A questionnaire survey was conducted and observations were made concerning the presence of livestock in the home. The survey protocol also covered other known risk factors for contamination of household stored water, particularly water treatment, storage and handling behaviours, sanitation, and hygiene: these data form the basis of this collection. At the end of the questionnaire interview, a sample of household stored water was collected for subsequent microbiological testing, forming a related data set. Evidence is emerging that close proximity of livestock and people in domestic environments may result in microbiological contamination of food, objects, water, and people's hands, thereby compromising people's health. However, there remains limited evidence quantifying the relationship between livestock contacts in the home and microbiological contamination. This longitudinal study aimed to quantify human-livestock contacts in rural Kenya and examine their relationship with bacterial contamination of household stored water. Participating households in rural Siaya County were visited twice in both wet and dry seasons.Diarrhoeal disease and lack of access to safe water remain significant public health issues in developing countries. There is also growing concern about the potential for disease, including diarrhoeal infections, to be transmitted from livestock to humans. This project addresses the potential drinking-water contamination risks to human health in rural sub-Saharan Africa, where people and livestock often live in close proximity. Preliminary fieldwork will be carried out in rural Kenya, building on an ongoing study that is simultaneously recording human and livestock disease in ten villages. The fieldwork will test different techniques to identify contamination hazards from livestock, alongside water quality testing and recording of diarrhoea in children. These techniques will include the use of GPS collars to track cattle movements, maps of hazardous areas created by the communities themselves, and also checklists for recording signs of livestock hazards at water sources and around water stored in the home. We will look at how feasible it is to record hazards using these techniques. We will also statistically assess whether we find greater water contamination and greater diarrhoea in children where there are more recorded hazards. Since measurement of water contamination used in such areas is based on bacteria found in both livestock and humans, the project will also work on affordable ways of testing for micro-organisms that are specifically found in livestock faeces versus those found in human faeces. If successful, such techniques could be used to investigate the importance of different sources of faecal contamination of drinking-water. This in turn could help manage the safety of rural water sources like wells and rainwater and better protect drinking-water stored in the home from contamination through livestock. Because this complex problem requires a wide range of expertise, during the project we will strength our academic team to include more disciplines, particularly specialists in child health and social sciences. The tools for identifying hazards from livestock will be made widely available at the end of the project and UK expertise in the microbiological laboratory techniques will be shared with Kenyan collaborators. The experience gained will be used to build up contacts and develop a plan and team for a larger-scale study of livestock hazards, water contamination, and diarrhoeal disease risk in several countries.

Fieldwork took place in ten villages in Siaya County, Kenya, a rural site on the shores of Lake Victoria, which hosts a Health and Demographic Surveillance System (Odhiambo et al. 2012) and where residents participate in several ongoing studies of livestock and human health (Thumbi et al. 2015). These villages are among a sub-sample of 33 that also participate in an ongoing Population and Population-Based Infectious Disease Surveillance platform and are the focus of an ongoing Population-Based Animal Syndrome Surveillance (PBASS) study (Thumbi et al., 2015). These data were collected through a longitudinal, observational questionnaire survey of livestock-related risk factors for contamination of point-of-consumption water with faecal indicator bacteria. Eligible study participants were adult members of households participating in the ongoing PBASS study, whose households included children aged 6-59 months as the cohort at greatest risk of diarrhoeal disease. The sample size was powered to detect differential proportions of contaminated drinking-water using preliminary effect size estimates from Ghana (Wardrop et al. 2018), in the absence of evidence from Siaya. In Ghana, approximately 70% of water samples were contaminated in non-cattle keeping households, 90% were contaminated in cattle-keeping households (H1) and the proportion of contamination variance in cattle-keeping households explained by other covariates was estimated at 0.3. Within the study population, 55% of households own cattle (Thumbi et al. 2015). Based on these assumptions and a Type 1 error rate of 0.05, and a desired power of 0.9, a power calculation using the G* Power software indicated a required sample size of 196 households, which we rounded to 240 to allow for households declining to participate or dropping out of the study. Eligible households were randomly selected from lists of those participating in the PBASS study. After seeking informed consent, questionnaire interviews were conducted in the Dholuo language with adult respondents during an initial and follow-up visit. To assess domestic contacts with animals, interviewers observed the presence of livestock (e.g. goats, poultry), dogs and cats in the compound during interview and observed evidence of animal presence in the home (e.g. faeces; feathers; footprints). Interviewers also assessed whether the drinking-water container could be accessed by any animals and where chicken coops, asked whether poultry were permanently confined in such coops. The respondent was also asked about whether drinking-water was stored separately for children versus adults. To measure other known risk factors for stored water contamination, water storage characteristics (e.g. whether containers were covered) were observed, and respondents were asked about any water treatment or cleaning of storage vessels. Interviewers also asked about sanitation facilities and handwashing behaviours, observing whether soap was available. After completing the questionnaire, the survey team requested that the respondent fetch a sample of the stored drinking-water, including any water stored separately for children. The interviewer observed and recorded how the respondent collected the sampled water. Samples were also tested in situ for their residual free chlorine levels using a strip test.