Abstract
Aims
To evaluate the relative effect of cyanosis, surgical interventions and family processes on neuropsychological and behavioural outcomes in 4‐year‐old survivors of serious congenital heart disease (CHD).
Methods
90 children with a range of cyanotic and acyanotic conditions, who underwent either corrective or palliative surgery, completed a neuropsychological and behavioural evaluation. Families of participants were also profiled by evaluation of maternal mental health, worry, social support, parenting style and family functioning.
Results
Compromised neuropsychological outcomes were associated with a combination of cyanotic conditions and open‐heart surgery, but this was not exacerbated by having a complex, palliative, status. Both cyanotic and acyanotic conditions were associated with specific sensorimotor delays, regardless of method of the correction. Only children with complex conditions and palliative interventions seemed at risk of poor behavioural outcomes; indeed, children with cyanosis with complete repair showed favourable behavioural outcomes compared with controls. Multivariate analyses highlighted the sometimes greater relevance of family processes (eg parenting style, maternal mental health and worry), rather than disease or surgical factors, in predicting especially behavioural outcomes.
Conclusions
The findings (1) suggest a more complex relationship between cyanosis, surgical methods of correction, neuropsychological and behavioural outcomes than previously charted, (2) highlight that family processes may be aetiologically more important than disease and surgical factors, and (3) indicate specific targets for secondary prevention programmes for this at‐risk population.
An emerging view suggests that children with cyanotic congenital heart disease (CHD) who have undergone surgical procedures involving circulatory arrest are at particular risk of both neuropsychological and behavioural problems.1,2,3,4,5 However, some studies have found no such evidence or indeed, in terms of behavioural adjustment, suggested better outcomes than controls.6,7,8 Moreover, where statistically significant differences have been found, clinical significance is sometimes dubious. Indices of intellectual functioning and behavioural adjustment in some studies have often been well within the normal range, with statistical significance sometimes determined by above‐average scores in control groups.3,9,10
Contributing to this ambiguity, studies have often used restricted samples of children with CHD (eg cases of transposition of great arteries) and have neglected to examine the mediating effects of maternal and family factors, which can covary with CHD.6,8,11,12,13 This study includes measures of maternal adaptation and family functioning, in addition to cyanosis and surgical factors, in examining determinants of neuropsychological and behavioural outcomes in 4‐year‐old survivors of severe congenital heart defects. Children with cyanosis are divided into subgroups of those who had corrective versus palliative surgery only. These subgroups have often been merged in previous studies, but important differences are likely. Children without acyanosis are also subgrouped into those who underwent open or closed surgery. In this way, the importance of surgical procedures involving cardiopulmonary bypass, independent of cyanosis, may be considered. Neuropsychological and behavioural profiles across groups are first considered, followed by multivariate analyses of outcome determinants.
Methods
Participants and groups
Four‐year‐old children and their families were recruited over a 3‐year period from a regional centre. This age group was chosen, as some families later participated in an early intervention trial to promote child and family adjustment at a key developmental period (ie start of formal education). Eligibility criteria included having undergone at least one invasive procedure for correction or palliation of a major heart defect. Children with developmental or psychiatric syndromes were excluded. Of the 149 families invited, 90 agreed to participate. Participants did not differ from non‐participants in terms of age, cyanotic status, surgical procedure, correction or palliative status, or socioeconomic status as measured by Townsend Index scores.14 All were assessed in the 36 months before starting formal education. Mean age at time of assessment was 4.6 years (standard deviation (SD) 0.3; range 4–5.1 years) and mean time since main operative procedure was 3.8 (SD 0.9) years.
Four groups were defined by diagnosis and surgical procedure as follows:
Acyanotic‐closed (n = 34): acyanotic conditions corrected by closed‐heart surgery or an interventional catheterisation (eg coarctation of aorta)
Aycanotic‐open (n = 25): acyanotic conditions corrected by open‐heart surgery (eg ventricular septal defect, atrial septal defect)
Cyanotic‐corrected (n = 19): cyanotic conditions corrected by open‐heart surgery (eg transposition of great arteries, tetralogy of fallot)
Cyanotic‐complex (n = 12): cyanotic conditions and combinations of conditions for which only palliative open‐heart surgery was possible (eg hypoplastic right and left heart conditions, double inlet left ventricle).
