Association between malaria infection and early childhood development mediated by anemia in rural Kenya

Malaria is a leading cause of morbidity and mortality among children under five years of age, with most cases occurring in Sub-Saharan Africa. Children in this age group in Africa are at greatest risk worldwide for developmental deficits. There are research gaps in quantifying the risks of mild malaria cases, understanding the pathways linking malaria infection and poor child development, and evaluating the impact of malaria on the development of children under five years. We analyzed the association between malaria infection and gross motor, communication, and personal social development in 592 children age 24 months in rural, western Kenya as part of the WASH Benefits environmental enteric dysfunction sub-study. Eighteen percent of children had malaria, 20% were at risk for gross motor delay, 21% were at risk for communication delay, and 23% were at risk for personal social delay. Having a positive malaria test was associated with increased risk for gross motor, communication, and personal social delay while adjusting for child characteristics, household demographics, study cluster, and intervention treatment arm. Mediation analyses suggested that anemia was a significant mediator in the pathway between malaria infection and risk for gross motor, communication, and personal social development delays. The proportion of the total effect of malaria on the risk of developmental delay that is mediated by anemia across the subscales was small (ranging from 9% of the effect on gross motor development to 16% of the effect on communication development mediated by anemia). Overall, malaria may be associated with short-term developmental delays during a vulnerable period of early life. Therefore, preventative malaria measures and immediate treatment are imperative for children’s optimal development, particularly in light of projections of continued high malaria transmission in Kenya and Africa. These data were collected during the WASH Benefits trial, which is a cluster-randomized controlled trial assessing the impact of WASH and nutrition interventions individually and combined on child health outcomes [44,45,46]. The study was conducted in Bungoma and Kakamega counties in western Kenya. The region is rural and the majority of the population are subsistence farmers. The WASH Benefits trial took place from November 2012 to July 2016. Households were selected to participate in the trial if they were in a rural community (defined as <25% of residents living in a rented home (most families in rural areas own homes compared with urban households in Kenya), with <2 gas stations, and <10 shops in the area) and had limited access to water and sanitation (defined as communities where >80% of residents did not have access to piped water). Female participants were eligible for enrollment into the study if they were pregnant, if they or their partner owned their house, and if they were not planning to move within the next 12 months. The infant that was in utero during the village census was the index child. Households were visited at enrollment and two follow-up periods, one and two years later. A sub-study assessing environmental enteric dysfunction, malaria infection, and hemoglobin concentrations from biological samples enrolled 2,304 households from four arms of the WASH Benefits trial (WASH, WASH and Nutrition, Nutrition, or Active Control) [36,44,47]. The nutrition intervention entailed nutrition education and all children between 6–24 months in the participating household receiving LNS with iron. In the second follow-up, 1,449 children were included in the sub-study [36,44,47]. This sample size was calculated based on having the power to detect differences in environmental enteropathy biomarkers [48] and this sample size was sufficient to determine the association between malaria infection and child development outcomes. An equal proportion of children were selected from each intervention arm. Caregivers brought study children to a central location for sub-study data collection between August 2015 and April 2016. Blood samples for malaria and anemia diagnoses were taken by trained phlebotomists and health surveys were completed. Caregivers provided written informed consent. Consent for blood collection and a sufficient blood sample was obtained for 699 children and therefore malaria and hemoglobin measurements only existed for this subset of children. An average of two to three months later, caregivers brought study children to a central location for main trial data collection, including a child development assessment. Local enumerators conducted most surveys in Kiswahili, the regional language. Surveys and data collection tools were developed from validated measures [49] and locally adapted. Venous blood samples were drawn from sub-study participant children in a central location within villages for biomarker analyses and drops of blood were used to assess malaria parasites and hemoglobin concentrations. Rapid diagnostic tests (RDTs) for P. falciparum, P. vivax, P. ovale, and P. malariae were conducted using rapid diagnostic kits (SD Bioline, Malaria Ag, P.f, P.f, P.v, Alere). The kits can detect parasite concentrations up to 28 days after infection. All children with a positive test result were defined as having malaria in these analyses. Axillary temperatures were taken for all children and if the malaria parasite test was positive and the child’s temperature was 37.5 degrees Celsius or higher, he/she was referred to the nearest clinic for malaria treatment. Hemoglobin concentrations were measured using a Hemocue (Hb 301) machine (Angelholm, Sweden). Anemia was defined as having a hemoglobin concentration below 11.0 g/dL [50]. A medical history form was also completed with caregivers about the child’s recent experiences with illness, prophylaxis, and treatment related to a variety of health conditions, including malaria. Household demographics, including asset ownership and maternal education, were