BACKGROUND: To assess functional disability in children born before 26 weeks of gestation at 11 years of age and the stability of findings in individuals between 6 and 11 years of age.
METHODS: Of 307 surviving children born in 1995, 219 (71%) were assessed at 11 years of age alongside 153 classmates. Children were evaluated by using standardized tests of cognitive ability and clinical condition at both ages.
RESULTS: Using classmate data to determine reference ranges, serious cognitive impairment (score of less than −2 SD) was present in 40% of extremely preterm children and 1.3% of classmates (odds ratio [OR]: 50 [95% confidence interval (CI): 12–206]) at 11 years of age. Overall, 38 (17%) extremely preterm children had cerebral palsy; moderate or severe impairment of neuromotor function, vision, and hearing was present in 10%, 9%, and 2% of these children, respectively. Combining impairment across domains, 98 (45%) extremely preterm children had serious functional disability compared with 1% of the classmates (OR: 61 [95% CI: 15–253]); this was more common in boys than girls (OR: 1.8 [95% CI: 1.0–3.1]) and in those born at 23 or 24 weeks' gestation compared with those born at 25 weeks' gestation (OR: 1.8 [95% CI: 1.0–3.1]). The prevalence of serious functional disability was 46% at 6 years of age and 45% at 11 years of age. Using multiple imputation to correct for selective dropout, it is estimated that 50% (95% CI: 44%–57%) of extremely preterm children are free of serious disability at 11 years of age.
CONCLUSIONS: Extremely preterm children remain at high risk for neurodevelopmental disability at 11 years of age compared with term peers. The prevalence of disability remained stable between 6 and 11 years of age, and large individual shifts in classification of disability were unusual.
Early population-based studies of school-aged survivors after extremely preterm birth have consistently identified an increased prevalence of neurodevelopmental and cognitive impairments compared with term peers. Studies investigating outcomes for infants born in the 1990s, after marked improvements in perinatal interventions, are relatively scarce but continue to document similar trends.1–5 Although neurosensory disabilities remain a significant risk, cognitive deficits are the most prevalent disability in this population.2 Cognitive deficits have greater functional impact later in childhood as children progress through school and must compete with their term peers.
To obtain outcome data relevant to modern neonatal care, we recruited a whole population of infants born at ≤25 weeks' gestation from March through December 1995 in the United Kingdom and Ireland (the EPICure Study). This is the largest prospective, population-based study of a gestational age–defined cohort of extremely preterm children to date. Neurodevelopmental outcomes for these children at 2 years6 and 6 years of age2 have previously been reported. We have reevaluated these children to assess cognitive and functional disability and to investigate stability and change in such impairments over middle childhood.
This cohort has been described previously.2 Of 308 survivors at 6 years of age, 2 subsequently died (both previously classified with severe cognitive disability). One additional child for whom no previous data were available was identified. Of 307 survivors at 11 years of age, 11 (4%) living outside the United Kingdom were not eligible, the parents of 18 (6%) children refused consent and the parents of 57 (19%) children did not respond to study invitations. Participants thus comprised 219 (71%) extremely preterm children (median age: 131 months; range: 121–145 months).
At 6 years of age, for each extremely preterm child in mainstream school, we identified a classmate born at term. These children were randomly selected from 3 classmates identified of the same gender and ethnic group and closest in age to the extremely preterm child.2 Of 160 classmates evaluated at 6 years of age, 110 (69%) were reassessed at 11 years of age. Where the extremely preterm child was at a different school to the original 6-year classmate, or where the original 6-year classmate declined to participate, a new classmate was selected at 11 years of age by using the same method as at 6 years of age. This resulted in 43 new classmates being selected. A total of 153 classmates were evaluated at 11 years of age (median age: 131 months; range: 117–147 months). Longitudinal data were available for 202 (92%) extremely preterm children and 110 (72%) classmates assessed at 6 and 11 years of age. There were no significant differences in age and gender between extremely preterm children and classmates.
