Fecal Contamination in Child Play Spaces and on Child Hands Are Associated with Subsequent Adverse Child Developmental Outcomes in Rural Democratic Republic of the Congo: REDUCE Prospective Cohort Study

ABSTRACT.

The objective of the Reducing Enteropathy, Undernutrition, and Contamination in the Environment (REDUCE) program is to identify exposure pathways to fecal pathogens that are significant contributors to morbidity among young children in the Democratic Republic of the Congo (DRC), and on developing and evaluating scalable interventions to reduce fecal contamination from these pathways. This prospective cohort study of 270 children under 5 years of age was conducted in rural South Kivu, DRC, to investigate the association between Escherichia coli in hand rinse, soil, food, object, surface, stored water, and water source samples and child developmental outcomes. Child developmental outcomes were assessed by communication, fine motor, gross motor, personal social, problem-solving, and combined scores measured by the Extended Ages and Stages Questionnaire (EASQ) at a 6-month follow-up. Children having E. coli present in the soil in their play spaces had significantly lower combined EASQ z scores (coefficient: −0.38 (95% CI: −0.73, −0.03)). E. coli on children’s hands was associated with lower communication EASQ z scores (−0.37 (95% CI: −0.0.10, −0.01), and E. coli in stored drinking water was associated with lower gross motor EASQ z scores (−0.40 (95% CI: −0.68, −0.12). In the REDUCE cohort study, E. coli in soil in child play spaces, on children’s hands, and in stored drinking water was associated with lower developmental outcome scores (communication, gross motor, fine motor, and problem-solving skills). These results suggest the need for interventions to reduce fecal contamination in the household environment to protect the cognitive development of susceptible pediatric populations in rural DRC.

INTRODUCTION

In 2017, an estimated 250 million young children in low- and middle-income countries were at the risk of not achieving their developmental potential. ¹ Diarrheal disease is a leading cause of death for young children, resulting in 500,000 deaths globally each year. ² Enteric infections, both symptomatic and asymptomatic, during early life are associated with malnutrition and impaired growth in children under 5 years of age. ^{3 – 7} This resulting malnutrition and impaired growth can lead to deficits in childhood cognitive development. ^{8 – 10} The first 2 years of life represent a critical window for child development when the brain is developing most quickly with increased neuron proliferation, axon and dendrite growth, synapse formation, and myelination. ⁸ Being raised in poverty adversely affects child development through multiple mechanisms including food insecurity, chronic infections, and poor access to healthcare resulting in an environment that prevents young children from thriving. ¹¹ Interventions are urgently needed to promote a healthy environment for child development for susceptible pediatric populations.

Delivery of water, sanitation, and hygiene (WASH) interventions has the potential to improve child developmental outcomes through reducing enteric infections that can lead to intestinal inflammation causing malnutrition through poor nutrient absorption. ^{6, 7, 9} However, the literature investigating the impact of household WASH infrastructure (e.g., improved sanitation) and WASH intervention programs on child developmental outcomes is limited. ^{12 – 16} A cohort study conducted in Ethiopia, India, Peru, and Vietnam found that children with access to improved water sources and toilets had higher vocabulary test performance. ¹⁴ A randomized controlled trial in Pakistan found that delivery of a handwashing with soap program increased child cognitive outcomes. ¹⁵ Recently, there were three large randomized controlled trials of community-based WASH interventions conducted in Bangladesh, Zimbabwe, and Kenya. ^{12, 13, 17} The WASH interventions for these trials included delivery of modules on water treatment, handwashing with soap, safe child feces disposal, and provision of an improved latrine. The study in Bangladesh found significant improvements in cognitive developmental outcomes with delivery of these WASH interventions. ¹³ However, in Kenya and Zimbabwe, these interventions alone did not substantially improve child developmental outcomes. ^{12, 16}

