Academic achievement and needs of school‐aged children born with selected congenital anomalies: A systematic review and meta‐analysis

Abstract Children with congenital anomalies have poorer intellectual and cognitive development compared to their peers, but evidence for academic achievement using objective measures is lacking. We aimed to summarize and synthesize evidence on academic outcomes and special education needs (SEN) of school‐aged children born with selected major structural congenital anomalies. Electronic databases (MEDLINE, EMBASE, Scopus, PsycINFO, CINAHL, ProQuest Natural Science and Education Collections), reference lists and citations for 1990–2020 were systematically searched. We included original‐research articles on academic achievement in children with non‐syndromic congenital anomalies that involved school test results, standardized tests and/or SEN data. Random‐effects meta‐analyses were performed to estimate pooled mean test scores in mathematics and/or reading where possible and pooled odds ratios (ORs) for SEN in children with severe congenital heart defects (CHDs) and children with orofacial clefts (OFCs). Thirty‐nine eligible studies (n = 21,066 children) were synthesized narratively. Sixteen studies were included in meta‐analyses. Children with non‐syndromic congenital anomalies were at a higher risk of academic underachievement than controls across school levels. Children with severe CHD (pooled OR = 2.32, 95% CI: 1.90, 2.82), and children with OFC (OR = 1.38 (95% CI: 1.20, 1.57), OR = 3.07 (95% CI: 2.65, 3.56), and OR = 3.96 (95% CI: 3.31, 4.72) for children with cleft lip, cleft palate and cleft lip/palate, respectively) had significantly higher ORs for SEN than controls. Children with non‐syndromic congenital anomalies underperform academically and have higher SEN rates compared to their peers. Early monitoring and development of differential SEN are important to promote academic progress in these children.


| INTRODUCTION
Long-term (beyond infancy) survival of children born with congenital anomalies (CAs) has improved over the last three decades due to advances in neonatal care and operative interventions (Cassina et al., 2016;Erikssen et al., 2015;Glinianaia et al., 2020;Shin et al., 2012) resulting in an increasing number of children reaching school age. The association with intellectual and learning disabilities has long been established for children with CAs associated with chromosomal and genetic syndromes. Current evidence suggests that children with non-syndromic/isolated CAs have a higher risk of lower academic achievement and special education needs (SEN) than the reference population. Poorer academic performance is not restricted to children with more severe CAs characterized by lower survival (e.g., severe congenital heart defects, CHDs (Mulkey et al., 2016;Olsen et al., 2011;Oster, Watkins, Hill, Knight, & Meyer, 2017) but is also reported for children with CAs with higher survival (e.g., isolated orofacial clefts, OFCs [Fitzsimons et al., 2018;Fitzsimons et al., 2021;Persson, Becker, & Svensson, 2012;Wehby et al., 2014]) compared to their classmates.
Earlier studies exploring intellectual and cognitive development of children born with CAs used data from parent-report questionnaires, including a school component, which lacked objective measures of a child's academic performance. Evidence is accumulating from studies using standardized tests measuring academic performance and school tests in larger populations of children with CAs. A summary of the existing evidence of academic performance and SEN in children with different types of CAs, along with the factors associated with educational outcomes, are important for families, health and social care professionals and school team members. This information will help identify timely and effective support during the child's school life to improve academic achievement in this growing population of children and young people.
We performed a systematic review and meta-analysis of observational studies to summarize and synthesize data on academic outcomes of school-aged children born with selected major non-syndromic structural CAs compared to controls or age-matched referent children. This work was undertaken as part of the European collaborative project EUROlinkCAT (https://www.eurolinkcat. eu/).

| Search strategy and selection
The reporting of this systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline 2009 (Table S1). The review was registered on the PROSPERO database (CRD42017080250). We conducted comprehensive literature searches using: (1) electronic bibliographical databases, (2) reference lists of included papers and relevant literature reviews, (3) citations of included studies (via Google Scholar).
