ADHD symptomatology of children with congenital heart disease 10 years after cardiac surgery: the role of age at operation

All procedures performed in this study were in accordance with institutional guidelines and the 1964 Helsinki declaration and its later amendments. The study protocol was approved by the Hungarian Medical Research Council (approval number: TUKEB 190/2008; 696–1/2014/EKU 461/2014) and was performed upon participants and parents giving informed consent. Altogether, 133 children were examined: 54 patients with congenital heart disease undergoing first cardiac surgery under 3 years of age, 26 operated at the age of 3 or later, and 53 healthy controls. Detailed demographic data about the study groups are reported in Table 1.

The surgery group enrolled a convenience sample¹ of parents and 97 school-aged patients with congenital heart disease who underwent their first cardiac surgery with extracorporeal circulation between 1995 and 2012 at the Paediatric Cardiac Centre of the Gottsegen György Hungarian Institute of Cardiology in Budapest. Exclusion criteria were prenatal injury (n = 2), cerebral palsy (n = 3), severe hypoxic-injury during peripartum periods (n = 2), genetic syndromes (Down or DiGeorge syndrome) associated with cognitive deficits (n = 2), lack of parental or children consent (n = 1), the incapability of completing questionnaires (n = 1), premature birth (n = 2), lack of perioperative data (n = 2), and missing ADHD questionnaire data (n = 2) – resulting in a sample size of 80. In the first subgroup of patients undergoing cardiac surgery, 54 children were enrolled if having the first surgery under 3 years of age, while the second subgroup included 26 patients who received the first operation at or above 3 years of age. Patients underwent 174 operations in total: 130 operations in the subgroup operated below the age of 3 for the first time and 44 operations in the other subgroup (diagnoses and surgical procedures can be found in a supplementary table, Table S1). The psychological examination was conducted in 2013 and 2014 during ambulatory check-ups, with an average of 10.5 ± 3.3 years (51% male) after the first surgery.

The control group included a convenience sample of 53 children (mean age at survey completion 11.79 ± 3.65 years, 45% male) who underwent an ambulatory investigation in 2013 and 2014 at the same center because of accidentally detected heart murmurs but later were identified as healthy. No study participants were born with extreme low weight (< 2500 g) or had known intrauterine growth retardation either in the control or the two intervention groups. In the control group, a single patient had a formal diagnosis of ADHD and treated by psychotherapy. In the surgery subgroups, no patients were formally diagnosed with ADHD. No participants of the study received pharmacotherapy for ADHD symptoms.

The demographic and perioperative medical data were retrieved retrospectively from the institutional medical record. Medical variables included the length of the cardiac intensive care unit- (ICU) and overall hospital stay; length of the surgery; and length of extracorporeal circulation (ECC) needed while surgical interventions were made. Further, length of aortic cross-clamp use and the risk adjustment system for congenital heart surgery (RACHS) score were also reported. The former is a surgical instrument used during interventions without extracorporeal circulation, which separates the aorta from the outflow deriving from the heart excluding it from the systemic circulation, while the latter refers to the complexity of the operation (ranging from 0 to 6, where 6 represents the most complex form of surgery) [21].

The Hungarian version [22‐24] of the Child Behaviour Checklist (CBCL; to be completed by parents), the Youth Self Report (YSR; to be completed by children) and the ADHD Rating Scale–IV (to be completed by parents) were used to assess ADHD symptomatology. The Child Behaviour Checklist (CBCL) and the Youth Self Report (YSR) measure a wide-range of emotional and behavioral problems of children [25]. The CBCL is to be completed by parents of children aged 6–18, while the YSR is to be completed by youth aged 11–18 themselves. Both scales are comprised of items measuring the same 6 domains, out of which the Attention Disorders Subscale (CBCL_AD and YSR_AD) was used in the present study. Internal consistency for scores on the CBCL_AD was good (Cronbach’s alpha = 0.82), while that of scores on the YSR_AD was suboptimal but acceptable (Cronbach’s alpha = 0.63) in the present sample. In both cases, higher scores indicate more severe ADHD symptomatology. Diagnostic cut off scores for the CBCL_AD were used considering T scores of 65 (94th percentile) based on gender- and age specific norms [26]. Using this cut-off score, the prevalence of ADHD was estimated in our sample (Table 1). Even though this method is not equivalent to an expert diagnosis, these prevalence data (Table 1) can provide helpful information for the purpose of inter-study comparability of samples.

