Population-based studies have consistently shown that extremely preterm or extremely low birth weight (ELBW) children are at risk for functional disabilities later in life.1^–3 Cognitive impairment is the most prevalent disability at school age1 and contributes to the excess of learning impairment and poor academic attainment in these populations, even among children without serious disability or cerebral palsy.1 4^–12 Although an underlying global cognitive deficit may account for much of the educational underachievement observed, differences in IQ scores do not account for all the learning difficulties in these children.10 13 14

Studies have shown that extremely preterm and ELBW children are more likely to require special school placement, repeat a school year (if applicable), have special educational needs (SEN) and receive learning support services than their term counterparts.6 7 15^–20 The prevalence of SEN and SEN resource provision provides an important measure of the totality of the functional deficit in this population. However, information regarding SEN resource utilisation in extremely preterm children is currently lacking.

We have studied educational outcomes in a whole population of extremely preterm children at 11 years of age. Firstly, we assessed general and specific academic attainment and the prevalence of learning impairment. Secondly, we investigated whether academic attainment and educational outcome are explained by a general cognitive deficit. Finally, we investigated educational outcomes for extremely preterm children entering compulsory education an academic year earlier than if they had been born at term.

METHODS

Participants

The derivation of the study population has been described previously.21 Of 307 survivors at 11 years of age, 11 (4%) were resident outside the UK and Ireland, the parents of 18 (6%) refused consent and 57 (19%) did not respond to invitations to participate. Thus, 219 (71%) extremely preterm children were assessed at 11 years of age (median age: 131 months; range: 121–145 months).

At the previous 6-year follow-up, for each extremely preterm child in mainstream school we identified a classmate born at term and matched for age, sex and ethnic group. Of 160 classmates evaluated at 6 years, 110 (69%) were re-assessed at 11 years. We selected 43 new comparator children using the same methodology1 to replace those now at a different school to the extremely preterm child or those declining participation. Thus a comparison group of 153 classmates was evaluated at 11 years of age (median age: 131 months; range 117–147 months).

Parents and children were given information leaflets and parents provided informed consent for their child’s participation. Three psychologists blind to study group allocation assessed children in schools, or at home or hospital (13%) if a school visit was not permitted. Study psychologists simultaneously scored the battery of standardised tests administered to non-study children prior to commencing EPICure study assessments and achieved excellent inter-rater reliability: there was >95% agreement across test items for both standardised measures. The study was approved by the Southampton and South West Hampshire Research Ethics Committee.

Measures

Cognitive ability (IQ) was assessed using the Kaufman-Assessment Battery for Children (K-ABC).22 A Mental Processing Composite (MPC) score of 39 was allocated to 16 children with severe cognitive deficit who did not participate in testing. Two children who did not complete the K-ABC because of blindness and behavioural problems were classified with normal cognition based on teacher report. Academic attainment was assessed using the Wechsler Individual Achievement Test 2nd Edition (WIAT-II^UK) for reading and maths.23 A score of 39 was allocated to 18 children with severe cognitive deficit. Reading (n = 7) and maths (n = 4) scores were not substituted for children who failed to complete the test for other reasons (communication difficulties, behavioural issues, blindness). Given the secular drift in IQ scores over time,24 25 cognitive and learning impairment was classified into four categories according to conventional standard deviation (SD)-banded cut-offs (none >−1 SD; mild −1 SD to −2 SD; moderate −2 SD to −3 SD; severe <−3 SD) using the mean and SD of the comparison group for each standardised test. A further test assessed intuitive mathematics (eg, estimating numbers of dots, lengths of lines).26 27 Scores ranged from 0 to 11 with higher scores indicating better performance.

Teachers in mainstream schools rated each child’s performance in seven subjects (English, maths, science, technology, geography, information technology and history). Scores were summed to yield a total academic attainment score (TAAS; range 1–5)28 which was used to classify performance below the average range (TAAS <2.5).