The objective of the Reducing Enteropathy, Undernutrition, and Contamination in the Environment (REDUCE) program is to identify exposure pathways to fecal pathogens that are significant contributors to morbidity for young children in the Democratic Republic of the Congo (DRC), and on developing and evaluating scalable interventions to reduce fecal contamination from these pathways. To achieve this objective, we collected data on fecal contamination in the household. In this prospective cohort study, we investigated the association between E. coli on child and caregiver hands, soil, surfaces and objects, and in food and subsequent child developmental outcomes. Our a priori hypothesis was that E. coli in water, soil, food, and on child and caregiver hands would be associated with adverse child developmental outcomes. Our causal model was that higher fecal contamination on the household compound would make young children more susceptible to enteric infections leading to poor nutrient absorption, and thereby inadequate nutrient intake for proper brain development resulting in adverse child developmental outcomes. We selected fecal contamination as our risk factor based on our a priori hypothesis, and to identify sources of fecal contamination on the household compound to focus on for subsequent REDUCE program WASH interventions.

MATERIALS AND METHODS

Ethical approval.

Informed consent was obtained from a parent or guardian of all study participants. Study procedures were approved by the research ethical review committee of the University of Kinshasa (Protocol 043-2017), and the Johns Hopkins School of Public Health (Protocol 8057).

Study design.

This prospective cohort study of 270 children under 5 years of age was conducted in rural Walungu Territory of South Kivu in the DRC. The study was part of a larger USAID/Bureau for Humanitarian Assistance funded DFSA award with the goal of improving food and nutrition security and economic well-being of vulnerable households in South Kivu and Tanganyika provinces of DRC. The REDUCE cohort study is a separate component from the larger DFSA program implementation activities. No households during their time enrolled in the cohort study received child health-related interventions from the DFSA program. To be eligible, households had to have a child under 5 years currently living in the household. For our sampling approach, a list of children under 5 years was obtained from health centers in 47 villages. These children were screened for eligibility and enrolled; our target was to enroll five households in each village. The sample size was based on the number of study participants that could be enrolled between July 2018 and January 2019 with child developmental outcome data and laboratory data available from food, surface/object, water, soil, and hand rinse samples. A 6-month follow-up was conducted in households between December 2018 and August 2019. All children with at least 5 months of surveillance data from baseline to follow-up that were younger than 5 years of age at follow-up were included in the analysis. Unannounced spot checks of the household compound for water, soil, surface/object, food, and hand rinse sample collection were performed after the baseline visit for study enrollment. Unannounced spot checks occurred between 8 am and 5 pm. There was no set time for spot checks because we did not want households to prepare for our arrival (e.g., washing hands with soap). The REDUCE cohort study included 47 villages with on average six households per village. Households tend to be very spread out in our study setting, making it unlikely that households had prior warning about our presence in the village to prepare for spot check visits. Participants put both hands in a Whirl-Pak bag containing a phosphate buffered saline solution for the hand rinse samples. Water samples were collected from the household water source and stored drinking water in the home. Households were asked to provide a sample of the drinking water given to young children. Soil samples were collected from areas where young children were reported by caregivers to most frequently play (defined as child play spaces). Samples were collected of the food given (or to be given) to children under 5 years in the household. Surface/object samples were collected from objects or surfaces that caregivers reported children most frequently put in their mouths during the day of the unannounced spot check. The participant ID and sample ID was recorded for each child’s food and surface/object sample. Hand rinse samples were collected from caregivers in the household and from children under 5 years of age. Because of the logistical challenges of working in eastern DRC, we were limited in our laboratory capacity and decided to collect a subset of environmental samples across all study households rather than intensive sampling in a small number of households. Households were randomly selected to receive a particular type of sampling (e.g., soil or hand rinse sampling) based on the day of the week their surveillance was conducted.

Fecal exposure pathway analysis.

All object/surface, food, hand rinse, soil, and water samples were enumerated using the most probable number (MPN) method with the IDEXX Quanti-Tray system with Colilert-18 reagents (IDEXX, Westbrook, ME) according to previously published methods. ¹⁸ All samples were stored on ice and transported to the laboratory at the Catholic University of Bukavu in Bukavu, DRC, where they were processed within 6 hours of collection. This method is approved by the US Environmental Protection Agency (EPA) as a standard method for the examination of water and wastewater (APHA, 2012). ¹⁹ Samples were incubated at 37°C for at least 18 hours for the detection of Escherichia coli. Two thousand five hundred MPN E. coli was the detection limit for water, food, surface, and object samples. The lower detection limit for all samples was zero. The current manuscript reports laboratory findings for children with child developmental outcome data available. The full laboratory results are described elsewhere. ²⁰ Samples were not analyzed in duplicate. Each day samples were analyzed a negative control was run. The negative control was made by adding 100 mL of phosphate buffered saline (PBS) to a 120-mL Whirl-Pak bag, then the procedure used to analyze water samples was followed.