Beyond the total score (ADHD RS_TOT), the 18-item ADHD Rating Scale–IV [27] provides two subscale scores pertaining to inattention (ADHD RS_IA) and hyperactivity-impulsivity (ADHD RS_HI) symptoms. This assessment tool is appropriate for use regarding children aged 4–20. In this study, the ‘home version’ of the instrument was used as parents were the raters (in contrast to the ‘school version’ where teachers report on the children), who characterized their child’s behavior considering the 6 months prior to assessment. Internal consistency in the present sample was very high both for the two subscales (Cronbach’s alpha = 0.95 in both cases) and the total (Cronbach’s alpha = 0.97) scores. Higher scores indicate more severe symptomatology in this case as well. To provide an approximate prevalence of ADHD in the sample (Table 1) according to this assessment tool, diagnostic cut off scores for the ADHD RS_TOT were used corresponding to T scores of 64.5 (93th percentile) based on gender- and age specific norms [28].

Statistical analyses were performed using the SPSS 27.0 software (SPSS, Chicago, IL, USA). On the univariate level, chi square tests (for categorical variables), F-tests (for continuous variables where data were applicable for all three groups), and t-tests (for continuous variables where data were relevant for the two surgically treated groups only) were used to compare participants across surgery status (no cardiac surgery, first cardiac surgery under 3 years of age, first cardiac surgery at or above 3 years of age).

On the multivariate level, analysis of covariance (ANCOVA) was used to investigate the relationship between surgery status and the five indictors of ADHD symptom severity (YSR_AD, CBCL_AD, ADHD RS_IA, ADHD RS_HI, ADHD RS_TOT), used as separate dependent variables. In addition to the omnibus tests, post-hoc pairwise comparisons were also performed to provide a more nuanced picture of the differences among the study groups. All multivariate analyses (including the post hoc tests) were controlled for sex and age at survey completion. To estimate statistical power of the multivariate models, post hoc power analysis was also carried for the main independent variable of interest (surgery status).

In case of both bivariate and multivariate analyses, estimates of effect size were also presented. We considered 0.2 as a threshold for small effect, 0.5 for moderate effect, and 0.8 for large effect in case of Cohen’s d; and 0.01 as a threshold for small effect, 0.06 for moderate effect, and 0.14 for large effect in case of η² [29]; while the corresponding thresholds for Cramer’s V were 0.05, 0.10 and 0.15, respectively [30].

Demographic, perioperative and ADHD-related characteristics of the sample are described in Table 1. There was no statistically significant difference among the three study groups in terms of participants’ sex. However, those undergoing their first cardiac surgery later in life were older at survey completion (medium effect size) than members of the control group or who were operated at or below the age of 3 for the first time. Between the two surgically treated groups, there were statistically significant differences in the length of hospital stay (medium effect size), the number of operations (medium effect size), and the frequency of perioperative cyanosis (large effect size); in all three cases, values were higher (less favorable) for children who were first operated below 3 years of age (Table 1).

In terms of ADHD symptomatology, those undergoing surgery later in life had higher ADHD symptom scores across all five indicators than the other two subgroups; however, the difference did not reach statistical significance in case of the CBCL_AD (Table 1). The magnitude of the differences was small (CBCL_AD) or moderate (YSR_AD, ADHD RS_IA, ADHD RS_HI, ADHD RS_TOT). In terms of the prevalence of diagnosable ADHD symptom severity, those undergoing surgery later in life had higher prevalence rates than the other two subgroups. The difference did not reach statistical significance in case of the CBCL_AD (although the effect size indicator suggested a moderate difference), while in the case of the ADHD RS_TOT, the difference was statistically significant and the effect size was large (Table 1).

Results of the omnibus tests from the multivariate analyses indicated that surgery status was significantly associated with most indicators of ADHD symptomatology even after controlling for sex and age at survey completion, with effect sizes in the small to medium range (Table 2). Similar to the bivariate analyses, the only ADHD symptom severity indicator regarding which surgery status was not a statistically significant correlate was the CBCL_AD (Table 2). Observed (post hoc) power for the surgery status variable ranged from 0.43 (CBCL_AD) to 0.75 (YSR_AD).

Regarding all ADHD symptom severity indicators, the post hoc tests indicated that the symptoms of those treated surgically at or above 3 years of age were more severe than not only the control group but also of those who were treated surgically at a younger age; and with some exceptions, these differences were statistically significant (Table 3). Interestingly, the control group and those treated surgically under 3 years of age did not differ significantly across any of the five ADHD symptom severity indicators according to the post hoc tests.

