Study design
This is an individually randomized facility-based controlled trial in which half of the newborns will be randomized to receive the intervention and the other half will receive the standard of care (dry cord care).
Setting
The trial will be conducted in three health facilities: Mukono Health Center IV, Kawaala Health Center III and Kitebi Health Center III. These three facilities have a monthly combined average of 2400 antenatal visits and 1200 deliveries. Mukono Health Center IV is located within Mukono district while Kitebi and Kawaala Health Centers are located in Kampala district, the capital city of Uganda. Kampala has an estimated resident population of two million while Mukono district, located 25 km from Kampala City has a largely rural population of almost 60,000 people.
Participants
The study will be carried out among newborns of HIV-1-negative mothers giving birth at the three clinics in Kampala and Mukono districts. Conducting this trial in a homogeneous group of HIV-unexposed (HU) neonates will substantially reduce the risk of imbalance of HIV exposure between study arms and thus mitigate any potential confounding from such an imbalance. Moreover, a finding of CHX protecting babies from severe illness and/or omphalitis among HU neonates can be cautiously generalized to similar populations elsewhere.
Inclusion criteria
Children born of HIV-1-negative mothers in the three study clinics.
Exclusion criteria
-
1.
Newborns with severe congenital anomalies
-
2.
Newborns with infection of the umbilical cord when born
-
3.
Severely ill infants requiring hospitalization immediately after birth
-
4.
Children of mothers who cannot appropriately give consent within 12 h of birth
-
5.
Babies born with a birth weight of less than 1500 g
Randomization
For each of the three health centres, a computer-generated random sequence list with permuted blocks of varying size (4, 6 or 8) was generated by CISMAC scientist Hans Steinsland, who is otherwise not involved in the trial. Based on this sequence, children are randomly allocated to intervention or comparison arms in a 1:1 ratio. Concealment is by a cell phone-based application that yields the trial arm allocation only after consent for inclusion has been confirmed.
Intervention
The intervention is umbilical cord stump cleansing with a single application of 4% CHX solution at birth. The watery solution was obtained from Galentic Pharma in India (http://www.galentic.com) and came in packages of 10-ml dropper bottles. The application is provided by a trained health worker (nurse or midwife) at the health facility as soon as possible and no later than 12 h after birth. The solution is applied using the dropper bottle onto the tip of the infant’s umbilical cord stump, its base and the area of skin surrounding it. A CHX-moistened cotton ball is then used to gently cleanse the base of the stump and the surrounding skin. Health workers have been trained by general physicians and pediatricians to correctly, consistently and safely apply CHX. Five CHX containers will be randomly chosen early, during and towards the end of the trial to check for CHX concentration and antibacterial activity. The chlorhexidine stock is securely kept below 30 °C, away from light in tightly closed/sealed bottles.
Comparator
Participants in the comparison arm of the trial will be given the standard of care (dry cord care).
Measurements
The main outcomes of this study are clinically diagnosed omphalitis and severe illness. Assessment for these outcomes is scheduled on days 1, 3, 7, 14 and 28 after birth. Further, the mothers are encouraged to visit our clinics if they observe any signs of umbilical cord stump infection or if they are worried that their child is otherwise ill. Severe illness is defined as illness that is associated with any of the following danger signs observed or verified by a study clinician: inability to feed or vomiting of all intake, lethargy or unconsciousness, severe lower chest in-drawing, axillary temperature of ≥37.5 °C or <35.5 °C, grunting, cyanosis, convulsions or a history of convulsions, and/or results in hospitalization and/or results in death. The data collection team records the presence or absence of omphalitis at each clinic visit. Signs for omphalitis include: pus, redness (inflammation) and swelling (edema) of the cord stump and the surrounding skin at its base. Swelling and redness is further broken down into four groups: none, mild, moderate and severe. No swelling or redness is defined as absence of visible swelling or redness; mild swelling or redness as that which is limited to the cord stump only; moderate swelling or redness as that extending less than 2 cm onto the abdominal skin at the base of the stump and severe swelling or redness as extending at least 2 cm or more around the abdominal skin at the base of the stump. Pus is characterized as being either present or absent. Cord infection is then defined based on combinations of these signs and their severity into four categories as follows: (1) redness extending to skin or pus, (2) moderate or severe redness, (3) moderate or severe redness with pus, or severe redness alone and (4) severe redness with pus [6, 7].