Teachers were asked to identify children with SEN, defined in the educational context as those with learning difficulties or disabilities that make it harder for them to learn or access education than most children of the same age, and were also asked to detail any SEN support the child received. The UK educational system adopts a three-stage approach to the provision of SEN resources that can culminate in a “Statement of SEN”, which is a legal document outlining a child’s SEN and what provision must be made within the school and by external bodies. Statements are awarded to children in special schools and to those in mainstream schools with complex needs typically requiring multi-agency support. (Data relating to grade retention were not recorded as the UK adopts an age-based educational system in which school year repetition is rare.)

Parental socio-economic status (SES) was classified into four categories (SES-I: professional/managerial; SES-II: intermediate; SES-III: routine/manual; SES-IV: long-term unemployed/never worked) using the UK National Statistics socio-economic classification system.29

Statistical analyses

Data were double-entered, verified for accuracy and analysed using SPSS (SJ, EH). Differences between groups in standardised test scores were analysed using independent samples t tests or one-way analysis of variance as appropriate. Differences in standardised test scores between extremely preterm and comparison children were also adjusted for SES, and WIAT-II^UK test scores were adjusted for general cognitive ability using multivariate regression. Estimation of the prevalence of serious cognitive (MPC <82) and learning (reading <74; maths <69) impairment and SEN in the whole cohort was done using multiple imputation to account for selective dropout.30 Variables used to predict 11-year outcomes were those independently and statistically significantly (p<0.05) associated with cognitive outcomes at discharge, 2.5 or 6 years.

RESULTS

Loss to follow-up

Extremely preterm children lost to follow-up at 11 years (n = 88) were more likely have a non-white ethnic origin, unemployed parents, lower cognitive test scores and more frequent cognitive impairment at 2.5 and 6 years of age than those assessed (n = 219) (see supplemental table s1).

Cognitive ability

Extremely preterm children had significantly lower MPC scores than classmates (table 1). Mean MPC scores were 82.9 (SD 21.2) at 23 weeks, 79.6 (SD 20.8) at 24 weeks and 86.1 (SD 15.3) at 25 weeks. In the extremely preterm group, boys had significantly lower scores than girls (−8 points; 95% CI −13 to −3). There was no sex difference among classmates. Excluding children with substituted scores, extremely preterm children had a mean MPC score 16.8 points (−19 to −14) lower than classmates. This group difference was −15.7 points in children for whom SES data were available (extremely preterm, n = 169; classmates, n = 135), and adjustment for SES reduced this by 0.2 points (−15.5 points; −18.2 to −12.8). Extremely preterm children also had lower scores than classmates for both sequential and simultaneous information processing, the largest deficit being in simultaneous processing (table 1). Extremely preterm children were more likely to have a relative deficit in simultaneous compared with sequential processing than their classmates (OR 7.2; 3.2 to 16.1).22

Overall, 40% of extremely preterm children had serious cognitive impairment (MPC <−2 SD) compared with 1.3% of classmates (OR 50; 12 to 206; table 2).,i Among extremely preterm children, boys were more likely to have serious impairment than girls (2.1; 1.2 to 3.7), but the prevalence of impairment was not significantly different for children born at 23/24 weeks (45%) and 25 weeks (36%) (1.48; 0.9 to 2.6). Using multiple imputation to correct for selective loss to follow-up of children with functional difficulties, the estimated proportion of extremely preterm children with serious cognitive impairment rose to 45% (95% CI 38% to 52%).

Academic attainment

Extremely preterm children had significantly lower reading and maths scores than classmates (table 1). Extremely preterm boys had significantly lower scores than extremely preterm girls in reading (−7 points; −1 to −13) but not in maths (−4 points; −1 to 10). No sex difference was found among classmates. Excluding children in whom severe cognitive impairment precluded test use (n = 17), extremely preterm children had a mean deficit of 15 points (−18 to −12) in reading and 25 points (−28 to −21) in maths. Adjusting for SES in those with data available reduced the reading and maths deficits by only 0.5 and 0.3 points, respectively. There were significant interaction effects of extremely preterm and MPC on both reading and maths with extremely preterm children having increasingly lower scores at lower MPC values than classmates. When adjusting for MPC, a significant effect of extremely preterm birth remained for maths (−8.1; −11.3 to −4.8) but not for reading (−2.5; −5.5 to 0.5).

