We did a randomised, double blind, placebo controlled clinical trial at the Dhaka Hospital of the International Centre for Diarrhoeal Disease Research, Bangladesh between November 2000 and June 2002.
Eligible patients were children of both sexes, aged 3-14 years, admitted to the hospital with cholera within 24 hours of onset, who had some dehydration or severe dehydration according to World Health Organization criteria,10
had a high purging rate (stool output more than 4 ml/kg/hour for the first six hours of observation after initial rehydration), and had positive dark field examination of stool (a direct observation of the movement of V cholerae
against a dark background under high power in a light microscope). We excluded patients with any of the following criteria: unable to rehydrate within six hours, negative dark field examination, systemic illness requiring immediate administration of antibiotics, receipt of antibiotics within 24 hours before admission to hospital, presented with bloody mucoid diarrhoea, unconscious, or had medical emergencies. The parents or guardians/caregivers of each patient gave written informed consent before enrolment into the study. We estimated the sample size needed to detect a 20% difference in the duration of illness between the zinc group and control group at 90% power and 5% level of confidence to be 78 in each group.11
Considering the 15% drop out according to Kirkwood, the total sample size was 180.12
Study groups and randomisation
We assigned patients to the study groups when they met the criteria for selection and recruited patients into the study when cholera was confirmed by stool culture. We used a block randomisation with a block length of six. We randomised enrolled children into two groups: those in the intervention group received zinc supplementation and those in the control group received a placebo for zinc; all children received erythromycin. For double blinding, we used identical bottles containing syrup of same colour and flavour with or without zinc and labelled the bottles only with the random number previously allocated to one of the two interventions.
Patients in the zinc group received 30 mg of elemental zinc as acetate each day in two equally divided doses, and those in the control group received placebo, until resolution of diarrhoea or for up to seven days. Base syrup was used as placebo for patients randomised to the control group, and zinc was added to this base syrup for patients randomised to the zinc group. The base syrup used for both intervention groups was of the same chemical composition. Cost of zinc treatment was $0.14 (£0.07; €0.10) for three days. All children received erythromycin in a dose of 12.5 mg/kg every six hours for three days (total of 12 doses). We used the disc diffusion method to test the sensitivity of V cholerae to erythromycin.
After screening for positive dark field results, we corrected dehydration with intravenous fluid, oral fluid, or both. Patients were rehydrated with intravenous polyelectrolyte solution (Dhaka Solution with 133 mmol/l sodium, 13 mmol/l potassium, 98 mmol/l chloride, and 48 mmol/l acetate) or rice based oral rehydration salts solution (with 90 mmol/l sodium, 20 mmol/l potassium, 80 mmol/l chloride, and 30 mmol/l bicarbonate), depending on their dehydration status and according to WHO guidelines. From the start of the study, we measured intakes of oral rehydration salts and fluids and output of stool and urine eight hourly until resolution of diarrhoea. Study physicians observed stools for watery, liquid, soft, or formed consistency, with daily validation by the investigators (SKR and JMH). All children had a standardised hospital diet, which included milk sujee (rice powder boiled in milk with added sugar), rice, chicken or fish, vegetables, dal (lentils), milk, bread, and banana. We cultured stools for isolation of V cholerae.
We withdrew patients from the study if they developed complications such as excessive vomiting, abdominal distension, pneumonia, hypokalaemia, or septicaemia or left the hospital. In some patients who had excessive vomiting, oral rehydration salts had to be stopped and intravenous fluids given to correct dehydration. Patients with abdominal distension due to ileus were kept nil by mouth and transferred to the special care unit for appropriate treatment with other antibiotics. Patients who developed hypokalaemia, pneumonia, or septicaemia were also transferred to the special care unit. These patients who dropped out could not follow the study protocol; we included them in the intention to treat analysis. The end point for these patients was drop-out rather than clinical recovery. The remaining patients followed the full procedure of the study to reach the end points, and we included them in the per protocol analysis.
Owing to resources limitations, we collected 2 ml of venous blood from every alternate randomised child for estimation of serum zinc on admission and on day 10 after the start of the intervention. We used an atomic absorption spectrophotometer (Pye Unicam, SP9) to measure serum zinc.4
We collected urine separately by the application of paediatric urine collection bags. We measured stool weight with a pre-weighed bucket into which stool was collected directly by means of a cholera cot with plastic sheets draining stool directly to the bucket. The weight of the stool was measured in a Toledo Scale (Toledo, USA) with a precision of 1 g.4 5
We started measurement of stool immediately after recruitment into the study and stopped when the last liquid stool was passed.
We used body weight after initial rehydration to determine nutritional status. We measured body weights every day between 9 am and 11 am during the hospital stay; we measured heights once at discharge after recovery rather than on admission when patients were unable to stand correctly. We calculated the nutritional status of the children as a percentage of the National Centre for Health and Statistics median.
The primary outcomes were diarrhoeal stool output and time to resolution of diarrhoea (defined as the time from the start of rehydration and collection of stool until the first formed stools or no stool for 24 hours; every eight hour period of measuring stool output was counted for defining the recovery period). Secondary outcome measures were the serum zinc status and the intake of oral rehydration salts or intravenous fluid.
The investigators checked data daily at recording, cleaned them, and then entered them on a personal computer with SPSS/PC+. The consistency check used logical programs. We tested normally distributed continuous data for comparison of group means by using Student’s t test. We used Kaplan-Meier survival analysis to compare differences in the time to resolution of diarrhoea between the two groups and the log-rank test to test for significance. We used χ2 to compare proportions. We considered differences to be significant at the 5% probability level.
We analysed the probability of having diarrhoea on an intention to treat basis, including all patients assigned to the trial. We did Kaplan-Meier survival analysis on intention to treat and per protocol bases, to see the effect of zinc with and without the dropout cases. We analysed outcome variables (duration of diarrhoea, total stool weight, and serum zinc concentration) only for patients who completed the study.