Evidence of impaired neurocognitive functioning in school-age children awaiting cardiac surgery
Abstract
Aim Children with congenital heart disease (CHD) are at risk of developing neurocognitive problems. However, as these problems are usually identified after cardiac surgery, it is unclear whether they resulted from the surgery or whether they pre-existed and hence might be explained by complications and events associated with the heart disease itself. The purpose of this study was to examine whether neurocognitive deficits commonly reported after cardiac surgery are present before surgery.
Method Forty-five children (22 males, 23 females; mean age 11y 6mo, SD 3y 0mo) with cyanotic and acyanotic heart diseases scheduled for elective cardiac surgery were compared with 41 healthy peers (17 males, 24 females; mean age 11y 10mo, SD 2y 10mo) for attention and processing speed, construction, motor speed, motor planning and fluency, and visual memory. Twenty-three children in the patient group were awaiting their first cardiac surgery and 22 were awaiting follow-up surgery.
Results The patients showed manifest neurocognitive difficulties. Their performance was inferior to that of the healthy comparison group for motor planning (p=0.02) and visual memory (p=0.01). The same neurocognitive profile was found in the group of patients awaiting their first cardiac operation.
Interpretation School-age children with various forms of CHD are at risk of neurocognitive impairments before cardiac surgery.
Children with congenital heart disease (CHD) are at risk of neurological and developmental delays.1 Neurocognitive research has attempted to provide a detailed description of problem areas in the functioning of these children. Areas of motor function and perceptual organizational abilities, such as visuospatial and visuomotor integration skills, are most vulnerable.1,2 In addition, these children have been shown to experience difficulties with attention and speed of information processing, and with verbal functions.3,4 At school age, a substantial proportion of these children have been found to lag behind the population norms in academic achievement and to require special educational services.5
Neurocognitive problems may result from injury to the central nervous system (CNS) and in children with CHD such injuries have been attributed to processes associated with surgical treatment. The use of cardiopulmonary bypass, in particular, is known to be associated with embolic complications, hypoperfusion, ischaemic reperfusion injury, or inflammatory reactions.6 Besides intraoperative risk factors, postoperative haemodynamic instability and complications such as cardiac arrest and infection may also affect the CNS, thereby increasing the risk of neurocognitive problems.7
As most children with CHD need cardiac surgery shortly after birth, the majority of studies have measured neurocognitive functioning after cardiac surgery and have attributed the difficulties they observed to the surgical and associated procedures used. However, despite the risks known to be associated with cardiopulmonary bypass, a previous study of ours showed that open-heart surgery in school-age children did not affect neurocognitive functioning.8 Therefore, the question arose of whether the neurocognitive difficulties frequently reported after cardiac surgery were present before surgery. Another question was whether, apart from surgical interventions, complications and events associated with the CHD itself should be taken into account as a risk factor for neurocognitive problems.
After all, a malfunctioning heart may subsequently affect the CNS. First, the heart condition may be associated with cyanosis or haemodynamic instability leading to hypoxia, acidosis, poor nutrition, and inadequate cerebral perfusion.6 Further, the heart defects may coincide with multiple other congenital abnormalities or genetic syndromes involving cerebral abnormalities.6 Thus, irrespective of surgery, children with CHD may exhibit numerous neurocognitive problems.
Although the difference between neurocognitive problems due to processes associated with the heart disease and those resulting from the ensuing cardiac surgery can only be ascertained by assessment of the child’s preoperative neurocognitive functioning, few such studies have been undertaken. Those that have been undertaken typically investigated neonates and preschool children7 or used only global measures of intelligence.9 To our knowledge, the present study is the first to look extensively into the neurocognitive performance of school-age children with CHD before cardiac surgery. We looked for evidence to suggest that the neurocognitive problems may indeed be explained by processes associated with the CHD rather than with its surgical treatment.
Method
Participants and procedure
All patients aged between 6 and 16 years scheduled for elective open-heart surgery at the Radboud University Nijmegen Medical Centre, the Netherlands between June 2002 and June 2006 were eligible for participation (n=89). Children unable to speak and understand the Dutch language adequately were excluded (n=3), as were children with severe family problems (i.e. major parental illness, n=1). Moreover, to facilitate an unambiguous interpretation of the neurocognitive findings, i.e. as being associated with the heart disease rather than with any physical or cognitive comorbidity, children with diagnosed genetic syndromes, cognitive impairments (full-scale IQ<70), visual/hearing/speech/motor impairments that precluded valid and reliable neurocognitive assessment, or severe learning difficulties, were excluded (n=23).