We searched seven electronic databases: MEDLINE, EMBASE, Scopus, PsycINFO, CINAHL, ProQuest Natural Science, and Education Collections. We used keywords and subject headings combining the keywords for the population (child/adolescent/school student/youth), exposure (CAs/birth defects, including specific anomalies, for example, spina bifida, cleft lip), outcome (school/ education/academic achievement/performance/special education) and study design (observational studies), incorporating elements of the PICOS (Population/ Patient; Intervention/Exposure; Comparator group; Outcome; Study design) framework into our search strategy (Moher et al., 2009) (Table S2). Authors were contacted if any clarification or additional information was needed.
Screening of all titles and abstracts to identify papers for full text review was performed by the first author, while a random 60% sample of records was screened independently by other authors, using the Rayyan software (Ouzzani, Hammady, Fedorowicz, & Elmagarmid, 2016) to ensure consistency in study selection.

| Eligibility criteria
Studies were included if they were: (1) observational, peer-reviewed reporting educational outcomes, that is, academic achievements and/or SEN of school-aged children (from 4-6 to 18 years old) born with a major structural CA (as defined by the European Surveillance of Congenital Anomalies, EUROCAT [EUROCAT, 2013a], https://eu-rd-platform.jrc.ec.europa.eu/eurocat_en); (2) reporting quantitative measurements of academic performance (e.g., school test scores or questionnaire-based scores using standardized tests of academic achievement) in children with CAs versus reference/control group or local normative data, or comparing SEN rates with reference groups; (3) published from January 1, 1990 to November 30, 2020 (from 1990 a more inclusive practice towards special education was encouraged [UNESCO, 1994]); and (4) published in the English language.
Studies were excluded if they were: (1) restricted to pre-school children or adult patients; (2) questionnairebased studies exploring quality of life, including school functioning component, executive function, intellectual, cognitive or speech/language development; (3) intervention studies, qualitative studies, case reports or small case series (≤10 cases); (4) reporting educational outcomes in children with conditions other than CAs (e.g., autism, cerebral palsy, intellectual disability) or in children with chromosomal, genetic or teratogenic syndromes known to be associated with lower academic achievement (e.g., Down syndrome, neurofibromatosis, skeletal dysplasia, fetal alcohol syndrome); (5) restricted to a specific patient sub-group (e.g., preterm births, heart transplant recipients). Based on the results of preliminary searches, we included studies on children with more common and well-studied CAs; spina bifida (with/without hydrocephalus), CHDs and craniofacial anomalies.

| Definitions
We followed the EUROCAT definition for isolated "minor" anomalies defined as "those which do not in themselves have serious medical, functional or cosmetic consequences for the child," and their definition, diagnosis and reporting vary considerably (EUROCAT, 2013a) to exclude them from eligible CA types for this review.
According to the EUROCAT criteria, we defined severe CHD as including the following CHD types: common arterial truncus, double outlet right ventricle, transposition of great vessels, single ventricle, atrioventricular septal defect, tetralogy of Fallot, pulmonary valve atresia, triscuspid atresia and stenosis, Ebstein anomaly, hypoplastic right heart, aortic valve atresia/stenosis, mitral valve anomalies, hypoplastic left heart, coarctation of aorta, aortic atresia/interrupted aortic arch, total anomalous pulmonary venous return (EUROCAT, 2013b). We consistently use the term "severe CHD" despite variations in the terms used in the included studies. Specific CHD subtypes included in the severe CHD groups in the included studies corresponded to the EUROCAT criteria (see Appendix 1 for studies' definitions).

| Data extraction and quality appraisal
Information was extracted on study location, year of publication, study design, study size for cases and comparison groups, and exclusion criteria. We also extracted results of school tests or standardized tests measuring educational outcomes, age of children or school year/grade at assessment and SEN data. The results of the analysis of the association with risk factors were also reported.