Water sampling.

For each water sample, 200 mL of water was collected and stored in a Whirl-Pak bag containing sodium thiosulfate for processing. The sample (undiluted) was analyzed using the IDEXX method as previously published. ¹⁸

Food sampling.

Each food sample was collected using a sterile spoon, then placed in a Whirl-Pak bag. Fifteen grams of each food sample and 150 mL of PBS was placed in a Whirl-Pak Sterile Filter bag, then put in a BagMixer 400 S Homogenizer (Interscience, France) for 1 minute and left to settle for 5 minutes. About 10 mL of the soft food supernatant was pipetted into a Whirl-Pak bag with 90 mL of PBS. The moisture content of each sample was measured by putting 5 g of food on an aluminum weigh boat and drying it in a high-temperature drying oven at 100–110°C for 72 hours. The moisture content was calculated using the following formula: (wet weight − dry weight)/dry weight.

Soil sampling.

For soil samples, 25 g of soil was collected (< 1 cm deep) using a sterile spoon within an area of 100 cm² using a plastic stencil and put into a Whirl-Pak bag. Twenty grams of soil was mixed with 100 mL of PBS in a Whirl-Pak bag and homogenized by hand for 1 minute, then left to settle for 5 minutes. Using the supernatant, 25 mL of the homogenized sample was pipetted into a Whirl-Pak bag containing 76 mL of PBS then mixed for a ¼ dilution. The sample was diluted again by pipetting 1 mL of the mixture into another Whirl-Pak bag containing 99 mL of PBS for a 1/400 dilution. The upper detection limit was 10⁶ MPN. The moisture content of the soil was measured as described earlier for food sampling.

Hand sampling.

To collect hand rinse samples, the caregiver or child placed one hand, and then the other, into the same 500-mL Whirl-Pak bag with 350 mL of PBS inside. Hands were dipped into the PBS solution for 1 minute (30 seconds of shaking followed by 30 seconds of research staff massaging the participant’s hand through the Whirl-Pak bag). One hundred milliliters of the sample was processed with bacterial counts being multiplied by 3.5 (to account for the initial water volume).

Surface/object sampling.

For surface sampling, a sterile swab was moistened using PBS to swab a surface area of 100 cm² in size using a plastic stencil and put into a Whirl-Pak bag. The entire surface was swabbed for object samples. The swab was put in a Whirl-Pak bag containing 100 mL of PBS, then massaged through the bag to elute bacteria into the PBS solution matrix for 30 seconds (no dilution).

Child developmental outcomes.

Communication, gross motor, fine motor, problem-solving, and personal social skills were assessed for children under 5 years using the Extended Ages and Stages Questionnaire (EASQ) at the 6-month follow-up. This questionnaire is a combination of direct tests and parental reports, which was adapted for use in low- and middle-income countries to assess cognitive and socioemotional development. ^{13, 21} Direct tests included drawing, naming items in a picture book, stacking blocks, and kicking a ball. Previous use of this tool in Bangladesh and in Kenya has shown results consistent with other measures of child development such as the WHO module, which also has direct assessments and parental reports of whether children can perform certain actions such as standing with support. ²² The domains for the tests were in 2-month age bands up until 24 months, then in 3-month age bands from 25 to 36 months, and 6-month age bands thereafter. The standardized z scores were constructed for each domain and all domains combined based on the mean and SD of each individual item in each age band using all observations for children within the REDUCE cohort with EASQ data available, including those that did not have laboratory data. This was done to include as large of a population as possible for standardizing z scores. Respondents could respond “yes” coded as 10, “sometimes” coded as 5, or “no” coded as 0 for EASQ items. Mean EASQ scores for individual domains and combined could range from 0 to 10. The mean EASQ scores by age band is reported in Supplemental Table 1. The EASQ questionnaire was reviewed by the study psychologist, translated and backtranslated and piloted according to previously published methods. ²³ The EASQ was piloted in a subset of 50 households prior to being administered for the cohort study, and households were read questions and asked to explain the meaning of the statements as a form of cognitive interviewing. Eighteen testers (university graduates) received a 10-day extensive hands-on training on how to administer the EASQ. We arranged refresher trainings every 3 months.