Of the 62 patients who met the inclusion criteria, 45 children (and their parents) agreed to participate (73%). No difference was found between the participants and the non-participants for age (t test, p=0.30), sex (χ2 test, p=0.78), and number of previous operations (Mann–Whitney U test, p=0.30). The mean time between neurocognitive assessment and surgery was 2 months 9 days (SD 1mo 16d).
Apart from the neurocognitive functioning of the total patient group, we were especially interested in the subgroup of patients who had not previously undergone open-heart surgery (n=23). This is because we were investigating neurocognitive functioning associated with the heart disease rather than with cardiac surgery.
As a comparison group, 41 healthy peers were recruited from mainstream primary and secondary schools. The comparison group were matched with the patient group for age, sex, educational level, general intelligence, and parental educational level.
Participants were scheduled for their neurocognitive assessment at the medical centre after written informed consent was obtained from the child and parent(s) (children ≥12y) and parent(s) only (children <12y). All participants individually completed the same test battery administered by a trained psychology assistant under supervision of a research psychologist. The study was approved by the Regional Committee on Research Involving Human Subjects.
Neurocognitive assessment battery
As motor functions, perceptual organizational abilities, attention, and speed of information processing are most vulnerable in children with CHD,2–4 the assessment battery focused on these neurocognitive domains. All measures were suitable for the age group under study and had sound psychometric properties.10–17
General intelligence was assessed by means of the Dutch version of the Wechsler Intelligence Scale for Children – 3rd edition (WISC-III),10 and used to match the patients and the healthy comparison group. Three measures were derived: verbal intelligence, performance intelligence, and full-scale intelligence.
Attention and processing speed were tested with four different tasks: the Letter Detection subtest of the Amsterdam Neuropsychological Test Battery, the Complex Reaction Time Task, a computerized drawing task, and the Bourdon–Vos test for sustained attention.11–14 For the first three tasks, the mean reaction times were derived; for the last task, the mean row-completion time was analysed. To ensure that differences in speed would not result from differences in accuracy, we inspected the accuracy standard scores of the Bourdon–Vos test and found no difference between the patients and the healthy comparison group (χ2 test, p=0.09).
Construction, i.e. the ability to assemble the separate parts of a picture and to copy it as a coherent whole, was reflected by the organization score of the Rey–Osterrieth Complex Figure (henceforth referred to as Rey’s figure) and the standard score of the Developmental Test of Visual–Motor Integration.15,16
Motor speed was measured using a basic finger-tapping task, with the mean number of taps being calculated for both hands. We also calculated mean movement times on the Complex Reaction Time Task and mean drawing speed on the computerized drawing task.12,13
Motor planning and fluency were also assessed from the computerized drawing task.13 Stimulus patterns, consisting of alternating small and large apexes, were displayed on a computer screen. The children were instructed to start drawing the patterns on a sheet of paper as soon as they were able to reproduce them from memory. Motor planning was expressed by the mean duration of pauses between line segments, and movement fluency by the mean number of velocity peaks within line segments.
Visual memory measures were derived from the immediate and delayed recall conditions of Rey’s figure.15 For both conditions, the number of structural elements (indicating memory of the basic structure) and the number of incidental elements (indicating memory for details) in the reproduction were analysed.
Behavioural and emotional functioning were evaluated as problems in these areas might affect neurocognitive functioning. Three versions of the Child Behavior Checklist were administered: the Dutch Child Behavior Checklist, the Dutch Teacher Report Form, and the Dutch Youth Self-Report.17 Scores indicating internalizing, externalizing, and total problems were computed.
Statistical analysis
Presurgical neurocognitive functioning in general was analysed by comparing the results of the total patient group with those of the healthy comparison group. In addition, to eliminate the possible influence of previous cardiac surgery, we particularly contrasted the results of the children scheduled for their first operation with those of the healthy comparison group.
To test adequate matching of the patients with the healthy comparison group, t-tests and χ2 tests were applied on matching variables. Binomial tests were used to compare the patients’ behavioural and emotional function scores with population norms. Besides matching the patients and the comparison group at group level, we controlled for individual differences in general intelligence by including full-scale intelligence as a covariate.
