Cognitive function and executive function (EF) impairments contribute to the long-term burden of congenital heart disease (CHD). However, the degree and profile of impairments are insufficiently described.
To systematically review and meta-analyze the evidence on cognitive function and EF outcomes in school-aged children operated for CHD and identify the risk factors for an unfavorable outcome.
Cochrane, Embase, Medline, and PsycINFO.
Original peer-reviewed studies reporting cognitive or EF outcome in 5- to 17-year old children with CHD after cardiopulmonary bypass surgery.
Results of IQ and EF assessments were extracted, and estimates were transformed to means and SE. Standardized mean differences were calculated for comparison with healthy controls.
Among 74 studies (3645 children with CHD) reporting total IQ, the summary estimate was 96.03 (95% confidence interval: 94.91 to 97.14). Hypoplastic left heart syndrome and univentricular CHD cohorts performed significantly worse than atrial and ventricular septum defect cohorts (P = .0003; P = .027). An older age at assessment was associated with lower IQ scores in cohorts with transposition of the great arteries (P = .014). Among 13 studies (774 children with CHD) reporting EF compared with controls, the standardized mean difference was −0.56 (95% confidence interval: −0.65 to −0.46) with no predilection for a specific EF domain or age effect.
Heterogeneity between studies was large.
Intellectual impairments in CHD are frequent, with severity and trajectory depending on the CHD subtype. EF performance is poorer in children with CHD without a specific EF profile. The heterogeneity in studied populations and applied assessments is large. A uniform testing guideline is urgently needed.
Because of a significant increase of survival rate in children born with congenital heart disease (CHD), neurocognitive sequelae have been recognized as the most prevalent long-term comorbidity in CHD.1 This has shifted the research focus toward the assessment and improvement of neurocognitive outcomes in this vulnerable population. Mild cognitive impairments, delays in language acquisition, and achievements of motor milestones are evident already in early infancy.2 Persistence of these problems into school age and adolescence have been described by a number of studies;3–7 however, they are characterized by a large heterogeneity of studied CHD populations, assessment instruments, and age at assessment. Furthermore, improvements in outcomes may occur over time with new perioperative and surgical treatment strategies. Thus, an overview of cognitive functioning and assessment of risk factors in CHD at school age is urgently needed. Moreover, deficits in higher-order cognitive functions, such as in executive functions (EFs), only become apparent with increasing cognitive demands at school age.8 The term EF summarizes a broad array of higher-order cognitive and behavioral core functions that enable goal-directed behavior. Furthermore, they are key to academic achievement9 and can have significant negative consequences on quality of life and psychosocial well-being.10 EFs have been identified as a particularly vulnerable cognitive domain in school-aged children with CHD,11–13 thus adding to the individual and societal burden of CHD.14
Importantly, it is unclear whether there is a specific cognitive and EF impairment profile and how this is impacted by severity of CHD. Consequently, therapeutic interventions targeting impairments in specific domains are lacking.15 Only with an in-depth understanding of the affected cognitive domains and identification of the most vulnerable subpopulation can we develop targeted therapeutic interventions that allow care takers and families to maximize the developmental opportunities of their children with CHD.
Thus, the aim of this systematic review is threefold. First, we aim to systematically review and meta-analyze the existing literature on cognitive and EF outcomes in school-aged children with CHD who underwent cardiopulmonary bypass surgery. Second, we aim to identify the risk factors for adverse cognitive and EF outcomes, and, third, we aim to identify potential knowledge gaps that need to be addressed to improve long-term developmental outcomez for this vulnerable population.
Methods
The methods of this systematic review were prespecified and summarized in a systematic review protocol adhering to the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols guideline.16 The protocol has been registered in the International Prospective Register of Systematic Reviews on September 1, 2019 (registration number CRD42018086568), and was published in BioMed Central Systematic Reviews.17
Eligibility Criteria: Inclusion Criteria
As predefined and described in detail elsewhere,17 any peer-reviewed original research study that reported intellectual and/or EF outcomes in school-aged children (5–17 years of age) who underwent surgical repair or palliation for CHD with a cardiopulmonary bypass were included. There is evidence suggesting that children with CHD with genetic syndromal disorders or after a heart transplant constitute a particularly poor performing CHD subgroup, with presumably different underlying mechanism and risk factors.18,19 Therefore, studies in which researchers exclusively report outcomes of children with CHD with genetic syndromal disorders or after heart transplant were excluded. Studies in which both genetic syndromal and nonsyndromal patient data were reported were only included if findings of the nongenetic group could be extracted separately or corresponding authors provided separate data upon request.
No restriction on study design, year of publication (earliest year of publication: 1946; limited by electronic databases), or publication language was applied. An assessment of eligibility of studies in foreign languages and data extraction was conducted with the help of acquainted researchers who were native speakers and received instructions on the evaluation of inclusion and exclusion criteria. A minimum sample size of 10 subjects per study was required for inclusion. Reference lists of reviews were searched for further suitable publications that were not identified through the search strategy.
Search Strategy
The following electronic databases were searched: Cochrane, Embase, Medline, and PsycINFO. The search strategy was developed by a librarian (M.A.G.P.) and conducted by using subject headings (MeSH, EMTREE, and PsycINDEX thesaurus) and free text words related to CHD and intellectual and EF outcomes in children and adolescents. The search was adapted to each database. The first literature search was conducted in July 2019, and an updated search was conducted in December 2020. The prepublished Medline search protocol17 is included as Supplemental Table 2.
Selection of Studies and Data Extraction
In a two-step procedure, 2 independent reviewers (C.U. and C.B.) screened the titles and abstracts first and then the full texts for eligibility. Furthermore, reference lists of included studies and any reviews were screened for additional relevant studies. Reviewers were not blinded to journal titles, authors, or author’s affiliations during the study selection process. Multiple reports of the same cohort were identified by juxtaposing authors and cohort characteristics (ie, number, year, and place of recruitment) as well as reported outcomes. In case of multiple reports of the same cohort, outcome, and time point, only the most extensive report in terms of sample size was considered for data extraction. Conversely, 2 separate data extraction forms were completed if the follow-up time points or reported outcomes differed. If needed, corresponding authors were contacted to clarify questions on overlapping reports of the same cohort. For studies in which presurgical and postsurgical cognitive assessments were performed, only the postsurgical data were extracted. Data were extracted by using a digital data extraction form that was predefined17 and refined during the data extraction process (Supplemental Table 3). If reported results were stratified by CHD subtype, data were extracted for each CHD subtype separately and pooled, if needed. If provided, healthy control data were extracted too. Data extraction was conducted by MF. In the absence of complete outcome reports or missing crucial information on the study population, the corresponding authors were contacted to obtain missing information. In cases in which the age at assessment was only given as a range from minimum to maximum age, a proxy for age was calculated as the mean over the age range. A proxy for surgical year was calculated by taking the average year across the indicated period of years of surgery.
Meta-analysis
For the meta-analysis, only studies reporting extractable location parameters (mean or median) and measures of dispersion (SD, SE, interquartile range [IQR], or range) for intellectual or EF outcomes were included. Therefore, if studies reported location parameters only in graphical but not numerical form, they could not be included in the meta-analysis. The same applied for studies that reported a location parameter but failed to report the corresponding measure of dispersion. Heterogeneity between studies was measured with Higgin’s I2 statistic. As for the majority of outcomes, the Higgin’s I2 indicated that >50% of variation between the sample estimates was due to heterogeneity rather than sampling error; random-effects models were used for all meta-analyses by using the metafor package in R.20 To account for multiple longitudinal reporting of the same cohort in different studies across multiple time points, the weights of those individual studies were adjusted to sum up to 1 for each cohort. Total subject counts for each meta-analysis were adjusted to account for longitudinal reports of the same cohort. For meta-analysis, all location parameters were recalculated to mean outcome measures and SDs were recalculated from median and IQR or ranges following the proposed formulae by Wan et al.21 SD was consecutively transformed to SE by using the sample size. For studies in which reporting of results was stratified by therapeutic interventions that were of no interest to this review (eg, different surgical approaches and EF training intervention) subgroup results were combined following the recommendation of the Cochrane Handbook for Systematic Reviews of Interventions.22 For outcomes in which a large number of studies also reported estimates of healthy controls as comparator group or the scale of effect estimates differed across studies, the standardized mean difference (SMD) was calculated, with the SD from the comparator group as the denominator. By using this so called Glass’ Δ as SMD as opposed to the Hedges’ (adjusted) g, we avoid underestimating the group differences because of an increase of interindividual variation in the CHD group.22 Reporting and publication bias for each meta-analysis was assessed by visual inspection of the funnel plots and formally tested with the regression test for plot asymmetry by Egger et al,23 as implemented in the metafor package in R. To test for the effect of including studies in languages other than English, we conducted a sensitivity analysis after excluding those studies from the main outcome analyses, such as intellectual functioning and EF summary estimates. All statistical analyses were performed in R Statistical Programming.24
Intellectual Functioning Analysis
To meta-analyze intellectual functioning in children with CHD, the following cognitive outcomes were considered: total IQ, verbal IQ, and if tested with a version of the Wechsler Intelligence Scale25 performance IQ. Because IQ measures were reported on a common standardized scale, and many studies additionally reported results in a healthy control group, 2 types of meta-analyses were performed for this outcome by (1) using the standardized test scores and (2) calculating the SMD.