The four groups did not differ significantly in terms of age, sex, weight, parental marital status, socioeconomic status, parental qualifications or family size (supplementary table A at http://adc.bmjjournals.com/supplemental). As expected, the cyanotic‐complex group had had more surgical procedures than the other groups (supplementary table A).
For analyses involving behavioural outcomes, an additional mild control group (n = 19) of children who had been diagnosed with mild, self‐correcting, heart defects (eg small ventricular heart murmurs and patent ductus arteriosus) in infancy were recruited. Although they were slightly older than the clinical groups at the time of assessment (mean age 5.1 (SD 0.3) years; range 4.9–5.2 years), exploratory analyses confirmed that age did not exert an independent effect on behavioural outcomes. They did not differ from the other groups in terms of the other demographic variables described earlier. For analyses involving neuropsychological outcomes, the acyanotic‐closed group (presumed to be without major neurological risk as neither cyanosis nor open‐heart surgery applied) served as a control group for statistical analyses, in addition to the comparisons made with the clinical norms for the scales used.
Recruitment and study procedures were approved by the local research ethics committee for the region, and families were recruited by letters of written informed consent.
Outcome measures
Summative IQ scores may be misleading by not detecting deficits in specific cognitive processes or domains. Consequently, five neuropsychological domains were assessed using factor scores from the complete 3–5‐year‐old version of NEPSY–A Developmental Neuropsychological Assessment.15 These included attention, language, sensorimotor, visuospatial and memory skills. Five additional domains were assessed using specific subscales from the Wechsler Preschool and Primary Scale of Intelligence (WPPSI‐R.UK).16 These included (with scale names added for reference) verbal reasoning (similarities), non‐verbal reasoning (block design), social reasoning (comprehension), cognitive speed (coding) and number skills (arithmetic). NEPSY factor scores have a reference group mean of 100 (SD 15), and WPPSI subscale scores have a reference group mean of 10 (SD 3). Tests were conducted with child participants on a one‐to‐one basis by chartered clinical psychologists, trained in neuropsychological evaluation and blinded to the diagnostic status of participants at the time of assessment.
Behavioural functioning was assessed by scores on the maternally completed Child Behaviour Checklist (CBCL).17 This study used the composite total problem behaviour index score from this screening inventory for behaviour problems. The percentage of each group who fell within the clinically significant range of the scales was also recorded.
Predictor factors
The effect of cyanosis, surgical interventions, palliative status and combinations thereof was assessed through group comparisons as described above. Length of time on cardiopulmonary bypass was additionally recorded from surgical records. In addition, family factors were assessed by having mothers complete several scales. Maternal mental health was assessed using the Brief Symptom Inventory.18 The composite “general severity index” score was used in analyses here. Maternal worry about the child's health status—which previous research has suggested may vary independently of actual illness severity4—was assessed using the Maternal Worry Scale.19 The Parenting Locus of Control Scale20 was used to profile the parenting style. Four of the five subscales were used to measure the degree to which parents hold beliefs and practice parenting behaviours that are likely to be effective in terms of child behaviour management and adjustment. Family functioning was assessed using selected subscales from the Family Environment Scale.21 Four subscales (45 items) were used which rated familial cohesiveness, conflict, expressiveness (open and direct communication styles) and extent of active or recreational family activities. Finally, social support available to the family was evaluated by the Significant Others Scale.22 This measure assesses the quantity and perceived quality of available social support, and two indices were used that reflect the degree of “emotional” and “practical” social support.
Socioeconomic status was measured as described earlier,14 and additional family and demographic factors (highest parental qualification, marital status, number of siblings, age and sex) were reported by mothers before psychometric assessment.
Statistical analysis
The data satisfied statistical assumptions for parametric analyses. Multivariate or univariate analyses of variance were used to compare group means on neuropsychological and behavioural outcomes for statistical significance. Effect size deviations from reference means were used to assess the clinical relevance.23 Half a SD below or above reference group means was taken as indicative of a medium effect size and three quarters high.23 The selection of predictor variables entered into the multivariate analyses was determined by preliminary exploratory analyses, with p<0.05 used as the threshold for multivariate inclusion. Power was calculated on the basis of 0.75 standardised mean deviation representing clinical relevance on both the neuropsychological (WPPSI and NEPSY) and behavioural (CBCL) measures. For the four‐group neuropsychological comparisons, this required a sample size of 76 to attain power of 0.8, and a sample size of 80 for the five‐group behavioural comparisons. The actual sample included was 90.