For variables that were not normally distributed (Amsterdam Neuropsychological Test Battery, reaction time; Complex Reaction Time Task, reaction time and movement time; computerized drawing task, reaction time; Bourdon–Vos, row completion time) reciprocal transformations were applied and used for analysis. For each of the neurocognitive domains, except for visual memory, group comparisons were made by multivariate analyses of covariance. The results for visual memory could not be transformed into normal distributions and were therefore analysed by means of Mann–Whitney U tests.
The sizes of the total patient group (n=45) and the comparison group (n=41) were fit to detect medium effects with a power of 0.80. However, for the comparison of the subgroup of patients awaiting their first operation (n=23) with the healthy participants, the power might drop owing to the smaller sample size. Therefore, we chose to report effect sizes (Cohen’s d18 for multivariate analyses of covariance, and estimated θ19 for Mann–Whitney U tests) in addition to p values. All statistics were analysed using a two-tailed probability level of p<0.05 with the Statistical Package for the Social Sciences (version 12; SPSS Inc, Chicago, IL, USA).
Results
Table I shows the demographic data for the patient and comparison groups; Table II lists the patients’ diagnoses. As Table II shows, cyanotic as well as acyanotic heart disease was present. Because children with cyanotic heart disease were likely to have undergone previous cardiac surgery we assumed that, at the time of neurocognitive assessment, cyanosis would not influence the results.
| Total patient group (n=45) | First surgery patients (n=23) | Comparison group (n=41) | |
|---|---|---|---|
| Age, mean (SD) y:mo | 11:6 (3:0) | 11:5 (3:0) | 11:10 (2:10) |
| Sex, n | |||
| Male/Female | 22/23 | 12/11 | 17/24 |
| Education of child, n (%) | |||
| Primary | 25 (56) | 13 (57) | 21 (51) |
| Lower secondary | 13 (29) | 6 (26) | 13 (32) |
| Higher and pre-university | 7 (16) | 4 (17) | 7 (17) |
| Education of father, n (%) | |||
| Low | 13 (29) | 6 (26) | 12 (29) |
| Intermediate | 11 (24) | 8 (35) | 12 (29) |
| High | 21 (47) | 9 (39) | 17 (41) |
| Education of mother, n (%) | |||
| Low | 16 (36) | 10 (43) | 11 (27) |
| Intermediate | 19 (42) | 6 (26) | 15 (37) |
| High | 10 (22) | 7 (30) | 15 (37) |
| General intelligence of child, mean (SD) | |||
| Verbal intelligence | 99.1 (14.2) | 100.1 (13.2) | 103.0 (12.7) |
| Performance intelligence | 98.6 (11.5) | 98.6 (11.3) | 102.1 (11.4) |
| Full-scale intelligence | 98.9 (13.4) | 99.3 (12.5) | 102.9 (12.1) |
| Behavioural and emotional problemsa of child, n (%) | |||
| Parent report | (n=45) | (n=23) | |
| Internalizing problems | 4 (9) | 3 (13) | |
| Externalizing problems | 1 (2) | 1 (4) | |
| Total problems | 3 (7) | 2 (9) | |
| Teacher report | (n=43) | (n=22) | |
| Internalizing problems | 3 (7) | 3 (14) | |
| Externalizing problems | 0 (0) | 0 (0) | |
| Total problems | 1 (2) | 1 (5) | |
| Youth self-report | (n=25) | (n=13) | |
| Internalizing problems | 1 (4) | 1 (8) | |
| Externalizing problems | 0 (0) | 0 (0) | |
| Total problems | 0 (0) | 0 (0) | |
- aData from the Dutch version of the Child Behavior Checklist, showing the number of children with clinically significant problems; 10% of children in the general population show clinical problems on the Child Behavior Checklist.17
| Diagnosis, n (%) | Total patient group (n=45) | First surgery patients (n=23) |
|---|---|---|
| Atrial septal defect | 5 (11) | 5 (22) |
| Ventricular septal defect | 2 (4) | 2 (9) |
| Atrioventricular septal defect | 3 (7) | 1 (4) |
| Aortic stenosis and incompetence or aortic pathology | 15 (33) | 8 (35) |
| Mitral valvar anomalies | 3 (7) | 2 (9) |
| Ebstein malformation | 1 (2) | 1 (4) |
| Pulmonary atresia with ventricular septal defect | 2 (4) | 0 (0) |
| Pulmonary venous abnormalities | 2 (4) | 2 (9) |
| Tetralogy of Fallot | 4 (9) | 0 (0) |
| Multiple cardiac anomalies (including transposition of great arteries) | 8 (18) | 2 (9) |
The patient and comparison groups were adequately matched, as t tests and χ2 tests revealed no significant difference for the characteristics in Table I. In addition, according to the Child Behavior Checklist outcome measures, the patients’ behavioural and emotional functioning was comparable to population norms (binomial tests, p>0.05).
