Data collection, management and analysis
Screening and randomisation
Both sites will be recruiting from community clinics. Surveillance systems are set up at clinics to identify all children presenting with diarrhoea. Trained study staff stationed at each health facility are responsible for identification of children who are potentially eligible. The study team is trained in good clinical practice as well as confidentiality and research ethics.
Written informed consent is sought from the caretaker of children meeting the eligibility criteria. A child can only be enrolled once into the study. The informed consent process includes telling the caretaker that they are free to withdraw from the study at any time, and that participation is entirely voluntary.
Following consent, each participant will be assigned a unique subject identification number. A card detailing the subject ID and the health personnel responsible for enrolment will be provided to the caregiver of enrolled children. The patient locator form/subject master log will be the only link between the subject ID and the participant’s name. This information will be stored together with the consent in a locked file cabinet and an online database accessible to only the Project Manager and required study staff for the purpose of patient tracking.
A baseline assessment, including a detailed physical examination (record of vital signs, axillary temperature, respiratory rate, pulse rate, hydration and nutritional status), will be performed by a research physician at the time of screening for enrolment. The child is evaluated for any Integrated Management of Childhood Illness general danger signs (unable/able to drink or breast feed, convulsions, vomiting, unconscious, stiff neck) and evidence of dehydration or pneumonia, as well as information on recent medical history (duration of diarrhoea, bloody stool/dysentery, recent fever). Vital signs (temperature/heart rate/respiratory rate) in dehydrated child are recorded after the child is rehydrated and stabilised and remaining clinical assessment/screening is completed thereafter. Anthropometry measurements will be taken after rehydration and stabilisation as required, before completing the screening for enrolling the participant. Weight will be measured using an electronic scale with a sensitivity of ±10 g by two independent observers: length (for children aged <24 months) or height (for children aged 24 months or older), which will be measured using a length board within 0.1 cm. Mid-upper arm circumference (MUAC) will be measured using non-stretchable tape to the nearest 0.1 cm. All anthropometric measures will be recorded in duplicate (India) and triplicate (Tanzania). Once these measurements and other eligibility criteria are met, the subject will get enrolled.
Baseline sociodemographic information and vaccination history
Parents of enrolled subjects will be interviewed at study clinic or at home for the baseline sociodemographic characteristics viz. parent’s literacy level and occupation, type of house, possession of assets and possible exposures (recent antibiotic use, travel history, water use and filtration and sanitation), breast feeding and vaccination history and information is recorded in the case record forms.
Follow-up visits
Scheduled follow-up visits begin from the third day of the enrolment and are generated daily via an automated visit generation module of the data management software. All children enrolled at the study clinic are to be followed up in the clinic or at home by a trained field worker on day 3, 5, 7, 10, 15, 30, 45 and 60 (online supplementary additonal files 1 and 2). In India, study staff will visit the household of enrolled children for all follow-up visitations. If the child is not available at home, a phone call will be made to follow-up the child. In Tanzania, a participant will be followed up at home or clinic on days 3, 7, 15, 30 and 60 and by phone on days 5, 10 and 45. Children still having diarrhoea after day 15 are treated according to standard management guidelines.8 At each visitation during active follow-up period, a study worker asks the mother or other caregiver about the following and records the responses in a predesigned form/data entry module:
Ongoing symptoms of diarrhoeal illness, including the number and consistency of stools.
Vomiting episodes including any episode within 30 min following zinc administration.
Adherence with zinc study regimen ascertained via mothers’ recall and/or reported use of zinc tablets by the child in the Diarrhoea Diary Card and cross-checked by determining the number of unused tablets.
Any adverse events, including hospital admissions and deaths of study participants.
Criteria for returning to clinic (eg, lethargy, poor oral intake, dehydration).
Any other intercurrent illness.
Reported use of antibiotics and intravenous/other fluids.
At days 30, 45 and 60 postenrolment, the study participants are assessed either at the study clinic, via telephone or with a home visit for presence of any illness, diarrhoea, fever, cough or fast or difficult breathing within last 24 hours and any time in past 2 weeks. On the day 60 study completion clinic visit, in addition to morbidity assessment, anthropometric measurements will be taken for each participant using the same procedure as mentioned in the baseline assessment.
