Statistics from Altmetric.com
Very preterm birth (<32 weeks gestation) has consistently been found to be predictive of poorer school performance and special educational needs.1 2 Recent studies have found slightly lower cognitive ability in children born late preterm (34–36 weeks gestation) or at early term (37–38 weeks gestation), compared with children born at around 40 weeks gestation.3,–,5 The results of cognitive tests, however, may not necessarily reflect the child's current or future school performance. Nevertheless, a recent Scottish study found that even children who were born late preterm or early term were at increased risk of having special educational needs compared with their full term counterparts.6 Early school performance is an indicator of future academic achievement, employment prospects and adult health.7 8 Other important determinants of academic achievement include social factors and early feeding;9 10 hence these factors need to be considered together.
What is known on this topic
▶ Very preterm birth (<32 weeks gestation) is associated with poorer school performance and special education needs.
▶ Some studies have observed poorer cognitive development in children born late preterm (34–36 weeks gestation) and early term (37–38 weeks gestation).
▶ Few studies have measured school performance in these children.
What this study adds
▶ For children born in England in 2000–2001, birth before full term increases the risk of poorer school performance at age 5 years.
▶ After adjustment for other factors, the excess risk remains even in children born at early term or late preterm.
▶ Birth before full term has a smaller effect than socio-demographic factors, but it is one more element that affects the child's risk profile.
We present data on the association between gestational age and school performance at age 5 years in a cohort of children at school in England.
Millennium Cohort Study
The Millennium Cohort Study (MCS) is a nationally representative UK longitudinal study of 18 818 infants born in the UK.11 A random two-stage sample of all infants born in England and Wales between September 2000 and August 2001, and in Scotland and Northern Ireland between November 2000 and January 2002, who were alive and living in the UK at age 9 months was drawn from Child Benefit registers that cover virtually all children. The MCS does not include children who died within the first 9–10 months after birth, but these constituted less than 1% of all births.12 Stratified sampling by electoral ward (defined geographical areas) was employed, with over-sampling of ethnic minority and disadvantaged areas. Parents were interviewed for the first time when the children were aged 9 months and again at 3, 5 and 7 years. Detailed information was collected on a range of socio-economic and health factors. This study uses data collected at ages 9 months and 5 years.
Gestational age in weeks was calculated using the mother's report of the expected due date, which corresponded well with data in routine hospital records.13 The only exception was for the post-term group (42–43 weeks), which had a relatively high misclassification rate,13 and hence has been excluded from this study. Gestational age was grouped as shown in table 1.
Foundation stage profile
The foundation stage profile (FSP) records the child's achievement as measured by their teacher at the end of their first school year, ‘foundation stage’. Teachers are trained in how to conduct the assessments, which are based on observations during the whole year. The FSP captures the ‘Early Learning Goals’ as a set of 13 assessment scales across six areas of learning (details in table 2).14 15 For each scale, the teacher gives the child 1–9 points according to the child's progress in achieving the learning goals. Children achieving a scale score of ≥6 points are classified as working securely within the Early Learning Goals and are classified as having achieved a good level of development. Children who achieve a score of ≥78 points across the 13 assessment scales (ie, an average of 6 points per scale) and a score of ≥6 in each of the three ‘personal, social and emotional development’ scales and the four ‘communication, language and literacy’ scales are classified as reaching a good level of overall achievement. It is a statutory requirement to submit FSP data to the Department for Education for children in government-maintained (state) schools in England, and for children in independent (fee-paying) schools who receive government funding.
Exclusions and missing data
We included only MCS children attending school in England since FSP is not used in the other UK countries. Of the 18 818 children recruited at age 9 months, 14 887 (79%) participated at age 5 years; 9319 were living in England and were eligible for inclusion (figure 1). For children whose parents gave consent, the child's FSP results were sought using record linkage between the MCS and the appropriate Department for Education FSP dataset.16 We excluded 1669 children (18% of those eligible, figure 1) which yielded a study population of 7650.
Thus, the largest sources of missing data were non-response at age 5 (21%) years and missing FSP (12%). Non-response at age 5 years was associated with socio-demographic factors such as younger maternal age, non-white ethnicity and lower levels of maternal education,17 and has been allowed for in the analysis using non-response weights. Non-response at age 5 was slightly higher in the very preterm group (26%) than in other gestational groups (20–22%) The most common reason for missing FSP was non-consent (4.7% of 8728 children). The second largest group of children with missing FSP were attending mainstream school (4.1%), but for some reason the assessment was not done or not submitted. The remaining missing values were in children who did not attend school (1.0%), or attended fee-paying (2.5%) or special schools (0.05%) that did not participate in FSP assessment. Where FSP results were available for children in fee-paying or special schools, these are included in our analysis.