Results
Neuropsychological outcomes
Table 1 summarises the mean (SD) scores and clinical effect sizes, on the 10 neuropsychological domains outlined above. As cohort means for “normals” on these scales have tended to increase by several points across time,24 the extent of these deviations from clinical norms (from 1998 for NEPSY and 1989 for WPPSI) is likely to be a conservative estimate. All four CHD survivor groups showed medium to high effect size deviations from reference group norms on the sensorimotor dimension. Otherwise, both acyanotic groups, regardless of cardiopulmonary bypass surgery, had all other neuropsychological competencies within the normal range.
Table 1 Mean neuropsychological domain scores (standard deviation) across the four congenital heart disease survivor groups.
| Dimension | Acyanotic‐closed (n = 34) | Acyanotic‐open (n = 25) | Cyanotic‐corrected (n = 19) | Cyanotic‐complex (n = 12) |
|---|---|---|---|---|
| Attention* | 101.7 (9.7) | 98.5 (11.4) | 99.1 (10.6) | 95.4 (18.2) |
| Language* | 98.9 (15.3) | 101.8 (14.1) | 97.3 (11.6) | 92.5 (11.6)† |
| Sensorimotor* | 89.7 (16.8)† | 89.2 (14)† | 80.6 (12.8)‡ | 86.1 (14.5)‡ |
| Visuospatial* | 101.3 (15.6) | 100 (13.1) | 92.4 (14.5)† | 96.2 (10.4) |
| Memory* | 95.3 (15.7) | 100.7 (12.2) | 94.9 (11.1) | 96.1 (14.4) |
| Verbal reasoning§ | 8.7 (3) | 9.6 (1.7) | 7.8 (2.5)‡ | 8.8 (3.4) |
| Non‐verbal reasoning§ | 9.7 (2.6) | 9 (2.5) | 8.1 (3.3)† | 9.6 (2.4) |
| Social reasoning§ | 8.8 (2.5) | 9.9 (1.9) | 8.1 (2.3)† | 8.7 (2.3) |
| Cognitive speed§ | 8.7 (2.7) | 8.7 (2.6) | 9.6 (3.1) | 9.6 (2.7) |
| Number skills§ | 8.8 (2.7) | 9.3 (2.3) | 7.9 (2.9)† | 7.9 (2)† |
*Developmental neuropsychological assessment (NEPSY) tests: reference group mean 100 (SD 15).
†Study group mean is at least half a SD below reference group mean, defined as a medium effect size.
‡Study group mean is at least three quarters of a SD below reference group mean, defined as a large effect size in terms of clinical significance.25
§Wechsler Preschool and Primary Scale of Intelligence (WPPSI) tests: reference group mean 10 (SD 3).
The cyanotic groups, both of whom underwent open‐heart surgery, showed a greater number of clinically relevant effect size deviations—on six of the domains in the cyanotic‐corrected group and on three domains in the cyanotic‐complex group. No evidence was suggested, therefore, that neuropsychological outcomes are worse in children with cyanosis with palliation only. Although the data suggest a clinical pattern of increasing problems being associated with cyanosis, regardless of corrective status, multivariate analyses of variance did not indicate significant differences among the four groups across all 10 dimensions together (Pillai's trace F = 1.03; p = 0.4).
For multivariate analyses, two neuropsychological factors (accounting for 56% of the variance) were extracted by principal component analysis: (1) a perceptual‐motor skills factor included non‐verbal reasoning, sensorimotor, visuospatial, cognitive speed and number skills; (2) a verbal skills factor included social reasoning, memory, verbal reasoning and language dimensions. Two multivariate analyses were subsequently conducted, with cyanotic status (cyanotic/acyanotic) and surgical procedure (open/closed) entered as primary variables of interest in both. The univariate criteria for inclusion of other variables are described above, and checks were completed to ensure that no other factors became statistically significant once these were accounted for (supplementary table B). Considering both regressions together, four factors emerged as important determinants for neuropsychological outcomes (supplementary table B). These were paternal education, sex, surgical procedure and a parenting style variable, parental efficacy. Being male and having a father with relatively reduced educational qualifications were the greatest risk factors for poorer performance on the perceptual skills dimension. Open‐heart surgery and low efficacy in parent management skills were risk factors for poorer performance on the verbal skills dimension.