EF Analysis
Two types of EF measures were considered in this systematic review: behavioral-rated and performance-based EF. As behavioral rating–based measures of EF, the Global Executive Composite Score of the parent- and self-reported Behavior Rating Inventory of Executive Function (BRIEF)26 was considered. This questionnaire measuring daily-life EF was commonly deployed in this population. BRIEF Global Executive Function Composite scores >50 indicate EF difficulties and >5 clinically significant executive dysfunction. To summarize and meta-analyze the performance-based EF measures, the reported subtests were categorized into 1 of the 3 main domains of EF27 : inhibitory control; working memory; and shifting or cognitive flexibility. Furthermore, the subdomains planning and fluency were included because these are frequently reported and tested in commonly used EF test batteries. Because EF results, unlike IQ, are not reported on a common standardized scale, only studies in which the SMD by comparison with a healthy control group could be calculated were included in the quantitative analysis. SMD of subtests that assessed EF performance on the basis of number of errors or reaction time were multiplied by −1. Thus, for all reported tests, negative SMDs represented poorer performance, compared with that of controls. Scores of Rey–Osterrieth complex figure tests were not included in the meta-analysis because there is currently no consensus on how to rate the results, hindering comparability between studies.28 Two recently published randomized controlled trials in which researchers examined the effect of an EF training intervention could not be included because no healthy control group was reported.29,30
Subgroup Analyses and Meta-Regression
If at least 3 studies could be grouped by CHD subtype, a separate summary estimate was calculated in the meta-analysis. The following CHD subtypes were reported: univentricular heart disease (UVH) as a mixed group of different types of single ventricle physiology CHD; total anomalous pulmonary venous return; tetralogy of Fallot (TOF); d-transposition of the great arteries (TGA); hypoplastic left heart syndrome (HLHS); atrial septum or ventricular septum defect (ASD/VSD); and mixed CHD (mixed) as any heterogenous group of different CHD subtypes. A modifying effect of CHD subtype on reported intellectual functioning scores were assessed, with ASD/VSD subjects serving as the reference group. In most centers, surgical correction in patients with ASD/VSD is performed beyond the neonatal period, which can be considered the most vulnerable for brain developmental disturbances. Cardiopulmonary bypass times are comparably short, with no need for deep hypothermia or cardiac arrest that might affect cerebral perfusion and neurologic outcome.31,32 Furthermore, postoperative length of respiratory support and length of intensive care is often shorter, in comparison with all the other CHD subtypes mentioned above. For surgical repair of VSD during infancy, early neurodevelopmental outcomes have been found to be within the normal range.33 Meta-regression was used to quantify the association of age at assessment, year of surgery, and year of publication with outcome measures. Interactions of these risk factors with CHD subtype were evaluated where appropriate.
Risk of Bias Assessment
The risk of bias in the included studies was assessed by means of a modified checklist, which was based on the criteria of the Scottish Intercollegiate Guidelines Network (SIGN) checklist for cohort studies (https://www.sign.ac.uk/what-we-do/methodology/checklists/) and the rating system of the Newcastle-Ottawa Scale for the quality of nonrandomized studies in meta-analyses (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp). In our final 10-item checklist, we incorporate adapted criteria of the SIGN checklist to assess bias in selection, attrition, detection, and analysis. By combining items with the star rating system adopted from the Newcastle-Ottawa Scale, the modified checklist allows for a transparent rating of the bias in each of the assessed domains. The bias assessment was performed by CB, after 2 authors modified and piloted the checklist (C.B. and M.F.). The final checklist, including an elaboration of the rating criteria, can be found in Supplemental Table 4. Studies were not weighted or excluded on the basis of the bias assessment, but the results were taken into account in the overall discussion of the results.
Results
A total number of 3275 studies were identified by our search strategy, and an additional 3 studies were identified through reference list screenings. After deduplication and the two-step process of title and abstract and full-text screening, 94 studies were eligible for inclusion (for a detailed flowchart, see Fig 1). The included studies had a median (IQR) sample size of 40.5 subjects with CHD per study (27.0 and 72.2) and reported intellectual and/or EF outcome in a total number of 4566 unique children and adolescents with CHD (after subtraction of longitudinal follow-up cohorts or multiple reports). Of all included studies, 3 were published in German and 1 in French. However, the latter could not be included in the quantitative analysis because of semiquantitative reporting of results. The median percentage of male children with CHD was 58.8% (IQR: 51.2%–68.3%), and a median percentage of 87.5% (IQR: 65.7%–90.1%) of study subjects were white. Children with CHD were assessed at a mean follow-up age of 9.1 (SD: 2.9) years and underwent cardiopulmonary bypass surgery at a median age of 3.0 (IQR: 0.4–19.2) months. The median percentage of cyanotic heart defects in the reported cohorts was 84.6% (IQR: 42.9%–100.0%). For an overview, characteristics of included studies can be found in Table 1.
Summary and Bias Assessment
Author . | Study Site . | Study Design . | CHD Subtype . | Age at Surgery, mo . | Age at Assessment, y . | N . | Type of IQ Test . | Type of EF Test . | Note . | Bias Assessment . | |||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Selection . | Attrition . | Detection . | Analysis . | ||||||||||
Aldén et al, 199869 | Gothenburg, Sweden | Cross-sectional | TGA | 19.1 | 13.2 | 31 | WISC-R | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Bellinger et al, 200370 | Boston, Massachusetts | RCT | TGA | 0.3 | 8.2 | 155 | WISC-III | Rey–Osterrieth Complex Figure | Same cohort as Bellinger et al,71 combined data extraction | ★★☆ | ☆☆☆ | ★★ ★★ | ★★ |
Bellinger et al, 200371 | Boston, Massachusetts | RCT | TGA | 0.3 | 8.2 | 155 | WISC-III | — | Same cohort as Bellinger et al,70 combined data extraction | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Bellinger et al, 20114 | Boston, Massachusetts | RCT | TGA | NA | 16.1 | 139 | — | DKEFS, BRIEF | Longitudinal follow-up of Bellinger et al70 and Bellinger et al71 | ★★☆ | ★☆☆ | ★★ ☆☆ | ★★ |
Bellinger et al, 20155 | Boston, Massachusetts | Cross-sectional | TOF | 3.7 | 14.6 | 68 | WISC-IV | — | — | ★★☆ | ★☆☆ | ★★ ☆☆ | ★★ |
Bellinger et al, 20156 | Boston, Massachusetts | Cross-sectional | UVH | 1.7 | 14 | 91 | WISC-IV | WISC-IV working memory, BRIEF | Authors provided additional data after exclusion of genetic cases | ★★☆ | ★☆☆ | ★★ ☆☆ | ★★ |
Bergemann et al, 201512 | Kiel, Germany | OSP | HLHS | 0.3 | 7.7 | 37 | Culture Fair Test nonverbal IQ | German version of the Auditory Verbal Learning Test, Corsi Block-Tapping test, Rey–Osterrieth Complex Figure test, WISC-IV working memory, Semantic fluency, Rey–Osterrieth Complex Figure test | — | ★★☆ | ★★★ | ★★ ☆☆ | ☆☆ |
Bordacova et al, 200772 | Bratislava, Slovakia | Cross-sectional | HLHS | 0.3 | 6.9 | 19 | SB-IV | — | — | ☆☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Brosig et al, 201373 | Wisconsin | Cross-sectional | HLHS | 0.2 | 5 | 34 | WPPSI-III short | — | — | ★★★ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Calderon et al, 201011 | Paris, France | Cross-sectional | TGA | NA | 7.3 | 21 | Columbia Mental Maturity Scale | Stroop, Nepsy, WISC-IV working memory, Corsi Block-Tapping test, Tower of London | — | ★★☆ | ☆☆☆ | ★★ ★☆ | ★☆ |
Calderon et al, 201274 | Paris, France | OSP | TGA | 0.2 | 5 | 90 | Columbia Mental Maturity Scale | Nepsy, Stroop, WISC-IV working memory, Corsi block-tapping test, Dimensional Change Card Sort test | — | ★★★ | ☆☆☆ | ★★ ★★ | ★☆ |
Calderon et al, 201475 | France, Paris | OSP | TGA | 0.2 | 7.3 | 38 | Columbia Mental Maturity Scale | — | Longitudinal follow-up of Calderon et al74 | ★☆☆ | ★★★ | ★★ ☆☆ | ★★ |