The Newcastle-Ottawa Quality Assessment Scale (NOS) for cohort studies (Wells et al., n.d.) and an amended version for cross-sectional studies (Modesti et al., 2016), were used to assess study quality. The scale assesses information bias, selection bias and confounding (Table S3). The detailed scores are provided in Table S4.
Full text reviewing, data extraction and quality appraisal of the included studies were performed by two independent reviewers, and identified discrepancies resolved by consensus.

| Statistical analysis
Where three or more studies of children with a specific CA reported academic performance in reading and/or mathematics measured by standardized tests, or the percentage with SEN, a random-effects meta-analysis was performed to better account for heterogeneity between studies. Further subgroup analysis was performed where possible and heterogeneity was quantified using the I 2 statistic, with I 2 > 50% indicating significant . heterogeneity (Higgins, Thompson, Deeks, & Altman, 2003). Publication bias was investigated using Egger's test and funnel plots. Sensitivity analyses, whereby individual studies or subgroups are removed from the analysis and the effect size re-calculated, were also performed where possible. All analysis was conducted in Stata v16 (StataCorp LLC, 2019). Figure 1 shows the selection of studies from a total of 11,303 records identified for screening titles and abstracts. From 129 studies eligible for full-text review, 39 met the inclusion criteria, covering a total population of 21,066 children. If publications included an overlapping cohort, the main paper reporting unique educational outcomes of interest (e.g., at different school ages/grades (Fitzsimons et al., 2018;Fitzsimons et al., 2021) or schools tests results in one and SEN data in another paper (Watkins et al., 2018;Watkins et al., 2019)) was included. Sixteen studies were eligible for inclusion in one or more of five meta-analyses. Table 1 shows the description of the 39 included studies and the NOS quality scores. Studies differed by source of CA ascertainment (population-based registries (n = 15), multi-center (n = 8) or single-center (n = 16)), sample size, type of assessment of academic achievement (school tests (n = 13), questionnaire-based using standardized tests (n = 19) or SEN data only (n = 7)), school age and type of CAs. The included studies reported results for children with the following CAs: spina bifida (with/without hydrocephalus) (n = 6), CHDs (n = 15) and craniofacial anomalies, including OFCs (n = 15), craniosynostosis (n = 2), and craniofacial microsomia (n = 1) (Table 1). Most studies included isolated or nonsyndromic CAs (n = 34), while some also included additional structural or syndromic anomalies, but analyzed them separately (n = 4). One study (Fletcher et al., 2005) did not specify whether they excluded children with syndromes; however, as this study excluded children with severe intellectual disability, we included it as well (Table 1). Some studies also excluded specific groups of children known to be associated with lower academic achievement, such as preterm birth (<37 (Collett et al., 2010;Sarrechia et al., 2016), <36 (Hiraiwa et al., 2020) or <34 (Wright & Nolan, 1994) weeks' gestation) or those with low birth weight (<1,500 g (Mulkey et al., 2016) or <2,000 g (Sarrechia et al., 2016;Wright & Nolan, 1994)), analyzed these groups separately (Olsen et al., 2011), or adjusted for these factors in their analyses (Lawley et al., 2019;Oster et al., 2017;Watkins et al., 2019).
The need for special education, both in primary and secondary schools, reported in a Dutch study was substantially higher for children with spina bifida associated with hydrocephalus compared to the reference population and those without hydrocephalus (Barf et al., 2004) ( Table S5).
Factors that significantly reduced academic achievement (reading and calculations) in 11-year-old children with spina bifida (with hydrocephalus) were upper lesion level and Hispanic ethnicity (Fletcher et al., 2005). Shunting for hydrocephalus, IQ ≤85, and wheelchair dependency, were independent predictors of SEN after Adjusted for maternal education, race/ethnicity, public pre-Kindergarten enrolment, and gestational age. f Adjusted for group differences in socioeconomic status, gender and maternal education. g All ORs adjusted for year of birth, maternal age, parity, and maternal education. h Adjusted for maternal education, race/ethnicity, and public pre-Kindergarten enrolment. i Adjusted for age (continuous), gender, socioeconomic status (continuous), maternal IQ (continuous). j Adjusted for age at assessment (continuous), sex, race/ethnicity (white non-Hispanic, Hispanic, other), income (categorical), and primary caregiver's highest level of education (categorical). adjustment for lesion level, annual number of surgical interventions and incontinence (Barf et al., 2004) (Table S6).
