Neighbourhood-level prevalence of teacher-reported Autism Spectrum Disorder among kindergarten children in Canada: A population level study

Autism Spectrum Disorder (ASD) is a commonly diagnosed neurodevelopmental disorder in Canada, with a national prevalence estimate of 1 per 66 children diagnosed in 2015 (Public Health Agency of Canada, 2018). ASD is a heterogeneous, behaviourally-defined neurodevelopmental disorder, which has been associated with multiple genetic conditions, but it has no unifying pathological or neurobiological etiology (Geschwind & Levitt, 2007). According to the Diagnostic and Statistical Manual of Mental Disorders – 5th Edition (DSM – 5), the core dysfunctions of ASD occur in two behavioural domains: difficulties in social communication and social interaction, as well as restricted, repetitive behaviours and interests (Lai, Lombardo, Chakrabarti, & Baron-Cohen, 2013). The characteristics and varying severity of ASD are now more widely recognized – thus, the diagnosis rate of this disorder has increased substantially over the years (McConachie & Diggle, 2007). Children with ASD have heterogeneous developmental trajectories (Fountain, Winter, & Bearman, 2012). Children who are high functioning in early years tend to improve in their development more rapidly over time (Fountain et al., 2012). However, even children who are very low functioning in childhood through middle years can also improve in their development substantially by adolescence to match outcomes comparable to high functioning children (Fountain et al., 2012). Research suggests that participation in Early Intensive Behavioural Intervention (EIBI) improves adaptive behaviour, communication, daily living skills, and socialization capabilities among children with ASD (Eldevik, Hastings, Jahr, & Hughes, 2012; Peters-Scheffer, Didden, Korzilius, & Sturmey, 2011). There is evidence indicating that earlier intervention for these children is better for their development, including improving behavioural and cognitive outcomes (Janus et al., 2018a, Janus et al., 2018b; McConachie & Diggle, 2007). While several studies have demonstrated that ASD can be reliably diagnosed by the age of 2 years, the median age of ASD diagnosis remains over age of 4 years (Janus et al., 2018a, Janus et al., 2018b; Monteiro et al., 2015). Thus, the identification of ASD among kindergarten children between the age of 4–6 years presents an optimal opportunity to target early interventions (Janus et al., 2018a, Janus et al., 2018b). Although epidemiological studies suggest that the prevalence of ASD is increasing, there are several challenges associated with the current methods available for estimating prevalence (Matson & Kozlowski, 2011; Rice et al., 2010). Estimates based on administrative databases (which depend on special education classifications, ASD service eligibility, or medical billing codes) have several limitations (Zablotsky, Black, Maenner, Schieve, & Blumberg, 2015). For example, they may underestimate prevalence among specific subpopulations who have reduced access to systems that generate administrative counts and therefore are not captured in these counts – leading to socioeconomic disparities in the prevalence of ASD (Zablotsky et al., 2015). Furthermore, criteria for special education or other ASD services can differ across jurisdictions (Zablotsky et al., 2015). Survey-based estimations also have limitations – including respondents' lack of fluency to respond in the dominant language, general population surveys not being designed to include sufficient numbers of individuals affected by rare conditions precluding analyses that will generate reliable estimates for these sub-populations, reliance on respondents’ ability to understand the questions asked and accurately recall the specific diagnosis assigned (Ouellette-Kuntz et al., 2012; Zablotsky et al., 2015). Determining the prevalence of ASD using multiple data sources remains a difficult task given different data maintenance and linkage procedures across systems and jurisdictions. In the Canadian context, the Early Development Instrument (EDI), a population-based developmental assessment tool, presents a unique opportunity to monitor the prevalence and developmental health of kindergarten-age children with ASD (Janus & Offord, 2007). The EDI is implemented across Canada and is completed by teachers for each child in kindergarten classes. It provides data on development in five domains: physical health and well-being, social competence, emotional maturity, language and cognitive development, and communication skills and general knowledge. While the EDI is completed for individual children, the data obtained are interpreted at different levels of aggregation (e.g. children attending a school, children living in a neighbourhood) to provide information on the strengths and weaknesses of children in a particular group/community. Given the wide-scale implementation, the EDI data provide valuable information on low-frequency populations such as children with special needs, who typically represent a small percentage of population at the community level (Janus et al., 2018a, Janus et al., 2018b). Although the National Autism Spectrum Disorder Surveillance System provides the prevalence of children with ASD living in some provinces and territories in Canada for 2015, including Newfoundland and Labrador, Prince Edward Island, Quebec, British Columbia, Nova Scotia, New Brunswick, and Yukon, our knowledge of prevalence of kindergarten children with this disorder in other provinces and territories, as well as the development of these kindergarten children across all provinces and territories remains limited (Public Health Agency of Canada, 2018). Examining the prevalence and development of kindergarten children with ASD at the neighbourhood level that represents a meaningful geographic unit of residence has particular value. There is growing evidence indicating that kindergarten children's health is influenced by neighbourhood characteristics (Curtis, Dooley, & Phipps, 2004; Kohen, Oliver, & Pierre, 2009; Minh, Muhajarine, Janus, Brownell, & Guhn, 2017). Furthermore, there is increasing consensus that inequalities in health outcomes of a population are usually not fully accounted for by combinations of individual level factors and may therefore be attributable to factors that operate at an aggregate level, such as the neighbourhood level (Pickett & Pearl, 2001). Pickett and Pearl (2001) noted that contextual factors may be the most important determinants of the health of a population. There is no consistency in literature regarding the association between neighbourhood socioeconomic status and prevalence of children with ASD (Emerson, 2012; Hock & Ahmedani, 2012; Li, Sjostedt, Sundquist, Zoller, & Sundquist, 2014; Thomas et al., 2012). In contrast, spatial clustering of children with ASD appears to be associated with neighbourhood resources that can facilitate diagnosis – including number of pediatricians, number of advocacy organizations, and regional center spending on ASD services (Mazumdar, Winter, Liu, & Bearman, 2013). Children who live in close proximity of other children previously diagnosed with ASD are more likely to be diagnosed with ASD as well – which has been attributed to the diffusion of information about ASD through social networks, a phenomenon that can lead to spatial clustering of children with ASD in neighbourhoods (Liu, King, & Bearman, 2010). In view of these findings, it is propitious to examine the prevalence of ASD, and the development of all children living in neighbourhoods with different levels of spatial clusters as this will allow to generate hypotheses about drivers of prevalence of ASD and its association with child development. Furthermore, identifying geographic and jurisdictional differences in the prevalence of ASD is integral for providing necessary early intervention services and community education in areas of need. Combination of availability of reliable developmental outcomes at a national, population level for kindergarten children with spatial and diagnostic information makes this task particularly promising. The objectives of this study are to determine, among kindergarten children across provinces and territories in Canada (1) the variability in prevalence of ASD at the neighbourhood level; (2) the prevalence of neighbourhoods with no children with ASD; (3) the prevalence of neighbourhoods with different levels of spatial clusters of children with ASD; and (4) the developmental health status of children living in neighbourhoods with no children with ASD in comparison to children living in neighbourhoods with different levels of spatial clusters of children with ASD.