Effects of individual and combined water, sanitation, handwashing, and nutritional interventions on child respiratory infections in Rural Kenya: A cluster-randomized controlled trial

Poor nutrition and hand hygiene are risk factors for acute respiratory infections (ARIs). Safe drinking water and sanitation can reduce exposure to pathogens and encourage healthy immune responses, reducing the risk of ARIs. Within a trial assessing impacts of water, sanitation, and handwashing (WASH), and nutritional interventions, we evaluated effects on ARIs. The WASH Benefits cluster-randomized trial enrolled pregnant women from Kenyan villages and evaluated health outcomes in children born to enrolled mothers 1 and 2 years after intervention delivery. Geographically adjacent clusters were block-randomized into a passive control (no promotional visits), a double-sized active control (monthly visits to measure mid-upper arm circumference), and six intervention groups: Chlorinated drinking water (W), improved sanitation (S), handwashing with soap (H), combined WSH, improved nutrition (N) through counseling and lipidbased nutrient supplementation (LNS), and combined WSHN. The main outcome was the prevalence of ARI symptoms (cough, panting, wheezing, or difficulty breathing) in children younger than 3 years. Masking participants was not possible. Analyses were intention-to-treat. Between November 2012 and May 2014, 702 clusters were enrolled, including 6,960 (year 1) and 7,088 (year 2) children with ARI data. The cluster-level intra-cluster correlation coefficient for ARIs was 0.026 across both years. Water, sanitation, and handwashing interventions with behavior change messaging did not reduce ARIs. Nutrition counseling and LNS modestly reduced ARI symptoms compared with controls in year 1 [prevalence ratio (PR): 0.87, 95% confidence interval (CI): 0.77-0.99], but no effect in the combined WSHN group weakens this finding. The Kenya WASH Benefits study was a cluster-randomized trial conducted in rural villages in Bungoma, Kakamega, and Vihiga counties in western Kenya. Study design details for the Kenyan study and accompanying trial in Bangladesh were previously published.21,22 Villages were eligible for the study if they were rural, primarily relied on communal water sources and unimproved sanitation facilities, and were not enrolled in ongoing WASH or nutrition programs. Households were eligible if there was a pregnant woman in her second or third trimester who had planned to reside in her current residence for the next 2 years and could speak Kiswahili, Luhya, or English. Study clusters included up to three adjoining villages with at least six eligible pregnant women. Children born of these pregnancies (including twins) were defined as index children. Mothers provided written informed consent for their children and themselves. Village clusters comprising 12 enrolled households, on average, were each randomized by blocks (groups of nine geographically adjacent clusters) into eight groups—passive control; active control; chlorinated drinking water; improved sanitation; handwashing with soap; combined WASH interventions; improved nutrition through infant and young child feeding counseling and LNS; and combined WASH and nutritional interventions—using a random number generator with a reproducible seed at the University of California, Berkeley. Community health promoters visited households monthly to measure child mid–upper arm circumferences in the randomized groups, excluding the passive control group. The purpose of the passive control group was to differentiate effects of household interaction with promoters from those of the interventions. In the six intervention groups, promoters delivered intervention-specific behavior change messaging, helped troubleshoot problems with hardware, and replenished supplies of chlorine, soap, and nutrient supplements. Masking participants was not possible because of the nature of the interventions, and participants were informed of their treatment assignment after baseline data collection. The data collection team was not informed of the treatment assignments but might have inferred assignments if they observed intervention materials in study communities. Investigators remained blinded to the treatment assignments until statistical analyses were replicated. The WASH Benefits trial is registered at ClinicalTrials.gov, number {“type”:”clinical-trial”,”attrs”:{“text”:”NCT01704105″,”term_id”:”NCT01704105″}}NCT01704105. The trial protocol was approved by the Committee for the Protection of Human Subjects at the University of California, Berkeley (protocol number, 2011-09-3654), Institutional Review Board at Stanford University (IRB-23310), and Scientific and Ethics Review Unit at the Kenya Medical Research Institute (protocol number, SSC-2271). Innovations for Poverty Action was responsible for participant enrollment, intervention delivery, and data collection. Details of the trial implementation will be published elsewhere (https://osf.io/26r59/).21 Community health promoters were nominated by their local communities and trained to provide intervention-specific behavior change messaging, instructions on hardware use, and consumable supplies (chlorine, soap, and nutrient supplements). In intervention groups, promoters engaged caregivers of index children, and other compound members, through intervention-specific key messages, visual aids, interactive activities, and provision of hardware or products. Based on a literature review, a theory-based approach,23,24 formative research, and the WASH Benefits pilot randomized controlled trial,25 the developed behavior change messages focused on themes of nurture, aspiration, and self-efficacy. Interventions considered convenience and cultural norms to encourage adherence. In water quality intervention groups, promoters advocated drinking water treatment with sodium hypochlorite using chlorine dispensers installed at communal water source collection points or bottled chlorine provided to households in study compounds. Promoters used chlorine test strips during monthly visits to determine stored water chlorine concentrations; negative results invoked discussions with households to address barriers to chlorination. In sanitation groups, existing latrines were improved by installing plastic slabs with tight-fitting lids. Households without access to a latrine or access to a poor latrine were provided new latrines with plastic slabs and lids. Promoters advocated