Calderon et al, 202029 | Boston, Massachusetts | RCT | Mixed | NA | 14.5 | 28 | WISC-V | — | — | ★★★ | ★★★ | ★★ ★★ | ★☆ |
Cassidy et al, 201513 | Boston, Massachusetts | RCT | TGA, TOF, UVH | 2.1 | 15.1 | 352 | — | DKEFS, BRIEF | Same cohort as Bellinger et al,4 Bellinger et al,5 and Bellinger et al6 but other outcome reported | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Cassidy et al, 201776 | Boston, Massachusetts | RCT, cross-sectional | TGA, TOF | NA | 15.6 | 298 | WISC-V | Children’s Memory Scale | Same cohort as Bellinger et al4 and Bellinger et al5 but other outcome reported | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Claessens et al, 201877 | Utrecht, Netherlands | RCT | Mixed | 0.3 | 5.9 | 60 | WPPSI-R | — | — | ★★☆ | ★★★ | ★★ ☆☆ | ★☆ |
Creighton et al, 200740 | Edmonton, Canada | OSP | TGA, UVH, TAPVR, mixed | NA | 5 | 53 | WPPSI-III | — | — | ★★☆ | ★☆☆ | ★★ ☆☆ | ★☆ |
Easson et al, 201978 | Montreal, Canada | Cross-sectional | UVH, BVH, mixed | 1.1 | 15.7 | 142 | Leiter scale | — | — | ★☆☆ | ☆☆☆ | ★★ ★☆ | ☆☆ |
Ehrler et al, 202079 | Zürich, Switzerland | OSP | Mixed | NA | 10.2 | 107 | WISC-V | — | — | ★★☆ | ★☆☆ | ★ ☆☆☆ | ★★ |
Eichler et al, 201980 | Erlangen, Germany | Cross-sectional | VSD | NA | 7.3 | 39 | IDS | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Finley et al, 197481 | San Francisco, California | OSP | TOF | 90.4 | 7.6 | 37 | SB-I | — | — | ★★☆ | ★☆☆ | ★ ☆☆☆ | ★☆ |
Fleisher et al, 200219 | Stanford, California | Cross-sectional | Mixed | NA | 6 | 14 | SB-I | — | — | ★☆☆ | ☆☆☆ | ★★ ★★ | ★☆ |
Forbess et al, 200182 | Boston, Massachusetts | OSR | UVH | NA | 5 | 27 | WPPSI-R | — | Historical Fontan subgroup excluded for age at assessment outside the target range | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Forbess et al, 200283 | Boston, Massachusetts | Cross-sectional | BVH | 3 | 5 | 69 | WPPSI-R | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Gerstle et al, 20169 | Cincinnati, Ohio | Cross-sectional | Mixed | NA | 12 | 143 | WISC-IV | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Gomelsky et al, 199884 | Baltimore, Maryland | OSP | TGA | 2.5 | 8.2 | 57 | SB-IV | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ☆☆ |
Guan et al, 201185 | Jinan, China | Cross-sectional | VSD | 46.9 | 9.1 | 16 | WISC-III | — | — | ★★☆ | ★★★ | ★★ ☆☆ | ★★ |
Hagemo et al, 200786 | Oslo, Norway | Cross-sectional | HLHS | NA | 8.5 | 15 | SB-IV | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Heinrichs et al, 201487 | Aachen, Germany | OSP | TGA | 0.2 | 16.9 | 56 | WAIS-R | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ☆☆ |
Heying et al, 199988 | Aachen, Germany | Cross-sectional | Mixed | 45 | 5.6 | 11 | Bühler-Hetzer-Test, HAWIK, Kramer Test | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ☆☆ |
Hiraiwa et al, 201989 | Toyama, Japan | OSP | UVH | NA | 7.5 | 35 | WISC-IV | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ★☆ |
Honzik et al, 196990 | San Francisco, California | OSP | Mixed | 78 | 10.5 | 118 | WISC | — | — | ☆☆☆ | ☆☆☆ | ★ ☆☆☆ | ★☆ |
Hövels-Gürich et al, 199791 | Aachen, Germany | Cross-sectional | TGA | 0.2 | 5.4 | 77 | K-ABC | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Hövels-Gürich et al, 200292 | Aachen, Germany | OSP | TGA | 0.2 | 10.5 | 60 | K-ABC | — | Longitudinal follow-up of Hövels-Gürich et al91 | ★★☆ | ★☆☆ | ★★ ☆☆ | ★★ |
Hövels-Gürich et al, 200693 | Aachen, Germany | Cross-sectional | VSD, TOF, Mixed | 8.4 | 7.4 | 80 | K-ABC | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ★☆ |
Iwamoto et al, 199094 | Miyazaki, Japan | OSR | Mixed | NA | 12 | 75 | Tanaka Group IQ | — | — | ☆☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Jedlicka-Köhler et al, 198795 | Vienna, Austria | OSR | TOF | 19.6 | 6.5 | 13 | HAWIVA, HAWIK, HAWIE | — | Published in German, older subgroup excluded because of age outside the target range | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Jedlicka-Köhler et al, 199595 | Vienna, Austria | Cross-sectional | TGA | 10.4 | 8.8 | 28 | Stanford–Binet, HAWIK, HAWIE | — | Published in German | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Jones et al, 201596 | Brisbane, Germany | OSP | Mixed | 1.7 | 6.6 | 20 | WPPSI | BRIEF | — | ★☆☆ | ★★☆ | ★★ ☆☆ | ★☆ |
Jordan et al, 202030 | Nashville, Tennessee | RCT | HLHS | 0.2 | 11.2 | 20 | WISC-V | — | — | ★☆☆ | ☆☆☆ | ★★ ★★ | ☆☆ |
Karl et al, 200497 | Melbourne, Germany | Cross-sectional | TGA | 0.4 | 9.1 | 74 | WISC-III, WPPSI-III | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Kirshbom et al, 200598 | Philadelphia, Pennsylvania | Cross-sectional | TAPVR | 1.2 | 11 | 30 | WAIS | — | — | ★★★ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Ma et al, 202099 | Nanjing, China | TOF | 25.2 | 10.1 | 10 | WISC-R | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ | |
Mahle et al, 2000100 | Philadelphia, Pennsylvania | Cross-sectional | HLHS | 0.2 | 8.9 | 28 | WISC-III | — | — | ★★★ | ★★★ | ★★ ☆☆ | ★☆ |
Mahle et al, 2004101 | Atlanta, Georgia | OSP | Mixed | 113 | 9.4 | 29 | WAIS | BRIEF | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Mahle et al, 2006102 | Atlanta, Georgia | Cross-sectional | HLHS | 0.2 | 12.4 | 26 | WAIS | BRIEF | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Majnemer et al, 20087 | Montreal, Canada | OSP | Mixed | 2.7 | 5.4 | 94 | WPPSI-R | — | — | ★★☆ | ★★☆ | ★★ ★★ | ★☆ |
Matos et al, 2014103 | Porto, Portugal | Cross-sectional | VSD, TGA, TOF | NA | 15.02 | 77 | — | Wechsler’s Digit Span direct and reverse, Rey–Ostherieth Complex Figure Test, key search test from the BADS-C, Color-Word Stroop Test | EF results could not be extracted because only nonparametric test statistic results were provided | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Messmer et al, 1976104 | Zürich, Switzerland | Cross-sectional | Mixed | 3 | 6.5 | 11 | SON-R, HAWIK, Buhler-Hetzer | — | — | ★★☆ | ★☆☆ | ★★ ☆☆ | ☆☆ |
Miatton et al, 2007105 | Ghent, Belgium | Cross-sectional | Mixed | NA | 8.7 | 43 | WISC-III | Nepsy-I | — | ★★☆ | ☆☆☆ | ★ ☆☆☆ | ★☆ |
Miller et al, 1994106 | Hershey, Pennsylvania | Cross-sectional | Mixed | 1 | 5.5 | 20 | SB-I | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Mittnacht et al, 2015107 | Heidelberg, Germany | RCT | Mixed | 7.2 | 11.1 | 56 | WISC-III | — | — | ★★☆ | ★☆☆ | ★★ ☆☆ | ☆☆ |
Moerbeke et al, 1986108 | Brussels, Belgium | Cross-sectional | Mixed | 7 | NA | 31 | WISC | — | Published in French | ☆☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Munoz-Lopez et al, 2017109 | London, United Kingdom | Cross-sectional | TGA | NA | 11.4 | 40 | WISC-IV | — | — | ★★★ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Murphy et al, 2017110 | Nashville, Tennessee | Cross-sectional | Mixed | 3 | 16.1 | 18 | WISC-IV, WAIS-IV | WISC-IV working memory | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Naef et al, 201718 | Zürich, Switzerland | OSP | Mixed | 2.7 | 6.3 | 169 | WPPSI-III, SON-R, K-ABC | — | — | ★★☆ | ★★★ | ★★ ☆☆ | ★☆ |
Newburger et al, 1984111 | Boston, Massachusetts | OSP | TGA | 19.2 | 5.8 | 38 | WISC-III, Leiter International Scale, Grace Arthur Adaption | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
O'Dougherty et al, 1983112 | Minneapolis, Minnesota | Cross-sectional | TGA | NA | 9.1 | 31 | WPPSI-R | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Oates et al, 1995113 | Camperdown, Australia | Cross-sectional | ASD, VSD, TGA, TOF | 25.9 | 10.6 | 168 | WISC-R | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Oberhuber et al, 2017114 | Linz, Austria | OSP | HLHS | NA | 10.1 | 43 | WISC-V | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Poncelet et al, 2011115 | Brussels, Belgium, | RCT | Mixed | 19.2 | 7 | 36 | WPPSI-R, WISC-IV | — | — | ★☆☆ | ★☆☆ | ★★ ★★ | ☆☆ |
Quartermain et al, 2010116 | Philadelphia, Pennsylvania | Cross-sectional | Mixed acyanotic | 142 | 11.8 | 35 | WAIS | — | — | ★★☆ | ★★★ | ★★ ★★ | ★☆ |
Ratzmann et al, 1991117 | Leipzig, Germany | Cross-sectional | Mixed | NA | 13.3 | 62 | WISC | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Sahu et al, 2009118 | New Delhi, India | RCT | TOF | 106 | 8.8 | 80 | WISC | — | — | ★☆☆ | ☆☆☆ | ★★ ★★ | ★☆ |
Sakamoto et al, 2006119 | Tokyo, Japan | Cross-sectional | Mixed | 10.2 | — | 30 | WPPSI | — | — | ★☆☆ | ★★ ☆ | ★★ ☆☆ | ☆☆ |
