Discussion

In our cohort, patients with CHD between the ages of 3 and 21 are at high risk for neurodevelopmental disorders, particularly attention-deficit/hyperactivity disorder, autism, and intellectual disability. Using more thorough clinical methods (e.g., clinical diagnostic interview and neuropsychological assessment versus parent report or survey), these findings are generally consistent with past research. We found that in our clinically referred sample, children, adolescents, and young adults with CHD were five times more likely to be diagnosed with an intellectual disability (5.9%), which is consistent with rates of intellectual disability from a national survey of patients with CHD using parent report (5.3%). ^{Reference Razzaghi, Oster and Reefhuis8} In our clinical sample, 27% of youth with CHD were diagnosed with attention-deficit/hyperactivity disorder (odds ratio 4.37). This is higher than rates reported in a national sample of youth with CHD based on parent report (10.3%; odds ratio 1.6), ^{Reference Razzaghi, Oster and Reefhuis8} though it is very similar to studies using more rigorous clinical methods (e.g., 17.8–44% in studies using rating scales; ^{Reference Holst, Kronborg and Jepsen10–Reference Sistino, Atz and Simpson13} 16–34% in studies using diagnostic interviews ^{Reference DeMaso, Calderon and Taylor15–Reference Holland, Cassidy and Stopp17} ). Research generally suggests rates of autism are higher in children with CHD (2.6% based on parent report, ^{Reference Razzaghi, Oster and Reefhuis8} 3.2% based on parent ratings, ^{Reference Bean Jaworski, Flynn and Burnham19} 5.9% based on diagnostic evaluation ^{Reference Tan, Semmel and Wolf20} ), and our results demonstrated even higher rates of autism (9.7%; odds ratio 4.37). Taken together, this pattern of results suggests more rigorous diagnostic methods (i.e., neurodevelopmental or neuropsychological evaluation) are necessary to appropriately identify children with autism or attentional disorders and supports previous assertions that regular evaluation, and not screening, is necessary for children with CHD. ^{Reference Marino, Lipkin and Newburger25,Reference Ilardi, Sanz and Cassidy32,Reference Ware, Butcher and Latal33} Additionally, characteristics typically thought to indicate risk for worse neurodevelopmental outcomes, such as single-ventricle disease, were not associated with higher risk for neurodevelopmental diagnoses. This suggests that regardless of their specific cardiac diagnosis, all children and adolescents with high-risk CHD per American Heart Association and American Academy of Pediatrics guidelines should receive evaluation for neurodevelopmental disorders across the course of development.

Our study suggests that genetic disorders and history of aortic obstruction may be important risk factors for neurodevelopmental diagnoses associated with more substantial delays (e.g., intellectual disability and autism). Of note, many genetic disorders included in our sample are independently associated with higher risk of intellectual disability (e.g., 22q11.2 deletion), and there are a smaller number where intellectual disability is a core characteristic of the syndrome (e.g., Down syndrome). Consideration should be given to specific literature associated with genetic syndromes as appropriate, as these confer additional, specific risks. Unexpectedly, other medical factors often associated with poor outcomes, including single-ventricle CHD, prematurity, and history of seizures or stroke, were not associated with neurodevelopmental diagnoses in our study. Of note, we did not systematically collect or interpret specific neuroimaging findings in our patient group; as such, our information about neurological findings is limited. It is likely that subtler findings related to brain development, such as white matter injury or indicators of immaturities in brain development, may be more helpful in understanding these associations. However, our results remain relevant from a clinical decision-making standpoint, as this information is most often available even in lower-resourced centres. While literature suggests that prematurity is associated with developmental delay and cognitive dysfunction, its relation to specific diagnoses, including autism and attention-deficit/hyperactivity disorder, is inconclusive and dependent on other factors (including birthweight or severity of prematurity). ^{Reference Peralta-Carcelen, Schwartz and Carcelen34} It is likely that specific medical factors contribute to specific neuropsychological outcomes in children with CHD (e.g., executive dysfunction), or more subtle dysfunction, even if they do not affect whether a child meets diagnostic criteria for a neurodevelopmental disorder. For example, patients with two-ventricle CHD, ^{Reference Gerstle, Beebe and Drotar35} aortic obstruction and prematurity, ^{Reference Sanz, Berl and Armour36,Reference Sanz, Wang and Berl37} have been found to have poorer executive functioning on rating scales. Additionally, Cassidy et al. found that some executive functioning skills are similar across CHD groups, while visuo-spatially mediated executive functioning skills were impaired in patients with tetralogy of Fallot and single-ventricle CHD, but within normal range for patients with transposition of the great arteries. ^{Reference Cassidy, White and DeMaso38} Viewed in the context of this prior research, our findings suggest that patients with CHD are at risk for adverse neurodevelopmental outcomes regardless of the severity of their heart defect or other neurological risk factors. That said, the presence of a genetic disorder increases the risk for more substantial global delays (i.e., intellectual disability). Furthermore, CHD with aortic obstruction may also present increased risk to neurodevelopment, perhaps by impacting pre-natal cerebral blood flow and subsequent foetal and perinatal brain development. ^{Reference Sanz, Wang and Berl37} For example, research suggests that children with smaller aorta diameter at birth show differences in white matter development in deeper brain regions; these types of abnormalities have been associated with neuropsychological and neurodevelopmental outcomes (e.g., cognition, motor, executive functioning). ^{Reference Zaidi, Newburger and Wypij39,Reference von Rhein, Buchmann and Hagmann40} Additional research is needed to understand other medical and demographic risk and protective factors.