Sixty four (30%) extremely preterm children and three (2%) classmates had serious impairment in reading (OR 22; 95% CI 6.6 to 70) (table 2). Serious impairment was marginally more common in extremely preterm boys than girls (1.7; 0.9 to 3.0). Ninety four (44%) extremely preterm children and two (1.3%) classmates had serious impairment in maths (59; 14 to 243). Using multiple imputation to correct for selective loss to follow-up, the estimated proportion of extremely preterm children with serious impairment in reading rose to 33% (27% to 40%) and to 50% (44% to 56%) in maths.

Teacher questionnaires were returned for 200 (91%) extremely preterm children and 146 (95%) classmates. Extremely preterm children in mainstream school had significantly poorer performance than classmates in all subjects rated, and 50% were classified with attainment below the average range compared with 5% of classmates (OR 18; 8 to 41).

SEN

SEN data were obtained for 215 (98%) extremely preterm children and 152 (99%) classmates. Twenty nine (13%) extremely preterm children attended a special school and were more likely to be boys (OR 3.6; 95% CI 1.5 to 8.6). Many more extremely preterm children than classmates had academic or behavioural SEN and utilised SEN provision, both overall and among those in mainstream schools only (table 3). Extremely preterm boys were more likely to have SEN and to utilise SEN provision than girls (table 3). Among classmates, boys were more likely to have SEN (3.8; 1.3 to 11.5) and marginally more likely to require SEN resources (2.8; 0.99 to 8.1) (p = 0.052). Using multiple imputation to account for loss to follow-up, the estimated proportion of extremely preterm children with SEN and those utilising SEN support rose to 66% (60% to 72%) and 64% (58% to 70%), respectively.

Overall, extremely preterm children in mainstream schools utilised SEN resources more than classmates (OR 10; 95% CI 6 to 18) (fig 1). The services most commonly used by extremely preterm children compared with classmates were educational psychologists (12; 2 to 95), one-to-one support (10; 4 to 26) and small group provision (9; 5 to 16). Extremely preterm children in receipt of SEN provision had lower MPC scores than extremely preterm children who did not receive support (−15.9 points; −19.1 to −12.7). For those extremely preterm children requiring SEN provision, 57% utilised one resource, 24% used two, 11% used three and 8% utilised four or more types of resource. In contrast, only 12% of classmates required two forms of support.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Type of special educational needs (SEN) resource utilised by extremely preterm children and classmates in mainstream schools at 11 years of age.

There were more extremely preterm children in receipt of support in mainstream schools than were identified by teachers as having academic or behavioural SEN (n = 13): 10 received minimal support (eg, “year 6 booster classes”, “programme for accelerated learning”, “help with spelling”, “lunchtime maths help”). Of the remaining three children, one had an outreach teacher for hearing impairment, one a classroom assistant due to visual and hearing impairment and one an occupational therapist; all three were on the local authority register for children with SEN. When the category for presence of SEN is extended to include these 13 children, 114 (61%) extremely preterm children and 21 (14%) classmates in mainstream schools had SEN (OR 9.9; 95% CI 5.7 to 17.1); this rises to 67% when children in special schools are included (12.4; 7.2 to 21.3).