Large studies from the USA, Denmark and Australia using school test results were consistent in reporting poorer academic performance in children with CHD compared to the reference children (Lawley et al., 2019;Mulkey et al., 2016;Olsen et al., 2011;Oster et al., 2017) (Table 2). Third-grade children (9-year-olds) with CHD had poorer performance in reading/literacy and mathematics/numeracy tests compared with their peers (Lawley et al., 2019;Mulkey et al., 2016;Oster et al., 2017). Moreover, when children with severe and non-severe CHD were analyzed separately, both CHD groups showed significantly poorer academic performance in reading and higher SEN rates (Oster et al., 2017) (Table S6).
Overall, studies using standardized tests were smaller than those using school tests or SEN data. Most of these studies reported poorer scores in reading and mathematics for children with severe CHD compared to controls (Bellinger et al., 2015;Mahle et al., 2000;Wright & Nolan, 1994). However, some reported lower scores in mathematics only (Wotherspoon et al., 2020) or comparable scores in any school subject in children with nonsevere CHD (Simons et al., 2010). In Switzerland, where a structured follow-up programme of children with CHD is well established, similar percentages of these children completed mandatory school at the high, medium or low  (Table 2); however, those with severe CHD were over-represented in the lower level education (p = .03) (Schaefer et al., 2016) (Table S6).
Presence of extracardiac CAs (Oster et al., 2017), genetic diagnosis (Bellinger et al., 2015) and CHD severity (Oster et al., 2017;Riehle-Colarusso et al., 2015;Schaefer et al., 2016;Wray & Sensky, 2001) were major risk factors of lower academic achievement or higher  (Table 2, Table S6). Other significant predictors of poorer academic achievement included factors related to surgery and hospitalizations: longer duration of hospitalization (Mulkey et al., 2016), >4 re-hospitalizations in their first six years (Lawley et al., 2019) and ≥2 complications at first surgery (Bellinger et al., 2015), adjusted in all studies for socioeconomic status (SES) or parental education. Receipt of free school meal as a proxy of lower SES (Mulkey et al., 2016), and low level of parental education (Lawley et al., 2019;Mulkey et al., 2016), were also significantly associated with lower academic achievement. Low birth weight was associated with underachievement in writing and numeracy after adjustment for parental education, student language at home and some clinical factors (Lawley et al., 2019) (Table S6).