Sanz et al, 2017120 | Washington | Cross-sectional | Mixed | NA | 9.1 | 91 | — | BRIEF | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Sarajuuri et al, 2007121 | Helsinki, Finland | Cross-sectional | HLHS | 0.1 | 5.7 | 26 | WPPSI-R | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Sarajuuri et al, 2012122 | Helsinki, Finland | OSP | UVH, HLHS | 0.2 | 5.1 | 34 | WPPSI-R | — | — | ★★☆ | ★★☆ | ★★ ☆☆ | ☆☆ |
Sarrechia et al, 2013123 | Ghent, Belgium | Cross-sectional | ASD, VSD | 22 | 8.7 | 15 | WISC-III | Nepsy-I | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Sarrechia et al, 2015124 | Ghent, Belgium | Cross-sectional | ASD, VSD | 16.2 | 8.9 | 46 | WISC-III | — | Subgroups of ASD and VSD after catheter intervention were excluded | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Sarrechia et al, 2016125 | Ghent, Belgium | Cross-sectional | UVH | 0.8 | 9.1 | 17 | WISC-III | Nepsy-I | IQ outcome of biventricular heart disease subgroup extracted from124 | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Schäfer et al, 20133 | Zürich, Switzerland | OSP | Mixed | 10.8 | 13.6 | 59 | WISC-IV | WISC-IV working memory, Rey–Osterrieth Complex Figure test | — | ★★☆ | ★★☆ | ★★ ☆☆ | ★☆ |
Simons et al, 2010126 | Wilmington, Delaware | Cross-sectional | VSD | 19.9 | 8.7 | 13 | WISC-IV | BRIEF | Younger subgroup was excluded because of age outside the target range, BRIEF data were extracted for entire cohort | ★★★ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Spijkerboer et al, 2008127 | Rotterdam, Netherlands | Cross-sectional | ASD, VSD, TGA | NA | 12.1 | 103 | WISC-R | — | — | ★☆☆ | ★★★ | ★★ ☆☆ | ★☆ |
Stavinoha et al, 2003128 | Dallas, Texas | OSP | ASD | 102 | 8.5 | 18 | Differential Ability Scale | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Sterken et al, 2016129 | Leuven, Belgium | OSP | Mixed | 1.9 | 7.2 | 100 | WPPSI-R | ANT | — | ★☆☆ | ★★★ | ★★ ☆☆ | ★★ |
Sugimoto et al, 2013130 | Shizuoka, Japan | OSR | UVH | NA | 9 | 70 | WPPSI-R, WISC-III, WISC-IV | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Utens et al, 1993131 | Rotterdam, Netherlands | Cross-sectional | Mixed | NA | 13.3 | 287 | WISC-R short form, Groninger Intelligence Test | — | — | ★★☆ | ★★ ☆ | ★★ ★★ | ☆☆ |
Uzark et al, 1998132 | San Diego, California | Cross-sectional | UVH | NA | 6.3 | 32 | SB-I | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Vahsen et al, 2009133 | Sankt Augustin, Germany | Cross-sectional | TGA | 0.6 | 7.6 | 30 | K-ABC | — | Published in German | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Vahsen et al, 2018134 | Sankt Augustin, Germany | OSR | UVH | NA | 8.6 | 104 | K-ABC | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
van der Rijken et al, 2008135 | Nijmegen, Netherlands | OSP | Mixed | NA | 11.6 | 43 | WISC-III | ANT, Rey–Osterrieth Complex Figure Test | — | ★★☆ | ★★☆ | ★★ ☆☆ | ☆☆ |
Venchiarutti et al, 2019136 | Udine, Italy | Cross-sectional | Mixed | 12 | 9.8 | 17 | WISC-IV | Nepsy-II | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ★☆ |
von Rhein et al, 2012137 | Zürich, Switzerland | Cross-sectional | Mixed | 13.2 | 10.4 | 117 | Raven‘s PM | Rey–Osterrieth Complex Figure Test | — | ★★☆ | ★★★ | ★★ ★★ | ★★ |
Wells et al, 1983138 | Brompton, United Kingdom | Cross-sectional | Mixed | 15.4 | 5.7 | 50 | McCarthy Scales of Children’s Abilities, WISC | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Wernovsky et al, 2000139 | Boston, Massachusetts | Cross-sectional | UVH | NA | 14.1 | 130 | WPPSI-R, WISC-III, WAIS-R | — | — | ★★★ | ☆☆☆ | ★★ ★★ | ★★ |
Whitman et al, 1973140 | Cleveland, Ohio | OSP | Mixed | NA | 9 | 11 | WISC | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Wolfe et al, 2020141 | Aurora, Colorado | Cross-sectional | UVH | NA | 9.55 | 68 | WISC-V, WAIS-IV, WISC preschool | Tower of London Drexel | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Wotherspoon et al, 2019142 | Brisbane, Australia | OSP | Mixed | 1.6 | 15.4 | 21 | WISC-IV, WAIS-IV | BRIEF | Longitudinal follow-up of Jones et al96 | ★★☆ | ★★☆ | ★★ ☆☆ | ★★ |
Wray et al, 2001143 | London, United Kingdom | OSP | Mixed | NA | 7.4 | 82 | BAS short | — | — | ★☆☆ | ★★★ | ★ ☆☆☆ | ★★ |
Wray et al, 2010144 | London, United Kingdom | Cross-sectional | Mixed | NA | 10.6 | 24 | BAS short | — | — | ★★☆ | ★★ ☆ | ★ ☆☆☆ | ★☆ |
Wright et al, 1994145 | Melbourne, Australia | OSR | Mixed | NA | 9.5 | 29 | WISC-R | — | — | ★★☆ | ☆☆☆ | ★★ ★★ | ★★ |
Yang et al, 1994146 | Changsha, China | Cross-sectional | Mixed acyanotic | NA | 8.1 | 39 | WPPSI-R, WISC-R | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Author . | Study Site . | Study Design . | CHD Subtype . | Age at Surgery, mo . | Age at Assessment, y . | N . | Type of IQ Test . | Type of EF Test . | Note . | Bias Assessment . | |||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Selection . | Attrition . | Detection . | Analysis . | ||||||||||
Aldén et al, 199869 | Gothenburg, Sweden | Cross-sectional | TGA | 19.1 | 13.2 | 31 | WISC-R | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Bellinger et al, 200370 | Boston, Massachusetts | RCT | TGA | 0.3 | 8.2 | 155 | WISC-III | Rey–Osterrieth Complex Figure | Same cohort as Bellinger et al,71 combined data extraction | ★★☆ | ☆☆☆ | ★★ ★★ | ★★ |
Bellinger et al, 200371 | Boston, Massachusetts | RCT | TGA | 0.3 | 8.2 | 155 | WISC-III | — | Same cohort as Bellinger et al,70 combined data extraction | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Bellinger et al, 20114 | Boston, Massachusetts | RCT | TGA | NA | 16.1 | 139 | — | DKEFS, BRIEF | Longitudinal follow-up of Bellinger et al70 and Bellinger et al71 | ★★☆ | ★☆☆ | ★★ ☆☆ | ★★ |
Bellinger et al, 20155 | Boston, Massachusetts | Cross-sectional | TOF | 3.7 | 14.6 | 68 | WISC-IV | — | — | ★★☆ | ★☆☆ | ★★ ☆☆ | ★★ |
Bellinger et al, 20156 | Boston, Massachusetts | Cross-sectional | UVH | 1.7 | 14 | 91 | WISC-IV | WISC-IV working memory, BRIEF | Authors provided additional data after exclusion of genetic cases | ★★☆ | ★☆☆ | ★★ ☆☆ | ★★ |
Bergemann et al, 201512 | Kiel, Germany | OSP | HLHS | 0.3 | 7.7 | 37 | Culture Fair Test nonverbal IQ | German version of the Auditory Verbal Learning Test, Corsi Block-Tapping test, Rey–Osterrieth Complex Figure test, WISC-IV working memory, Semantic fluency, Rey–Osterrieth Complex Figure test | — | ★★☆ | ★★★ | ★★ ☆☆ | ☆☆ |
Bordacova et al, 200772 | Bratislava, Slovakia | Cross-sectional | HLHS | 0.3 | 6.9 | 19 | SB-IV | — | — | ☆☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Brosig et al, 201373 | Wisconsin | Cross-sectional | HLHS | 0.2 | 5 | 34 | WPPSI-III short | — | — | ★★★ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Calderon et al, 201011 | Paris, France | Cross-sectional | TGA | NA | 7.3 | 21 | Columbia Mental Maturity Scale | Stroop, Nepsy, WISC-IV working memory, Corsi Block-Tapping test, Tower of London | — | ★★☆ | ☆☆☆ | ★★ ★☆ | ★☆ |
Calderon et al, 201274 | Paris, France | OSP | TGA | 0.2 | 5 | 90 | Columbia Mental Maturity Scale | Nepsy, Stroop, WISC-IV working memory, Corsi block-tapping test, Dimensional Change Card Sort test | — | ★★★ | ☆☆☆ | ★★ ★★ | ★☆ |
Calderon et al, 201475 | France, Paris | OSP | TGA | 0.2 | 7.3 | 38 | Columbia Mental Maturity Scale | — | Longitudinal follow-up of Calderon et al74 | ★☆☆ | ★★★ | ★★ ☆☆ | ★★ |
Calderon et al, 202029 | Boston, Massachusetts | RCT | Mixed | NA | 14.5 | 28 | WISC-V | — | — | ★★★ | ★★★ | ★★ ★★ | ★☆ |
Cassidy et al, 201513 | Boston, Massachusetts | RCT | TGA, TOF, UVH | 2.1 | 15.1 | 352 | — | DKEFS, BRIEF | Same cohort as Bellinger et al,4 Bellinger et al,5 and Bellinger et al6 but other outcome reported | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Cassidy et al, 201776 | Boston, Massachusetts | RCT, cross-sectional | TGA, TOF | NA | 15.6 | 298 | WISC-V | Children’s Memory Scale | Same cohort as Bellinger et al4 and Bellinger et al5 but other outcome reported | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Claessens et al, 201877 | Utrecht, Netherlands | RCT | Mixed | 0.3 | 5.9 | 60 | WPPSI-R | — | — | ★★☆ | ★★★ | ★★ ☆☆ | ★☆ |
Creighton et al, 200740 | Edmonton, Canada | OSP | TGA, UVH, TAPVR, mixed | NA | 5 | 53 | WPPSI-III | — | — | ★★☆ | ★☆☆ | ★★ ☆☆ | ★☆ |
Easson et al, 201978 | Montreal, Canada | Cross-sectional | UVH, BVH, mixed | 1.1 | 15.7 | 142 | Leiter scale | — | — | ★☆☆ | ☆☆☆ | ★★ ★☆ | ☆☆ |
Ehrler et al, 202079 | Zürich, Switzerland | OSP | Mixed | NA | 10.2 | 107 | WISC-V | — | — | ★★☆ | ★☆☆ | ★ ☆☆☆ | ★★ |