Statements of SEN

Seventy two (34%) extremely preterm children had a statement of SEN compared with one (0.7%) classmate (OR 76; CI: 10 to 552); this rose to 39% (33% to 45%) using multiple imputation to estimate prevalence in the full cohort. In mainstream schools, 43 (23%) extremely preterm children had a statement (46; 6 to 339) (table 3).,ii Among extremely preterm children for whom statement data were available, of those with MPC <70 (n = 28), 22 were in a special school and seven in a mainstream school and all had a statement; among the children with MPC ⩾70, having a statement was independently associated with MPC, reading scores and cerebral palsy. However, once maths and SES were included in the model, both MPC and reading scores were no longer significant. After adjustment for each other, for those with MPC ⩾70 the OR for having a statement for a 10-point increase in maths score was 0.42 (0.30 to 0.59), for having cerebral palsy 7.1 (1.8 to 27) and for SES class IV 9.9 (2.1 to 47).

Early school entry

A subgroup of 68 (36%) extremely preterm children in mainstream school had dates of birth and gestational ages that would have placed them in an earlier academic year than if they had been delivered at term (23 weeks, n = 4, 22%; 24 weeks, n = 24, 40%; 25 weeks, n = 40, 36%). There were no significant differences in standardised test scores for these children compared with the remainder of the extremely preterm cohort. However, more children who would have entered compulsory education in an earlier academic year had SEN and statements and a greater proportion utilised SEN resources than the rest of the extremely preterm children (table 4).

DISCUSSION

After initial anxieties regarding survival during the neonatal period, parents are increasingly concerned about how likely their child is to progress well at school and what kind of support their child may need to realise his or her potential. We found that extremely preterm birth places children at high risk for cognitive and learning deficits affecting their schooling in middle childhood. Up to 44% had a serious impairment in the core subjects of reading and maths, and 50% had performance below the average range expected for their age across the spectrum of subjects studied. Extremely preterm children also had a 13-fold increased risk of SEN requiring additional learning support and were 77 times more likely to have an educational statement at 11 years of age.

Such a high prevalence of cognitive deficits in extremely preterm/ELBW survivors is a robust finding.1 2 31^–33 Although we have reported a higher prevalence than other population-based studies, we included only extremely immature children in whom a higher level of impairment would be expected given the gestational age-related gradient in cognitive function.34^–36 Furthermore, we classified impairment relative to the performance of the comparison group which yields higher rates but reflects reality as children are compared with the peers against whom they are judged in school.25 37

Comparable with our findings, other studies have shown that extremely preterm and ELBW survivors have poorer academic attainment than their term peers on teacher’s ratings6 11 and standardised tests,2 5 7 11 15 with rates of serious impairment of up to 50%.19 38 We also confirm maths as an area of specific difficulty in half of this population when loss to follow-up is accounted for.1 6 7 11 In contrast, the deficits in reading scores were no longer evident after adjustment for general cognitive ability.10 11 14 31 Thus some learning difficulty associated with extremely preterm birth, particularly in language and reading, can be accounted for by a general cognitive impairment, while mathematical difficulties appear to be a specific deficit. It has been suggested that this pattern of findings is largely indicative of a disruption to global brain development and imaging studies have provided confirmatory evidence of reduced cortical volume, size and complexity in preterm populations.39 40 The neuropsychological deficit appears to be worse in simultaneous rather than sequential processing of complex information.1 14 Specific deficits in maths skills may be a result of more specific impairment of regional brain areas.41 Such abilities are related to working memory, executive function, attentional control, and perceptual and visuo-spatial skills, which are also selectively impaired in preterm populations.13 42 43 Interventions targeted at enhancing executive control and motor function may therefore attenuate the effects of prematurity on educational outcomes. Furthermore, behavioural and emotional problems44 45 may impact on scholastic performance in this population and may be amenable to intervention.