Children not available on the visitation day will be visited the next day; if they are still not available, they are followed-up until found or deemed lost to follow-up. Caregivers are asked to contact the study physicians if they feel that their children are sick between the visits. Children sick on the day of visitation will be referred to the study clinic for a detailed morbidity assessment as well as treatment by the study physician. During each visit, a record of all the medicines and intravenous fluids given, information on number and consistency of stools and data on adherence is recorded in the predesigned forms. Close monitoring will be accomplished through repeated follow-ups during the 14-day regimen period. Every effort will be made to track and successfully follow-up all randomised children.
Study withdrawal or discontinuation
The participants have the right to withdraw from the study at any time. The subjects will be considered withdrawn from the trial in case of death or lost to follow-up. However, every effort will be made to track and successfully follow-up all enrolled children. The information on withdrawn or discontinued participants will be recorded through a withdrawal slip by field worker and clinic supervisor.
Blood sampling and laboratory investigations
To allow the measurement of the potential differential impact of the varying zinc doses on plasma zinc levels during the follow-up period, each study participant will provide two blood samples. One-third of the study participants will have blood draws at days 1 and 15, one-third at days 3 and 21 and the remaining one-third on days 7 and 30 postenrolment for assessment of plasma zinc levels.
Blood drawing will follow the protocols of the International Zinc Nutrition Consultative Group (IZINCG). These include use of powder-free gloves, zinc-free syringes and butterfly, needles and trace element-free blood tubes.14 Plasma is separated at 15 min after blood collection and aliquots are transferred into trace element-free Eppendorf plastic tubes for temporary storage at −20°C and then shipped to central lab for final storage at −80°C in India. In Tanzania, plasma samples are stored in site refrigerators at 2°C–8°C and then shipped to central laboratory storage at −80°C by the end of the day. Plasma zinc concentrations will be measured by atomic absorption spectroscopy (AAS-400 PerkinElmer, USA).
Quality assurance
The persons with at least a secondary school degree will be recruited for data collection team. The following quality assurance checks will be done regularly to ensure highest level of data integrity and quality:
An extensive 1–3 weeks training session for data collectors to ensure the interdata collector reliability is above 95%.
Pilot field data testing is monitored closely by the research scientist(s) responsible for quality assurance.
Real-time electronic data capture will ensure data validation such as range, logical checks and data integrity.
Interim aggregate data analysis will be carried out periodically for data pattern, heaping, missing values, etc. Based on the feedback, effective measures will be taken to eliminate/minimise the errors.
For overall visit supervision, quality control (QC) supervisors will perform quality checks by visiting at least 10% of the households, re-interviewing 5% respondents and cross-checking these data with concerned field worker. QC supervisors are primarily responsible for overall field-related project activities (data quality and integration at all levels, resolving community issues), and training. Computer randomised list of enrolled children will be used by QC team for random household visits.
The QC team works in close coordination with the investigative team and provides regular feedback on quality issues. Quarterly meetings between QC team and senior investigative team are scheduled to discuss overall quality issues and related corrective measures to be undertaken.
Data management
Data are collected in real-time using either tablets or laptop computers. Each site uses an in-house designed data management software with user friendly graphical user interface for data collection. Generic case report forms and data dictionary are used to maintain data consistency across sites. In-built range, logical and skipping pattern will ensure data quality and integrity.
The database is password protected with restricted access. Each user is assigned a unique user identifier to access the software and an audit trial is maintained for database operations. The centralised server located at each site synchronises the data on daily basis. Data are shared monthly with WHO data quality monitoring team who is responsible for data quality checks and provide feedback to sites. They generate query forms which are returned to the study sites for verification.
Data security and storage
A backup of the data is kept in the server at WHO protected by a specific password and accessible to only authorised users. An additional backup of the data is kept in a password-protected external hard drive at a site away from WHO at the recruiting site central offices.
Record retention and archival
All the study documents including participant’s source data and documents will be archived by the study sites after the completion of the study, till the time WHO informs in writing to the study sites that they no longer need to maintain the study documents.
Outcome measures
The primary efficacy outcomes are 1) the percentage of enrolled children who have duration of diarrhoea of >5 days, 2) the mean number of loose or watery stools after enrolment and 3) the percentage of children vomiting within 30 min of administration of the zinc tablet.
