2016 Royal Society Pfizer Award Seminar

The study of language development in diverse settings is possibly one of the most crucial for us to understand how children in general develop language. I will talk about three areas of language development that we have investigated in children learning language in Kilifi District, Kenya.

Assessing children’s language is challenging in most settings, but in particular where children are unused to meeting strangers, and especially to formal education contexts. Parent report methods – such as the CDI (Communicative Development Inventory) are an ideal solution. The CDI is an internationally renowned, widely used set of measures of communication development in young children. I will describe our development of this measure for the Kenyan Bantu languages, and its use in applied studies.

Data collected using the CDIs are not just of interest in such studies. Theories of children’s early vocabulary suggest that early words are primarily nouns – and that this is because children have a bias towards these, either due to what they hear or what they bring to the task of learning words.
But children do not just say words – they also understand them. CDI methods can collect data on children’s comprehension much more easily than many other methods. Looking at universals across languages – especially in the under-studied Bantu languages – I conclude that children’s knowledge overall is more balanced across word classes than their spoken vocabulary.

Cross-linguistic studies of grammar are the classic method of investigating how children learn the abstract structures of language. Between languages, there is increasing evidence that children learn supposedly difficult structures, such as the passive construction, more readily in languages where these structures are more available to them. In three Kenyan dialect contexts, we examine evidence that these differences in availability also exist between children and families, as well as between languages.

Background: There are ten million epilepsy sufferers in sub-Saharan Africa (SSA) and 50% are children. Children with epilepsy (CWE) in the West are known to have a high prevalence of externalising psychopathology including ADHD (Attention Deficit Hyperactivity Disorder) and conduct disorder, in SSA the impact may be greater. Older anti-epileptics used to treat Kenyan children may also increase the risk of externalising disorders. This case-control study uses the Swahili Strengths and Difficulties Questionnaire to investigate epilepsy as a risk factor for externalising problems in rural Kenyan Children. 

Methods: We have screened 113 CWE and 151 unaffected children from the community using a new Swahili version of the internationally recognised Strengths and Difficulties Questionnaire (SDQ) to probe children’s behaviour and mental health. A random sample of 89 of these children (36 screen positive and 55 screen negative for externalising disorders) were fully assessed by a child psychiatrist (blind to screening and epilepsy status) to determine whether the SDQ is valid in this setting and if diagnostic criteria were met. For those children taking anti-epileptics we took serum drug levels to determine level of concordance with medication (n=20).

Results and Discussion: The psychometrics of the SDQ indicated good internal consistency, and Receiver Operator Characteristic Curves using the clinical assessment as a gold standard showed fair to good discrimination between those with and without psychopathology (area under the curve (AUC) of 0.74) and those with and without externalising problems (AUC of 0.79). The hyperactivity scale of the SDQ performed poorly against clinical diagnosis of ADHD, but well when combined with the conduct scale. This may indicate that rural Kenyan parents perceive their children with ADHD as delinquent rather than having a mental health problem. Overall, CWE were more likely than controls to be perceived as having conduct problems by their parents (p<0.001, β=1.61 CI .63-1.70). There also was a statistical trend for children taking anti-epileptic drugs to also show higher levels of conduct and emotional problems than CWE not taking drugs.

In 2007, it was estimated that more than 200 million children in low and middle income countries fail to reach their developmental and cognitive potentials due to exposure to a host of risk factors such as infectious disease, malnutrition and chronic poverty. More recently, it has been estimated that 1 in 3 children are delayed in achieving their milestones.

In many low and middle income countries the lack of adequately standardised tools for monitoring child development has been a barrier to estimating the impact of both communicable and non-communicable disease on childhood outcome, identifying children in need of early intervention and monitoring intervention.

In this presentation, I will highlight some of the ways in which we have contributed to addressing this challenge in our context. In Kilifi, we have pioneered the development of a set of measures to evaluate child development. These measures have presented with good psychometric properties including reliability (internal consistency, test-retest and inter-rater) and validity (concurrent, criterion, convergent, discriminant and predictive). Additionally, through working with diverse teams in various work sites, I have shared my expertise to help those teams develop measures suitable to their needs. As a result, this is contributing to improved access to appropriate measures of neurodevelopment for low and middle income countries.  

I will conclude this talk by highlighting the next steps we would like to take towards refinement of the neurodevelopmental tools in Kilifi, Kenya.  Based on feedback from previous usage both in Kilifi and in other contexts as well as advancements in technology, we plan to update our measures to a more user-friendly mobile based version, and reduce time associated with training test administrators by developing multimedia training materials that are open access.