Table III shows the neurocognitive results for the patient and the comparison groups. As can be seen in Table IV, the total patient group performed significantly worse than the healthy comparison group on the domains of motor planning and fluency and visual memory. On the other domains, no difference was found between the patients and their healthy peers.
| Total patient group (n=45) | First surgery patients (n=23) | Comparison group (n=41) | |
|---|---|---|---|
| Attention and processing speed | |||
| ANT: reaction time, s | 1.44 (0.57) | 1.52 (0.64) | 1.22 (0.41) |
| CRT: reaction time, s | 0.38 (0.10) | 0.38 (0.09) | 0.35 (0.08) |
| Drawing task: reaction time, s | 1.22 (0.45) | 1.20 (0.41) | 1.09 (0.38) |
| Bourdon–Vos: row completion time, s | 17.28 (5.60) | 16.30 (4.83) | 16.26 (4.79) |
| Construction | |||
| Rey copy: organization score | 6.84 (3.61) | 6.96 (3.61) | 6.90 (3.02) |
| VMI: standard score | 92.76 (9.70) | 92.30 (9.39) | 93.90 (13.84) |
| Motor speed | |||
| Finger-tapping: taps preferred hand | 43.73 (7.24) | 43.98 (7.13) | 47.88 (10.76) |
| Finger-tapping: taps non-preferred hand | 37.66 (7.85) | 37.42 (9.00) | 40.91 (8.34) |
| CRT: movement time, s | 0.25 (0.07) | 0.24 (0.05) | 0.22 (0.05) |
| Drawing task: drawing speed, cm/s | 3.03 (0.96) | 2.89 (0.71) | 3.33 (0.96) |
| Motor planning and fluency | |||
| Drawing task: time between elements, s | 0.63 (0.30) | 0.63 (0.29) | 0.47 (0.23) |
| Drawing task: number of velocity peaks | 4.87 (0.67) | 5.01 (0.72) | 4.72 (0.76) |
| Visual memory | |||
| Rey immediate recall: structural elements | 19.11 (5.80) | 18.91 (6.40) | 21.85 (4.28) |
| Rey delayed recall: structural elements | 19.71 (5.27) | 19.57 (5.65) | 22.05 (3.94) |
| Rey immediate recall: incidental elements | 26.82 (8.29) | 28.57 (8.30) | 28.41 (5.46) |
| Rey delayed recall: incidental elements | 27.04 (7.87) | 28.43 (7.41) | 28.83 (5.36) |
- Values are means (SD). ANT, Amsterdam Neuropsychological Test Battery; CRT, Complex Reaction Time Task; Rey, Rey–Osterrieth Complex Figure; VMI, Developmental Test of Visual–Motor Integration.
