Menu
Menu
Menu
Menu
Background: Young children who contract Ebola Virus Disease (EVD) have a high case fatality rate, but their sources of infection and the role of breastfeeding are unclear. Methods/Principal Findings: Household members of EVD survivors from the Kerry Town Ebola Treatment Centre in Sierra Leone were interviewed four to 10 months after discharge to establish exposure levels for all members of the household, whether or not they became ill, and including those who died. We analysed a cohort of children under three years to examine associations between maternal illness, survival and breastfeeding, and the child’s outcome. Of 77 children aged zero to two years in the households we surveyed, 43% contracted EVD. 64 children and mothers could be linked: 25/40 (63%) of those whose mother had EVD developed EVD, compared to 2/24 (8%) whose mother did not have EVD, relative risk adjusted for age, sex and other exposures (aRR) 7·6, 95%CI 2·0–29·1. Among those with mothers with EVD, the risk of EVD in the child was higher if the mother died (aRR 1·5, 0·99–2·4), but there was no increased risk associated with breast-feeding (aRR 0·75, 0·46–1·2). Excluding those breastfed by infected mothers, half (11/22) of the children with direct contact with EVD cases with wet symptoms (diarrhoea, vomiting or haemorrhage) remained well. Conclusion/Significance: This is the largest study of mother-child pairs with EVD to date, and the first attempt at assessing excess risk from breastfeeding. For young children the key exposure associated with contracting EVD was mother’s illness with EVD, with a higher risk if the mother died. Breast feeding did not confer any additional risk in this study but high risk from proximity to a sick mother supports WHO recommendations for separation. This study also found that many children did not become ill despite high exposures.
In July-September 2015, interviews were sought with the household members of all individuals who were discharged from the Ebola Treatment Centre in Kerry Town, Sierra Leone (“Ebola survivors”) from November 2014 to March 2015. Contact was made through members of the survivor support team who were involved in their reintegration into the community. An initial approach was made to explain the study. If the household head agreed, an interview was arranged at a community centre or other meeting place and all who were in the household at the time that members of the households had Ebola were encouraged to attend. At the interview, individual informed written consent to participate in the study was sought from all adults, and from parents or guardians for children (< 18 years), with assent from children of 12 years or older. An inventory was drawn up of all household members who had been present in the household at the time that one or more household members were ill with EVD, including any who had died or were not present at the interview. For each member we asked whether they had had Ebola. We asked relatives whether any deceased had died of Ebola. Household members were asked to describe what happened when Ebola came to their household, including who became ill first, whether those with Ebola had any diarrhoea, vomiting or bleeding while they were at home, and who looked after them. They were encouraged to tell the narrative in their own words, with probing questions to clarify who had been exposed and how. For each household member (including those who had died, but excluding any absent members or those who refused consent) we sought to establish the highest-risk exposure. Reported exposures were ranked a priori from highest to lowest as: contact with the body of someone who died of Ebola; direct contact with body fluids of someone with Ebola, including breastfeeding, or other direct contact with “wet” cases (i.e. those with diarrhoea, vomiting or bleeding); direct contact with “dry” cases (i.e. those without diarrhoea, vomiting or bleeding); indirect contact with a wet case (e.g. washing their clothes); indirect contact with a dry case; minimal contact (e.g. shared utensils); and no known contact. For each mother-baby pair who both had EVD we attempted to ascertain from the narratives who was affected first. All survivors from the Kerry Town Ebola Treatment Centre had EVD confirmed by PCR. We did not have laboratory data for those from other treatment centres or for those who died, so have relied on the families’ reports. For individuals who were not reported as having had Ebola we asked about symptoms at the time that Ebola was in the household. For the analysis they were classified as not having had Ebola if they were asymptomatic or had symptoms that did not fulfil the Sierra Leone Ministry of Health and Sanitation case definition for “probable” Ebola,[10] or had had a negative test; and as having had Ebola if they were symptomatic and fulfilled the case definition for probable Ebola and were not tested. The case definition was contact with a case plus fever or miscarriage or unexplained bleeding; or contact plus three or more symptoms (of fatigue, headache, loss of appetite, nausea or vomiting, abdominal pain, diarrhoea, muscle or joint pain, sore throat or pain on swallowing, hiccups). In this analysis we concentrate on risks to children aged less than three years at the time Ebola reached their household in order to include all those who were breast fed, and examine attack rates, case fatality rates and the role of breast feeding. Proportions were compared using Χ2 or Fisher’s exact test. Analyses used multivariable logistic regression. Because the outcome is very common we have presented the results as risk ratios (RR) using marginal standardization to estimate RRs, and the delta method to estimate 95% confidence intervals (95%CI).[11–13] We repeated the analysis calculating risk ratios using Poisson regression with robust error variance.[14] Crowding (number of people per room) and sanitation (access to water, soap and latrine) were considered as possible confounders, in addition to age, sex and the exposure variables. The effects of clustering by household were explored using generalised estimation equations in logistic regression: the results were very similar to analyses ignoring clustering so clustering is not included in the models. Analyses used STATA 14. The study was approved by the Sierra Leone Ethics and Scientific Review Committee and the Ethics Committee of the London School of Hygiene & Tropical Medicine. At the interview, individual written informed consent to participate in the study was sought from all adults, and from parents or guardians for children (< 18 years), with assent from children of 12 years or older.