Parental socioeconomic status was classified into 4 classes by using UK National Statistics: (1) professional/managerial; (2) intermediate; (3) routine/ manual; and (4) unemployed.7 Classmates were slightly more likely to be in class 1 (57% vs 44%), and extremely preterm children were slightly more likely to be in class 4 (24% vs 15%; P = .041).
The study was approved by the Southampton and South West Hampshire Research Ethics Committee. Parents and children were given information leaflets and parents provided written consent. Children were assessed at school (87%) or at a hospital or their home (13%) for 1 day by a study psychologist and pediatrician. Study assessments were scheduled by an administrator and assessors were blind to group allocation. Parents were sent a feedback letter detailing the child's test results.
Cognitive ability was assessed by using the Kaufman-Assessment Battery for Children (K-ABC)8 at 6 and 11 years of age. This yields a mental processing composite (MPC) score for global cognitive ability (mean: 100 [SD: 15]). Sixteen children did not complete testing because of severe cognitive disability and were allocated a nominal value (39) 1 point below the basal test score. Two additional children were not allocated scores because of blindness and behavior problems, but cognitive function was categorized on the basis of teachers' reports. Cognitive impairment was categorized by using conventional SD-banded cutoffs. The first edition of the K-ABC was used to provide directly comparable longitudinal assessments, and the scores of the comparison group were used as reference data (mild: 82–92; moderate: 71–81; severe: ≤70), rather than the obsolete test norms, given the secular trends in IQ over time.9,10 Psychologists achieved >95% agreement across test items before commencing study assessments.
Disability in neuromotor function was assessed by using a standard pediatric evaluation and presence and type of cerebral palsy (CP), independent of degree of disability, was classified retrospectively (Dr Fawke and Dr Marlow) by using clinical information obtained at the study assessment. Objective ratings of neuromotor function were made by using the Gross Motor Function Classification System (GMFCS)11 and the Manual Abilities Classification System (MACS).12 In children with CP, the highest GMFCS or MACS level was used to classify neuromotor impairment into 3 categories (severe: level 4/5; moderate: level 3; mild: level 1/2). Sensory impairment was also assessed by clinical examination as at 6 years of age.
Composite functional disability was categorized by using the most severe classification in any domain. For children with multiple impairments, the most severe classification was used to categorize composite disability. Categories for moderate and severe impairment were combined to classify “serious” functional disability impacting on daily living.
Data were double-entered, verified, and analyzed (Dr Johnson and Mrs Hennessy) by using SPSS (SPSS Inc, Chicago, IL) and Stata9 (Stata Corp, College Station, TX). Differences between groups were analyzed by using independent samples t tests, and the effect of gender and gestational age on cognitive outcomes was assessed by using regression analyses. The risk of disability is reported as odds ratios (ORs) with 95% confidence intervals (CIs). Cohen's κ was computed to determine stability in disability classifications. Estimation of the prevalence of serious cognitive impairment (MPC score < 82) and functional disability at 11 years of age was analyzed by using multiple imputation to account for selective dropout.13 Variables used to predict 11-year outcomes were those independently and statistically significantly (P ≤ .05) associated with these outcomes at discharge, 2.5 years, and 6 years of age.
Examination of longitudinal data for extremely preterm children reassessed (n = 219) and not assessed (n = 89) at 11 years of age (Table 1) revealed that those not assessed were more likely to be of nonwhite ethnic origin, had an operation for necrotizing enterocolitis, had unemployed parents, had lower cognitive scores, and had more frequent cognitive impairment (score of less than −2 SD) at 2.5 and 6 years of age. Classmates not reassessed at 11 years of age also had lower cognitive scores (Table 2). However, the mean difference in cognitive scores between extremely preterm children and classmates at 11 years of age would change by only 0.5 points had the dropouts been assessed.