| Assessment | Total patient group vs comparison group (n=45, n=41) | First surgery patients vs comparison group (n=23, n=41) | ||
|---|---|---|---|---|
| p valuea | Cohen’s d b | p valuea | Cohen’s d b | |
| Attention and processing speed | ||||
| ANT: reaction time | 0.44 | 0.45 | 0.11 | 0.61 |
| CRT: reaction time | 0.33 | 0.36 | ||
| Drawing task: reaction time | 0.31 | 0.28 | ||
| Bourdon–Vos: row completion time | 0.20 | 0.01 | ||
| Construction | ||||
| Rey copy: organization score | 0.98 | 0.02 | 0.97 | 0.02 |
| VMI: standard score | 0.10 | 0.13 | ||
| Motor speed | ||||
| Finger-tapping: taps preferred hand | 0.22 | 0.47 | 0.14 | 0.41 |
| Finger-tapping: taps non-preferred hand | 0.40 | 0.41 | ||
| CRT: movement time | 0.50 | 0.40 | ||
| Drawing task: drawing speed | 0.31 | 0.50 | ||
| Motor planning and fluency | ||||
| Drawing task: time between elements | 0.02c | 0.60 | 0.04c | 0.64 |
| Drawing task: number of velocity peaks | 0.21 | 0.39 | ||
| p value | Est θ | p value | Est θ | |
| Visual memory | ||||
| Rey immediate recall: structural elements | 0.01c | 0.34 | 0.04c | 0.35 |
| Rey delayed recall: structural elements | 0.03c | 0.36 | 0.07 | 0.37 |
| Rey immediate recall: incidental elements | 0.71 | 0.48 | 0.58 | 0.46 |
| Rey delayed recall: incidental elements | 0.51 | 0.46 | 0.77 | 0.48 |
- a p value, significance of multivariate analyses of covariance and Mann–Whitney U tests, respectively. aSignificant test results. bd, Effect size measure based on Cohen’s d: 0.2–0.49, small; 0.5–0.79, medium; ≥0.80, large. cEst θ: effect size measure based on U statistic: <0.29, large; <0.36, medium; <0.44, small; <0.50, no difference between groups. ANT, Amsterdam Neuropsychological Test Battery; CRT, Complex Reaction Time Task; Rey, Rey–Osterrieth Complex Figure; VMI, Developmental test of Visual-Motor Integration.
The univariate analyses of covariance (ANCOVA) of the motor planning and fluency data on the drawing task revealed that the patients had taken more time to plan the subsequent movement: they had paused longer between consecutive elements (ANCOVA, p=0.004, medium effect size d=0.60). The number of velocity peaks did not differ between the two groups, indicating comparable movement fluency (ANCOVA, p=0.31, small effect size d=0.21). The visual memory test showed that the patients’ recall of the structural elements of Rey’s figure was significantly inferior to that of the comparison group. No group difference was obtained for the recall of the incidental elements.
To eliminate the possible influence of previous open-heart surgery on the neurocognitive results, the results of the patients awaiting their first operation were compared with those of the healthy comparison group. Table IV again shows significant group effects for motor planning and fluency, and for visual memory. Moreover, the effect sizes for this specific subgroup were almost equivalent to the effect sizes for the total patient sample.
Discussion
Research has shown neurocognitive problems and learning difficulties in children with CHD after cardiac surgery.2–5 However, a previous longitudinal study of ours found no negative changes in neurocognitive functioning between preoperative and postoperative assessment.8 We therefore wondered whether the neurocognitive difficulties commonly reported after cardiac surgery would already be present before surgery. If so, the neurocognitive problems might be due to processes associated with the CHD, irrespective of its treatment. To explore this possibility, we conducted a thorough analysis of the neurocognitive outcomes of school-age children obtained before cardiac treatment. To our knowledge, the present study is the first to do so.
Relative to their healthy peers and general intelligence being controlled for, the cardiac patients showed neurocognitive deficits in motor planning and visual memory before cardiac surgery. As the measures used for these domains (i.e. the computerized drawing task and Rey’s complex figure) both require the encoding of visual patterns, the findings suggest that our patients had difficulty encoding figures as a coherent whole. The longer between-element pauses in the drawing task and the recall of less structural elements in Rey’s figure indicate that they tended to get more distracted by detailed information than their healthy peers, preventing them from recognizing the underlying main structure. The other performance results showed that the patients did not need more time to process information and that their constructional abilities and motor execution were comparable to the levels observed in the comparison group. The patients were thus able to accomplish relatively simple and structured visuomotor tasks, but encountered difficulties when tasks required more higher-order integration skills.
Our results support earlier studies describing deficiencies in motor functioning, perceptual organizational abilities, higher-order problem solving, and planning in paediatric cardiac patients.2,20 However, these studies all concerned deficits measured after cardiac surgery, whereas the present study demonstrated that some neurocognitive difficulties already exist before surgery. Hence these specific neurocognitive problems might be explained by processes associated with the heart disease itself rather than with its surgical treatment.