Eichler et al, 201980 | Erlangen, Germany | Cross-sectional | VSD | NA | 7.3 | 39 | IDS | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Finley et al, 197481 | San Francisco, California | OSP | TOF | 90.4 | 7.6 | 37 | SB-I | — | — | ★★☆ | ★☆☆ | ★ ☆☆☆ | ★☆ |
Fleisher et al, 200219 | Stanford, California | Cross-sectional | Mixed | NA | 6 | 14 | SB-I | — | — | ★☆☆ | ☆☆☆ | ★★ ★★ | ★☆ |
Forbess et al, 200182 | Boston, Massachusetts | OSR | UVH | NA | 5 | 27 | WPPSI-R | — | Historical Fontan subgroup excluded for age at assessment outside the target range | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Forbess et al, 200283 | Boston, Massachusetts | Cross-sectional | BVH | 3 | 5 | 69 | WPPSI-R | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Gerstle et al, 20169 | Cincinnati, Ohio | Cross-sectional | Mixed | NA | 12 | 143 | WISC-IV | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Gomelsky et al, 199884 | Baltimore, Maryland | OSP | TGA | 2.5 | 8.2 | 57 | SB-IV | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ☆☆ |
Guan et al, 201185 | Jinan, China | Cross-sectional | VSD | 46.9 | 9.1 | 16 | WISC-III | — | — | ★★☆ | ★★★ | ★★ ☆☆ | ★★ |
Hagemo et al, 200786 | Oslo, Norway | Cross-sectional | HLHS | NA | 8.5 | 15 | SB-IV | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Heinrichs et al, 201487 | Aachen, Germany | OSP | TGA | 0.2 | 16.9 | 56 | WAIS-R | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ☆☆ |
Heying et al, 199988 | Aachen, Germany | Cross-sectional | Mixed | 45 | 5.6 | 11 | Bühler-Hetzer-Test, HAWIK, Kramer Test | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ☆☆ |
Hiraiwa et al, 201989 | Toyama, Japan | OSP | UVH | NA | 7.5 | 35 | WISC-IV | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ★☆ |
Honzik et al, 196990 | San Francisco, California | OSP | Mixed | 78 | 10.5 | 118 | WISC | — | — | ☆☆☆ | ☆☆☆ | ★ ☆☆☆ | ★☆ |
Hövels-Gürich et al, 199791 | Aachen, Germany | Cross-sectional | TGA | 0.2 | 5.4 | 77 | K-ABC | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Hövels-Gürich et al, 200292 | Aachen, Germany | OSP | TGA | 0.2 | 10.5 | 60 | K-ABC | — | Longitudinal follow-up of Hövels-Gürich et al91 | ★★☆ | ★☆☆ | ★★ ☆☆ | ★★ |
Hövels-Gürich et al, 200693 | Aachen, Germany | Cross-sectional | VSD, TOF, Mixed | 8.4 | 7.4 | 80 | K-ABC | — | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ★☆ |
Iwamoto et al, 199094 | Miyazaki, Japan | OSR | Mixed | NA | 12 | 75 | Tanaka Group IQ | — | — | ☆☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Jedlicka-Köhler et al, 198795 | Vienna, Austria | OSR | TOF | 19.6 | 6.5 | 13 | HAWIVA, HAWIK, HAWIE | — | Published in German, older subgroup excluded because of age outside the target range | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Jedlicka-Köhler et al, 199595 | Vienna, Austria | Cross-sectional | TGA | 10.4 | 8.8 | 28 | Stanford–Binet, HAWIK, HAWIE | — | Published in German | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Jones et al, 201596 | Brisbane, Germany | OSP | Mixed | 1.7 | 6.6 | 20 | WPPSI | BRIEF | — | ★☆☆ | ★★☆ | ★★ ☆☆ | ★☆ |
Jordan et al, 202030 | Nashville, Tennessee | RCT | HLHS | 0.2 | 11.2 | 20 | WISC-V | — | — | ★☆☆ | ☆☆☆ | ★★ ★★ | ☆☆ |
Karl et al, 200497 | Melbourne, Germany | Cross-sectional | TGA | 0.4 | 9.1 | 74 | WISC-III, WPPSI-III | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Kirshbom et al, 200598 | Philadelphia, Pennsylvania | Cross-sectional | TAPVR | 1.2 | 11 | 30 | WAIS | — | — | ★★★ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Ma et al, 202099 | Nanjing, China | TOF | 25.2 | 10.1 | 10 | WISC-R | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ | |
Mahle et al, 2000100 | Philadelphia, Pennsylvania | Cross-sectional | HLHS | 0.2 | 8.9 | 28 | WISC-III | — | — | ★★★ | ★★★ | ★★ ☆☆ | ★☆ |
Mahle et al, 2004101 | Atlanta, Georgia | OSP | Mixed | 113 | 9.4 | 29 | WAIS | BRIEF | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Mahle et al, 2006102 | Atlanta, Georgia | Cross-sectional | HLHS | 0.2 | 12.4 | 26 | WAIS | BRIEF | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Majnemer et al, 20087 | Montreal, Canada | OSP | Mixed | 2.7 | 5.4 | 94 | WPPSI-R | — | — | ★★☆ | ★★☆ | ★★ ★★ | ★☆ |
Matos et al, 2014103 | Porto, Portugal | Cross-sectional | VSD, TGA, TOF | NA | 15.02 | 77 | — | Wechsler’s Digit Span direct and reverse, Rey–Ostherieth Complex Figure Test, key search test from the BADS-C, Color-Word Stroop Test | EF results could not be extracted because only nonparametric test statistic results were provided | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Messmer et al, 1976104 | Zürich, Switzerland | Cross-sectional | Mixed | 3 | 6.5 | 11 | SON-R, HAWIK, Buhler-Hetzer | — | — | ★★☆ | ★☆☆ | ★★ ☆☆ | ☆☆ |
Miatton et al, 2007105 | Ghent, Belgium | Cross-sectional | Mixed | NA | 8.7 | 43 | WISC-III | Nepsy-I | — | ★★☆ | ☆☆☆ | ★ ☆☆☆ | ★☆ |
Miller et al, 1994106 | Hershey, Pennsylvania | Cross-sectional | Mixed | 1 | 5.5 | 20 | SB-I | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Mittnacht et al, 2015107 | Heidelberg, Germany | RCT | Mixed | 7.2 | 11.1 | 56 | WISC-III | — | — | ★★☆ | ★☆☆ | ★★ ☆☆ | ☆☆ |
Moerbeke et al, 1986108 | Brussels, Belgium | Cross-sectional | Mixed | 7 | NA | 31 | WISC | — | Published in French | ☆☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Munoz-Lopez et al, 2017109 | London, United Kingdom | Cross-sectional | TGA | NA | 11.4 | 40 | WISC-IV | — | — | ★★★ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Murphy et al, 2017110 | Nashville, Tennessee | Cross-sectional | Mixed | 3 | 16.1 | 18 | WISC-IV, WAIS-IV | WISC-IV working memory | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Naef et al, 201718 | Zürich, Switzerland | OSP | Mixed | 2.7 | 6.3 | 169 | WPPSI-III, SON-R, K-ABC | — | — | ★★☆ | ★★★ | ★★ ☆☆ | ★☆ |
Newburger et al, 1984111 | Boston, Massachusetts | OSP | TGA | 19.2 | 5.8 | 38 | WISC-III, Leiter International Scale, Grace Arthur Adaption | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
O'Dougherty et al, 1983112 | Minneapolis, Minnesota | Cross-sectional | TGA | NA | 9.1 | 31 | WPPSI-R | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Oates et al, 1995113 | Camperdown, Australia | Cross-sectional | ASD, VSD, TGA, TOF | 25.9 | 10.6 | 168 | WISC-R | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Oberhuber et al, 2017114 | Linz, Austria | OSP | HLHS | NA | 10.1 | 43 | WISC-V | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Poncelet et al, 2011115 | Brussels, Belgium, | RCT | Mixed | 19.2 | 7 | 36 | WPPSI-R, WISC-IV | — | — | ★☆☆ | ★☆☆ | ★★ ★★ | ☆☆ |
Quartermain et al, 2010116 | Philadelphia, Pennsylvania | Cross-sectional | Mixed acyanotic | 142 | 11.8 | 35 | WAIS | — | — | ★★☆ | ★★★ | ★★ ★★ | ★☆ |
Ratzmann et al, 1991117 | Leipzig, Germany | Cross-sectional | Mixed | NA | 13.3 | 62 | WISC | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Sahu et al, 2009118 | New Delhi, India | RCT | TOF | 106 | 8.8 | 80 | WISC | — | — | ★☆☆ | ☆☆☆ | ★★ ★★ | ★☆ |
Sakamoto et al, 2006119 | Tokyo, Japan | Cross-sectional | Mixed | 10.2 | — | 30 | WPPSI | — | — | ★☆☆ | ★★ ☆ | ★★ ☆☆ | ☆☆ |
Sanz et al, 2017120 | Washington | Cross-sectional | Mixed | NA | 9.1 | 91 | — | BRIEF | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
Sarajuuri et al, 2007121 | Helsinki, Finland | Cross-sectional | HLHS | 0.1 | 5.7 | 26 | WPPSI-R | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Sarajuuri et al, 2012122 | Helsinki, Finland | OSP | UVH, HLHS | 0.2 | 5.1 | 34 | WPPSI-R | — | — | ★★☆ | ★★☆ | ★★ ☆☆ | ☆☆ |
Sarrechia et al, 2013123 | Ghent, Belgium | Cross-sectional | ASD, VSD | 22 | 8.7 | 15 | WISC-III | Nepsy-I | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Sarrechia et al, 2015124 | Ghent, Belgium | Cross-sectional | ASD, VSD | 16.2 | 8.9 | 46 | WISC-III | — | Subgroups of ASD and VSD after catheter intervention were excluded | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Sarrechia et al, 2016125 | Ghent, Belgium | Cross-sectional | UVH | 0.8 | 9.1 | 17 | WISC-III | Nepsy-I | IQ outcome of biventricular heart disease subgroup extracted from124 | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Schäfer et al, 20133 | Zürich, Switzerland | OSP | Mixed | 10.8 | 13.6 | 59 | WISC-IV | WISC-IV working memory, Rey–Osterrieth Complex Figure test | — | ★★☆ | ★★☆ | ★★ ☆☆ | ★☆ |
Simons et al, 2010126 | Wilmington, Delaware | Cross-sectional | VSD | 19.9 | 8.7 | 13 | WISC-IV | BRIEF | Younger subgroup was excluded because of age outside the target range, BRIEF data were extracted for entire cohort | ★★★ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Spijkerboer et al, 2008127 | Rotterdam, Netherlands | Cross-sectional | ASD, VSD, TGA | NA | 12.1 | 103 | WISC-R | — | — | ★☆☆ | ★★★ | ★★ ☆☆ | ★☆ |