use of improved latrines for defecation and safe disposal of children’s and animals’ feces. All households in study compounds were provided plastic potties for each child younger than 3 years and sani-scoops with paddles for feces removal. In handwashing groups, promoters advocated handwashing with soap before preparing food and after defecating (including assisting a child). Households were provided with two handwashing stations, one each near the food preparation area and the latrine. Stations were constructed with two-foot pedal-operated jerricans that dispensed soapy and rinse water. Households provided rinse water, but promoters added pieces of bar soap to the soapy water container every 3 months. Promoters helped participants identify compound members to refill tippy taps and manage barriers to use, such as running out of soap. In nutrition groups, promoters delivered key messages for maternal, infant, and young child feeding around dietary diversity during pregnancy and lactation; early initiation of breastfeeding; exclusive breastfeeding from 0 to 6 months and continued breastfeeding through 24 months; introduction of appropriate and diverse complementary foods at 6 months; feeding frequency; and feeding during illness. Index children and siblings aged between 6 and 24 months were provided with two 10-g sachets per day of LNS (Nutriset, Malaunay, France), and caregivers were instructed to mix LNS into complementary foods twice a day. After consent and enrollment, a baseline questionnaire was administered to collect data on household demographics; socioeconomics; water, sanitation, and hygiene behaviors; and food insecurity using the Household Hunger Scale.26 One and two years after intervention delivery, data were collected on adherence to interventions and child health outcomes (including ARI symptoms). Respiratory infection symptoms were collected for all children younger than 3 years in study households. The mean ages of index children and siblings younger than 3 years were 14.2 months (SD: 6.77 months) and 22.9 months (SD: 5.70 months) at years 1 and 2, respectively. The primary outcome in this study is ARI symptoms—defined as having caregiver-reported cough or difficulty breathing, including panting or wheezing, within 7 days before the interview—in children younger than 3 years. Prespecified secondary outcomes in this study include difficulty breathing, including panting or wheezing, in the past 7 days (a more specific indicator of respiratory infection than a cough alone); ARI symptoms presenting with fever in the past 7 days (a potentially more severe infection); and enumerator-observed runny nose (an objective outcome). Enumerator-observed runny nose was a rare outcome, so caregiver-reported runny nose was examined in a post hoc analysis. The sample size calculation for the main trial is described elsewhere.21 Sample size was chosen to detect a difference of 0.15 in LAZ and relative risk of diarrhea of ≤ 0.7 for comparing any intervention with the double-sized active control group; both calculations assume a type I error (α) of 0.05, power (1−β) of 0.8, and 10% loss to follow-up after the baseline. Two masked researchers independently replicated the analyses following the prespecified analysis plan (https://osf.io/jre7x). We considered the intention-to-treat, unadjusted differences between each intervention group and a control group as our primary inference. The passive and active control groups were combined (hereby referred to as the control group); there were no differences in the prevalence of ARI symptoms between the two groups (data not presented). We compared outcomes in the combined water, sanitation, and handwashing group (WSH) to the individual intervention groups and the combined water, sanitation, handwashing, and nutrition group (WSHN) to the 1) nutrition group and 2) WSH group. We combined data from years 1 and 2 for the primary analyses. We conducted secondary analyses with year 1 data only because adherence to water treatment and handwashing interventions was higher.22 We used generalized linear models with robust standard errors and fixed effects for geographically pair-matched clusters. No adjustments for multiplicity were applied. We assessed effect modification of interventions on ARI symptoms for the following prespecified characteristics: index child status, child gender, and malaria seasonality. The peaks in ARI and pneumonia cases have been observed during malaria seasons, which occur shortly after wet seasons, in rural Kenya.27 We defined two malaria seasons: January–February and June–August. The sample size of non-index children (older siblings and children born during the study) was small, so results for the index child status subgroup analysis will not be presented; 10% (1,347/8,508) of children with ARI symptom data were non-index children. Analyses were done with Stata (version 14.2, StataCorp LLC, College Station, TX) and R (version 3.3.2, The R Foundation, Vienna, Austria). After we unmasked the prespecified analyses’ results, we performed a post hoc analysis to interpret the main findings and examine whether intervention uptake differed between the single nutrition and combined WSHN arms. Previously, we reported that the prevalence of achieving minimum dietary diversity, defined as consuming at least four of seven key food groups (grains or tubers, legumes or nuts, dairy products, animal flesh foods, eggs, vitamin A–rich fruits and vegetables, and other fruits and vegetables) in the past 24 hours, was higher in the nutrition group than in the control group in the WASH Benefits trial but not when nutrition interventions were combined with improved water quality, sanitation, and handwashing.28 To test whether this difference persisted in the population of index children with ARI symptom data, we used generalized linear models (with robust standard errors and fixed effects for clusters) to compare the prevalence of achieving minimum dietary diversity in the nutrition versus WSHN group. We examined differences between the nutrition and WSHN groups in breastfeeding rates for index children for three indicators: any breastfeeding in the past 24 hours, early initiation of breastfeeding (i.e., caregiver reported putting the child to the breast immediately or within the first hour after birth), and exclusive breastfeeding for 6 months (i.e., caregiver reported exclusive breastfeeding for 6 months, corrected by the reported age of cessation and complementary foods eaten).