Behavioural and emotional comorbidities in school-aged children with neurological conditions in Kilifi, Kenya, and their long-term consequences

Background: Neurological conditions and mental health problems are common in children in low- and middle-income countries, but the risk factors and downstream impact of these problems on children with neurological conditions are not reported. Objective: To determine the association of neurological conditions with behavioural and emotional problems in children, the prevalence and risk factors of behavioural and emotional problems, and long-term impact of these conditions. Methods: Data on multiple neurological conditions and mental health problems were available for 1,616 children (aged 6–9 years) from Kilifi, Kenya. Neurological conditions were diagnosed using standardised tools and clinical examination. Behavioural and emotional problems assessed using Child Behaviour Questionnaire for Parents. Long-term outcomes were obtained from census data of the Kilifi Health and Demographic Surveillance System. Logistic and linear regression were used to measure associations. Results: Mental health problems were higher in those with any neurological condition compared to those without (24% vs. 12%, p < 0.001). Cognitive (odds ratio (OR) = 2.39; 95% CI: 1.59–3.59), motor (OR = 3.17; 95% CI: 1.72–5.82), hearing (OR = 2.07; 95% CI:1.12–3.83) impairments, and epilepsy (OR = 4.18; 95% CI: 2.69–6.48), were associated with mental health problems. Prevalence of any mental health problem was 15%, with externalizing problems more common than internalizing problems (21% vs. 17%, p = 0.004). Longitudinal follow-up indicated that the disorders affected an individual’s future schooling (e.g. OR = 1.25; 95% CI: 0.14–1.46 following cognitive impairments), occupation (OR = 2.44; 95% CI: 1.09–5.44 following mental health problems), and access to household assets (OR = 2.78; 95% CI: 0.99–7.85 following epilepsy). Conclusions: Neurological conditions in school-aged children in Kilifi are associated with mental health problems, and both disorders have long-term consequences. Preventive and therapeutic measures for these conditions are needed to improve outcomes of these children. An epidemiological survey was carried out between June 2001 to March 2002 on neurological disabilities, impairments, and mental health problems in children aged 6–9 years in a rural area in Kilifi County [9]. The county is located along the coast of Kenya and the area residents are mainly Mijikenda, a Bantu group of nine tribes with the Giriama dominating. Kilifi County has a population distribution of 1.5 million residents, whereby 0–14 year olds constitute 42% of the total population [14]. In this study, we selected children aged 6 years and above owing to the difficulty in identifying and assessing mental and neurological disorders in younger children especially hearing, visual, and cognitive impairments. Additionally, having survived early childhood, which is a period associated with high adversity and mortality in sub-Saharan Africa [15], studying children above 6 years of age enabled understanding the impact of early-life negative impacts of neurological conditions, including being able to access education. Assessments were performed in two stages during the 2001 epidemiological survey (Figure S1). To screen for neurological impairment and epilepsy in stage I, five trained field interviewers fluent in the local Kigiryama language administered Ten Questions Questionnaire (TQQ) [16], to parents or guardians of 10,218 children who agreed to participate. The TQQ consists of ten items (with a yes or no response) designed to detect moderate-to-severe impairments and disorders; including five questions addressing cognitive development, two questions relating to motor ability, and one question each regarding vision, hearing, and seizures. Those who tested positive on the TQQ, and a random sample of those who tested negative (10.3%), were invited to participate in stage II. In this stage, a team of clinicians and psychological assessors performed clinical history, examination, and psychological assessments to detect cognitive, motor, hearing, visual impairments, and epilepsy (Table S1). Perinatal and postnatal adverse occurrences were documented from medical history and maternal recall of pregnancy and delivery events, a method that was shown to be relatively reproducible and accurate in previous reports [17,18]. Data on perinatal events included pregnancy complications, place and mode of delivery, and birth trauma/difficulties, while postnatal occurrences assessed were history of neonatal insults, neonatal jaundice, developmental problems, child’s immunization, and neurological deficit. During assessments in stage II, a Child Behaviour Questionnaire for Parents (CBQFP) was administered to assess behavioural and emotional problems in the children. The CBQFP was administered to a parent or guardian in a conversational manner comprising of 15 items to assess various aspects of behaviour and emotion including reaction to change, independence, mood, worries, fears, and habits. The severity and frequency of behaviour described in the questionnaire were rated and scored depending on the parents’ response with a higher overall score signifying a higher level of total behaviour or emotional problems [19]. The CBQFP had been previously adapted and validated for use in this setting, demonstrating a high degree of interrater reliability (r = 0.92), and fair internal reliability (standardized item α = 0.61) in the assessment of neuropsychological outcomes of cerebral malaria [19]. Questions in the CBQFP on anxiety, temper, mood, worries, fears, and empathy were classified as indicators for internalizing or emotional problems; while those on the child’s concentration span, social relationships, social dependency, and behaviour in public assessed for externalizing problems. Assessment data from questions on appetite, habits, self-care, and wetting/soiling oneself were not included in the analysis as they were not categorized as indicators for internalizing or externalizing problems. We followed up the children assessed in stage II through the Kilifi Health and Demographic Surveillance System (KHDSS) from June 2001 to May 2008, to assess any long-term consequences of neurological conditions, behavioural and emotional problems on their education and socio-economic status. The KHDSS is a database that was established to create longitudinal community-based records of births, deaths, pregnancies, migration events, and additional sociodemographic information including socio-economic status and educational achievement [20]. The surveillance region includes an estimated population of 280,000 residents [21] living in an area covering 891 km2, which is in reference to the area served by the county’s main referral hospital. Follow-up of the participants was done through 4-monthly household visits during which data on their schooling, education level and years completed, economic status (measured by access to a source of lighting), assets acquired (e.g. a working mobile phone, radio) at a household level, and occupation were collected by trained field workers using standardised data collection tools. All statistical analyses were performed using STATA version 15 (StataCorp, College Station, TX, USA). Demographic characteristics between those who tested positive or negative on the TQQ test were compared using Pearson χ2 test (for categorical measures) and Student’s t-test (for continuous measures). Associations between neurological conditions and total mental health problems were illustrated using Pearson χ2 test, while Fisher’s exact test was used for comparison of infrequent observations [22]. The associations were further evaluated through building of age and sex adjusted linear and logistic regression models. The cut-off score for behavioural and emotional problems was derived from the 90th percentile of total behavioural and emotional scores among children who screened negative for neurological conditions, with the resultant cut-off of the mental health problems applied to all participants in the dataset. This was done following guidelines provided by Richman et al [23], and similar criteria were used to derive cut-offs for internalizing and externalizing problems, respectively. Age and sex stratified prevalence of total behavioural and emotional problems were computed, and linear regression applied to identify significant risk factors for total mental health problems scores, total internalizing, and externalizing scores. A logistic regression model was also used to determine potential risk factors associated with internalizing and externalizing problems as categorical variables, and in determining long-term impact of neurological conditions and mental health problems on schooling, occupation, and asset ownership. Risk factors with a univariable p ≤ 0.25 were fitted into a sex and age adjusted multivariable linear and logistic regression model to further identify independent factors.