The children were followed up in hospital for 8 days. The main outcomes were the
clinical response on day 7 (well or continuing illness), the parasitological response, the time from
first treatment to last unformed stool, and mortality by day 8. Results are presented here for the
HIV-seronegative subgroup only (n = 50). [EL 1+]
Twenty-five HIV-seronegative children were allocated to each treatment group, although three
from the placebo group were subsequently excluded as no Cryptosporidium parvum oocytes
were detected in their stool at baseline. The groups were similar for sex, age, weight, malnutrition
status, laboratory abnormalities and stool frequency. Children receiving placebo had a shorter
mean duration of diarrhoea prior to enrolment than the nitazoxanide group, but this reflected
one child in the nitazoxanide group reporting 91 days of diarrhoea prior to enrolment. Eleven
children in each group had physical signs of malnutrition. Oedema indicative of kwashiorkor or
marasmic kwashiorkor was detected in 22 children (10/25 nitazoxanide, 12/22 placebo).
Fourteen of 25 children receiving nitazoxanide were ‘well’ (rather than having ‘continuing illness’)
at day 7 compared with five of 22 of those receiving placebo. ‘Well’ was defined as having no
symptoms, no watery stools and no more than two soft stools within the previous 48 hours.
‘Continuing illness’ was defined as not fulfilling the definition of ‘well’. This difference was
statistically significant (P = 0.037). At day 7, significantly more children in the nitazoxanide group
(13/25) had a parasitological response (defined as two negative stool examinations) compared
with the placebo group (3/22) (P = 0.007). There were no deaths in the nitazoxanide group but

95