Extremely preterm children had significantly lower MPC scores (mean: 83.7 [SD: 18.0]) than the classmates (mean: 104.1 [SD: 11.1]), with a mean difference of 20 points (95% CI: −17 to −23). Excluding children with substituted MPC scores reduced the mean deficit only slightly to −16.8 points (95% CI: −19 to −14). This group difference was −15.7 points in children for whom socioeconomic status data were available (extremely preterm children: 169; classmates: 135), and adjustment for socioeconomic status reduced this only by 0.2 points (−15.5 points [95% CI: −18.2 to −12.8]). Among classmates there was no gender difference in MPC scores. In contrast, extremely preterm boys had scores 8 points lower than girls (95% CI: 3 to 13). Mean (SD) MPC scores at each gestational week were: 23 weeks, 82.9 (21.2); 24 weeks, 79.6 (20.8); and 25 weeks, 86.1 (15.3). There was an independent effect of male gender (−9.2 [95% CI: −13.9 to −4.5]; P < .001) and gestational age (per week) (4.5 [95% CI: 1.4–7.7]; P = .005). Excluding 16 children with substituted MPC scores, the effects were −5.1 for male gender (95% CI: −8.8 to −1.3; P = .008) and 2.0 for gestational age (95% CI: −0.5 to 4.4; P = .12).
Severity of cognitive impairment was classified into 4 categories by using classmate scores as reference data (Table 3). Serious cognitive impairment (severe/moderate; MPC score of <82) was found in 40% (95% CI: 33%–47%) of extremely preterm children and 1.3% of classmates (OR: 50 [95% CI: 12–206]).* Serious impairment was more common in extremely preterm boys (50%) than girls (31%) (OR: 2.1 [95% CI: 1.2–3.7]) and in children born at 23 to 24 weeks' gestation (45%) compared with 25 weeks' gestation (36%) (OR: 1.48 [95% CI: 0.9–2.6], not significant). This prevalence remained stable from 6 (41%) to 11 (40%) years of age among the full cohort and those with longitudinal data (n = 202; 36% to 40%). Using multiple imputation to correct for selective loss to follow-up, the estimated proportion of extremely preterm children with serious cognitive impairment rose to 45% (95% CI: 38%–52%).
Thirty-eight (17.4%) extremely preterm children had CP compared with none of the classmates (Table 4), and 21 (9.6%) extremely preterm children had CP with serious neuromotor impairment (Tables 3 and 5). The prevalence of cognitive impairment was highest in children with spastic bilateral CP (73%), compared with other types of CP (50%) or those without CP (34%; P < .001). CP was more common among boys (25%) than girls (11%) (OR: 2.7 [95% CI: 1.3–5.5]), and among children born at 23 to 24 weeks' gestation (22%) than 25 weeks' gestation (14%) (OR: 1.64 [95% CI: 0.8–3.3]). The prevalence of CP remained stable over time (2.5 years of age: 17.5%; 6 years of age: 20.3%). The 11-year prevalence decreased slightly from 6 years of age, but differential loss to follow-up and changes in classification of children with mild CP make comparison difficult. Of the 49 children with CP seen at 6 years of age, 37 (75%) were reevaluated at 11 years of age, of whom 6 no longer had demonstrable neurologic signs.
In extremely preterm children, the prevalence of serious hearing impairment declined from 6% at 6 years of age to 2% (n = 4) at 11 years of age (Tables 3 and 5). Thirteen of 14 children with serious impairment at 6 years were reassessed at 11 years, of whom 9 were reallocated to a less-impaired category.
The prevalence of serious visual impairment was 7% and 9% at 6 and 11 years of age, respectively. The risk of serious sensory impairment did not differ significantly by gender or gestational week. Of 97 (44%) extremely preterm children with mild visual impairment, 72 (74%) had squints, with or without refractive errors. Of 27 (19%) classmates with mild impairment, 22 (81%) had refractive errors alone. There was an increase in mild visual impairment in both groups (28% of extremely preterm children and 4% of classmates had mild impairment at 6 years of age).