Behavioural outcomes
For behavioural outcomes, data from a mild‐control group were available and included as described above. Table 2 summarises the mean (SD) t scores, together with the percentage in each group whose t scores on the total problem index were in the range of likely clinical significance (t>60).17
Table 2 Group outcomes on the Child Behaviour Checklist Total Problem Index Score.
| Scale | Acyanotic‐closed (n = 34) | Aycanotic‐open (n = 25) | Cyanotic‐corrected (n = 19) | Cyanotic‐complex (n = 12) | Mild‐control (n = 19) |
|---|---|---|---|---|---|
| Total Problem Index Score | 53.7 (10.9) | 52.3 (13) | 45.4 (7.8) | 54.2 (9.7) | 45.6 (13.6) |
| Percentage in clinically significant range | 21 | 20 | 5 | 33 | 16 |
Values are mean (SD).
The frequency of clinically relevant t scores in the mild‐control group in this study is comparable to those referenced for “normals” in the standardisation sample.17 Frequencies are comparable in both acyanotic groups, but seem to vary between the cyanotic groups—being twice as high in the cyanotic‐complex group and reduced in the cyanotic‐corrected group. An overall significant difference was seen in mean scores across groups (F = 2.95; p = 0.02), although retrospective Tukey tests did not suggest any significant pairwise comparisons (p>0.05) to indicate the specific source of this difference. Inspection of means and clinical frequencies suggest that the most likely interpretation is of comparable, or reduced, levels of behaviour problems in the cyanotic‐corrected group compared with the mild‐control group, and raised levels in the other groups, but most clearly in the cyanotic‐complex group.
Group comparisons clearly do not suggest a simple additive effect of cyanosis and surgical procedure as risk factors for poorer behavioural adjustment. A multivariate analysis, which incorporated these disease and surgical factors, but which also included the family and cognitive factors described above, was conducted. This was conducted for the four study groups of interest only because (1) relative comparisons of disease, surgical and maternal worry about illness did not apply to the mild‐control group and (2) it was the relative importance of these disease and surgical factors versus psychosocial factors that was of interest. As the preliminary univariate analyses suggested that many more psychosocial factors contributed to behavioural outcomes than was found for the neuropsychological outcomes, a backward regression analysis was conducted in this case to derive the most parsimonious final regression model. The final model indicated that five factors were statistically significant; together, these accounted for 59% of the variance on behavioural outcomes (supplementary table C). These were parenting style (parental control), marital status, maternal worry, cyanotic status and maternal mental health. Specifically, risk factors for poorer behavioural adjustment included poor parental control skills, lone parent status, high maternal worry about the child, and raised levels of maternal psychological symptoms. Although cyanosis was associated with poorer neuropsychological outcomes, acyanotic status was more of a risk factor for behavioural outcomes. Neither surgical procedure nor length of time on bypass contributed to outcomes here.
Discussion
Group comparisons
Group profiles provide some evidence that children with cyanosis who have undergone open‐heart surgery are at greater risk of generally poorer neuropsychological outcomes than children without cyanosis. However, this risk does not seem to be exacerbated by having a complex cyanotic condition with palliative interventions only. This may suggest that the key neurodevelopmental period in the association between cyanosis and compromised cognitive functioning occurs at the prenatal, early postnatal or surgical stage.25 Concern that ongoing cyanosis across development exacerbates this risk26 is not substantiated by the group patterns evident in table 1.
Secondly, group patterns suggested that all children with a history of severe CHD, regardless of surgical procedure or cyanotic status, are at risk of sensorimotor problems. This finding suggests that cyanosis is not the only important factor in operation in explaining neuropsychological outcomes after CHD. Generally, heart defects may be a marker for other neurological deficits related to sensorimotor ability.25 Alternatively, environmental pathways involving reduced stimulation and physical restrictions on movement, created by physical limitations or indeed by an anxious parent, at a key period in infancy may compromise the later development of these skills.