Given the necessity of the comparison arm to receive advice and education on dry cord care (the standard of care), this is an unblinded study. The study nurses are not made aware of the hypotheses that are being tested, and undergo rigorous training and standardization with respect to the assessment of omphalitis and severe illness to maintain a high validity in outcome assessment.
Background characteristics and potential confounders
Using questionnaires and available records, data will be collected on several potential individual-level confounders. These include: maternal age, maternal education, antenatal care attendance, iron supplementation during pregnancy, antibiotic use during the last weeks of pregnancy, premature rupture of membranes, signs of chorioamnionitis, wealth, household size, parity, tetanus vaccination, gestational age, singleton or multiple birth, sex of the infant, birth weight, use of a clean delivery kit, prelacteal feeds, breastfeeding initiation time, infant’s receipt of colostrum, and Bacille Calmette-Guérin (BCG) vaccination. To further describe the population from which our participants are derived and to adjust for any contextual level confounders, we collect data using questionnaires on group/contextual-level factors such as the presence of electricity in the community, and residence (rural versus urban).
Sample size
Sample size for primary objective 1: with type 1 error set at 5%, type 2 error at 0.2 (i.e., 80% power), and accounting for 5% attrition, a sample size of 4760 children will be required to detect a 30% relative reduction in the incidence of severe illness following a single cleansing of the umbilical cord stump with the 4% CHX solution. We assumed a 6% risk of severe illness in the control group.
Power calculation for objective 2: setting the type 1 error at 5% and accounting for 5% attrition, a total sample size of 4760 children will yield 96% power to detect a 30% relative reduction in the incidence of omphalitis following a single cleansing of the umbilical cord stump with the 4% CHX solution. The assumed risk of omphalitis in the dry cord care group (comparator) is 11.5% based on recently published data from the trial in Pemba [10].
Data collection and management
Data collection: primary objectives
Our study staff in the three study clinics identifies HIV-1-negative pregnant women in the delivery rooms. Potential participants are approached and informed about the study. Following a live birth, study procedures are explained to the women, and babies of those providing informed consent are enrolled and then randomized to either the intervention or comparison arm by research assistants, using the cell phone-based application described above. For enrolled infants, data will be collected using questionnaires on day 1, and on days 3, 7, 14 and 28 after birth (Fig. 1).
Using a standardized approach (distance, light, etc.), at each of the above-defined visits and at any additional contact, data will be collected on umbilical cord stump infection and digital pictures are captured of the umbilical cord stump with its surrounding skin. Participating women will be invited to bring their children to the study clinics on the scheduled dates for data collection. Home visits will be conducted for women unable to come to the health facilities for the scheduled interviews (Fig. 1).
To capture severe illness, mothers and other caregivers will be informed about its symptoms and signs, and will be encouraged to contact the study clinic in case the baby develops any of the symptoms. Each mother is provided with a durable note book in which she or an attending health care worker are requested to note down all relevant events for the child. A child who comes to a clinic with such symptoms or signs will be examined by a study nurse/midwife and will receive the appropriate treatment. If the child does not attend after the mother/caregiver has informed the study team of such symptoms/signs or for a scheduled visit, the study nurse/midwife will make a visit to the child’s house. If the nurse/midwife classifies the condition as possible severe illness, a study physician will be called to examine the child. If the physician classifies the child as having severe illness, they will collect a blood specimen for blood culture (using BACTEC) and for a septic screen (C-reactive protein (CRP), total lymphocyte count, differential count, band cell: neutrophil count ratio) as soon as possible. A repeat specimen will be collected again for CRP between 24 and 48 h. The study team will do its utmost to obtain specimens for such a septic screen from as many infants with signs of severe illness as possible. With the consent of the parents, all children with neonatal sepsis are admitted to the National Referral and Teaching Hospital (Mulago Hospital) where they are managed according to the national guidelines. In the event of a neonatal death, it is our experience that a mother may call spontaneously to inform the study team of why she is not coming to the next scheduled visit. However, in most cases, she will simply not turn up; in that case, the death will be captured when she is contacted because of a missed visit. A “verbal autopsy” for all neonatal deaths will be conducted as soon as it is socially acceptable to maximize recall without socially offending the participants. We will assess the cause of death using a standard World Health Organization (WHO) Verbal Autopsy Questionnaire that has been validated in Uganda. The questionnaire has both open-ended questions (for verbatim narratives) and closed-ended questions, and we will employ a standard algorithm to determine the likely cause of death [12]. Two independent reviewers will examine the verbal autopsies and assign a likely cause of death. In event of divergent conclusions, a third reviewer will be requested to provide a third review.