Composite Functional Disability
Ninety-eight (45%) extremely preterm children had a serious disability compared with 2 (1.3%) classmates (OR: 61 [95% CI: 15–253]) (Fig 1). Serious disability was more common in extremely preterm boys (53%) than girls (38%) (OR: 1.8 [95% CI: 1.0–3.1]) and in those born at 23 to 24 weeks' gestation (53%) than 25 weeks' gestation (39%) (OR: 1.8 [95% CI: 1.0–3.1]) (Table 5).† In those with a serious disability, 75% had serious impairment in 1 domain, 17% in 2 domains, and 8% in 3 domains; none had impairments in all 4 domains. Imputing composite disability in the entire cohort gave the following proportions for none/mild, moderate, and severe disability, respectively: 50% (95% CI: 44%–57%); 31% (95% CI: 24%–37%); and 19% (95% CI: 14%–24%).
Cognitive deficit was the most prevalent impairment: 88% of extremely preterm children were classified with a serious disability on the basis of a cognitive impairment alone (n = 65) or in combination with other impairments (n = 21). In those without cognitive impairment, 7 (7%) were classified on the basis of serious impairment in vision, 2 (2%) in hearing, 2 (2%) in neuromotor function, and 1 (1%) with both visual and neuromotor impairments.
Stability in Functional Disability
Using longitudinal data, the prevalence of serious disability in extremely preterm children (n = 202) remained stable from 6 (42%) to 11 years of age (45%) (Fig 1); these figures are similar to the point prevalence at both ages (46% and 45%, respectively). The lower prevalence at 6 years in those assessed longitudinally can be accounted for by selective dropout of children with a disability (Table 1). Cohen's κ between classifications at 6 and 11 years of age was 0.6 (95% CI: 0.5–0.7; P < .001).
There was considerable variation in individual trajectories (Table 6). Of 84 extremely preterm children who shifted category over time, 38 (45%) became less impaired and 46 (55%) more impaired, resulting in similar prevalence over time. The majority of movement occurred between neighboring categories (92%) with a minority shifting 2 (7%) or 3 categories (1 child with serious hearing impairment at 6 years of age but hearing without aids at 11 years). Prediction of outcome at 11 years was enhanced with increasing severity of disability at 6 years (Table 6).
In this population of children born at ≤25 weeks' gestation, we found serious functional disability in 45% compared with 1% of term classmates at 11 years of age, which, after adjustment for dropouts, is estimated to be 50% (95% CI: 43%–56%) in the entire cohort. Cognitive deficit was the most prevalent impairment, and the majority of children with serious disability were classified as such on the basis of cognitive impairment. The 40% prevalence of cognitive impairment is somewhat higher than reported for other population-based cohorts in middle childhood.1,4,5,14 Similarly, the prevalence of functional disability is higher than the 20% to 30% prevalence typically reported in other studies.4,5 Drawing parallels is difficult because of variability in populations and methodologies. Furthermore, other studies apply standardized test norms to classify cognitive function. Because the application of obsolete test norms underestimates impairment,2,10,15,16 we classified cognitive ability according to reference data obtained from the group of classroom peers. When the 1983 K-ABC norms are applied to this population, the 14% and 22% prevalence of serious cognitive and functional disability, respectively, are comparable with previous reports. We and others2,10 advocate the use of contemporaneous controls for classifying impairment as children are compared with current peers for educational decisions and treatment provision.