Considering the non-trivial differences in standard deviation of some ADHD indicators across study groups (Table 1), the multivariate analyses were also rerun on a trimmed data set eliminating outliers, using the criterion of a z score > |3.29|. In this case, surgery status was only associated with YSR_AD scores (F = 3.126, p = 0.047, η² = 0.047) but not with the other four ADHD symptomatology indicators (CBCL_AD: F = 1.093, p = 0.339, η² = 0.017; ADHD RS_IA: F = 0.777, p = 0.462, η² = 0.012; ADHD RS_HI: F = 1.887, p = 0.156, η² = 0.029; ADHD RS_TOT: F = 1.720, p = 0.183, η² = 0.027).

Given the importance of sex differences in the phenomenology of ADHD, further multivariate analyses were completed (in the untrimmed data set) to test the moderator role of sex: an interaction term of sex and surgery status was added to the default models. This interaction term was significant in the model for the YSR_AD (p = 0.031) and the CBCL_AD (p = 0.036); while non-significant for the model for ADHD RS_IA (p = 0.073), ADHD RS_HI (p = 0.355), and the ADHD RS_TOT (p = 0.284). Considering the significant interactions, sex-stratified analyses were also conducted for scores on the YSR_AD and the CBCL_AD. These analyses indicated that surgery status was significantly associated with YSR_AD (F = 6.248, p = 0.003, η² = 0.159) and CBCL_AD (F = 4.441, p = 0.016, η² = 0.122) scores in girls but not in boys (YSR_AD: F = 0.659, p = 0.521, η² = 0.022; CBCL_AD: F = 1.040, p = 0.360, η² = 0.035).

The worldwide prevalence of ADHD in pediatric populations is approximately 5.29% [31]. ADHD is often co-morbid with behavioral disorders (36–40%), major depressive episodes (18%), anxiety problems (26%) [32] and learning disabilities (30%) [33]. Quality of life of affected patients is lower, even if in half of the cases, ADHD symptoms disappear upon reaching adulthood [8, 34].

The aim of this study was to shed light on the long-term consequences of pediatric cardiac surgery in terms of attention-deficit and hyperactivity symptoms. To the best of our knowledge, this is the first study reporting an age-related comparison of patients with congenital heart disease undergoing cardiac surgery and a healthy population. The results indicated a significantly higher severity of attention deficit and hyperactivity symptoms in children who underwent cardiac surgery at or above the age of three compared to non-operated individuals or even children operated at a younger age, an outcome which was largely consistent across assessment tools and data providers (patient- vs. parent-report). Interestingly, the subgroup operated at a younger age underwent a larger number of operations and was more frequently diagnosed with a cyanotic malformation, which suggests that factors other than cyanotic malformation or number of operations (cf. repeated anesthetic exposure) should explain the more severe ADHD symptomatology observed in the subgroup operated later in life.

A potential explanation for our findings is that repair mechanisms and neuronal plasticity in newborns and younger children are more efficient than in children of older ages. Brain development consists of a cascade of different stages composing of cell birth, cell migration, differentiation, maturation (dendrite and axon growth), synaptogenesis, cell death, synaptic pruning, and myelogenesis – all following an age-dependent pattern [20]. Due to the uncertainty in the number of neurons reaching their predicted and appropriate destination, the brain overproduces both neurons and synapses with a peak of formation occurring between the ages of one and 2 years [20, 35]. During this stage, unneeded or damaged neurons are removed by synaptic pruning and cell death, while stem cells lining the subventricular zone are activated and parallel systems initiate to form new cell connections. Additionally, dendrites and spines show particularly high plasticity in response, forming synapses in just a few hours after injury [20]. Due to these facts, developmental outcomes from different injuries will vary according to the exact stage of development [35, 36] with individuals below the age of three having better chances for full recovery. Further research is needed to further confirm these explorative findings suggesting age-dependent behavioral outcomes in relation to pediatric cardiac surgery.

It is also worthy of note though that the results of the primary analyses of this study were not particularly robust: the sensitivity analyses indicated that the trends observed in the primary analyses emerge more consistently among females than males and that outliers might be responsible for a disproportionately large amount of variance. These considerations suggest that further investigations with a more explicit focus on sex differences and larger samples need to confirm the findings of this exploratory study and investigate potential moderators of the relationship between age at cardiac surgery and ADHD symptomatology.