The impact of extreme prematurity is most evident in that two thirds of these children have academic and behavioural SEN compared with 11% of their classmates and 24% of school children in England.46 The lower prevalence of special school placements than in other extremely preterm/ELBW populations19 32 38 reflects the UK policy of integrating children with SEN in mainstream schools. Indeed, 36% of extremely preterm children in mainstream schools had serious functional neurodevelopmental disability (Johnson S, Fawke J, Hennessy E, et al. Neurodevelopmental disability through 11 years in extremely preterm: the EPICure Study (submitted for publication)). SEN were not confined to those with severe disability, as a further 21% of extremely preterm children in mainstream schools had SEN. The prevalence of educational statements was remarkably high in this population: 1/3 of all extremely preterm children, and almost 1/4 of extremely preterm children in mainstream schools, compared with 2.3% of school children in England.46 The marked increase reflects the greater severity and complexity of learning support required, including physical and medical therapies, as indicated by the greater variety of special SEN services utilised in this population.

Most often SEN support comprised small group tuition or one-to-one support, but extremely preterm children also utilised services from a range of allied health professionals that were rarely accessed by classmates. Extremely preterm children were also more likely to need support provided by external agencies and to require multiple support services, impacting further upon the cost to schools and the local government for the education of these children. However, we have not investigated the severity of SEN, and, while the vast majority of those with SEN are in receipt of some degree of provision, this support may be insufficient6 32 and some may also receive additional professional help at home that we have not recorded.

Schools in the UK adopt an age-based educational system. Local authorities vary in their policies for entry into nursery (preschool) and virtually all of this cohort will have started preschool in the academic year in which they reach 5 years of age. Entry to formal schooling is compulsory in the term after the child’s fifth birthday. As such, children will vary in age and length of schooling on admission. Children whose date of birth and expected date of delivery cross the cut-off date for school entry (1 September) will ultimately be entered an academic year earlier than if they had been born at term. Concern is frequently expressed regarding educational outcomes for these children as the disadvantage already conferred by summer birth, due to age and length of preschool education,47 48 may be compounded by premature school entry, a disparity that is never rectified throughout schooling.

When outcomes for extremely preterm children who would have entered compulsory education a year early were compared to the rest of the extremely preterm cohort there were no significant differences in mean age-standardised scores; similarly there were no significant differences between summer born and non-summer born classmates on these measures and neither did we find significant differences in WIAT-II^UK raw scores or on non-standardised tests. However, extremely preterm children entered for school a year early had more SEN and educational statements. The additional support and resources utilised by these children may thus have contributed to their comparable academic attainment and if academic attainment is the outcome of concern, then early school entry does not appear to disadvantage extremely preterm children. It may be advantageous to delay school entry, and to adopt admission rules based on corrected age, as these children may then require less SEN provision, with the personal and social advantages of attenuating the impact of delayed cognitive and psychomotor development, emotional immaturity and social difficulties that are associated with extremely preterm birth.44 49 A preschool curriculum that emphasises language and social development, play and exploration may be more developmentally appropriate for extremely preterm children at this age. Although these data provide some preliminary results, more systematic investigation is required that is designed to study the impact of premature school entry.

We did not recruit comparison children for those in special schools and it may therefore be argued that impairment was overestimated in this study. However, we do not believe we have a high-achieving comparison group as classmates performed as would be expected in the normal population on standardised measures. We have also compared educational outcomes to current national statistics where possible. The small effects of SES on outcomes were expected as the selection of classmates typically adjusts for these factors. It is more likely that we have underestimated impairment in this population given the selective loss to follow-up of children with serious cognitive deficits and functional disability. Multiple imputation was used to account for selective dropout on educational outcomes. Using these techniques, the prevalence of cognitive impairment rose to 45% and the prevalence of those with SEN to 66%.

In summary, we identified a high prevalence of learning deficits that impacted significantly upon the school performance and educational needs of extremely preterm children. By 11 years of age, around 60% of extremely preterm children require additional support in school and 1/3 have an educational statement indicative of complex SEN. The impact of these impairments is likely to increase over time19 46 50 and existing difficulties may be exacerbated in secondary school when cognitive demands increase in parallel with progressively complex academic studies. The cost to society of extreme prematurity lies increasingly within the sphere of education51 as these children grow older and approach the transition to secondary school.