Commonalities and specificities between attention deficit/hyperactivity disorder and autism-spectrum disorders: can epidemiology contribute?

Abstract

The first step in searching the causes of a disorder is to provide a definition of the phenomenon under study. This may seem obvious, yet different forms of measurement have different properties and entail strengths and limitations which are not always acknowledged. Data on children’s mental health problems come either from clinical records or from specifically designed studies. The use of clinical records offers the advantage of providing a diagnosis established by a health professional knowledgeable of the disorder. Yet while this is the best approach when assessing an individual, its validity at a population level is not always guaranteed. As shown by the study conducted by Ashwood et al. [3] in 66 clinical centres in 31 countries in Europe, tools used to systematically ascertain ASD vary across countries with specific differences between Western and Eastern Europe. This complicates the comparisons of studies examining risk factors of ASD in different settings and makes it difficult to merge different datasets to study ASD in large samples. This leads the authors to call for greater scientific collaboration to enable ‘the translation, adaptation and validation of ASD diagnostic tools in multiple languages and cultures as well as evaluation of existing free/open access measures to determine their screening and diagnostic accuracy in different communities’ [3]. Another issue regarding the use of clinical records is that, as pointed out by Lehti et al. [15], they are by definition only available among individuals who have accessed the health care system. This is an unlikely source of major bias when studying severe mental health problems, but may be a concern in less severe disorders. In a study of over 7000 youths living in Finland, Lehti et al. find no association between youths’ migrant background and their likelihood of Asperger syndrome diagnosis, which goes against research showing that children of immigrants have higher levels of autism than native children [16]. One possible explanation of this unexpected finding is that children of immigrant parents are less likely to be referred to mental health services in case of Asperger syndrome than those of non-immigrant parents, and therefore less likely to count as cases. The degree to which migrant status as well as other socio-demographic characteristics can influence individuals’ access to services can vary according to the characteristics of the health care system, and Lehti et al.’s surprising finding calls for additional research on the links between migrant status and ASD in different settings. An alternative to the use of clinical records is the implementation of studies which examine the presence of mental health symptoms in children in a systematic way—either through diagnostic schedules [e.g. Kiddie-Schedule for Affective Disorders and Schizophrenia (K-SADS) [27] or the Autism Diagnostic Interview (ADI) [3]] or through the scales which have been validated against clinical diagnosis [e.g. Child Behavioral Checklist (CBCL) [1] or the Strengths and Difficulties Questionnaire (SDQ) [9]]. In a study of over 4500 children recruited in 6 European countries, Kovess et al. [14] use the SDQ to examine a controversial research question—is maternal prenatal smoking a risk factor of ADHD? Combining parent and teacher reports, a more valid approach than single-informant reports, the authors find evidence of an association even after controlling for multiple covariates (adjusted OR 1.45). This is an important contribution to an area marked by widely divergent findings: while several studies set in community samples have found independent effects of maternal prenatal smoking on children’s symptoms of ADHD—in an era when many women smoked during pregnancy [12] or controlling for multiple risk factors of behavioral problems [19, 29], studies conducted among siblings [23] or among children born to genetically unrelated mothers [24] have reported no association. However, study designs based on siblings or on children born using Assisted Reproductive Technologies may be limited by selection bias which makes participants different from the general population and could explain why findings differ from those of community samples [13]. Cross-country comparisons such as the one published by Kovess et al. are an elegant way of comparing settings in which levels of maternal smoking as well as ADHD rates differ. This study contributes to the ongoing debate about the long-term consequences of in utero exposure to nicotine on children’s behavior [25]. Interestingly, studies examining the association between maternal prenatal smoking and ASD have generally reported no association [18], but given secular changes in diagnostic criteria of ASD and increased case recognition with time, this topic that requires additional research.