Stavinoha et al, 2003128 | Dallas, Texas | OSP | ASD | 102 | 8.5 | 18 | Differential Ability Scale | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Sterken et al, 2016129 | Leuven, Belgium | OSP | Mixed | 1.9 | 7.2 | 100 | WPPSI-R | ANT | — | ★☆☆ | ★★★ | ★★ ☆☆ | ★★ |
Sugimoto et al, 2013130 | Shizuoka, Japan | OSR | UVH | NA | 9 | 70 | WPPSI-R, WISC-III, WISC-IV | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Utens et al, 1993131 | Rotterdam, Netherlands | Cross-sectional | Mixed | NA | 13.3 | 287 | WISC-R short form, Groninger Intelligence Test | — | — | ★★☆ | ★★ ☆ | ★★ ★★ | ☆☆ |
Uzark et al, 1998132 | San Diego, California | Cross-sectional | UVH | NA | 6.3 | 32 | SB-I | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Vahsen et al, 2009133 | Sankt Augustin, Germany | Cross-sectional | TGA | 0.6 | 7.6 | 30 | K-ABC | — | Published in German | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Vahsen et al, 2018134 | Sankt Augustin, Germany | OSR | UVH | NA | 8.6 | 104 | K-ABC | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ★★ |
van der Rijken et al, 2008135 | Nijmegen, Netherlands | OSP | Mixed | NA | 11.6 | 43 | WISC-III | ANT, Rey–Osterrieth Complex Figure Test | — | ★★☆ | ★★☆ | ★★ ☆☆ | ☆☆ |
Venchiarutti et al, 2019136 | Udine, Italy | Cross-sectional | Mixed | 12 | 9.8 | 17 | WISC-IV | Nepsy-II | — | ★★☆ | ★★ ☆ | ★★ ☆☆ | ★☆ |
von Rhein et al, 2012137 | Zürich, Switzerland | Cross-sectional | Mixed | 13.2 | 10.4 | 117 | Raven‘s PM | Rey–Osterrieth Complex Figure Test | — | ★★☆ | ★★★ | ★★ ★★ | ★★ |
Wells et al, 1983138 | Brompton, United Kingdom | Cross-sectional | Mixed | 15.4 | 5.7 | 50 | McCarthy Scales of Children’s Abilities, WISC | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Wernovsky et al, 2000139 | Boston, Massachusetts | Cross-sectional | UVH | NA | 14.1 | 130 | WPPSI-R, WISC-III, WAIS-R | — | — | ★★★ | ☆☆☆ | ★★ ★★ | ★★ |
Whitman et al, 1973140 | Cleveland, Ohio | OSP | Mixed | NA | 9 | 11 | WISC | — | — | ★★☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Wolfe et al, 2020141 | Aurora, Colorado | Cross-sectional | UVH | NA | 9.55 | 68 | WISC-V, WAIS-IV, WISC preschool | Tower of London Drexel | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ☆☆ |
Wotherspoon et al, 2019142 | Brisbane, Australia | OSP | Mixed | 1.6 | 15.4 | 21 | WISC-IV, WAIS-IV | BRIEF | Longitudinal follow-up of Jones et al96 | ★★☆ | ★★☆ | ★★ ☆☆ | ★★ |
Wray et al, 2001143 | London, United Kingdom | OSP | Mixed | NA | 7.4 | 82 | BAS short | — | — | ★☆☆ | ★★★ | ★ ☆☆☆ | ★★ |
Wray et al, 2010144 | London, United Kingdom | Cross-sectional | Mixed | NA | 10.6 | 24 | BAS short | — | — | ★★☆ | ★★ ☆ | ★ ☆☆☆ | ★☆ |
Wright et al, 1994145 | Melbourne, Australia | OSR | Mixed | NA | 9.5 | 29 | WISC-R | — | — | ★★☆ | ☆☆☆ | ★★ ★★ | ★★ |
Yang et al, 1994146 | Changsha, China | Cross-sectional | Mixed acyanotic | NA | 8.1 | 39 | WPPSI-R, WISC-R | — | — | ★☆☆ | ☆☆☆ | ★★ ☆☆ | ★☆ |
Summary table and risk of bias assessment of all included studies. ANT, Amsterdam Neuropsychological Task; ASD, atrial septum defect; BADS-C, Behavioral Assessment of the Dysexecutive Syndrome; BAS, British Ability Scale; DKEFS, Delis-Kaplan Executive Function System; HAWIK, Hamburg Wechsler Intelligenztest für Kinder; IDS, Intelligence and Development Scales; K-ABC, Kaufman Assessment Battery for Children; OSP, observational study, prospective data collection; OSR, observational study, retrospective data collection; Ravens PM, Ravens Progressive Matrices; RCT, randomized controlled trial; SB, Stanford–Binet Intelligence Scales; SON-R, Snijders–Oomen nonverbal intelligence test; TAPVR, total anomalous pulmonary venous return; VSD, ventricular septum defect; WAIS, Wechsler Adult Intelligence Scale; WISC, Wechsler Intelligence Scale for Children; WPPSI, Wechsler Preschool and Primary Scale of Intelligence; —, not applicable.
Intellectual Functioning
Total IQ data were assessed in 85 studies, of which 74 could be included in the quantitative meta-analysis (reasons for exclusion: data were extracted from another report of the same cohort [n = 2]; IQ estimates were only reported in graphical form or only semiquantitatively [n = 4]; measure of dispersion was not reported [n = 4]), assessing a total number of 3645 children with CHD. The median age at assessment in the reported studies was 8.7 (IQR: 5–16.9) years. The overall estimate of total IQ in children with CHD was 96.03 (95% confidence interval [CI]: 94.91 to 97.14), falling within the normal range from 85 to 115. A high degree of heterogeneity between studies was found (Q = 1627.89; degrees of freedom [df] = 87; P < .0001; I2 = 94.7%; Fig 2). We found no evidence for a reporting and publication bias (z = −1; P = .3). The CHD subgroup had a modifying effect on the reported total IQ (QM = 35.26, df = 6, P < .0001), with studies in children with HLHS (mean total IQ: 88.47; 95% CI: 84.39 to 92.55) and UVH (mean total IQ: 92.65 [95% CI 90.00 to 95.29]) reporting significantly lower total IQ scores, compared with that of ASD/VSD subtypes (mean total IQ: 98.51 [95% CI: 95.83 to 101.20]; β = −10.04 [95% CI: −15.5 to −4.58, P = .0003]; β = −5.87 [95% CI: −11.07 to −0.66; P = .027]; Supplemental Table 5). We found evidence for a negative association of age at assessment with total IQ scores in TGA (β = −1.62 [95% CI: −2.9 to −0.33; P = .014]) but in no other CHD subgroups (Fig 3). No evidence was found for an association between age at surgery or year of surgery and reported total IQ scores. When only considering studies that included healthy controls as a comparison group (n = 22; 874 subjects with CHD), we found a SMD in total IQ of −0.85 (95% CI :−1.08 to −0.62). Given the mean SD of 11.62 in healthy controls, this corresponded to a decrease of ∼9.9 IQ points in children with CHD, compared with that of healthy controls (Supplemental Fig 6). The subgroup effect was also evident (QM = 39.73; df = 5; P < .0001) with cohorts with HLHS and TOF scoring poorer on total IQ tests, compared with ASD/VSD cohorts (β = −1.67 [95% CI: −2.38 to −0.96; P < .0001]; β = −1.04 [95% CI: −1.8 to −0.27; P = .008]).
Forest plot showing the results of studies assessing total IQ in school-aged children and adolescents with CHD. Studies are grouped by reported CHD subtype and sorted by increasing age at assessment. Subgroups within the same study are denoted by consecutive numbers after the publication year. Black squares and lines for each study correspond to reported mean and 95% CI. If >2 studies were available for 1 subtype, black diamonds for each CHD subtype represent the estimated mean and 95% CI based on the random-effects model. The blue diamond corresponds to the overall estimated mean IQ. Weights were adjusted to account for longitudinal reports of the same cohort. “Mixed” indicates a heterogenous group of different CHD subtypes. TAPVR, total anomalous pulmonary venous return.
Forest plot showing the results of studies assessing total IQ in school-aged children and adolescents with CHD. Studies are grouped by reported CHD subtype and sorted by increasing age at assessment. Subgroups within the same study are denoted by consecutive numbers after the publication year. Black squares and lines for each study correspond to reported mean and 95% CI. If >2 studies were available for 1 subtype, black diamonds for each CHD subtype represent the estimated mean and 95% CI based on the random-effects model. The blue diamond corresponds to the overall estimated mean IQ. Weights were adjusted to account for longitudinal reports of the same cohort. “Mixed” indicates a heterogenous group of different CHD subtypes. TAPVR, total anomalous pulmonary venous return.
Meta-regression assessing the effect of age at assessment on reported total IQ scores in cohorts with TGA. The size of the circles corresponds to the size of the studies. Solid lines correspond to regression lines and dashed lines indicate the upper and lower boundary of the 95% CI.
Meta-regression assessing the effect of age at assessment on reported total IQ scores in cohorts with TGA. The size of the circles corresponds to the size of the studies. Solid lines correspond to regression lines and dashed lines indicate the upper and lower boundary of the 95% CI.