A major strength of this study was the consideration of both self- and collateral reports on the same construct (ADHD symptomatology) using multiple, parallel assessment tools. While overall the findings were homogeneous, certain differences did emerge across assessment tools (e.g., difference in effect sizes, prevalence of diagnosable severity of ADHD symptomatology), which indicates that future studies with a similar focus should carefully select which measure(s) to use considering, for instance, the goals of the study (e.g., focus on inattention versus the full range of ADHD symptoms) or the statistical power intended to achieve.

Limitations of this study need to be acknowledged as well. First, the sample was relatively small limiting our ability to reliably detect small differences among the study groups or conduct adequately powered sex-stratified analyses. This shortcoming of the study was also highlighted by the suboptimally low (< 0.80) post hoc power values. Further, the convenience nature of the sampling procedure limits the generalizability of the findings and reduces the strength of our hypothesis regarding neural plasticity. For example, it is possible that the subsample of those treated surgically at a younger age had a more favorable ADHD symptom profile in our study compared to those operated on later not because of more effective neural repair mechanisms but because only the most resilient individuals from the first group survived their early years (cf. the higher number and complexity of cardiac operations and more frequent cyanotic malformation in this group suggesting a higher risk of death) raising the possibility of selection bias.²

Another sampling-related limitation of the study design was the heterogeneity of the sample, that is, congenital heart disease was conceptualized as a homogenous disease despite the quite diverse conditions this umbrella term covers. The large variability in medical procedures used to correct this diverse group of medical conditions makes it challenging to design studies with adequate sample size while focusing on a single condition or surgical procedure. Most likely not independently of this fact, the omnibus approach used in the present study is not unique and was also used in other studies investigating the psychological aspects of congenital heart diseases or their treatment [37, 38].

Yet another limitation of the present study is that certain surgical interventions – from surgical technology to treatment guidelines for surgery and postoperative clinical management – have evolved considerably since the study period; and therefore, our findings might be less generalizable to clients receiving treatment according to these more recent protocols. Further studies following the most recent treatment guidelines and focusing on a single or smaller group of congenital heart diseases and their treatment could contribute to a more nuanced understanding of how congenital heart diseases and the surgical procedures aiming to correct them influence neurocognitive development.

A further limitation of the study was the lack of access to detailed medical information on the prenatal and the peripartum periods (e.g., intrauterine growth retardation, smoking behavior during pregnancy and other teratogenic factors), which could also have influenced the cognitive and behavioral development of the offspring [39‐41]. Data on families’ socio-economic status, life style factors, or parents’ psychological characteristics were not collected either, all of which may also contribute to the development, prognosis, or reporting of ADHD symptoms [42, 43]. In addition, the present study did not include any detailed genetic analyses either, thus the overlap with non-cardiac congenital conditions presenting only later in life (post-surgery) also remained unknown. Considering these variables in future research could increase the reliability of findings in studies with a similar focus. Finally, due to the methodology of the current study, we cannot rule out the possibility that the reported ADHD symptoms could have been better explained by mental health conditions other than ADHD (e.g., anxiety, depression); using diagnostic evaluations by mental health experts could have contributed to a more reliable and valid ADHD assessment protocol.

The results of the present study raise the possibility that the age of children at cardiac surgery has a prognostic role in the emergence of a future attention disorder. The clinical relevance of routine ADHD screening and early diagnosis in pediatric cardiac rehabilitation could help both patients and their social environment to recognize early signs and prevent severe consequences of ADHD such as poorer school performance, social isolation, depression, or substance abuse. Our study also calls attention to the importance of the large variety of factors potentially influencing behavioral outcomes in pediatric cardiac populations (number of surgeries, possible surgical complications, type of malformations etc.), which variables should be made easily accessible to health care providers in (electronic) health records to facilitate both high-quality clinical service delivery and research.

The results of the present study suggest that the negative psychological effects of pediatric cardiac surgery might not be equally salient in different ages and thus certain age groups might be more at risk to develop ADHD symptoms after cardiac surgery. This notion, on the one hand, could help better target ADHD screening, assessment, and treatment procedures in pediatric populations, while on the other hand, it raises the possibility that other cardiac-surgery-related psychological outcomes are also dependent on the timing of cardiac interventions, which possibility deserves further investigations.