Table 1 shows the percentage of children with missing FSP according to selected factors. FSP was missing for 21% of very preterm children, 7% of moderately preterm children and 12% of those born at 34–41 weeks. The children who were most likely to have missing FSP were of non-white ethnicity, living in households where languages other than English were spoken and their mothers had the highest level of education or overseas qualifications. Missing FSP was more common in children with a longstanding limiting illness, a low cognitive development score (in the preterm groups) and a high cognitive development score (in the full term group).
The primary outcome was the proportion of children not having reached a good level of overall achievement (defined as shown in table 2). Secondary outcomes focused on not working securely within each of the six areas of learning (table 2).
As the study outcomes were not rare, the risk ratio (RR) was the preferred effect measure. Modified Poisson regression18 was used to estimate RR for these outcomes across gestational age groups compared with term children, with adjustment for multiplicity and the following factors which were significantly (p<0.05) associated with the primary outcome: child's sex, ethnicity, whether the child was the firstborn for the mother, breastfeeding duration, month of birth (ie, age within the school year) and the mother's age at delivery, marital status, education, social class and languages spoken in the child's home.
All analyses allowed for the clustered, stratified sampling using the ‘survey commands’ in Stata version 11 (Stata Corporation, College Station, Texas, USA). Hence, all CI and p values account for clustering, and all proportions and RR are weighted using sample weights and non-response weights.11 17 Ethical approval for the MCS was granted from the Multi-centre Research Ethics Committee.
Among the 7650 children in our study, 8.4% were born preterm and 1.1% were born very preterm. The median gestational age in the very preterm group was 29 weeks, and 63% of this group had a gestational age of 29–31 weeks. Table 3 shows the descriptive characteristics of all children according to gestational age. Increasing prematurity was associated with multiple births, caesarean section, lower birth weight, longer length of stay in hospital and shorter duration of breastfeeding. Some of the maternal characteristics also varied according to gestational age, but there was no strong dose-response effect across gestational age.
The FSP outcomes in MCS children in England were remarkably similar to the national statistics for the same year (table 2). Although only 45% of children in the national statistics had reached a good level of overall achievement, the figure was much higher for the individual areas of learning and was above 70% for four of the six areas of learning.
At the end of foundation stage, 51% of term MCS children had not reached a good level of overall achievement. This percentage increased with prematurity, from 51% in full term children to 66% in very preterm children (table 4). After adjustment for other factors (see table 4 footnote), a slightly elevated risk remained even in early term (adjusted RR 1.05, 95% 1.00 to 1.11) and late preterm children (adjusted RR 1.12, 95% CI 1.04 to 1.22). These effects were small in comparison with other risk factors in the same model: adjusted RR=1.89 (95% CI 1.68 to 2.13) for youngest versus oldest in class (ie, born in month 12 vs month 1), adjusted RR=1.35 (95% CI 1.29 to 1.42) for boy versus girl and adjusted RR=1.32 (95% CI 1.22 to 1.44) for mothers having no qualifications versus university degree. The relative size of these different effects is shown in figure 2, in which the predicted percentage of children who had not reached a good level of overall achievement is shown according to the gestational age, stratified by age within the school year, mother's education and child's sex. However, the association between gestational age and not working securely at the end of foundation stage was consistent across all six areas of learning (table 4). There were striking dose-response effects of prematurity and increased risk of not working securely in mathematical development, physical development and creative development.
In our study, the percentage of children not reaching a good level of overall achievement at the end of their first school year increased as prematurity increased. There was a 5% increased risk of poorer achievement in early term children and a 12% increase in late preterm children. Birth before full term has a smaller effect than the child's sex and age within the school year, and their mother's level of education, but it is one more element that affects the risk profile of the child. For example, the percentage of children who had not reached a good level of achievement was 38% in children whose mothers had a university degree, 55% in children whose mothers had other qualifications and 76% in children whose mother had no qualifications. The additional impact of late preterm birth (ie, RR=1.12) would increase these risks to 42%, 62% and 85%, respectively.