Secondary outcomes of the trial include: i) proportion of children experiencing SAEs (life-threatening or hospitalisation); ii) proportion with intermediate duration of diarrhoea (diarrhoea continuing beyond 3 days); iii) proportion of guardians with positive attitude towards treatment; iv) treatment adherence; v) mean serum plasma zinc concentration at days 1, 3, 7, 15, 21 and 30; vi) illness symptoms between day 15 and 60 after the treatment 2-week period prevalence and number of days with diarrhoea, fever or respiratory symptoms; vii) 60-day change in height/length, weight, MUAC, height/length-for-age z-score, weight-for-length/height z-score, weight-for-age z-score, MUAC z-score, stunting, wasting, underweight; viii) adherence to regimen; ix) study dropouts; x) maternal report of ease in supplement administration; xi) maternal report of child acceptability; xii) maternal recommendation of use for other children; xiii) maternal report of change in child skin condition, child appetite, activity levels, mood and diarrhoea severity.
Statistical analysis
All primary analyses of non-inferiority tests will be conducted on the principle of intention-to-treat (ITT). All non-inferiority margins will be defined a priori. We will also conduct sensitivity analyses using a per-protocol (PP) approach.15 The analysis population for the ITT approach will include all randomised patients with assessable data, regardless of whether the participant is known to have received a full 14-day dose of zinc or not. The analysis population for the PP approach will include study participants who are documented to have taken zinc supplementation for the first 5 days after randomisation regardless if they vomited the dose. The PP definition was selected due to the primary outcome being diarrhoea duration >5 days. Descriptive statistics will be used to summarise background demographic and baseline clinical characteristics by trial arm.
Analysis of primary outcomes
Primary outcome 1
Number (%) of subjects with diarrhoea episodes >5 days. The last day of diarrhoea will be defined as the day in which there is passage of the last liquid or semi-liquid stool prior to two diarrhoea-free days (days when less than three loose stools/day). The duration of diarrhoea will be defined as the number of days between randomisation (not the reported day of onset of diarrhoea symptoms before randomisation) and the last day with diarrhoea. Children who are lost to follow-up, die or withdraw before 5 days will be excluded from this analysis. For the primary analysis, we have set the threshold for non-inferiority of lower doses of zinc (10 or 5 mg/day) as compared with the standard 20 mg/day at <4% in absolute risk (risk difference) compared with the standard dose in the proportion of subjects with prolonged diarrhoea. The proportions of participants with a prolonged diarrhoea episode (defined as >5 days) will be calculated in all treatment arms, and the risk differences (and their 95% CIs) between the lower doses (5 and 10 mg/day) and the standard dose (20 mg) will be estimated. A Cochran-Mantel-Haenszel approach will be used to take stratifying randomisation factors, for example, country and age groups, into account. If the upper limit of the 95% CI for the risk difference is <4%, we will conclude non-inferiority for one or both lower dose arms. We will also examine the RR of prolonged diarrhoea episodes in both lower dose arms compared with the standard dose arm using RR regression which accounts for factors included in the stratified randomisation. In secondary analyses, Cox proportional hazard models will be used to estimate HRs and their 95% CIs for recovery from diarrhoea in each of the lower dose arms compared with the standard dose arm. The time to recovery is defined as the duration of diarrhoea as noted above.
Primary outcome 2
Total number of loose or watery stools after enrolment. An analysis of covariance (ANCOVA) approach will be used to estimate the mean differences and their 95% CI in the total number of loose or watery stools after enrolment for each lower dose arm compared with the standard dose arm, while accounting for the stratified randomisation design. The non-inferiority margin was set at two stools, therefore if the upper limit of the 95% CI is less than two stools, we will reject the null hypothesis that the lower dose zinc is inferior to the standard dose in terms of reducing the total number of loose or watery stools.
Primary outcome 3
Vomiting within 30 min after administration of the trial regimen. We hypothesise that the lower dose of zinc will be associated with a reduced risk of vomiting and therefore a superiority test will be used. Vomiting within 30 min of each daily zinc dose is repeatedly measured during the 14-day administration period. Generalised estimating equations (GEE) with the log link, binomial variance and exchangeable correlation matrix will be used to assess the RR of vomiting among the three treatment arms. The exchangeable working covariance matrix and robust estimators of the variances will be used to construct CIs. The robust estimators are consistent estimators of the variances even if the working correlation matrix is misspecified. Stratified randomisation factors will be considered in the model. A secondary analysis concerning severe vomiting (defined as vomiting requiring hospitalisation or vomiting requiring intravenous fluids) will also be performed in similar fashions. We will also examine the proportion of children ever vomiting between the treatment arms and examine whether the RR of vomiting differs by follow-up day.