Verbal IQ data were assessed in 37 studies, of which 32 were included in the quantitative analysis (reasons for exclusion: data extracted from another report of the same cohort [n = 1]; measure of dispersion not reported [n = 3]; outcome not reported on standardized scale [n = 1]), assessing a total number of 1795 children with CHD. Median age at assessment in the reported studies was 8.18 (range: 5–16.9) years. The overall estimate of verbal IQ in children with CHD was 95.5 (95% CI: 92.14 to 98.86) and within the normal range with high heterogeneity between studies (Q = 1105.92; df = 39; P < .0001; I2 = 96.5%) No evidence for a modifying effect of CHD subgroup on the reported verbal IQ was identified (QM = 8.43; df = 6; P = .21). Age at assessment, age at surgery, and year of surgery were not associated with verbal IQ. No evidence for publication bias was found (z = 0.84; P = .4)
Performance IQ data were assessed in 27 studies, of which 23 were included in the quantitative meta-analysis (reasons for exclusion: data extracted from another report of the same cohort [n = 1]; measure of dispersion not reported [n = 2]; outcome not reported on standardized scale [n = 1]) assessing 1258 children with CHD. The median age at assessment in the reported studies was 8.2 (range: 5–16.9) years. The overall estimate of performance IQ in children with CHD was 96.61 (95% CI 94.59 to 98.64) and within the normal range with high heterogeneity between studies (Q = 147; df = 30; P < .0001; I2 = 79.6%). Evidence for a modifying effect of CHD subgroup on reported performance IQ was identified (QM = 14.93; df = 6; P = .021) with children with HLHS (mean performance IQ: 88.3 [95% CI: 83.33 to 93.35]) performing worse than ASD/VSD CHD subtypes (mean performance IQ: 100.46 [95% CI: 99.28 to 101.64]; β = −12.12 [95% CI: −20.17 to −4.07; P = .003]; Supplemental Table 5). Performance IQ improved with year of publication (β = 1.55 [95% CI: 0.27 to 2.82; P = .017]) and mean year of surgery (β = 2.5 [95% CI 0.76 to 4.25; P = .0049]) in cohorts with UVH, whereas no association with age at assessment or surgery was found for UVH or any other CHD subtype. For performance IQ outcomes, no evidence for publication bias was found (z = −1.76; P = .079).
Executive Functioning
Behavior rated EF by means of the self-reported BRIEF was assessed in 6 studies, of which 4 were included in the quantitative meta-analysis (reasons for exclusion: no global score reported [n = 1]; test performed but results not reported [n = 1]), assessing an overall number of 301 subjects. The median age at assessment in the reported studies was 15.12 (range: 14.6–16.06) years. The overall estimate of the self-reported Global Executive Composite Score in children with CHD was 51.31 (95% CI: 49.07 to 53.55), falling in the EF difficulties range, with higher scores indicating more EF difficulties. Moderate heterogeneity between studies (Q = 6.15; df = 3; P = .1; I2 = 51.3%) and weak evidence for publication bias (z = 1.84; P = .065) was found (Supplemental Fig 7). The parent-reported BRIEF was used in 12 studies, of which 10 were included in the quantitative analysis (reasons for exclusion: no Global Executive Composite Score reported [n = 1]; test performed but results not reported [n = 1]) reporting outcomes of 652 subjects with CHD. The median age at assessment in the reported studies was 13.2 (range: 6.6–16.06) years. The overall estimate of the parent-reported Global Executive Composite Score was 55.63 (95% CI 54.26 to 56.99), indicating EF difficulties. Heterogeneity (Q = 16.01; df = 9; P = .067; I2 = 43.8%) was moderate, and no evidence for publication bias was found for parent-reported BRIEF (z = −1.44; P = .15) (Fig 4).
Forest plot showing the results of studies assessing parent-rated behavioral EFs evaluated with the BRIEF. Studies are sorted by increasing age at assessment. Global Executive Function Composite scores >50 indicate EF difficulties and >65 indicate clinically significant executive dysfunction. Black squares and lines for each study correspond to the reported mean and 95% CI of composite T scores. The black diamond represents the estimated summary mean and 95% CI based on a random-effects model.
Forest plot showing the results of studies assessing parent-rated behavioral EFs evaluated with the BRIEF. Studies are sorted by increasing age at assessment. Global Executive Function Composite scores >50 indicate EF difficulties and >65 indicate clinically significant executive dysfunction. Black squares and lines for each study correspond to the reported mean and 95% CI of composite T scores. The black diamond represents the estimated summary mean and 95% CI based on a random-effects model.
Performance-based EF and a respective healthy control group were reported in 16 studies, of which 13 were included in the quantitative analysis (reasons for exclusion: only Rey–Osterrieth complex figure tested [n = 2]; only nonparametric test statistic values reported, which did not allow for the calculation of the SMD [n = 1]), assessing a total number of 774 unique children with CHD. The median age at assessment was 9.76 (range: 5–16.1) years. The mean SD in healthy controls was 14.21, and the SMD of EF was found to be −0.56 (95% CI −0.65 to −0.46) in CHD in comparison with healthy controls. No difference in the examined EF subdomain was evident (QM = 2.28, df = 4, P = .69) (Fig 5). The heterogeneity of the analysis was moderate (Q = 206.2; df = 73; P = 1.2e−14; I2 = 64.5%). No evidence was found for an association of age at assessment with the SMD, other risk factors were not tested because of the small number of included studies. No evidence for publication bias was found (z = 0.14; P = .89). When only assessing the main domains of EF (ie, working memory, inhibition, and flexibility), the moderate effect or poorer EF performance in CHD remained, with a SMD of −0.55 (95% CI: −0.66 to −0.45).
Forest plot showing the results of studies assessing performance-based EF in school-aged children and adolescents with CHD, in comparison with that of healthy peer controls. Studies are grouped by assessed EF domain and sorted by increasing age at assessment. Subgroups and subtests within the same study are denoted by consecutive numbers after the publication year. Black squares and lines for each study correspond to the SMD and 95% CI. Black diamonds for each EF subdomain represent the estimated SMD and 95% CI based on the random-effects model. The blue diamond corresponds to the overall estimated SMD. Weights were adjusted to account for multiple subtests being reported in specific cohorts and domains.
Forest plot showing the results of studies assessing performance-based EF in school-aged children and adolescents with CHD, in comparison with that of healthy peer controls. Studies are grouped by assessed EF domain and sorted by increasing age at assessment. Subgroups and subtests within the same study are denoted by consecutive numbers after the publication year. Black squares and lines for each study correspond to the SMD and 95% CI. Black diamonds for each EF subdomain represent the estimated SMD and 95% CI based on the random-effects model. The blue diamond corresponds to the overall estimated SMD. Weights were adjusted to account for multiple subtests being reported in specific cohorts and domains.
Sensitivity Analysis for the Inclusion of Articles in Foreign Language
When excluding the 3 non-English publications (all German) that were included in the quantitative analysis, the estimate of total IQ was 96.1 (95% CI: 94.96 to 97.24), verbal IQ was 95.21 (95% CI 91.76 to 98.66), and performance IQ was 96.6 (95% CI: 94.53 to 98.67). None of the non-English publications assessed EF.
Risk of Bias Assessment
The risk of bias assessment revealed large differences in risk of bias ratings between studies and also within studies and rating categories, with risk of bias being evident in all domains without a specific predilection. The majority of studies had a cross-sectional study design (50.3%). Details on risk of bias assessment for individual studies can be found in Table 1.
Discussion
In this systematic review and meta-analysis, summarizing a large body of literature, we found consistent evidence for an impairment in cognitive and EF outcomes in school-aged children with complex CHD. The severity and trajectory of intellectual functioning was modified by CHD subtype. Although total and performance IQ were significantly lower in HLHS and UVH cohorts, compared with that of ASD/VSD subtypes, UVH cohorts showed a significant secular trend, with improvement of performance IQ in more recent publications. An older age at assessment in cohorts with TGA was associated with lower total IQ scores. Furthermore, impairments in all subdomains of daily-life behavioral-rated and performance-based EF were found without a specific impairment profile.
In comparison with the normative mean of 100, intellectual functioning in children with CHD was within the normal range but reduced by 4 to 5 IQ points. These results are in line with a previous review on developmental outcomes in children with CHD that reported a reduction in 5 IQ points and that only included studies from the last 30 years with moderate to high quality.2 Although a few IQ points might seem like a small effect, one needs to consider the large lifetime cost of lower IQ for the individual, families, and, cumulatively, society resulting from reduced academic attainment and professional productivity.34 The clinical significance of the cognitive deficit in children with CHD was further substantiated by the comparison of intellectual functioning to healthy peer controls, revealing a large effect size in SMD, corresponding to a reduction of 9.9 IQ points in children with CHD. On the one hand, the direct comparison with healthy peer controls might bear the risk of overestimating the IQ deficit by comparison with “supernormal” controls, resulting from a potential selection bias toward well-functioning healthy controls.35 On the other hand, matching children with CHD to healthy peers enables controlling for confounders, such as sociodemographic factors or the Flynn effect.36 Despite the inability to finally determine the true effect size, the clinical significance of the reduction in IQ points is further underscored by the increased need for remedial services and educational support in children with CHD.37,38 Of course, concomitant problems in other developmental domains, such a behavior or EF,8 may accentuate the effect of IQ reduction on the schooling situation of children with CHD. More studies that recruit healthy peer controls matched for socioeconomic factors are needed to further explore intellectual functioning of children with CHD in the context of their daily life.
Conducting the prespecified subgroup analysis, we found that the overall effect of reduced IQ scores was modified by CHD subtype, with significantly lower total IQ and performance IQ scores reported in the most severe types of CHD (ie, HLHS and UVH scoring in the lower normal range from 88 to 93, compared with milder CHD subtypes, such as ASD/VSD scoring between 99 and 100). Accordingly, the proportion of children with HLHS and UVH performing in the abnormal IQ range (IQ: <85) is larger, likely corresponding to a higher need for educational support.39 The observed cognitive impairment in HLHS and UVH subtypes is in accordance with previous reports on a cohort40,41 and systematic review level,42 reporting a mean IQ between 85 and 96. However, our analysis is the first direct CHD subgroup comparison on a meta-analytical level, thus adding further evidence to the previously described increased risk for cognitive deficits in the most severe CHD subtypes. Previous studies revealed that aberrant brain development and an increased rate of brain injury4,43–48 might underly the increased risk for adverse cognitive development in children and adolescents with CHD. Notably, the most detrimental effects on brain development and brain structural integrity have been observed in children with HLHS49 and might explain the here observed CHD subtype dependent differences in long-term neurocognitive outcome.
