An epidemiological cohort study of children and adolescents investigating neurobehavioural effects from pesticide exposure and e-media use in South Africa

Background
Developmental neurotoxicity is currently referred to as the “silent pandemic” amongst children worldwide. Changes over time in the Global Burden of Disease (GBD) estimates of mental and substance use disorders have contributed to a shift in epidemiology. Although genetic factors play a causal role, environmental and behavioural factors are amongst the highest risk factors regarding disability-adjusted life years (DALYs) in the Global Burden of Disease. Industrial chemicals, air pollution, residential radon, and alcohol use fall within these environmental and behavioural risk factors. The associated neurodevelopmental disorders to these toxicants include Autism, Attention Deficit Hyperactivity Disorder and other cognitive disabilities. Several studies reveal associations to featured symptoms or partial characteristics of these disorders, including behavioural and or learning difficulties. These neurobehavioral difficulties may go unnoticed, yet impact on the cognitive functioning, learning and mental health of children in everyday life. Limited studies investigate risk factors related to chronic exposure at sub-toxic level on these outcomes. This study aims to investigate chronic neurotoxic effects of pesticides on the neurobehavior of children in the Western Cape of South Africa by including potentially relevant co-exposures from the use of electronic-media and maternal alcohol use.
To determine the effects on neurobehavioural difficulties in children living in the Western Cape, South Africa, this study is embedded in the South African-Swiss Bilateral SARChI Chair in Global Environmental Health, an ongoing pesticide exposure project, and the Child Health Agricultural Pesticide Cohort Study in South Africa (CaPSA). This project has a longitudinal design with a cohort of 1,000 school-going children between 9-16 years old. The study was conducted in three different agricultural farmland areas of the Western Cape region of South Africa, from which learners and their guardians were randomly selected from seven schools and categorised into the farm and non-farm residence. The study design conducted over three years included measurements at baseline (2017), follow-up (2019), and repeatedly in between to capture variation. Exposure assessment is measured through participant, guardian and farmer questionnaires on pesticide exposure, co-exposures, and relevant confounders. Objective measures using environmental sampling of air, water, and soil samples as well as biomonitoring of short-term biomarkers in urine, long-term biomarkers in hair, and geographical coordinates were collected over time to validate self-reported data and determine individual exposure variation.
Methods
Cross-sectional analysis using self-reported exposure data with standardised health outcome tools were employed in this thesis to assess the hypothesis testing under study: children who report exposure to the environmental factors in this study, will have lower neurocognitive functioning and higher health symptoms compared to those who report no exposure.
The primary health outcome on neurocognition was assessed using the software assessment tool, the CANTAB, (with the ability to measure small cognitive changes between groups ) and secondary outcome measures included the headache impact screening tool (HIT-6), sleep disturbance questions, and the Health Related Quality of Life tool (HRQol).
We aimed to determine the association between pesticide exposure and neurobehavioural health outcomes by including co-exposures of e-media and maternal alcohol use through the following objectives in each chapter of this thesis:
Chapter 2: The protocol paper which describes the methodology and the prospective cohort study in detail
Chapter 3: The association between e-media use on symptoms and neurocognitive outcomes
Chapter 4: The association between maternal alcohol use exposure on neurocognitive outcomes
Chapter 5: The association between the main environmental exposure to agricultural pesticides on the headache symptom and neurocognitive outcomes
Linear regression analysis was conducted with exposure proxies and health outcome scores as described in each chapter.
Results
Of the 1,001 grade two to nine students assessed at baseline from seven schools, participants are equally distributed across three study areas, equally proportionate in gender, with the youngest age group, 9-11 years more highly represented (60%) than the 12-16 year olds (40%) in the cohort. An almost equal percentage of the cohort report farm (46%) and village (56%) residence. 66% report having a family member including a sibling, parent, grandparent or other who works on a farm. 80% of the cohort report seeing pesticide spraying activities in nearby fields and just over 20% report having ever helped with cleaning farm equipment and assisting with pesticide storage in the past seven days. Of the 32% who report using a mobile phone and other electronic media devices with a smaller percentage connecting with the internet, the majority live in the area closest to the city. Other lifestyle covariates of this cohort include 33% who report a head injury and 15% who report to smoke and drink alcohol. Of the 482 subset guardian surveys collected, 10% of the mothers report gestational drinking, 29% past drinking and 27% report current drinking. Additional socio-demographics describing this sub-set of the cohort include 36% maternal unemployment, 41% maternal education at primary school level or lower, 38% live in a household of 5-6 members, 24% with seven or more members in a household, and 68% are qualified for government child support grants.