Although associated with poorer neuropsychological outcomes, cyanosis in itself did not seem to be associated with poorer behavioural adjustment as previous research has suggested.10 Indeed, the pattern of results found in this study highlight the importance of differentiating cyanotic samples by repair status. Findings suggested that the cyanotic‐corrected group had fewer behaviour problems than the acyanotic and mild‐control groups. In comparison, the cyanotic‐complex group had statistically the highest mean score on the CBCL, and the highest frequency of clinically relevant scores. Childhood survivors of CHD with ongoing medical, physical and associated psychosocial challenges seem to be most at risk of behavioural difficulties.
What is already known on this topic
Contradictory evidence in studies using restricted defect or surgical samples suggests that children with congenital heart disease may be at risk of negative neuropsychological and behavioural outcomes.
Knowledge is limited on how important surgical and cyanotic factors are in predicting these in comparison with psychosocial and familial influences.
What this study adds
Cyanosis and open‐heart surgery together lead to compromised neuropsychological outcomes, but this does not seem to be exacerbated by having an uncorrected, palliative, status—the latter does, however, tend to be associated with worse behavioural outcomes.
Family factors may be more important than disease factors in predicting behavioural outcomes especially.
Sensorimotor delays in early childhood may occur for these children regardless of cyanosis or open‐heart surgical status.
Mutivariate determinants of risk
The multivariate analyses of outcome have highlighted that the relevance of disease and surgical factors must be considered within a wider, systemic, context. The relative size and statistical significance of the β coefficients of predictor variables in the final regression models for both neuropsychological and behavioural outcomes (supplementary tables B and C) suggest that, although cyanotic and surgical factors have aetiological relevance, family and social variables make an often greater contribution to risk and outcome. Especially when considering behavioural outcomes, parenting style, marital status and maternal mental health difficulties are seen to have greater predictive importance than the disease factor of cyanosis (supplementary table C). Maternal worry about the child also exerts a considerable effect. Although neuropsychological outcomes might be expected to be less influenced by environmental processes, even here, parenting style, paternal education and sex exerted at least a comparable influence to the surgical procedure variable in predicting outcomes. Such findings place the effect of disease and surgical factors on subsequent child adjustment in a psychosocial context.
Clinical implications
In addition to informing prognosis, these findings have important implications for intervention. It may be difficult or impossible to change many of the disease, surgical and social variables of importance documented above (eg sex, marital status and educational level in the family). However, for some of the other risk factors described above—maternal worry, mental health difficulties and parenting style—early interventions of a psychological nature may improve outcomes.27 Moreover, the sensorimotor difficulties evidenced by all groups may suggest that some form of movement treatment or sensory integration may be useful.28 A clinical trial that tests these propositions is under way at this centre.
Supplementary Material
Abbreviations
CBCL - Child Behaviour Checklist
CHD - congenital heart disease
NEPSY - developmental neuropsychological assessment
WPPSI - Wechsler Preschool and Primary Scale of Intelligence
Footnotes
Funding: This work was sponsored by a research project grant from The National Lottery Charities Board.
Competing interests: None.