It is important to also highlight that rates of specific learning disorders were not necessarily higher in our clinical sample of children with CHD when compared to the general population, which may, on the surface, contradict prior research. That is, previous studies have shown high levels of academic service utilisation over time, with increasing rates as students get older. ^{Reference Riehle-Colarusso, Autry and Razzaghi41} Of note, in this clinical setting, neuropsychological evaluations did not always include full academic evaluations given limitations in insurance coverage or time (though they were often completed if parents endorsed academic concerns, or data were sometimes provided by school districts); therefore, rates may represent an underestimate of specific learning disorders in this population. Regardless of whether there are differences in rates of specific learning disorder, children with CHD have higher rates of attention-deficit/hyperactivity disorder, intellectual disability, and autism, which will likely impact academic performance. Additionally, other aspects of their neuropsychological profile (e.g., attention, executive function, language) may be important factors to consider for understanding academic outcomes. In this way, we might conceptualise children with CHD being at greater risk for more broadly termed “learning disorders,” or neurodevelopmental issues that affect academic learning, as opposed to specific learning disorders such as dyslexia, dysgraphia, or dyscalculia. ^{Reference Peterson, Boada and McGrath42}

Our findings also suggest that children within different age groups present with different neurodevelopmental profiles. Specifically, preschool-age children are more likely to receive diagnoses that reflect more global impairment (e.g., global developmental delays). Problems with later developing skills, such as attention, executive functioning, and cognitive skills needed for learning, may start to become more apparent as children enter school. Given this developmental trajectory, formal diagnoses such as attention-deficit/hyperactivity disorder and specific learning disorder may not be readily detected in early childhood. This pattern is also reflected in longitudinal research that suggests children may “grow into” certain deficits, which emerge over the course of development. ^{Reference Brosig, Bear and Allen43} This highlights the need for early monitoring for more globally impacted diagnoses. However, even if early evaluation is normal, continued evaluation over the course of development is needed as cognitive skills develop and as environmental expectations change as children grow. Furthermore, this suggests that researchers should work towards development of measures that can better predict later developmental outcomes (e.g., executive functioning screening, evaluation of phonological skills, and other predictors of academic skills) and that evaluations should go beyond formal diagnosis and include measurement of these specific skills. It is important that children with CHD continue to be evaluated for neurocognitive and neurodevelopmental problems throughout childhood and adolescence. Additionally, research suggests that up to 70% of children with CHD benefitted from neurodevelopmental evaluations (e.g., received a diagnosis, were referred for more testing, were referred for a service or therapy), which highlights the value of neurodevelopmental and neuropsychological evaluations in regard to differential diagnosis and treatment planning. ^{Reference Glotzbach, Ward and Marietta44}

It is important to note that our work examined diagnoses based on criteria in the Diagnostic and Statistical Manual of Mental Disorders. Identification of diagnoses is important, as it often leads to initiation of services and therapy and is a quick and efficient way to communicate with treatment providers in the community and schools to determine what services are needed. However, more subtle differences in functioning may be missed by looking at categorical data, and some neurodevelopmental outcomes in CHD may not be neatly captured by diagnostic criteria in the Diagnostic and Statistical Manual of Mental Disorders. Future research should look at continuous variables of neurodevelopmental outcomes and skills (e.g., neuropsychological profile, psychosocial, and quality of life measures). Looking at patterns of cognitive data and symptom report more broadly will provide better statistical power and allow for more nuanced understanding of risk and protective factors in CHD for neuropsychological factors that do not readily fit into diagnostic categories. ^{Reference Ilardi, Sanz and Cassidy32,Reference Ware, Butcher and Latal33,Reference Sanz, Berl and Armour36}