assessed using standardized questionnaires administered at enrollment. Socioeconomic status was defined with an asset index using enrollment survey results for all sub-study participants. The asset index was created using the first principal component from a principal components analysis, which has been shown to be an appropriate proxy for household wealth and is correlated with consumption expenditures [51,52,53]. Items included in the asset index were housing structure, electricity, and ownership of a radio, television, mobile phone, clock, bicycle, motorcycle, gas stove, and car. Maternal education was categorized into the following categories: incomplete primary school, completed primary school, and any secondary school. Three domains of child development were assessed: gross motor, communication, and personal social. The corresponding subscales of the Extended Ages and Stages Questionnaire (EASQ) were used to measure each developmental outcome among all WASH Benefits trial participants during the last data collection round (at approximately age 24 months). The EASQ is a modified version of the Ages and Stages Questionnaire (ASQ), which is widely used for the developmental screening of children under five years of age [54,55,56] and has been used in LMICs [57]. The questionnaires contain a series of age-specific items assessing the achievement of developmental milestones and tracking the child’s progress. The EASQ was piloted to assess respondent bias, maternal accuracy in reporting children’s abilities, cultural appropriateness, and feasibility according to standard procedures [49]. The Gross Motor subscale evaluates body and muscle movement, including tasks like standing, walking, and balancing. The Communication subscale assesses language development and the use of words or sounds to express feelings. The Personal Social subscale reflects emotional responses and social interactions. The EASQ was translated and adapted for this study by using local culturally appropriate items, examples, and tasks, but no substantial changes were made to the original questionnaires. Continuous scores for each subscale were calculated and converted to age-adjusted Z-scores based on the control group within the larger WASH Benefits trial. Children were considered at risk for delay in this population if their Z-scores were below the 25th percentile of control group Z-scores for each domain. This method of defining risk for delay was used as a conservative strategy because there was no prior research informing a culturally appropriate clinical cut-off in this study population. Binary variables were used to incorporate a quantitative definition of poor development and to be consistent with analyses used in the main WASH Benefits publications [46,58]. All statistical analyses were conducted in Stata 14 (StataCorp, College Station, TX, USA, 2015). Three sets of logistic multivariate analyses were undertaken to test the hypotheses corresponding to each child development outcome (gross motor, communication, and personal social). Two regression models were built for each child development outcome to assess the pathways between malaria infection (binary) and being at risk for gross motor, communication, and personal social delays (binary). The first model included malaria infection (binary) and the child development outcomes (three separate binary outcomes). Anemia status (binary; anemia defined as hemoglobin concentration <11.0 g/dL) was added to the second model. The interaction between malaria status and participation in the nutrition intervention was also analyzed in multivariate models to assess whether being in a nutrition treatment arm differentially affected malaria infection. All models were adjusted for child characteristics (sex [binary] and age [continuous, months]), household demographics (asset index [continuous] and maternal education [categorical]), study cluster, and whether the household received the nutrition intervention (binary). Analyses controlled for arms that received nutrition interventions because they were likely confounders. Mediation analyses were then undertaken to assess whether anemia was a mediator along the pathway between malaria infection and being at risk for gross motor, communication, and personal social delays. The binary_mediation method in Stata [59,60,61] was employed. Like the Sobel–Goodman test for continuous outcomes, the binary_mediation method shows that mediation occurred if: 1) malaria infection was significantly associated with the mediator (anemia), 2) malaria infection was significantly associated with developmental delay in the absence of the mediator (anemia), 3) the mediator (anemia) was significantly associated with developmental delay, and 4) the effect of malaria infection on developmental delay decreased upon the addition of the mediator (anemia) to the model. The proportion of the total effect of malaria infection on the risk for gross motor, communication, and personal social delay that is mediated by anemia was also calculated and reported. Children with missing malaria data or child development assessments were removed from analyses. Nine percent of children had missing data from one of these measurements and thus were not included in the final analyses. The final sample size was therefore 592 children. Differences between those who provided consent for blood collection and had a sufficient blood sample, and those who did not were assessed due to the low response rate and no significant differences or systematic attrition were found [36]. The WASH Benefits trial is registered as a clinical trial with the U.S. National Institutes of Health (https://clinicaltrials.gov/ct2/show/{"type":"clinical-trial","attrs":{"text":"NCT01704105","term_id":"NCT01704105"}}NCT01704105). Ethical approval was obtained from the University of California, Berkeley Center for the Protection of Human Subjects and the Kenya Medical Research Institute Ethical Review Committee. Adult participants provided written consent for themselves and their children prior to enrollment.