References
- 1.Sharma R, Choudhary S, Mohan M.et al Neurological evaluation and intelligence testing in the child with operated congenital heart disease. Ann Thorac Surg 200070575–581. [DOI] [PubMed] [Google Scholar]
- 2.Bellinger D C. Neuropsychological complications after cardiac surgery. Crit Care 2000416 [Google Scholar]
- 3.Wray J, Sensky T. Congenital heart disease and cardiac surgery in childhood: effects on cognitive function and academic ability. Heart 200185687–691. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.De Maso D, Campis L, Wypij D.et al The impact of maternal perceptions and medical severity on the adjustment of children with congenital heart disease. J Pediatr Psychol 199116137–149. [DOI] [PubMed] [Google Scholar]
- 5.Spurkland I, Bjornstad P, Lindberg H.et al Mental health and psychosocial functioning in adolescents with congenital heart disease. A comparison between adolescents born with severe heart defect and atrial septal defect. Acta Paediatr 19938271–76. [DOI] [PubMed] [Google Scholar]
- 6.Utens E, Verslusis‐Den Bieman H, Witsenburg M.et al Cognitive and behavioural and emotional functioning of young children awaiting elective cardiac surgery or catheter intervention. Cardiol Young 200111153–160. [DOI] [PubMed] [Google Scholar]
- 7.Bellinger D, Rappapport L, Wypij D.et al Patterns of developmental dysfunction after surgery during infancy to correct transposition of the great arteries. J Dev Behav Pediatr 19971875–83. [DOI] [PubMed] [Google Scholar]
- 8.Visconti K, Saudino K, Rappaport L.et al Influence of parental stress and social support on the behavioral adjustment of children with transposition of the great arteries. J Dev Behav Pediatr 200223314–321. [DOI] [PubMed] [Google Scholar]
- 9.Aram D, Ekelman B, Ben‐Shachar G.et al Intelligence and hypoxemia in children with congenital heart disease: fact or artefact? J Am Coll Cardiol 19856889–893. [DOI] [PubMed] [Google Scholar]
- 10.Delamanter A. Cardiovascular disease. In: Roberts M, ed. Handbook of pediatric psychology. New York: Guildford Press, 1995403–424.
- 11.Rutter M. Psychosocial resilience and protective mechanisms. In: Rolf J, Masten D, Cicchetti K, et al eds. Risk and protective factors in the development of psychopathology. Cambridge: Cambridge University Press, 1990181–214.
- 12.Wallander J, Thompson R. Psychosocial adjustment of children with chronic physical conditions. In: Roberts M, ed. Handbook of pediatric psychology. New York: Guildford Press, 1995124–141.
- 13.Thompson R, Gustafson K, Hamlett K.et al Psychosocial adjustment of children with cystic fibrosis: the role of child cognitive processes and maternal adjustment. J Pediatr Psychol 199217741–755. [DOI] [PubMed] [Google Scholar]
- 14.Townsend P, Phillimore P, Beattie A.Health and deprivation: inequality and the north. London: Croom Helm, 1988
- 15.Korkman M, Kirk U, Kemp S.NEPSY: a developmental neuropsychological assessment. San Antonio: The Psychological Corporation, 1998
- 16.Wechsler D. Wechsler Preschool and Primary Scale of Intelligence—revised UK edition. New York: The Psychological Corporation, 1989
- 17.Achenbach T M.Manual for the child behaviour checklist/4–18 and 1991 profile. Burlington, VT: University of Vermont Department of Psychiatry, 1991
- 18.Derogatis L R.Brief symptom inventory: administration, scoring and procedures manual. Minneapolis, MN: National Computer Systems, 1993
- 19.DeVet K A, Ireys H T. Psychometric properties of the maternal worry scale for children with chronic illness. J Pediatr Psychol 199823257–266. [DOI] [PubMed] [Google Scholar]
- 20.Campis L K, Lyman R D, Prentice‐Dunn S. The parental locus of control scale: development and validation. J Clin Child Psychol 198615260–267. [Google Scholar]
- 21.Moos R, Moos B.Family environment scale manual. Palo Alto, CA: Consulting Psychologists Press, 1981
- 22.Power M J, Champion L A, Aris S. The development of a measure of social support: the significant others scale (SOS). Br J Clin Psychol 198827349–358. [DOI] [PubMed] [Google Scholar]
- 23.Cohen J.Statistical power analysis for behavioural sciences. Hillsdale, NJ: Laurence Erlbaum, 1988
- 24.Flynn J. Searching for justice: the discovery of IQ gains over time. Am Psychol 1999545–20. [Google Scholar]
- 25.Limperopoulous C, Majnemer A, Shevell M.et al Neurodevelopmental status of newborns and infants with congenital heart defects before and after open heart surgery. J Pediatr 2000137638–645. [DOI] [PubMed] [Google Scholar]
- 26.Stieh J, Kramer H, Harding P.et al Gross and fine motor development is impaired in children with congenital heart disease. Neuropediatrics 19993077–82. [DOI] [PubMed] [Google Scholar]
- 27.McCusker C, Doherty N, Molloy B.et al
- 28.Vargas S, Camill G. A meta‐analysis of research on sensory integration treatment. Am J Occup Ther 199953189–198. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.