Analysis of change in plasma zinc and other secondary outcomes
Due to the sampling design of participants having paired blood sampled on three schedules (days 1 and 15; days 3 and 21; days 7 and 30), we will assess the effect of the randomised zinc regimen at both the population level and individual level. For the analysis of population-level zinc, we will use a mixed model approach to compare patterns of plasma zinc concentrations over time between the randomised treatment arms. The main exposure variables in the model will include treatment arm, study day and an interaction term between treatment arm and study day. In case there is not a linear relationship between study day and plasma zinc concentrations, restricted cubic splines will be used to examine the non-linear relationship between these two variables. The interaction term between zinc dose and study day in the model measures whether the trajectory of plasma zinc differs by zinc supplementation over the full period. If the overall test for this interaction term is statistically significant, group means at each assessment day will then be compared using least-square means with the Tukey-Kramer adjustment. To be consistent with our stratified randomisation study design, stratification factors will be included in the model.
For all other secondary outcomes (see clinicaltrial.gov NCT03078842), we will use approaches similar to the primary analyses, including an ANCOVA approach for non-repeated continuous outcomes, RR regression for non-repeated binomial outcomes and GEE or mixed effect models for repeated longitudinal outcomes.
Sensitivity analyses and modification of treatment effect
Although randomisation is designed such that, on average, treatment groups will be balanced with respect to all baseline prognostic factors, this may not be true in all cases. As a result, we will re-assess treatment effects as observed in the original analysis after adjusting for factors that are potentially imbalanced across randomised arms. If imbalance occurs, we will compare and present the results for both unadjusted and adjusted analyses.
In addition, the study was not designed to detect effect modification of the treatment effect among subgroups of children. As a result, we acknowledge that we are likely underpowered to detect differences in the magnitude of a treatment effect among subgroups unless there is strong effect modification. The potential baseline effect modifiers we plan to explore include: study site (India/Tanzania), child age (<12 vs >12 months), child sex (male vs female), prior receipt of rotavirus vaccine (yes vs no), maternal HIV status (yes vs no), dysentery at enrolment (yes vs no), baseline dehydration status (yes vs no), baseline axillary temperature (<38°C vs >38°C), baseline respiratory rate, recent antibiotic use (yes vs no), intercurrent illness (cough, difficulty breathing or earache), current breastfeeding status (yes vs no) and prior breastfeeding status (yes vs no). Additional variables that we will test for effect modification will include child anthropometric measures (length/height for age, weight for age and weight for length/height z-scores and MUAC), family socioeconomic status (<median wealth index, >median wealth index), maternal education, household water quality (improved vs unimproved), household sanitation quality (improved vs unimproved), baseline plasma zinc levels (<median, >median) and trial regimen adherence (<median, >median). Factor analysis will be used to derive wealth index from family possession variables. WHO/Unicef Joint Monitoring Programme definitions will be used to clarify improved versus unimproved for the household water quality and sanitation.16
Monitoring
Data and Safety Monitoring Board
A Data Safety and Monitoring Board (DSMB) has been established to monitor SAEs and to review the statistical analysis plan and associated stopping rules for benefit, futility or harm determined using O’Brien-Fleming stopping boundaries. The DSMB includes five members with expertise in clinical trials, statistics, child mortality assessment, paediatric care in resource-limited settings or practical experience from each study country. The DSMB meets electronically or in person at least every 6 months. SAEs related to study participation are monitored in real-time, are summarised and reported to study investigators, WHO and relevant institutional review boards (IRBs) within 24 hours of notification in India and 72 hours of notification in Tanzania. Both sites are required to report to WHO within 72 hours of notification. Frequencies and descriptions of SAEs are pooled each month by the data management team at WHO and circulated to investigators, DSMB and IRBs. When more than half of the person-time is accrued in the study, the DSMB will review an interim data analysis by arm to determine whether stopping boundaries have been crossed.
Auditing
The sites are responsible for internal quality checks. WHO coordinators conduct at least one or two site monitoring visits per year to each site.