| Craniofacial anomalies-orofacial clefts
In 15 studies of children with OFCs, nine used school tests or teacher's assessment at different age groups, four used standardized tests (Chapman, 2011;Collett et al., 2010;Hentges et al., 2011;Saervold et al., 2019), two analyzed SEN data only (Watkins et al., 2019;Yazdy et al., 2008), with five reporting both academic and SEN results (Collett et al., 2010;Fitzsimons et al., 2018;Fitzsimons et al., 2021;Hentges et al., 2011;Wehby et al., 2014) (Table 2, Table S5). Studies in Europe, USA and Australia using school test results were consistent in reporting lower achievement in children with OFC in various school age groups compared to controls (Bell et al., 2017b;Clausen et al., 2017;Fitzsimons et al., 2018;Fitzsimons et al., 2021;Grewal et al., 2020;Persson et al., 2012;Watkins et al., 2018;Wehby et al., 2014), however, the significant differences were not consistent for all academic domains and OFC types (Table 2). Findings from most studies of children with isolated OFC agreed that poorer academic outcomes were associated with cleft type, reporting a lower risk for children with cleft lip only, who often performed similarly to controls (Bell et al., 2017b;Clausen et al., 2017;Fitzsimons et al., 2018;Fitzsimons et al., 2021;Persson et al., 2012;Watkins et al., 2018;Wehby et al., 2014). There was less consistency for cleft palate and cleft lip with palate (CLP), with significantly higher risks for both groups reported in some studies (Fitzsimons et al., 2018;Fitzsimons et al., 2021), but for cleft palate only in others (Persson et al., 2012;Wehby et al., 2014) or CLP (Watkins et al., 2018). Longitudinal studies showed that children with isolated OFC were at higher risk of lower performance across all academic areas and grade levels, from elementary to high school (7-17 years), in particular for children with cleft palate (Wehby et al., 2014) (Table S6). In a recent Danish study, the OFC type was concluded to be more important for academic performance than timing and number of exposures to surgery and anesthesia (Clausen et al., 2017). Recent population-based English studies reported that academic achievement in 5-year-old (Fitzsimons et al., 2018) and 7-year-old (Fitzsimons et al., 2021) children with isolated OFC, particularly in those with palate involvement, was significantly lower in all school subject areas compared to the national average (Table 2). They also reported a significantly higher SEN rate in both age groups, with higher rates for children with a cleft involving the palate (Fitzsimons et al., 2018;Fitzsimons et al., 2021) (Table S5, Figure 4b). Both cleft type and school absence were independent risk factors for lower academic achievement after adjustment for SES and child's sex (Fitzsimons et al., 2021). SEN rates in children with OFC were also significantly higher compared to F I G U R E 5 Forest plots showing the odds ratios for special education needs (SEN) for children with severe congenital heart defects (CHD) versus controls controls in US studies (Collett et al., 2010;Hentges et al., 2011;Watkins et al., 2019;Wehby et al., 2014;Yazdy et al., 2008) (Figure 4b), ranging between 19.7% (Wehby et al., 2014) and 35.4% (Watkins et al., 2019) for all OFCs, rising to 56.5% for CLP and 41% in cleft palate (Collett et al., 2010). The higher SEN rates were consistent for children with non-syndromic OFC across all school levels (Wehby et al., 2014), remaining significant after exclusion of those receiving speech and language services (Watkins et al., 2019).
Children with OFCs from six studies included in the meta-analysis (n = 7,145) (Collett et al., 2010; F I G U R E 6 Forest plots showing the odds ratios for special education needs (SEN) for children with isolated/non-syndromic orofacial clefts (OFCs), by OFC type (cleft lip (CL), cleft palate (CP), cleft lip and palate (CLP) and mixed) versus controls. *The overall number of children by OFC type is given, as the number with known SEN status, which is lower by a total of 36 cases, is not reported and could not be obtained from the authors et al., 2021;Hentges et al., 2011;Watkins et al., 2018;Wehby et al., 2014;Yazdy et al., 2008) had significantly higher SEN rates than controls (OR 2.74, 95% CI 2.06, 3.65) ( Figure 6). Odds of SEN were significantly higher for all OFCs types; cleft lip: 1.38 (95% CI 1.20, 1.57; I 2 = 0.00%), cleft palate: OR 3.07 (95% CI 2.65, 3.56; I 2 = 23.58%), CLP: OR 3.96 (95% CI 3.31, 4.72; I 2 = 45.33%) with relatively low heterogeneity between studies. The high heterogeneity of the overall effect size estimate (I 2 = 95.23%) confirms the importance of distinguishing the risk between different OFC types. These results were robust to the exclusion of individual studies and subgroups in sensitivity analysis, and there was no evidence of publication bias (Table S7(2)).