The main strengths of our study are that we used a large, population-based cohort to assess the association between birth before full term and educational achievement across the full spectrum of gestational ages. We adjusted for many factors which were potentially associated with preterm birth and FSP, and compared the effect size for preterm birth with that of other known risk factors for poor educational achievement. The educational outcome used in our study, the FSP, is a statutory requirement in state schools and schools receiving some state funding in England. Teachers are trained in how to perform the assessments and the results are routinely submitted to the government. The fact that the FSP is based on observations by the teacher during the whole year means that, unlike many other standard assessments, it does not matter if the child has an ‘off day’.
A limitation of our study is that 12% of children had FSP missing. The children with FSP missing tended to have a different socio-demographic profile to children with non-missing data, but this included features of social advantage and disadvantage, and poorer health but better cognitive development. Within the very preterm and late preterm groups, however, the children with missing FSP tended to have poorer health and cognitive development, which means that our observed effects of preterm birth may underestimate the true effects. Children with severe disability may be under-represented in our data. This may explain why the results in our very preterm children are not as extreme as in other studies.6 19,–,21
Our results are consistent with a characteristic phenotype described for preterm children22 that includes overall poorer achievement and specific problems in mathematical abilities,23 social relationships and emotional development,24 25 and impaired physical development. Our study also found poorer creative development with creativity requiring mental manipulation and visual spatial coordination, another area described as impaired in very preterm children.26 In contrast, findings for late preterm or early term children have been variable, with some studies finding worse reading and mathematical performance,4 some finding only increased reading and spelling difficulties27 and others finding no association with cognitive or educational achievement.28 Our study demonstrates that the characteristic problems associated with prematurity show a dose-response relationship across the whole gestation spectrum.
Our study highlights the fact that educational achievement tends to be lower in children who are the youngest in the class.29 For a child already ‘disadvantaged’ by preterm birth or having a summer birthday, or the compounded problem of starting school a year earlier than if they had been born at term, it may be appropriate to consider delaying school entry. A UK study of children born extremely preterm found that those who started school a year early due to preterm birth had similar academic attainment, but were more likely to have special education needsstatements, compared with those who started in the appropriate year.2
Perhaps, the most striking effect of late preterm and early term birth observed to date is the higher proportion of children having special educational needs in Scotland.6 Interestingly, this study included children from age 5 to 18 years thus suggesting a long-term impact of being born earlier than full term. However, three other large studies that have assessed longer-term educational outcomes observed only very small effects in children born late preterm and early term for attending university,19 for completing high school20 and for completing compulsory basic schooling.21
One possible explanation for our findings is that it could be the direct result of a smaller and less mature brain at birth. At 34 weeks, the brain weighs only 65% of the full term brain and even at 36 weeks, the brain weighs about 80% of the full term brain.30 The volume of total grey matter increases by about 1.4% per week from 29 to 41 weeks, and a fivefold increase in white matter volume occurs between 35 and 41 weeks.30 Hence, even the late preterm and early term brain are immature and vulnerable to insults.
Alternatively, the pathology which causes birth before full term may lead to poorer health, attention problems or poorer cognitive development, which in turn affects educational achievement. In our study, children born late preterm or early term were more likely to be a multiple birth and be delivered by caesarean section. Unfortunately, we did not have data on maternal or fetal pathology, and could not determine whether particular causes of preterm birth were associated with educational achievement. Nevertheless, in other studies, children born late preterm had an increased risk of morbidity.31 32 In our study, children born late preterm and early term were more likely to have asthma, a limiting longstanding illness and increased numbers of hospital admissions compared with children born at full term.33 This increase in morbidity may be associated with a delay in achieving early developmental milestones, or being absent from school due to sickness, both of which may affect educational achievement.
Another potential explanation for our findings is that the late preterm and early term children may have different early formative experiences compared with the term children. For example, in our study and other studies,34 35 children who were born late preterm or early term had a shorter breastfeeding duration compared with children born at full term. Breastfeeding is also associated with improved cognitive development.36 37 Adjusting for breastfeeding and other factors reduced our effect estimates, but the elevated risks did not disappear. It is possible that there are other differences between the groups for which we have not adjusted.
For children born in England at the start of the new millennium, birth before full term increases the risk of not reaching a good level of overall achievement at the end of the first school year, even in children who are born early term or late preterm. Children who are at risk of not reaching their full potential may benefit from additional support at school or at home, or occasionally by delaying school entry. However, preterm birth can only be considered in the risk profile of a child if teachers are aware of a child's gestational age and the effects of prematurity on child health, behaviour and cognitive development. Further research should determine whether the elevated risks associated with late preterm birth are due to particular causes, and whether they remain as the children progress through school.
See Editorial, p F158
Funding This study was funded by a grant from the Bupa Foundation (Grant number TBF-08-007).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.