Our findings underscore the need for neurocognitive follow-up, particularly in severe types of CHD. Yet, it is important to note that, whereas children with TGA performed predominantly on or even above average on total IQ tests, meta-regression revealed a decrease in total IQ scores with older age at testing. This effect may be due to increasing demands on processing speed and working memory proficiencies that increasingly determine total IQ test results at older ages. However, caution is warranted when interpreting this finding because only a few studies assessed cognitive functioning in TGA children beyond the age of 11. More research in TGA adolescents is needed to confirm this result; in particular, longitudinal cohort studies are needed in which individual trajectories can address the question whether there is a decrease in intellectual functioning with older age. A similar age effect was found in children with HLHS in another review42 but could not be reproduced in our analysis, likely because of a smaller age range of cohorts in our review. Our finding highlights that sequential and long-term follow-up of all complex CHD subtypes into adulthood is imperative to better understand the long-term trajectories of cognitive outcomes in this vulnerable population and early identify the need for support. Assessing the change in performance IQ with later year of publication and surgery, we found that outcomes in cohorts with UVH improved over the last 2 decades. A similar finding has previously been reported in early infancy outcomes in CHD50 and, although only evident in this severe subgroup, might reflect the significant progress in surgical and intensive care management during the last decades.
Daily-life EF difficulties were substantially elevated in parental proxy reports by one-half an SD, in comparison with the normative mean of 50, whereas there was no evidence for compromised daily-life EF in self-reports of adolescents with CHD. Higher scores in parental, compared with that of self-reported, EF functioning have been described previously,51 indicating that parents perceive their child’s difficulties as more severe than the child or adolescent itself. Yet, despite the importance of daily-life EF for academic achievement and psychosocial well-being,9,10 the number of studies remains scarce. More research is needed, particularly focused on self-reported EF in older school-aged children and adolescents.
For studies measuring EF performance objectively by standardized neuropsychological tests, our review revealed that performance-based EF was lower in patients with CHD, compared with that of healthy peers. There was no evidence for a specific EF impairment profile.
EF is an umbrella term for a multidimensional set of higher-order cognitive and behavioral skills that are key to goal-directed behavior. However, the set of functions that are considered as EF has not been conclusively defined, and the terminology and underlying definition vary between the different neuropsychological test batteries. Nevertheless, there has been consensus to some extent: inhibition, working memory (also known as updating), and cognitive flexibility (also known as shifting) are often considered core domains of EF,27,52 which are correlated yet separable.53 Fluency and planning are of higher complexity and tap multiple EF processes.54 We therefore only considered EF measures that could be categorized in 1 of the 3 main subdomains of EF (ie, inhibition, working memory, and cognitive flexibility)27 and included more complex measures (ie, fluency and planning) in a subsequent analysis step. Nonetheless, given the differences in the conceptualization and operationalization of EF and the task impurity problem, the categorization is not perfect, with subdomains not being independent from each other.55 This might explain why we did not find subdomain differences.
Recently, it has been demonstrated that cognitive impairments in patients with CHD, including executive dysfunction, persist into adulthood.56–58 A number of studies have provided valuable insights into the potential underlying mechanisms of impaired EF in CHD. Global EF have been shown to be associated with altered white matter microstructural integrity48,59 and reduced cerebellar57 and total brain volume.60 Furthermore, working memory, a core component of EF, was associated with lower total and hippocampal volume,61,62 lower prefrontal microstructural integrity,48 and altered frontoparietal blood-oxygen-level-dependent signal.63 Whereas impairments in EF in CHD and their underlying neuronal correlates become more and more recognized, to date there are only 2 studies in which researchers investigated the feasibility and efficacy of a training intervention to improve EF (specifically, working memory performance).29,30 Although both studies demonstrated the feasibility of a working memory intervention with a computerized training program (Cogmed; http://www.cogmed.com), Calderon et al29 found no evidence for a postinterventional improvement of working memory and the initial progress reported in Jordan et al30 was not maintained at 6-months follow-up. This indicates that more research is urgently needed to assess how higher-order cognitive skills can be modified, resulting in long-lasting effects on downstream outcomes, such as academic and professional achievement.
Socioeconomic status is one of the strongest predictors of long-term neurocognitive outcome in CHD64,65 and other at-risk populations.66,67 However, despite the importance of this factor and our intention to analyze an association with reported outcomes,17 the lacking consensus on how to conceptualize and measure socioeconomic status hindered further analysis in this study.
The body of evidence of the 2 main outcomes, cognitive functioning and EF in CHD, reviewed in this study can be rated following the Grading of Recommendations Assessment, Development, and Evaluation guidelines.68 Most studies were observational, with prospective, cross-sectional, and retrospective study designs, which is inevitably linked to the observational nature of cognitive and EF outcome in a congenital condition. This results in a low baseline rating of the quality of evidence. For intellectual functioning, the observed effects were large and included a gradient effect, with the most severe cohorts of CHD scoring the lowest on the cognitive tests. The heterogeneity between studies was large, which could only be partly explained by effects of the CHD subgroup and a broad age range. Taking those aspects into consideration, the quality of the evidence on intellectual functioning in CHD should be upgraded to moderate. For the primary outcome of EF, the effects were also large with high precision, in sum resulting in an upgrade from a low baseline to a moderate degree of evidence. The above-mentioned heterogeneity of studies was also reflected in the risk of bias assessment, with large rating differences between studies but also within studies and between different risk of bias categories. The majority of studies were cross-sectional and selection and attrition bias were present in many studies, resulting in a risk of bias toward well-functioning children with CHD. Therefore, further high-quality prospective research studies are needed to better estimate the effects of CHD on intellectual and EF outcomes.
This systematic review has several strengths and limitations that need to be considered. The protocol for this systematic review was prepublished and preregistered, thus providing a high degree of methodologic transparency and reproducibility of the results. No restriction on the year of publication or publication language was applied, resulting in a rigorous and complete review of the existing body of literature. As a limitation, screening, data extraction, and bias assessment were not performed in duplicate by 2 independent reviewers for all studies because of the large number of eligible studies. However, processes were predefined, thoroughly piloted, and refined. There was a large methodologic heterogeneity between studies regarding EF assessment. Thus, comparison between studies could only be meta-analyzed by using the scale-free comparison of SMD, consequently limiting the analysis to studies that reported a healthy control group. The vast majority of studies included in this systematic review were conducted in high-income countries, which profoundly limits the generalizability of the results to the global prevalence and severity of cognitive impairments in CHD.
Conclusions
In our systematic review and meta-analysis, we demonstrated that intellectual and EF outcomes in school-aged children with CHD who underwent cardiopulmonary bypass surgery are significantly impaired without an identifiable specific EF impairment profile. Our analysis is the first direct CHD subgroup comparison on a meta-analytical level and reveals that more severe CHD subtypes show poorer outcomes. Older age at assessment was associated with poorer total IQ in TGA cohorts, underscoring the need for long-term comprehensive follow-up assessments. A positive secular trend could be observed for children with UVH, showing improvements in reported performance IQ for more recent years of publication and surgery, suggesting a positive effect of improved surgical and intensive care management during the last decades. Importantly, there was a large methodologic heterogeneity, in particular for studies reporting EF, with few reports including healthy peer controls. Therefore, it is imperative to establish a uniform definition of EF. Furthermore, recently published assessment guidelines56 for school-aged children with CHD need to be implemented and will enable comparability across future studies. This will be key to promoting research in the field and our current understanding of the complex phenotype of cognitive impairments in school-aged children with CHD.
Acknowledgments
We thank all the corresponding authors who provided additional information after request and helped completing the quantitative analysis. We are particularly grateful to our colleagues Dr. Milka Pringsheim, German Heart Center Munich; Dr. Anna-Maria Szela, DRK Kliniken Berlin Westend; Nil Dülek, Faculty of Medicine, University of Zürich; Dr Daniel Drozdov, University Children’s Hospital Zürich; Dr Tanja Kakebeeke, University Children’s Hospital Zürich; Leila Thalparpan, University Children’s Hospital Zürich; Jessie Guo, PhD, SickKids Hospital Toronto; Dr Min Sheng, SickKids Hospital Toronto; Dearit Tesfay, Faculty of Medicine, University of Oslo; Camille Ammann, Faculty of Medicine, University of Zürich; Daniel Gero, PhD, Universitätsspital Zürich, who helped us screen articles in foreign language for their eligibility and extract the data.
Dr Feldmann drafted the manuscript, helped conceptualizing and designing the study, and conducted the data extraction and analysis; Ms Bataillard drafted the manuscript, performed the screening and bias assessment, and helped conceptualizing and designing the study; Ms Ullrich helped conceptualizing and designing the study and screen the eligible studies; Ms Ehrler and Dr Knirsch helped interpreting the results and finalizing the manuscript; Dr Gosteli-Peter set up the search strategy and conducted the literature search; Dr Held helped conceptualizing the study and conducting the statistical analyses; Dr Latal helped conceptualizing and designing the study, interpreting the results, and finalizing the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
This trial has been registered with the International Prospective Register of Systematic Reviews (https://www.crd.york.ac.uk/prospero/) (identifier CRD42019118736).
FUNDING: No funding was secured for this study. Maria Feldmann was funded by the Anna Mueller Grocholski Foundation. The funding had no influence on study design, analysis or interpretation of the results.
- ASD/VSD
atrial septum or ventricular septum defect
- BRIEF
Behavior Rating Inventory of Executive Function
- CHD
congenital heart disease
- CI
confidence interval
- df
degrees of freedom
- EF
executive function
- HLHS
hypoplastic left heart syndrome
- IQR
interquartile range
- SIGN
Scottish Intercollegiate Guidelines Network
- SMD
standardized mean difference
- TGA
d-transposition of the great arteries
- TOF
tetralogy of Fallot
- UVH
univentricular heart disease
References
Competing Interests
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
Comments