Studies using standardized tests of academic performance in 7-year-old children with cleft lip with/without cleft palate reported significantly lower scores in spelling and mathematical reasoning compared to controls, but not in reading (Hentges et al., 2011) (Table 2). The metaanalysis of three studies (n = 149) (Chapman, 2011;Collett et al., 2010;Hentges et al., 2011) revealed a higher mean reading test score in children with CLP or cleft lip with/without cleft palate than in controls (101.83, 95% CI 99.31, 104.34; I 2 = 3.54%), but the difference was not statistically significant (Figure 7).
In addition to OFC type and school absence, other factors negatively affecting school attainment in children with OFC were: presence of associated anomalies (Bell et al., 2017b), male sex (Broder et al., 1998;Clausen et al., 2017), late timing of cleft repair (at 3-4 months versus neonatal) (Hentges et al., 2011), delayed speech (Chapman, 2011) and lower SES (Fitzsimons et al., 2018). Reduced speech intelligibility, poor oral health and ≥3 functional problems were associated with lower average point score in 7-year old children with unilateral CLP after adjustment for birth month, child's sex and SES (Grewal et al., 2020). SES measures such as area-based deprivation index (Bell et al., 2017b;Fitzsimons et al., 2018;Fitzsimons et al., 2021;Grewal et al., 2020), free school meal (Fitzsimons et al., 2018;Fitzsimons et al., 2021), family social class (Collett et al., 2010) or parental education (Clausen et al., 2017;Watkins et al., 2018;Wehby et al., 2014) were included in all multivariate analyses of the association between different OFC types and education outcomes (Table S6). Differences in SEN rates by ethnicity reported by two US studies were inconsistent (Watkins et al., 2019;Yazdy et al., 2008).

| Other craniofacial anomalies
A multi-center US study of children with single-suture craniosynostosis reported significantly lower scores in reading composite test and mathematics (Speltz et al., 2015). Unicoronal type of craniosynostosis was a risk factor for significantly lower scores in reading, spelling and mathematics compared to sagittal type after adjustment for age, sex, SES and maternal IQ (Speltz et al., 2015) (Table S6).
Children with craniofacial microsomia scored lower than test norms in reading, writing and mathematics; differences remained significant for reading and writing after adjustment for confounders (Speltz et al., 2017) ( Table 2).

| DISCUSSION
Most included studies of children with spina bifida with hydrocephalus, CHDs and OFCs found some degree of academic underperformance or higher SEN rates in these children compared to the referent children regardless of settings and measures of academic outcomes. The meta-F I G U R E 7 Forest plots showing the mean test scores in reading in children with non-syndromic orofacial clefts (cleft lip and palate (CLP) or cleft lip (CL) ± cleft palate (CP)) versus controls. TERA-3, Test of Early Reading Ability, third edition; WJTA-R, Woodcock-Johnson Tests of Achievement-Revised analysis results showed significantly higher pooled ORs for SEN in children with severe CHDs and children with OFCs, in particular for those involving palate, than in controls, with relatively low heterogeneity between studies. Lower academic achievement was reported across a number of academic domains, showing consistency across the school grades and levels in longitudinal studies and in studies analyzing different school ages. Longer school absence, specific anomaly type resulting in higher severity, presence of associated anomalies, some surgeryrelated factors, socioeconomic deprivation and lower parental education were the leading factors negatively affecting academic outcomes.