Findings
Preliminary patterns of association indicate an overall detrimental health effect on sleep disturbance, headache severity and lower HRQoL from high exposure-related behaviours of mobile phone calls (≥6minutes a day->1 hour/day), night-time mobile phone awakenings (≥1 time/week up to 7 times/week) and mobile phone addiction (36-91 score). Amongst those reporting regular night-time awakenings (≥1 times per week) from mobile phones, HRQol declined by 2.9 (95% CI: −6.1, 0.3), the sleep disturbance score increased by 2.0 (1.1, 2.9) units and headache impact score significantly increased by 5.4 (2.6; 8.2) units compared to the non-exposed group. Mobile phone ownership was a significant predictor of socio-economic status. Contrary to literature, we observe beneficial health effects on neurocognitive performance in all three domains, across all exposure proxies, especially for the moderate-media users, even after stratifying by age and socio-demographic factors. We observe significant predictors of area, alcohol use, head injury, and farm residence in these associations. Non-significant negative associations were found between maternal drinking behaviours and executive functioning. Significant predictors include the child’s age, sex, home-language, maternal employment, and household-size on executive functioning.
We identified high-risk exposure groups for pesticide-related activities, eating and picking crops off the nearby field, vineyard or orchard, as well as for storing pesticide equipment, on headache severity and lowered neurocognitive scores in memory and attention. About 50% of the cohort report engaging in behaviors related to pesticide exposure including work activities, eating crops directly from the field and leisure activities of playing, swimming or bathing in nearby water. Headache severity was consistently increased in relation to the three behaviors related to pesticide exposure, work activities (Beta: 1.99 [95% Confidence Interval: 0.86, 3.12]) eating crops (1.52 [0.41, 2,67]) and leisure activities (1.25 [0.18, 2.33). For neurocognitive outcomes, we observed an overall non-significant negative trend with pesticide exposure-related activities. Picking fruits directly from the vineyard or orchard was associated with lower paired associates learning (-0.88 [-1.60; -0.17]) and spatial working memory (-0.29 [-0.56; -0.03]) compared to those who do not pick crops off the field. Smaller differences were associated for eating fruits directly from the vineyard or orchard with lower motor screening -0.06 (-0.11,-0.01) and reaction speed -0.13 (-0.28, 0.10). We observed significant predictors of head injury and alcohol use for lowered neurobehavioural outcomes in these associations and learners who repeated a grade. Socio-demographic predictors include, home-language other than Afrikaans, maternal employment and learners who repeated a grade.
These negative trends are observed across regression models, in line with the hypothesis under study and coherent with literature.
Conclusions and recommendations
This study has gained insight into the environmental and behavioural determinants of pesticide exposure and e-media use on the neurobehavioral functioning of children in the rural LMIC of South Africa. We observe cross sectional associations in support of our hypothesis, warranting further investigation through longitudinal analysis with objective measures. in the next steps of the CapSA study.
Preliminary evidence on associations suggest adverse health effects and that several structures are at play, requiring further understanding of this multi-layer-framework. Lifestyle behavioral aspects related to the exposures, are crucial determinants to include in the steps toward causal analysis on neurobehavior. Our findings uncover that behaviors of e-media use, indicating dependency through longer duration calls, being woken up several times at night in the week due to mobile phone use and additional problematic aspects implicating decisions for social and emotional well-being, mediates the exposure-outcome relation. Since we observe beneficial results on neurocognitive development amongst moderate users, promotion of a healthy balance to e-media use is essential in this cohort, with the screening for high-risk dependency behaviors and associated health symptoms.
Overlapping environmental determinants in a rural LMIC setting with a history of apartheid and perpetuated disadvantage, poverty, and inequality in education and employment, create a highly complex socio-economic setting. This setting presented a suitable diverse context to unpack the complexity due to multiple risk factors. An epidemiological approach to understanding the cause of neurobehavioral difficulties in a rural LMIC, by exploring co-exposures, and through categorised exposure assessment, supports the efforts to unmask the true contextual health effects, and consequent confounding which may hinder comparability to findings from international studies if not adjusted for.
Our findings highlight that behaviors related to environmental pesticide exposures in children and adolescents should be investigated in health impact assessment, to identify high-risk groups who may reinforce the hazardous exposure to pesticides through ongoing dermal contact. While there are several hierarchical factors which contribute to pesticide exposure, from the most distant SES, implicating the disadvantage in education and language of safe use, to perpetuating industry sales of HHPs, behaviour related to the exposure should be considered in future research for comprehensive effective strategies of health intervention.
The CANTAB proved high quality data in its ability to identify chronic exposure health effects in a rural LMIC context and should be promoted in future studies for consistent findings on these topics. Strategies to raising awareness of health symptoms should be implemented in existing school intervention programmes, through parent workshops, and by means of e-media. An empowering approach may upskill this community and the children themselves on digitization and alternate methods of farming for safe use practices. Regular screening of hazardous work and leisure behaviours should be prioritised by all role players including parents, government, and industry. Qualitative research is an additional element for in-depth insight and understanding of this community’s perceptions on health and risks. The lack of infrastructure to carry out effective strategies should be managed with continuous health impact assessment and the investment of industry and government to address the issue at secondary level by testing all chemicals for neurotoxicity, providing guidelines on public health pesticides, removing HHP sales, enforcing restrictions, and regular screening for children at risk of environmental toxicity. Primary prevention should include alternate farming methods, alternate household and agricultural pesticide use and applications, and at a policy level, to enforce stricter regulations regarding the proximity of households and schools to farms in agricultural settings.