Statistical analysis.

This a priori analysis was conducted to achieve the aim of the REDUCE program to identify exposure pathways to fecal pathogens that are significant contributors to morbidity for young children in the DRC. The primary objective was to determine whether E. coli on child and caregiver hands, soil, surfaces, objects, and in food and water was associated with subsequent child developmental outcomes. The mean E. coli concentration was used if more than one of each type of sample was collected. To assess the association between hand, food, and environmental contamination and child developmental outcomes, linear regression models were fit using generalized estimating equations to account for clustering at the household level and to approximate 95% CI. Child developmental outcomes (combined, communication, fine motor, gross motor, problem-solving, and personal social z scores) were modeled separately as continuous outcomes and hand, food, and environmental contamination (binary and continuous) were the predictors. Models were adjusted for caregiver formal education (household education), number of individuals in the household (household size), and household wall type (housing type). These were the measures of socioeconomic status we had available, and were selected based on their association with EASQ scores in previous studies. ²⁴ Analyses were performed in SAS software (Version 9.4, Cary, NC).

RESULTS

A total of 303 children under 5 years of age residing in 47 villages were enrolled at baseline and had laboratory data available and were thereby included in this study. Sixty-seven children enrolled at baseline had no laboratory data. Eighty-nine percent of children with laboratory data at baseline (270/303) had EASQ data available at the 6-month follow-up. The median baseline age for children was 2.5 years (median) ±1.1 (SD) (range: 0.6–4.5) (Table 1). Fifty percent of children (135/270) were female. Forty-nine percent of children (129/265) were stunted (height for age z score < −2), 4% (11/269) were wasted (weight for height/length z score < −2), and 15% (40/269) were underweight (weight for age z score < −2). The median household size was six individuals ±2.5 (2–17). Seventy-six percent of children (200/262) resided in households with at least one household member with any level of formal education. Sixty-one percent of children (159/262) resided in households with mud walls, 8% (20/262) wood walls, 6% (16/262) concrete walls, 6% (15/262) wood and mud walls, 5% (12/262) biomass walls (dung and mud), 5% (14/262) brick walls, and 3% (8/262) wood and concrete walls.

Table 1.

Baseline demographic characteristics among children under 5 years in the REDUCE Cohort Study

	%	n	N
Children < 5 years of age			270
Baseline age (years)
Mean ± SD (Min–Max)	2.5 ± 1.1 (0.6–4.5)		270
Gender
Female	50	135	270
Household wall type
Mud walls	61	159	262
Wood walls	8	20	262
Concrete walls	6	16	262
Wood and mud walls	6	15	262
Biomass walls	5	12	262
Brick walls	5	14	262
Wood and concrete walls	3	8	262
Other	7	18	262
Household member with any formal education	76	200	262
Household size
Mean ± SD (Min–Max)	6.4 ± 2.5 (2–17)		262
Baseline growth measurements
Stunted (HAZ<−2)	49	129	265
Wasted (WHLZ<−2)	4	11	269
Underweight (WAZ<−2)	15	40	269

n represents the number of children with the individual or household level characteristic; HAZ = height for age z score; WHLZ = weight for height/length z scores; WAZ = weight for age z score.