While our research contributes to understanding of neurodevelopmental outcomes in CHD, there are several study limitations that should be considered. First, our work was cross-sectional; additional longitudinal research is needed to understand possible developmental cascades and improve prediction of later outcomes. Additionally, our study was a retrospective chart review and included patients who were referred for and attended a neuropsychological evaluation. Though all families who meet high-risk criteria are referred for participation in neurodevelopmental follow-up, parents who had more concerns about their child’s neuropsychological functioning may have been more likely to attend visits, which may lead to an overestimate of diagnoses. Despite this, our prevalence rates were remarkably similar to those presented in community-based research using surveys or other less sensitive techniques and likely reflect rates that are seen in clinically referred samples. Furthermore, prevalence rates were compared to the National Health Interview Survey data. ^{Reference Zablotsky, Black and Maenner27} This comparison was selected due to its large sample size, inclusion of a similar age range of patients, and inclusiveness of several similar neurodevelopmental diagnoses. However, limitations of this comparison include its reliance on parent report of diagnoses, which may underestimate the prevalence of some disorders requiring more comprehensive assessment. Fortunately, prevalence rates from Zablotsky et al. are generally consistent with prevalence rates reported in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition. ⁶ Our research also evaluated data from one outpatient centre, which may limit the generalisability of findings; multi-site or multi-centre studies and data registries will allow us to improve generalisability and statistical power. These larger samples will also allow researchers to more thoroughly examine specific medical factors and broader neuropsychological profiles.

There were also limitations with medical risk factors. Our work did not include several important risk factors, including length of hospital stay or perioperative risk factors, and sample sizes for specific risk groups might not have been large enough to detect differences. Additionally, while perioperative MRI is currently standard of care, MRIs may not have been completed for some patients, particularly older teens, which may underestimate other neurological abnormalities (such as white matter injuries, immaturity, encephalomalacia, periventricular leukomalacia) beyond overt stroke or seizures. Using a wider variety of neuroimaging approaches (such as electroencephalogram [EEG], diffusion tensor imaging [DTI], MRI) will also help us better understand neurologic risk factors and merits investigation in future studies. More work is needed to better operationalise other medical risk factors, which will also require larger sample sizes. Furthermore, other factors, such as socioeconomic status, distance from hospital, or other barriers to care, including insurance, may limit access to evaluations and representation in clinical research. Future research should examine sociodemographic factors that may contribute to health disparities and neurodevelopmental outcomes. It will also be important for future research to examine modifiable risk and protective factors for neurodevelopmental outcomes, including access to early intervention services and family factors (e.g., parent stress).

Despite these limitations, our work has several clinical implications. First, clinicians completing neurodevelopmental evaluations should consider that children, adolescents, and young adults with CHD are at risk for several neurodevelopmental disorders, including autism, attention-deficit/hyperactivity disorder, and intellectual disability, and therefore, comprehensive evaluations should assess social cognition, attention and executive functioning, overall cognitive functioning, and adaptive skills. Additionally, clinicians should recognise that some diagnoses, such as attention-deficit/hyperactivity disorder and specific learning disorder, may not become apparent until later elementary school, when there is an increase in the workload, more expectations for independence, and higher demands for learning and executive functioning (e.g., focused attention, organisation, and planning). In addition, children, adolescents, and young adults with CHD are at high risk for neurodevelopmental disorders and as such should be referred for evaluation over the course of development if they meet American Heart Association and American Academy of Pediatrics high-risk criteria, ^{Reference Marino, Lipkin and Newburger25} regardless of specific cardiac diagnosis or medical risk factors. If psychologists and/or neuropsychologists are not readily available, clinicians are encouraged to partner with community agencies, such as early intervention programmes and schools, and with ancillary service providers (such as speech-language, occupational, and physical therapists), which can help provide some evaluation that will help identify children at risk and connect them with services, even in centres with fewer resources.

Neuropsychological evaluations are an important aspect of cardiac care. Our research demonstrates the importance of monitoring neurodevelopmental outcomes. Additionally, it is important for all children and adolescents to be assessed, regardless of the severity of their CHD, though children with aortic obstruction or genetic disorders may have greater risk for neurodevelopmental disorders. Additionally, children should continue to be monitored over time, even if early evaluation is normal, as some diagnoses may not be apparent until later in development. Data registries and multi-centre studies will improve our ability to understand risk factors for poor outcomes. Work that examines resiliency and protective factors will also allow for better models of care and targets for intervention.