Commensurate with other reports, the pattern and prevalence of serious neurodevelopmental disability was remarkably stable throughout middle childhood and was related to underlying stability in classifications of cognitive impairment.17 There seems to be less stability in disability from infancy to middle childhood,2,18 but studies are based on infant developmental tests that are poor predictors of later IQ.18 Reporting group outcomes may mask variation in individual developmental trajectories and, although the prevalence of disability was comparable between 6 and 11 years of age, individual stability was only moderate to high over this period. Changes in classifications are to be expected, because children with MPC scores close to category boundaries may fall on either side of the border at different assessments given the confidence limits of the test8 and variation in performance over time.19 Large shifts in outcome were highly exceptional: most moved only to a neighboring category. Disability was most stable in children with severe impairment and least stable in those with no impairment at 6 years of age. Although we have shown stability over time in cognitive ability and neurosensory disability, assessments of other outcomes such as academic attainment, socioemotional problems, and neuropsychological abilities, may yield less stability, because the nature of functional impairments may evolve or diminish over time.20
The implications of these results are twofold. Given the high continuity observed, follow-up to 6 years of age is arguably sufficient to provide a valid estimate of the prevalence of general disability later in childhood. However, individual variation reasserts the need for long-term follow-up, particularly for those who were free of significant disability at 6 years in whom the greatest change was observed.
The relatively poor stability in children without impairment at 6 years of age is mainly accounted for by an increased prevalence of mild visual impairments at 11 years (28%–44%). A concomitant increase among classmates (4%–18%) indicated that the increased prevalence in extremely preterm children was not cohort-specific, but a result of a generally increased prevalence of refractive errors and better identification of visual impairments at this age. Squints were more common in extremely preterm children than classmates at both ages. Improved early identification of minor visual problems may help prevent secondary problems from arising, such as increased social victimization.21 Serious hearing disability decreased over time (6%–2%). This is unlikely to be a result of selective dropout and, thus, reflects improved hearing.
The 17% prevalence of CP at 11 years of age is comparable to other studies of extremely preterm/extremely low birth weight survivors in middle childhood.4,5 In a recent study, the prevalence of CP was 6%, but these diagnoses were based on a review of health records.4 Our classification was made posthoc from clinical signs detected using a standardized neurologic examination and was therefore more inclusive. CP was associated with a higher rate of cognitive impairment, but only 10% of the extremely preterm children had serious neuromotor impairment at 11 years of age compared with 12% at 6 years of age. The prevalence of CP in this cohort has remained highly stable from 2.5 years of age. The slightly decreased prevalence at 11 years may be accounted for by the selective dropout of children with a disability, but at each age we classified the presence or absence of CP dependent on the results of neurologic evaluation at that time. This process was conducted to maintain consistency in diagnostic criteria between the 2 assessments, and the earlier assessment at 30 months of age, but may result in changes in categorization for children with minimal neurologic signs.
Because classmates comprised solely children in mainstream school, it may be argued that this is a higher-achieving group than the normal population. However, in the United Kingdom and Ireland, children with special educational needs are largely integrated into mainstream schools. The prevalence of special school placements is 1.1% in England,22 hence the inclusion of classmates for extremely preterm children in special schooling would unfairly bias the comparison group. In addition, on recently standardized tests of academic attainment the full group of classmates achieved mean scores expected for the normal population (reading: mean: 99 [SD: 12]; math: mean: 99 [SD: 15]), thus we would suggest that our classification accurately reflects the degree of impairment in the school setting.23 Furthermore, given loss to follow-up of children with significant disabilities, it is likely we have underestimated the true extent of impairment, as is common in longitudinal studies.24 Analysis of missing data using multiple imputation suggests that serious cognitive impairment, and thus serious functional disability, in the entire cohort is 5% greater than measured. The increased risk of CP and cognitive and functional disability in extremely preterm boys was evidenced previously2,6 and in other populations.3,25–27 This is in the context of lower survival and increased neonatal morbidity among boys, and we have speculated that gender differences in intrauterine development may render the male fetus particularly vulnerable to perinatal adversity and developmental sequelae.27,28
Extremely preterm children remain at high risk for neurodevelopmental disability at 11 years of age, and cognitive deficits continue to be the most prevalent impairment. The prevalence of serious disability remained stable between 6 and 11 years of age and large individual shifts in severity of disability were rare. Follow-up to 6 years of age may be sufficient to identify the prevalence of neurodevelopmental disability in middle childhood and to provide guidance relating to outcomes for clinicians and parents. In addition to the disabling conditions we have described, extremely preterm children are also at risk for more subtle difficulties such as executive dysfunctions and social and emotional problems29,30 that require further investigation in this population.