A total of 1,076 water, soil, food, object, surface, and hand rinse samples were collected in households from children with EASQ data available (Table 2). Forty-one percent (65/158) of water source samples had E. coli and 68% (110/161) of stored water samples. The mean water E. coli level was 17 MPN/100 mL ± 139 (0–1,733) for source water samples and 234 MPN/100 mL ± 667 (0–2,500 MPN) for stored water samples. Seventy-three percent of soil samples (102/140) in child play spaces had E. coli. The mean soil E. coli level was 4,564 MPN/g dry weight ± 11,724 (0–50,158). Fifty-three percent of food samples (69/129) had E. coli. The mean food E. coli level was 326 MPN/g dry weight ± 777 (0–2,500 MPN). Thirty-six percent of object and surface samples (70/197) had E. coli. The mean object and surface E. coli level was 95 MPN/object or 100 cm² surface area ± 434 (0–2,500 MPN). Seventy-eight percent of child hand rinse samples (153/195) for children under 5 years had E. coli and 64% (61/96) of hand rinse samples from primary caregivers. The mean E. coli level for child hand rinse samples was 598 MPN for both hands ±2,284 (0–22,540 MPN) and 398 MPN for both hands ±1,259 (0–6,952) for hand rinse samples from primary caregivers.

Table 2.

Escherichia coli concentrations in hand rinse, food, and environmental samples (N = 1076)

Sample type	n	N	E. coli positive (%)	Mean ± SD (Min–Max)*
Water source E. coli > 1 MPN/100 mL	65	158	41	17 ± 139 (0–1,733)
Water source E. coli MPN/100 mL (Log10)
Stored household water E. coli > 1 MPN/100 mL	110	161	68	234 ± 667 (0–2,500)
Stored household water E. coli MPN/100 mL (Log10)
Soil in child play spaces E. coli > 1 MPN/gram dry weight	102	140	73	4564 ± 11,724 (0–50,158)
Soil in child play spaces E. coli MPN/gram dry weight (Log10)
Child food E. coli > 1 MPN/gram dry weight	69	129	53	326 ± 777 (0–2,500)
Child food E. coli MPN/gram dry weight (Log10)
Object/surface child mouthed E. coli > 1 MPN per object/100 cm2 surface area	70	197	36	95 ± 434 (0 − 2,500)
Object/surface child mouthed E. coli MPN per object/100 cm2 surface area (Log10)
Child hand rinse E. coli > 1 MPN/100 mL (both hands)	153	195	78	598 ± 2,284 (0–22,540)
Child hand rinse E. coli MPN (both hands) (Log10)
Caregiver hand rinse E. coli > 1 MPN/100 mL (both hands)	61	96	64	398 ± 1,259 (0–6,952)
Caregiver hand rinse E. coli (both hands) (Log10)

MPN = most probable number.

^*2,500 MPN E. coli was the detection limit for water, food, surface, and object samples.

All regression models were adjusted for household education, household size, and housing type. The overall mean EASQ z score for all children in the cohort was 0.1 (SD: 0.9 [−2.3 to 2.2]). Children with E. coli present in the soil in their play spaces had significantly lower combined EASQ z scores (binary: presence or absence of E. coli) (coefficient: −0.38 [95% CI: −0.73, −0.03]), lower fine motor z scores (binary: −0.64 [95% CI: −1.04, −0.24]), and lower problem-solving z scores (binary: −0.49 [95% CI: −0.83, −0.15]) (Table 3). A difference in z score of 0.38 is equivalent to an increase from 50% to 65% of milestones met, relative to peers. E. coli on children’s hands was associated with lower communication z scores (binary: −0.37 [95% CI: −0.72, −0.03]), and E. coli in stored drinking water associated with lower gross motor z scores (binary: −0.40 [95% CI: −0.68, −0.12]). The continuous outcomes are also reported in Table 3.

Table 3.

