School-age outcomes of children after perinatal brain injury: a systematic review and meta-analysis

Background Over 3000 children suffer a perinatal brain injury in England every year according to national surveillance. The childhood outcomes of infants with perinatal brain injury are however unknown. Methods A systematic review and meta-analyses were undertaken of studies published between 2000 and September 2021 exploring school-aged neurodevelopmental outcomes of children after perinatal brain injury compared with those without perinatal brain injury. The primary outcome was neurodevelopmental impairment, which included cognitive, motor, speech and language, behavioural, hearing or visual impairment after 5 years of age. Results This review included 42 studies. Preterm infants with intraventricular haemorrhage (IVH) grades 3–4 were found to have a threefold greater risk of moderate-to-severe neurodevelopmental impairment at school age OR 3.69 (95% CI 1.7 to 7.98) compared with preterm infants without IVH. Infants with perinatal stroke had an increased incidence of hemiplegia 61% (95% CI 39.2% to 82.9%) and an increased risk of cognitive impairment (difference in full scale IQ −24.2 (95% CI –30.73 to –17.67). Perinatal stroke was also associated with poorer academic performance; and lower mean receptive −20.88 (95% CI –36.66 to –5.11) and expressive language scores −20.25 (95% CI –34.36 to –6.13) on the Clinical Evaluation of Language Fundamentals (CELF) assessment. Studies reported an increased risk of persisting neurodevelopmental impairment at school age after neonatal meningitis. Cognitive impairment and special educational needs were highlighted after moderate-to-severe hypoxic-ischaemic encephalopathy. However, there were limited comparative studies providing school-aged outcome data across neurodevelopmental domains and few provided adjusted data. Findings were further limited by the heterogeneity of studies. Conclusions Longitudinal population studies exploring childhood outcomes after perinatal brain injury are urgently needed to better enable clinicians to prepare affected families, and to facilitate targeted developmental support to help affected children reach their full potential.


Results This review included 42 studies. Preterm infants
with intraventricular haemorrhage (IVH) grades 3-4 were found to have a threefold greater risk of moderate-tosevere neurodevelopmental impairment at school age OR 3.69 (95% CI 1.7 to 7.98) compared with preterm infants without IVH. Infants with perinatal stroke had an increased incidence of hemiplegia 61% (95% CI 39.2% to 82.9%) and an increased risk of cognitive impairment (difference in full scale IQ −24.2 (95% CI -30.73 to -17.67) . Perinatal stroke was also associated with poorer academic performance; and lower mean receptive −20.88 (95% CI -36.66 to -5.11) and expressive language scores −20.25 (95% CI -34.36 to -6.13) on the Clinical Evaluation of Language Fundamentals (CELF) assessment. Studies reported an increased risk of persisting neurodevelopmental impairment at school age after neonatal meningitis. Cognitive impairment and special educational needs were highlighted after moderate-tosevere hypoxic-ischaemic encephalopathy. However, there were limited comparative studies providing school-aged outcome data across neurodevelopmental domains and few provided adjusted data. Findings were further limited by the heterogeneity of studies.
Conclusions Longitudinal population studies exploring childhood outcomes after perinatal brain injury are urgently needed to better enable clinicians to prepare affected families, and to facilitate targeted developmental support to help affected children reach their full potential. Perinatal brain injuries can have wide-ranging deleterious consequences for children, families and broader society. [1][2][3][4] Over 3000 infants experience perinatal brain injury in England annually 1 and the Department of Health and Social Care (DHSC) has committed to halving the rate of perinatal brain injuries by 2030 as part of the national maternity ambition. 5 To monitor progress towards this goal, a standardised definition of perinatal brain injury was developed. 6 The degree to which this definition captures and represents true perinatal brain injuries is unclear and requires us to look beyond the neonatal period. 6 Focusing on the childhood outcomes of infants with perinatal brain injury provides a fuller understanding of the population captured by the DHSC definition. Despite their importance to families, school-age outcomes following neonatal care have been an overlooked research priority. Neonatal studies typically focus on 2-year composite outcomes, which may mask the true neurodevelopmental burden of injuries, and are WHAT IS ALREADY KNOWN ON THIS TOPIC ⇒ Thousands of children suffer a brain injury around the time of birth every year. Many of these injuries are associated with neurodevelopmental impairment at 2 years of age. However, 2-year outcomes are not necessarily representative of later childhood outcomes and function, which are a priority for parents.