Neurodevelopmental and cognitive impairment found in toddlers and pre-school children with spina bifida and/or hydrocephalus that manifest in poor school achievement originates from fetal abnormal brain development typical for these CAs (Dennis & Barnes, 2010;Juranek & Salman, 2010). Indicators of brain injury resulting in poorer neurodevelopmental and educational outcomes are also described for such CAs requiring corrective surgeries in infancy such as severe CHDs (Gaynor, 2014;Griffin, Elkin, & Smith, 2003;Wray, 2006) and OFCs (Gallagher & Collett, 2019). Earlier studies of children with CHDs linked poorer neurodevelopmental outcomes with anesthesia-related neurotoxicity and other surgery-related factors, including cardiopulmonary bypass, therefore focusing on intra-and perioperative neuroprotection and neuromonitoring of infants with severe CHDs to prevent brain injury in these children (Hirsch et al., 2012). However, there is accumulating evidence that factors preceding surgery, that is, delayed intrauterine brain maturation and white matter injury resulting from impaired fetal hemodynamics due to CHD, consequent brain immaturity at birth and longer time before surgery, are primary major risk factors underlying hypoxic brain injury and subsequent poor neurodevelopmental outcomes after surgery (Bolduc, Lambert, Ganeshamoorthy, & Brossard-Racine, 2018;Gaynor, 2014). Therefore, preventive strategies are suggested to be more effective if initiated antenatally (Gaynor, 2014). Large recent population-based studies reported that while CHD severity was a major predictor of lower educational achievement (Oster et al., 2017;Schaefer et al., 2016;Wray & Sensky, 2001), non-severe CHDs were also associated with poorer academic achievement in two different age groups of children (Olsen et al., 2011;Oster et al., 2017). This suggests that children with mild/moderate CHDs also need assessment, monitoring and support from early childhood to adolescence.
Recent studies are strengthened by use of populationbased data and longitudinal design allowing follow up of school attainment of children with CAs compared to earlier smaller cross-sectional studies. In a longitudinal study, children with OFC, particularly with cleft palate, showed consistently lower achievement across school ages for reading, language and mathematics (Wehby et al., 2014) with persistent, low achievement trajectories after adjusting for SES (Wehby, Collett, Barron, Romitti, & Ansley, 2015), in contrast to earlier reports of lower rates of learning disability in adolescents compared to younger children (Richman, McCoy, Conrad, & Nopoulos, 2012). Higher odds of receiving lower graduation grades and not receiving leaving certificates by 16-year-old Swedish adolescents with OFCs compared to the general population (Persson et al., 2012) confirm persistent low achievement and the need for early screening and academic support in these children. Thus, receipt of SEN services decreased the likelihood of third-grade retention (9-year-olds) of children with OFC in an US study (Watkins et al., 2019).
There was strong evidence across the included studies that children with OFC involving palate are at a consistently higher risk of lower education outcomes than children with cleft lip only, irrespectively of school age or measures used (Bell et al., 2017b;Broder et al., 1998;Chapman, 2011;Clausen et al., 2017;Collett et al., 2010;Fitzsimons et al., 2018;Fitzsimons et al., 2021;Persson et al., 2012;Watkins et al., 2018;Wehby et al., 2014). Both cleft type and school absence were independent predictors for underachievement in children with OFC (Fitzsimons et al., 2021). Neither longer school absence, nor socioeconomic differences could explain poorer school attainment in an English study of 7-year old children with cleft palate and CLP compared with children with cleft lip (Fitzsimons et al., 2021). An Australian study found that school absence adversely affected academic performance in all secondary school children, not being differentially worse for children with OFCs (Bell et al., 2017a). Surgery-related factors, such as exposure to anesthesia and larger number of operations for OFC correction, had little impact on poorer academic performance compared to OFC type (Clausen et al., 2017). This was recently confirmed in large samples of healthy children that neither early (<2 years) exposure to anesthesia nor multiple exposures were major risk factors for adverse neurodevelopmental outcomes (Graham, 2017).
Much higher SEN rates, especially for children with OFC involving palate, were also consistently reported (Collett et al., 2010;Fitzsimons et al., 2018;Fitzsimons et al., 2021;Watkins et al., 2018;Wehby et al., 2014) and confirmed by the results of our meta-analysis. Although the most common was speech, language and communication services associated with developing reading skills (Collett et al., 2010;Fitzsimons et al., 2018;Watkins et al., 2019), after exclusion of children with this SEN type, the SEN rate in third-grade children (9-year-olds) with non-syndromic OFC was still higher than in controls (Watkins et al., 2019). More research is needed to explore the etiology of educational underachievement in children with non-syndromic OFCs, as a recent genomewide association study meta-analysis of non-syndromic clef lip/palate found little evidence for shared genetic etiology or causal relationship between this OFC type and educational attainment (Dardani et al., 2020).