This assumption was verified by comparing only the children awaiting their first cardiac operation with the healthy comparison group. As their neurocognitive profile closely resembled the profile of the total patient sample, we conclude that all children with CHD should be considered at risk of neurocognitive difficulties, irrespective of cardiac surgery. This is in line with the findings of Limperopoulos et al.,7 who observed neurodevelopmental problems in neonates before surgery that generally persisted after surgery. The same research group recently reported that the neurodevelopmental problems even persisted to school entry and that children with all types of heart lesion requiring open-heart surgery appeared to be at risk.1 Our findings are also indirectly supported by Bellinger et al.,2 who demonstrated that, although visuospatial and visuomotor skills were the most prominent deficits after cardiac surgery, they were not strongly associated with characteristics of the surgical treatment used. Studies that did find associations between neurocognitive deficits and specific aspects of the surgical treatment only used postoperative measurements.3,20 Based on the present study, we suggest that the neurocognitive problems observed might already have been present before surgery.
CHDs may exist in combination with multiple other congenital abnormalities or genetic syndromes, which may involve neurocognitive impairments.6 In this study, we excluded all children with severe cognitive impairments, visual/hearing/speech/motor impairments, severe learning problems, and those with diagnosed genetic syndromes. However, as no genetic screening was performed on any of the patients, we cannot rule out a possible influence of unidentified genetic abnormalities.
The reported neurocognitive difficulties in children with CHD may result from processes associated with the heart condition itself. Reduced cerebral blood flow, oxygen deprivation, and altered brain metabolism have been shown to be associated with both cortical and subcortical lesions. Studies using magnetic resonance imaging in newborn infants with CHD identified structural brain anomalies before cardiac surgery, the most common being periventricular leukomalacia.21,22 In preterm children, periventricular leukomalacia has been linked to poorer functioning in the visuomotor and visuospatial domains and with problems in encoding new information into working memory.23,24 Our and the above-mentioned findings, therefore, suggest that the neurocognitive problems our sample exhibited before their cardiac surgery might be associated with cerebral anomalies due to the heart condition. Future research including detailed genetic screening and functional neuroimaging is needed to substantiate the genetic and/or haemodynamic nature of the neurocognitive deficits in children with CHD.
In addition to the medical risk factors described, merely growing up with a CHD may put children at risk for neurocognitive deficits. The illness and its treatment may be associated with physical limitations and frequent absence from school. Moreover, the illness may cause the parents to be anxious and overprotective towards the child. As a result, certain developmental experiences, such as physically exploring the environment and social interactions with peers, may be limited, thus impeding the child’s development.25 Further research is needed to differentiate between disease-specific and general effects of chronic illness on neurocognitive development.
Limitations
Some aspects of our study design merit further reflection. First, because most neurocognitive problems become manifest at school and extensive neurocognitive assessment is unfeasible in very young children, we specifically selected children scheduled for cardiac treatment in the 6 to 16 years age group, which, by definition, is a comparatively small pool. The small sample size might explain why for the patient subgroup, (i.e. the children without surgical histories), and the healthy comparison group the level of significance decreased. However, the inspection of the effect sizes in the subgroup revealed a neurocognitive profile comparable to that of the total patient sample. Second, the small sample size prohibited further group comparisons based on the aetiology of the heart condition. Third, to increase the likelihood that any neurocognitive findings would have resulted from processes associated with the heart disease rather than from any other diminished physical or cognitive capacity, we opted for very strict selection criteria. We are aware that, consequently, we examined a group of relatively well-functioning children, which limits generalization of the results to the entire patient population and complicates comparisons with previous research. It is highly likely that we would have found more neurocognitive deficits if we had adopted less stringent criteria. However, the fact that even relatively well-functioning children with various forms of CHD demonstrate specific neurocognitive difficulties before cardiac surgery once more underlines the importance of monitoring the neurocognitive functioning of all children with CHD.
Conclusions
The present study revealed specific neurocognitive deficits before cardiac surgery in a group of school-age children with various forms of CHD. The children investigated exhibited neurocognitive deficits in the areas of motor planning and visual memory, which seem to result from a tendency to focus on details rather than on the global structure, impeding their ability to encode information as a whole effectively.
Health professionals and teachers working with school-age children with CHD as well as the children’s parents should be aware of these potential neurocognitive deficits as early identification of such problems and targeted remedial actions may help prevent the children from falling (further) behind in school.