WHAT THIS STUDY ADDS
⇒ This review provides an overview of existing evidence of childhood outcomes after perinatal brain injury. It indicates that there is some evidence of ongoing impairment throughout childhood for different types of perinatal brain injury but that there are considerable gaps in knowledge.
Open access known to be poorly predictive of future functioning. [7][8][9][10] As such, our understanding of childhood developmental trajectories after brain injuries-and whether any sequelae are fixed, stable or amenable to interventionsis limited. We therefore undertook a systematic review to explore school-age neurodevelopmental outcomes following perinatal brain injury.

Study selection
The review was conducted as per the pre-registered protocol (CRD42021278572) and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 11 We included observational comparative studies exploring neurodevelopmental outcomes of children over 5 years of age after perinatal brain injury, published between 2000 and September 2021 (table 1). The DHSC definition of perinatal brain injuries used includes intraventricular haemorrhage (IVH), preterm white matter injury (WMI), stroke, central nervous system infection, hypoxicischaemic encephalopathy (HIE) and kernicterus diagnosed during the neonatal period. 6 12 We did not include seizures in isolation. For inclusion, studies were required to have a non-brain injured comparator group. The primary outcome was neurodevelopmental impairment; secondary outcomes included motor, cognitive, speech and language, behavioural and neuropsychological, visual and hearing outcomes and seizures. A search strategy incorporating 99 key terms and mesh headings was developed in Medline Ovid, adapted and run across 10 databases (online supplemental files 1; 2). Snowballing techniques were used to augment search sensitivity. All titles were screened independently by two reviewers. The full texts of all potentially relevant titles were retrieved, reviewed and their risk of bias assessed by two trained reviewers independently (PR, CC, MV, JD and SS). Disagreements were arbitrated by a third reviewer.

Data extraction and synthesis
Studies were stratified by brain injury type, substratified by age of outcome assessment and outcome type, and summarised in a narrative synthesis. Where sufficient suitable data were available from contextually and clinically comparable studies, data were pooled in random effects meta-analyses using RevMan V.5.4. Continuous data were pooled using the inverse variance method; dichotomous data were pooled using the Mantel-Haenszel method; and analysis data from studies which did not provide raw data were pooled with dichotomous data from other studies using the generic inverse variance method. 13 Where studies provided insufficient comparative data for a particular outcome, the combined incidence figures for that outcome within the brain injured population was calculated across studies using the Fisher's exact test for binomial data. 14 Statistical heterogeneity was assessed using the I 2 statistic and substantial heterogeneity (>85%) was explored further in subgroup analyses.

Quality assessment
The Newcastle-Ottawa Tool was used to assess risk of bias across three domains: population selection, the comparability of the 'brain injured' and 'non-brain injured' comparator groups, and outcome assessment. 15 Studies were classed as poor, fair, or good for each domain and given an overall risk of bias classification.