For children with either CHD or OFC, CA severity was a major risk factor of lower academic achievement and higher SEN rates consistently reported in the included studies after adjustment for SES. The adjustment for SES is a valid approach recommended for any research on school performance due to a well-established SES-achievement relationship and the importance of social and economic context in understanding school achievement (Sirin, 2005).
Major strengths of our systematic review include a rigorous search strategy and comprehensive literature searches using multiple sources. Our search strategy incorporated elements of the PICOS framework based on strict inclusion criteria, was piloted using Medline, then refined and retested to ensure appropriate inclusiveness. We also manually searched the reference lists and citations of included papers, thus increasing the identification of relevant papers (Papaioannou, Sutton, Carroll, Booth, & Wong, 2010). Screening of titles and abstracts of identified records was performed by several authors and the results were compared to enable consistency in study inclusion. Detailed reviewing of the full texts of potentially suitable papers and data extraction were performed independently and in duplicate. We used an established quality assessment tool as part of the critical appraisal process, which was amended for cross-sectional studies, following advice from an information scientist.
We started literature searches from 1990 to ensure relative consistency among included studies towards special education, as in the mid-1990s more inclusive education for children with intellectual disability and SEN was encouraged following the 1990 World Conference on Education for All and the 1994 UNESCO statement (UNESCO, 1994). We limited the included studies to those with isolated/non-syndromic anomalies to avoid bias, as multiple anomalies, chromosomal and genetic syndromes are known to be associated with poorer academic achievement. Our inclusion criteria restricted measurements of academic attainment to the school tests or standardized tests of education achievement, as questionnaire-based subjective measures do not always coincide with the objective measures, because parents of children with a major CA (e.g., severe CHD) tend to overestimate their child's school performance due to their lower expectations (Mahle et al., 2000).
However, there were several limitations. Due to substantial study heterogeneity in measures of academic achievement for a specific academic domain or school age, we had to restrict the meta-analysis to studies on relatively common CAs (e.g., OFCs) with sufficient data for a specific subgroup (e.g., CLP) that used age-standardized tests with unified test norms (mean and SD) in a specific academic domain (e.g., reading). Due to the small number of studies suitable for meta-analysis, subgroup analysis was rarely feasible, and exploration of possible publication bias with funnel plots and Egger's test is generally discouraged in meta-analyses that include fewer than ten studies. Similarly, the small number of included studies precluded the use of meta-regression to explore the high heterogeneity seen in some analyses. The included studies did not always give substantial detail of their definition and selection criteria of isolated and nonsyndromic CAs, therefore we cannot exclude a proportion of children with associated CAs in some included studies which may have affected the results. In addition, we were unable to assess the contribution of such potential pathways of academic underachievement in children with specific CAs as surgery and anesthesia or potential mediation effects via some psychological factors (selfconfidence or self-efficacy in school) or school absence for medical reasons due to lack of this information in the included studies.

| CONCLUSION
This systematic review reports that children with selected non-syndromic CAs are at a higher risk of academic underperformance across several fundamental school subjects, which may remain persistent over the school levels, and their need for special education services is significantly higher than in reference populations. Evidence from population-based studies of non-syndromic CAs confirms that lower academic achievement in children with CAs is not limited to those with chromosomal or genetic syndromes. As the risks significantly differ for specific CA types, early screening, identification and development of differential SEN are important to support children and families to promote their academic proficiency across all school levels. Further population-based studies should aim to involve high-quality register-based data of isolated CAs, including more rare CA types, and linking with objective longitudinal pre-school and academic outcome data. This would help further understanding of the origin of their academic underperformance, the association with pre-school development, the identification of specific groups of children at risk and the introduction of timely and targeted interventions to inform education and social services and plan appropriate resources.