Association between hand, food, and environmental contamination and extended ages and stages z scores in a cohort study in the Democratic Republic of the Congo

Risk factor	N	Combined z score coefficient (95% CI)			Communication z score coefficient (95% CI)			Gross motor z score coefficient (95% CI)			Fine motor z score coefficient (95% CI)			Problem solving z score coefficient (95% CI)			Personal social z score coefficient (95% CI)
Water source E. coli > 1 MPN/l00 mL	218	−0.14	−0.47	0.20	−0.09	−0.41	0.22	−0.12	−0.41	0.18	−0.11	−0.41	0.19	−0.11	−0.45	0.22	−0.14	−0.46	0.17
Water source E. coli MPN/100 mL (Log10)		−0.05	−0.12	0.01	−0.03	−0.10	0.03	−0.02	−0.08	0.03	−0.06	−0.11	0.00	−0.05	−0.12	0.02	−0.05	−0.12	0.01
Stored household water E. coli > 1 MPN/100 mL	217	−0.25	−0.58	0.08	−0.20	−0.52	0.13	−0.40	−0.68	−0.12	−0.17	−0.48	0.15	0.19	−0.51	0.13	−0.07	−0.41	0.26
Stored household water E. coli MPN/100 mL (Log10)		−0.03	−0.12	0.06	−0.03	−0.11	0.04	0.02	−0.06	0.10	−0.03	−0.11	0.05	−0.02	−0.11	0.06	−0.04	−0.12	0.04
Soil in child play spaces E. coli > 1 MPN/gram dry weight	161	−0.38	−0.73	−0.03	−0.13	−0.44	0.18	−0.25	−0.61	0.10	−0.64	−1.04	−0.24	−0.49	−0.83	−0.15	−0.11	−0.48	0.27
Soil in child play spaces E. coli MPNI/gram dry weight (Log10)		−0.03	−0.07	−0.002	−0.01	−0.04	0.01	−0.02	−0.06	0.01	−0.06	−0.10	−0.02	−0.04	−0.07	−0.01	−0.01	−0.04	0.03
Child food E. coli > 1 MPN/gram dry weight	115	0.11	−0.22	0.45	0.18	−0.15	0.51	0.08	−0.27	0.43	0.12	−0.19	0.42	−0.06	−0.41	0.30	0.07	−0.29	0.43
Child food E. coli MPN/ gram dry weight (Log10)		0.02	−0.02	0.06	0.04	−0.01	0.08	0.02	−0.02	0.06	0.01	−0.04	0.05	−0.01	−0.05	0.03	0.01	−0.04	0.06
Object/surface child mouthed E. coli > 1 MPN per object/100 cm2
Surface area	154	0.02	−0.25	0.28	0.12	−0.15	0.40	0.05	−0.25	0.34	0.03	−0.28	0.33	−0.10	−0.38	0.19	0.04	−0.24	0.33
Object/surface child mouthed E. coli MPN per object/100 cm2
Surface area (Log10)		0.01	−0.03	0.05	0.02	−0.03	0.06	0.00	−0.05	0.05	0.02	−0.03	0.06	−0.01	−0.05	0.03	0.02	−0.02	0.06
Child hand rinse E. coli > 1 MPN/100 mL (both hands)	162	−0.23	−0.58	0.12	−0.37	−0.72	−0.03	−0.08	−0.39	0.24	−0.15	−0.55	0.26	−0.19	−0.55	0.16	−0.19	−0.56	0.18
Child hand rinse E. coli MPN (both hands) (Log10)		−0.04	−0.09	0.01	−0.06	−0.10	−0.01	−0.01	−0.05	0.03	−0.03	−0.09	0.02	−0.03	−0.07	0.02	−0.04	−0.08	0.01
Caregiver hand rinse E. coli > 1 MPN/100 mL (both hands)	151	−0.03	−0.43	0.36	0.01	−0.38	0.39	0.03	−0.33	0.39	0.05	−0.30	0.40	−0.09	−0.47	0.29	−0.08	−0.46	0.29
Caregiver hand rinse E. coli (both hands) (Log10)		0.00	−0.05	0.05	0.01	−0.04	0.06	0.01	−0.04	0.05	0.02	−0.03	0.06	−0.01	−0.06	0.03	0.00	−0.05	0.05

MPN = most probable number.

^*Models are adjusted for wall type, number of household members, and household educational level.

DISCUSSION

In the REDUCE prospective cohort study conducted in rural DRC, E. coli in soil in child play spaces, on child hands, and in stored drinking water was associated with poorer developmental outcomes in young children, including lower communication, gross motor, fine motor, and problem-solving skills. This is the first study, to our knowledge, to assess the association between E. coli contamination in the household environment and child cognitive and socioemotional developmental outcomes. These results demonstrate the urgent need for interventions to reduce fecal contamination in the household environment to protect the health of susceptible pediatric populations in rural DRC.