Project management
The study will be conducted as a collaboration between the College of Health Sciences, Makerere University and the Center for International Health, University of Bergen. The study will be guided by a Steering Committee which will include the principal investigator (PI) and the co-principal investigator (Co-PI) supported by the other co-investigators and a newly established Participatory Scientific Advisory Group (PSAG). The Steering Committee (VN and HS) is responsible for the conduct and coordination of the study. The Steering Committee will be the decision-making body for all scientific and administrative aspects. It will send reports to the funding agency, ethical committees and regulatory bodies. The SC will meet on regular conference calls with at least two face-to-face meetings annually. The study is registered with ClinicalTrials.gov (identifier: NCT02606565).
Quality control
All data is collected using standardized questionnaires and forms on an ODK platform using Android cell phones. The study nurses were trained extensively in all the routines and how to fill the questionnaires and Case Report Forms before study implementation. In particular, they received training on the assessment and grading of omphalitis using audio-visual aids and pictorials by faculty members at the College of Health Sciences, Makerere University. Newborns with moderate to severe omphalitis or severe illness are immediately referred to the nearest government health unit. At least one scientist/medical officer is employed by the project to supervise the field team and ensure good-quality data. To minimize losses to follow-up, the following measures are taken: (1) careful screening of study participants for eligibility before enrollment, (2) consenting participants are provided with a transport refund that covers their costs to and from the health facility, (3) contact telephone numbers are obtained from each consenting participant and reminders are made prior to each scheduled date and (4) each study site has a dedicated tracer with good knowledge of the study site geography. Importantly, detailed information on the physical location/address of the participants’ homes is collected at enrollment to help with tracing mothers who do not attend for study visits. Research assistants (who are licensed nurses or midwives) have been, and are being, rigorously trained to obtain informed consent from study participates. They also receive training on all study tools which have been pretested and standardized to minimize information biases. The capture of digital images of the umbilical cord stump and surrounding skin will enhance the diagnostic specificity of reported omphalitis cases.
Data analyses
Baseline characteristics of the mothers, their households and the newborns will be compared between the intervention and comparison arms to check for comparability and identify potential confounders. Continuous variables will be summarized using means, medians, standard deviations and interquartile ranges while percentages will be used for categorical variables.
Analysis for both primary objectives will be based on intention-to-treat. Therefore, all infants randomized and enrolled into the study will be included in the final analysis regardless of whether CHX was applied to their umbilical cord stump and surrounding skin or not (Fig. 2: flow chart and Additional file 1: Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) Checklist). The risks of omphalitis and severe illness in the intervention arm will be compared to those in the comparison arm using binomial regression with a log link to obtain risk ratios (RR). Multivariable analysis will be used to take into account potential confounding. Only confounders whose inclusion in the model results in a substantial (>5%) change in the estimate of the main outcome will be maintained in the final model. Percent protective efficacy will be calculated as (1 − RR) × 100.
Subgroup analysis: subgroup analyses will be conducted. Associations between a single cord cleansing with 4% CHX and omphalitis will be stratified on relevant variables such as gestational age (±37 weeks (completed) of gestation) and birth weight (±2.5 kg). If substantial stratum-specific differences in effects are observed, we will proceed with analyses to identify effect measure modification. This may provide insight into possible interaction between the intervention and individual level as well as group- level contextual variables. Such effect measure modification analysis may provide potential causal explanations for our findings.
Equity impact analysis: further secondary analysis will compare the effectiveness of the intervention across strata of education levels, wealth quintiles and residence in order to assess the impact of the intervention on health and economic equity. All the data required for this analysis will be collected prospectively and, if the sample size allows, we shall estimate the effects of focusing this intervention on the most vulnerable groups. The Concentration Index, a bivariate measure of the distribution of important outcomes, will be used to measure any improved equity in outcomes.