Patient and public involvement
Patients or the public were not involved in the design or conduct of this review. However, the review's findings will be used to shape the larger CHERuB study in partnership with our parent advisory panel.
Cognitive outcomes at school age after preterm brain injuries were reported by 16 studies using 25 different cognitive assessment tools -limiting the potential for meta-analysis (online supplemental file  Children with a diagnosis of brain injury occurring at or around the time of birth (including during the neonatal period) as defined by the DHSC (including those with any white matter injury but not including those with isolated seizures).
Studies which include infants with brain injuries diagnosed during the neonatal and infancy period where most were diagnosed outside of the neonatal period.
Studies including infants with moderate to severe HIE born in the post-therapeutic hypothermia era (ie, where infants received therapeutic hypothermia).
Studies including infants with moderate to severe HIE born during the pre-therapeutic hypothermia era or in low or middle income countries that do not offer therapeutic hypothermia.    Open access Table 2 Overview of key findings for school-age outcomes of infants with perinatal brain injury compared with those without brain injury     3). 16 17 21 22 24-35 Educational outcomes were reported by five studies. 21 22 26 30 35 Studies consistently reported lower cognitive scores at school age following IVH grade 3-4. 16 22 26 van de Bor and den Ouden reported increased special educational needs at 5, 9 and 14 years: the adjusted risk at 14 years of age was marked, adjusted OR 3.99 (95% CI 1.36 to 11.69). 22 Studies reported no significant differences in language scores after IVH grades 3-4. 21 22 However, an association with reading OR 3.62 (95% CI 1.59 to 8.24), spelling OR 4.48 (95% CI 1.8 to 11.2), and arithmetic OR 2.79 (95% CI 1.2 to 6.48) impairment was demonstrated. 26 Most studies highlighted cognitive effects after WMI. 17 30 33 35 Studies exploring behavioural outcomes after IVH 3-4 did not find any associations with attention deficits, conduct issues or autism spectrum disorder (table 2). 16 25 36 However, there was conflicting evidence around the mental health effects of WMI. 17 37 Studies exploring hearing impairment after IVH and/ or WMI were small or not comparable. Ten studies explored visual impairment after IVH or WMI, four provided meaningful outcome data. 16 21-23 27 28 33 34 38 39 An increased prevalence of visual impairment after IVH grades 3-4 (45.4% and 90.9%) compared with controls (7.5%) was reported in addition to significantly lower visual motor integration scores. 27 Perinatal stroke Eight comparative studies explored school-age outcomes after perinatal stroke, these included 177 children with perinatal stroke (100 left sided and 54 right sided-not all studies specified laterality) and 232 comparator children (online supplemental file 3). [40][41][42][43][44][45][46][47] Infants' gestational age was largely unspecified. Five studies presented a combined incidence of childhood seizures after perinatal stroke of 40.1% (95% CI 26.8% to 53.3%; 5 studies; 115 subjects) I 2 =56% (online supplemental file 5). 40 43 44 46 47 The combined incidence of hemiparesis after perinatal stroke was 61% (95% CI 39.2% o 82.9%, I 2 =88%). There was considerable heterogeneity across studies, and likely detection bias (online supplemental file 6). 40 42-45 Five studies identified a significant combined mean difference in full scale IQ scores at 7-13 years of age after perinatal stroke: −24.2 (95% CI -30.73 to -17.67; 5 studies; 296 subjects) I 2 =80% (figure 4). 40 42 45-47 There was heterogeneity across studies in terms of assessment timing, assessment tools and combining those with leftsided and right-sided strokes.
Kolk et al reported significantly lower scores compared with controls across most NEPSY language domains following perinatal stroke. 43 Significantly lower receptive and expressive mean language scores on the CELF assessment were also reported across studies: −20.88 (95% CI -36.66 to -5.11; 2 studies; 137 subjects) I 2 =88% and −20.25 (95% CI -34.36 to -6.13; 2 studies; 137 subjects) I 2 =87%, respectively (online supplemental files 7,8). 40 45 Statistical heterogeneity may have been as a result of studies combining left-sided and right-sided strokes and the varying age of outcome assessment. Studies highlighted that deficits in receptive language scores present at 7-8 years persisted at 10-12 years but that expressive language scores improved (p=0.012). 40 41 Meningitis Studies consistently reported an increased risk of neurodevelopmental impairment after neonatal meningitis (table 2). [48][49][50] An increased likelihood of neuromotor disability at 5 years of age (n=45/274, 16%) compared with controls (n=2/1391, 0.1%) was reported (online supplemental file 3). 48 On reassessment of the same population at 9-10 years, this increased risk of severe disability persisted (n=12, 10.8% compared with n=0, 0%). 50 An increased risk of any neurodevelopmental impairment at 5 years after neonatal group B Streptococcal meningitis was also reported in the Netherlands, RR 5.30 (95% CI 2·57 to 10·89), and in Denmark, RR 7.80 (95% CI 4·42 to 13·77). 49 This increased risk persisted on subsequent assessment: at 11 years of age in the Netherlands, RR 2.99 (95% CI 1.83 to 4.88) and at 15 years of age in Denmark RR, 3.15 (95% CI 1.82 to 5,46). 49 Hypoxic-ischaemic encephalopathy Two comparative studies (of the same cohort) explored outcomes of term-born infants with moderate-to-severe HIE, but without cerebral palsy, at school age (online supplemental file 3). 51 52 They highlighted significantly lower full scale IQ scores after HIE (mean difference −13.62 (95% CI −20.53 to −6.71)). 51 This difference in cognition was also seen for perceptual reasoning, working memory and processing speed. Children with HIE were also more likely than controls to receive additional classroom support: OR 10 (95% CI 1.16 to 86) although the CI for this risk estimate was wide. 51 Children with HIE (without cerebral palsy) also had significantly lower motor scores (mean difference −2.12 (95% CI −3.93 to -0.30)) and verbal comprehension scores (mean difference −8.8 (95% CI −14.25 to -3.34)). 51 They were also noted to have higher behavioural difficulty scores especially for emotional problems. 51

