Summary of findings
To the best of our knowledge, this is the first study to examine whether early childhood onset epilepsy is associated with a greater risk of mid childhood onset ADHD in children with ASD. We found a near twofold increase in risk of developing ADHD for ASD children if they had been hospitalised with epilepsy before the age of 7, an association which still persisted once associated physical conditions, family history, and ID were accounted for.
Previous studies have identified children with epilepsy as at increased risk for neurodevelopmental disorders. For instance, Reilly et al. (2019) reported that children with epilepsy before age 7 were at high risk for reaching clinical thresholds for ADHD (41%) or ASD (18%) [
32]. Our study extends these findings using longitudinally collected data within a large ASD clinical sample. We showed that early childhood onset epilepsy had an association with a later clinical record of ADHD diagnosis. Furthermore, by adjusting for known potential causes of epilepsy and ADHD (such as neurodegenerative, metabolic and other conditions listed in Table S1), our findings suggest that idiopathic epilepsy may be an important risk factor for the ADHD phenotype within ASD. As observational, this study cannot definitely prove that early onset epilepsy causes mild childhood ADHD. However, we hope these novel findings encourage future studies to further test this hypothesis, e.g. by comparing the risk of ADHD in well-controlled longitudinal cohorts of ASD children with and without epilepsy. Various theories have been proposed to explain the association between idiopathic epilepsy and ADHD, which include a shared genetic basis causing underlying brain abnormalities, deregulation of the noradrenergic system, a common psychosocial environment, the effect of seizures and antiseizure medication use [
33]. However, evidence focussing on the impact of other neurodevelopmental abnormalities in the occurrence of co-morbid epilepsy and ADHD is still limited [
34]. Lo-Castro and Curatolo [
8] suggest that there is evidence of a co-occurrence of an epilepsy/ASD phenotype or an epilepsy/ADHD phenotype that has a complex and heterogeneous pathogenesis, resulting from several altered neurobiological mechanisms involved in early brain development, and influencing synaptic plasticity, neurotransmitter transmission and functional connectivity. It is likely that rare clinically relevant genetic abnormalities, in addition to other social and environmental factors, may confer an increased risk for ASD/ADHD associated with epilepsy [
35,
36].
Other findings of interest include the hospitalisation rates of epilepsy in children with ASD, which was 3.7% of the sample diagnosed before the age of 7 years. Past studies suggest a lifetime prevalence between 2.7% and 44.4% [
37‐
39], thus our results are in the lower range relative to previous estimates. This is likely due to the narrow time window of early childhood to determine epilepsy diagnoses. In addition, we coded hospitalised cases of epilepsy only, thus excluded lower severity cases that may be exclusively managed in the community. Also, of interest is our finding that females were over twice as likely to have epilepsy as compared to males with ASD. This is in line with previous studies reporting that females with ASD tend to be more severely mentally disabled [
12]. Nonetheless our findings are consistent with a recent meta-analysis by Lax-Pericall, Bird [
40], which suggests that females may have a greater likelihood of shared ASD-epilepsy pathogenesis than males. Further, ID is also highly co-morbid with both ASD [
41] and epilepsy [
42], and ID is associated with an increased risk for both epilepsy and ASD.
Within our large clinical sample of 3237 children with ASD, 25% were diagnosed with ADHD after their 7th birthday. This is lower than findings from whole population Scandinavian register studies, which found within a sample of 28,468 individuals with ASD (age range 3–30), nearly 50% had ADHD [
43]. However our findings were consistent with similar paediatric age community samples and study observation periods for detecting diagnosis, where 31% of the sample met full ADHD criteria [
7]. This study also demonstrates that the medical risk factors associated with ADHD from the general population, including head-injury, perinatal complications, and endocrine/metabolic disorders, are consistent with the risk factors observed in the ASD population. Our multivariable model demonstrates that early onset epilepsy is a risk factor for later ADHD diagnosis even when these physical risk factors are controlled for.
When examining the demographics of the cohort, our model might suggest that belonging to non-White British ethnic groups or residing in more deprived areas may be associated with a lower risk of ADHD diagnosis. However, caution is warranted when interpreting these results, as ethnicity and socio-economic deprivation were included as confounders, and their estimates of effect should not be intended as casual. Further, these results could potentially reflect barriers in receiving health care. Further research is needed within diverse ethnic populations of ASD patients to explore how patient ethnicity and social background may affect diagnostic practices in CAMHS.