A possible mechanism by which soil, water, and hand contamination impacted child developmental outcomes in our study population is through increasing enteric infections, and thereby intestinal inflammation, leading to poor nutrient absorption, malnutrition, and poorer brain development. ^{6, 25 – 27} Consistent with this hypothesis, our previous prospective cohort study in rural Bangladesh found that the presence of pathogenic E. coli in soil, caregiver reports of children ingesting soil, and enteric infections were associated with increased intestinal inflammation in young children. ^{6, 26} A previous cross-sectional study in Bangladesh also found that intestinal inflammation was associated with vitamin A and iron deficiency among young children. ²⁸ The first few years of life represent a critical period for brain development. ⁸ Nutrient deficiencies during this period can negatively impact brain development and result in cognitive deficits later in life. Delivering WASH interventions during the first few years of life has the potential to reduce exposure to fecal pathogens leading to longer term developmental benefits for children that can last into adulthood. ^{8, 15}

Only soil, water, and child hand contamination were associated with adverse child cognitive and socioemotional outcomes. This finding is likely attributed to these routes being predominate pathways for exposure to fecal pathogens for our study population, given the large proportion of these samples that were positive and had high E. coli concentrations. Soil in child play spaces was the pathway for exposure to fecal pathogens that had the most adverse outcomes across cognitive and and socioemotional domains (combined EASQ, fine motor, and problem-solving). Previous modeling studies have found that exposure to soil through direct ingestion and through mouthing hands, objects, and surfaces can serve as an important exposure route to fecal pathogens, which can be higher than dietary exposure routes (water and food). ^{29 – 31} Therefore, one potential explanation for this finding is that soil contributed the greatest to exposure to fecal pathogens for our study population, resulting in it being associated with the greatest adverse impacts to child cognitive and socioemotional outcomes. Future studies should be conducted similar to Kwong et al. to quantify the exposure to E. coli from each transmission pathway. ²⁹ It is not clear why fecal contamination from different pathways affected some cognitive and socioemotional domains and not others. EASQ is a broad measure intended to capture several measures of cognitive and socioemotional development; more sensitive tests of child cognitive and socioemotional development might show even larger effect sizes arising from exposure to environmental fecal contamination. Future research is needed to explore these associations in more detail using multiple assessments of cognitive and socioemotional domains using different assessment tools.

Our results align with previous findings from WASH trials demonstrating the importance of water quality, sanitation, and hygiene on child developmental outcomes. Previous randomized controlled trials in Bangladesh and Pakistan have shown that delivery of WASH interventions can improve cognitive outcomes in young children. ^{13, 15} The WASH Benefits Bangladesh trial found that delivery of WASH interventions including safe drinking water, handwashing with soap, and sanitation both individually and combined increased overall EASQ scores at 2 years of age and reduced diarrhea prevalence. ¹³ The trial conducted in Pakistan administered the Battelle Developmental Inventory II, and found higher global developmental quotients at 5–7 years of age for those who had received a handwashing intervention. ¹⁵ However, in trials in Kenya and Zimbabwe, intensive WASH interventions did not improve child developmental outcomes nor diarrhea, potentially because of low adherence to the WASH interventions recommended. ^{12, 16} However, when a subanalysis was performed of HIV-exposed children in the trial in Zimbabwe, it was found that combining WASH and improved infant and young child feeding significantly improved cognitive, motor, and language development among these children. ³² WASH intervention studies for young children are needed in eastern DRC to investigate these associations in our study setting.

In our study, we measured development as an age-standardized score, so our effect sizes are interpretable similar to other z scores in terms of the number of milestones reached that were queried by the EASQ. One way to interpret the effect sizes is also as a percentile change: a difference of 0.64 in EASQ development z score (soil E. coli in child play space finding for fine motor function) is equivalent to a change in percentile from 50% to 75% of milestones reached relative to those similar in age. Therefore, the impacts of these risk factors appear to be of potential clinical relevance.