DISCUSSION
This review brings together the existing evidence on the later childhood outcomes of infants with perinatal brain injury. Although 42 studies are included, small study populations, limited data on injury severity and laterality, and the heterogeneity of studies limited the potential power of results. However, studies demonstrate a threefold higher risk of moderate-to-severe neurodevelopmental impairment at school age following IVH grades 3-4. Studies consistently report cognitive impairment after IVH grades 3-4 but suggest that speech and language is relatively preserved. A higher risk of hemiplegia, cognitive impairment and poorer academic performance after perinatal stroke is reported in addition to poorer receptive and expressive language scores. Studies report a higher risk of persisting neurodevelopmental impairment after neonatal meningitis -however, few studies address this question. Few comparative studies explore school-age outcomes after HIE.
In following our a priori protocol, only comparative studies were included. This was with a view to enabling inferential analyses and adjustment for key confounders such as gestation. Unfortunately due to this strict inclusion criterion, many pertinent non-comparative studies were excluded. Additionally, our searches were conducted in September 2021, more recent studies would therefore have been missed.
Heterogeneity in terms of outcomes assessed, outcome assessment tools, and timing of outcome assessment limited the comparability of studies and the potential for meta-analyses. Several meta-analyses included low numbers of studies, reducing the reliability of the I 2 statistic. 53 This review was also limited by the size of available studies and how studies presented data for extraction. Few studies presented adjusted data or explored childhood trajectories after perinatal brain injury. Previous reviews were limited by a lack of comparable studies, heterogeneity, the inclusion of much older cohorts or by the inclusion of non-comparative studies. 4 54-56 While this review was also limited by studies' heterogeneity and the quality of available data, new and important findings -for example, the risk of neurodevelopmental impairment at school age after IVH 3-4were identified. Our finding of a higher risk of cerebral palsy after IVH grade 3-4 and motor impairments after preterm brain injuries is echoed by previous studies. 54 55 57 Lynch and Nelson highlight that 60% of infants have neurological sequelae that emerge over time following perinatal stroke. This was in-keeping with our findings of a higher risk of hemiparesis, cognitive impairment Open access and speech and language impairment. 58 Several noncomparative population-based studies also mirror these findings. [59][60][61][62] Although previous reviews highlight an increased risk of various neurodevelopmental impairments after neonatal meningitis in early childhood -we are unaware of any focusing on school-age outcomes after neonatal meningitis. 4 63 The review's findings of potential ongoing impairments across cognitive, speech and language, and behavioural domains -in addition to a need for increased school support -after HIE are mirrored by other studies. [64][65][66][67][68] Shankaran et al and Azzopardi et al highlight ongoing neurodevelopmental sequelae at school age among children who received therapeutic hypothermia for moderate to severe HIE. 64 65 67 Implications Considerable gaps in the evidence are highlighted, particularly around the risk of specific outcomes following different types of injury, the precision around risk estimates, the impact of different factors (such as injury laterality) and the developmental trajectories of these children. This information is key to prepare families for the future, inform enhanced developmental surveillance, and enable targeted multidisciplinary support to help affected children to reach their full potential. As such, this review highlights a pressing need for high-quality, comparative studies which use the 'Core Outcomes In Neonatology' to explore long-term outcomes after perinatal brain injury and permit future meta-analyses. 10 Additionally, to meet the DHSC ambition to reduce perinatal brain injury, real-time longitudinal population data, extending beyond the neonatal period to childhood, are needed. This could be achieved through linkage of existing population datasets within the UK which is a key objective of the CHERuB study.

CONCLUSION
This review provides an overview of existing evidence of the impact of perinatal brain throughout childhood. Studies' heterogeneity significantly limited the potential for evidence synthesis.
Data availability statement All data relevant to the study are included in the article or uploaded as online supplemental information.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Open access This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/ licenses/by/4.0/.