Strengths and limitations
This study has several strengths. First, we analysed a clinical cohort of children with neurodevelopmental disorders including both community and inpatient CAMHS settings; therefore, this represents a ‘real world’ clinical population and results may generalise to other clinical populations. Second, our large sample size led to more precision when assessing the associations between exposures and outcomes. Third, we were able to examine the association between childhood epilepsy and ADHD diagnosis in longitudinally collected data. Finally, we measured outcomes associated with epilepsy up to five years after the initial diagnosis of ASD, a longer observation period than that of other comparable studies.
Limitations should also be considered. First, the diagnosis of epilepsy and that of the associated physical risk factors were reliant on clinical observations within a hospital setting, rather than a gold standard diagnostic assessment. Second, we identified epilepsy cases from hospital records only, therefore we may have missed those exclusively investigated and managed in the community, e.g. with lower severity epilepsy. Similarly, we only included children with ASD that were open cases in our CAMHS, thus we may have potentially excluded more severely ill children that were diagnosed with epilepsy but died before a diagnosis of ADHD could be made. Hence the sample may not fully represent the full range of children with ASD who received a diagnosis of early onset epilepsy. However, under-recording children with epilepsy may increase the rate of false negatives, which could underestimate (not overestimate) the strength of the association between epilepsy and ADHD comorbidity. Third, within the HES data there is limited information on the severity and type of epilepsy syndrome, hence we could only provide an average effect of epilepsy on ADHD, but different epilepsy syndromes may have a differential association with ADHD. Fourth, despite this being one of largest cohort studies of ASD in childhood, it may have been underpowered to control for some of the rarer physical health conditions associated with epilepsy/ADHD. Fifth, as this is an observational study, residual (known and unknown) confounders could represent a potential explanation for the association between epilepsy and ADHD. For example, we were unable to include medication use or family history of ASD/ADHD among confounders and therefore to estimate their effects on our results. However, these effects are likely to be small. Prior findings show that an association between epilepsy and ADHD persists following adjustment for family history of psychiatric disorders and in medication naïve groups [
44]. For instance, a population-based case–control study showed that ADHD was 2.5-fold more common among children with newly diagnosed seizures than among controls [
45]. Furthermore, as anti-epileptic medications would be on the pathway from the exposure to the outcome, adjustment would pose the risk to underestimate the total effect of epilepsy on ADHD. Finally, it is possible that children with ASD and epilepsy may have been under increased clinical scrutiny, which may potentially lead to a greater awareness of their co-morbidities, hence an increased rate of ADHD diagnoses. Unfortunately, these potential biases in clinical classification cannot be fully accounted for.
Implications and future directions
Our findings suggest that children who present to mental health services with a history of early childhood epilepsy and ASD may also have ADHD. Clinicians should therefore be conscious of attributing a child’s hyperactive or inattentive behaviours to their ASD only, especially in the context of early childhood epilepsy. Early consideration of ADHD co-morbidity in children with ASD and early onset epilepsy could promote better detection and interventions, which are desirable as ADHD treatment in epilepsy is associated with improved outcomes [
46]. General practitioners/family physicians should also be aware of ADHD symptoms being more prominent during primary school, especially in children known to have epilepsy and ASD. Finally, more comprehensive assessments of medical history should be undertaken as early childhood epilepsy (especially if occurring and remitting in the first year of life) may not be volunteered by the patient or parent/carer without specific probes offered at initial consultation. Our study requires replication, ideally using both large scale retrospective cohorts based on electronic health records and prospective cohort studies within clinical and general populations. Longitudinal studies investigating brain development are needed to clarify the aetiological pathways to the combined ASD-ADHD phenotypes and their potential treatment targets. For example, it is still unclear what potential genetic components underpin ASD-ADHD epilepsy subtypes [
47]. Furthermore, future studies are needed to investigate whether antiseizure medication may play a role in increasing ADHD risk. To minimise the effects of confounding by indication, these studies should be conducted in an ASD-epilepsy only cohort and include detailed information on epilepsy type and severity. In sum, further longitudinal studies are needed to enable us to understand how epilepsy may be associated with attentional, hyperactive and impulsivity related impairments within ASD, and what treatments may help these impairments.