Few WASH programs focus on reducing child contact with fecal contamination in their play spaces. Play spaces are areas on the household compound where children frequently play. In Zambia and Zimbabwe, play yards were introduced as part of formative research to reduce child exposure to fecal pathogens, however, fecal contamination in child play spaces was not measured. ^{33, 34} Most recently, in Bangladesh and at our REDUCE cohort study in DRC, we conducted formative research on delivery of an intervention promoting the use of a play mat to reduce child contact with fecal contamination in their play spaces. ³⁵ There has only been one randomized controlled trial to date that has delivered a WASH intervention focused on limiting child contact with fecal contamination in their play spaces. ¹⁷ This study was conducted in rural Zimbabwe and provided a play yard for young children. This study found that delivery of a WASH intervention that included handwashing with soap, an improved latrine, water treatment, and provision of a play yard resulted in improvements in the MacArthur-Bates Communicative Development Inventories grammar checklist at 2 years. ¹⁶ However, this was the only significant child developmental outcome with delivery of these WASH interventions. Future intervention studies are needed that target environmental contamination in child play spaces and assess developmental outcomes in young children.

Based on our findings from this present cohort study, the REDUCE program developed through formative research behavior change communication modules to promote locally made play mats to keep young children off the ground when they are playing, treatment of stored drinking water using chlorine tablets, and frequent handwashing with soap for young children. ³⁶ These modules are named “Baby” WASH communication modules since they target young children. We piloted these Baby WASH communication modules in South Kivu, DRC. These modules were well-received by beneficiaries, who reported repairing play mats when they became worn and constructing and teaching neighbors to construct tippy taps (the handwashing station promoted in the module). These REDUCE program Baby WASH communication modules are currently being rolled out to over 1 million beneficiaries through the Care Group Model in South Kivu and Tanganyika, DRC. ³⁷

This study has some limitations. First, we relied on EASQ outcomes at a single time point, and therefore cannot conclude on trends over time. Second, we focused on assessing child developmental outcomes in children younger than 5 years of age, and therefore cannot conclude on longer term impacts of fecal contamination in the household on child development. Third, potential confounders for our study not measured include the micronutrient status of the child, stimulation in the home, maternal education, and birthweight of the child. These factors may have been associated with both child cognitive and socioemotional developmental outcomes and fecal contamination in the household. Therefore, it is important that future studies should include these covariates, and detailed measurement of socioeconomic variables for study households. Fourth, we are using an MPN method to quantify E. coli. Future studies should quantify E. coli concentrations through culturing methods and include other enteric pathogens in sampling such as viruses and protozoa and other bacterial pathogens such as Campylobacter and Shigella. Finally, we examined child developmental outcomes across all six domains for all selected exposures; we estimated a total of 84 associations, which increased the likelihood of spurious results because of type one error. However, out of these 84 estimates, nine were identified as statistically significant, which is above the 4 (5%) that would be expected by chance. It should also be noted that no significant associations found were contrary to the hypothesized direction. Future studies with a larger sample size are needed that investigate the significant associations found in this publication.

This study has multiple strengths. First, the prospective design of the study allowed us to assess the association between E. coli contamination and subsequent child developmental outcomes. Second, performing E. coli analyses on soil from child play spaces and on objects and surfaces children frequently came into contact with, instead of focusing only on dietary transmission routes (e.g., food and water). Most WASH studies focus on E. coli contamination in water, hands, and food, and rarely include other transmission pathways. Third, this study evaluated the use of individual level food and object and surface data rather than relying on household level data. This approach allowed us to characterize individual level exposure to E. coli from these exposure pathways.

The findings from the REDUCE prospective cohort study have shown that E. coli in the soil in child play spaces, on child hands, and in stored drinking water were associated with poorer communication, gross motor, fine motor, and problem-solving child developmental outcomes in rural DRC. Interventions are urgently needed to reduce fecal contamination in the household environment to improve child developmental outcomes. The REDUCE program design presents a model that can be used for the development of evidence-based Baby WASH intervention programs to improve child health.

Note: Supplemental table appears at www.ajtmh.org.