This study was funded by the Medical Research Council. We are indebted to the EPICure Study Group, which includes pediatricians in 276 maternity units in the United Kingdom and Ireland who identified the original cohort, contributed perinatal data to the study, and whose help was invaluable.
Co-investigators were Neil Marlow (Nottingham; chief investigator), Kate Costeloe (London), Enid Hennessy (London), Janet Stocks (London), and Elizabeth Draper (Leicester). Study progress was monitored by a steering group chaired by Peter Brocklehurst (Oxford). Assessments were performed by psychologists Rebecca Smith, Rebecca Trikic, and Samantha Johnson and pediatricians Joseph Fawke, Susan Thomas, and Victoria Rowell. Heather Palmer was the study manager.
We are also indebted to the schools and teachers who supported the study assessments throughout this follow-up, and to the many children and parents for their continued participation in the EPICure Study (www.epicure.ac.uk).
- Accepted March 25, 2009.
- Address correspondence to Samantha Johnson, PhD, CPsychol, Institute for Women's Health, 86-96 Chenies Mews, London WC1E 6HX, United Kingdom. E-mail: or
↵* Using the K-ABC normative data (published in 1983), 30 (13.7% [95% CI: 9.5%–19.1%]) extremely preterm children had serious cognitive impairment compared with none of the term classmates. Allowing for loss to follow-up, the estimated proportion with serious cognitive impairment rose to 17.4% (95% CI: 12.4%–22.4%).
↵† Using the K-ABC normative data to classify cognitive impairment resulted in 48 (22%) extremely preterm children with serious disability compared with none of the term classmates.
Financial Disclosure: The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject:
Extremely preterm children are at high risk for long-term neurodevelopmental disability, with the most common disability being cognitive impairment at 6 years of age.
What This Study Adds:
Extremely preterm children remain at high risk for neurodevelopmental disability at 11 years of age, and cognitive deficits remain the most prevalent impairment. The prevalence of serious disability remains remarkably stable between 6 and 11 years of age.
- ↵Mikkola K, Ritari N, Tommiska V, et al. Neurodevelopmental outcome at 5 years of age of a national cohort of extremely low birth weight infants who were born in 1996–1997. Pediatrics.2005;116 (6):1391– 1400
- ↵Farooqi A, Hagglof B, Sedin G, Gothefors L, Serenius F. Chronic conditions, functional limitations, and special health care needs in 10- to 12-year-old children born at 23 to 25 weeks' gestation in the 1990s: a Swedish national prospective follow-up study. Pediatrics.2006;118 (5). Available at: www.pediatrics.org/cgi/content/full/118/5/e1466
- ↵Office for National Statistics. The National Statistics Socio-economic Classification User Manual. New York, NY: Palgrave Macmillan; 2005
- ↵Kaufman AS, Kaufman NL. Kaufman Assessment Battery for Children. Circle Pines, MN: American Guidance Service, Inc; 1983
- ↵Royston P. Update: multiple imputation of missing values. Available at: www.stata-journal.com/software/sj5-4. Accessed July 15, 2008
- ↵Saigal S, den Ouden L, Wolke D, et al. School-age outcomes in children who were extremely low birth weight from four international population-based cohorts. Pediatrics.2003;112 (4):943– 950
- ↵Gross SJ, Slagle TA, D'Eugenio DB, Mettelman BB. Impact of a matched term control group on interpretation of developmental performance in preterm infants. Pediatrics.1992;90 (5):681– 687
- ↵Doyle L, Casalaz D. Outcome at 14 years of extremely low birthweight infants: a regional study. Arch Dis Child Fetal Neonatal Ed.2001;85 (3):F159– F164
- ↵Hack M, Taylor HG, Drotar D, et al. Poor predictive validity of the Bayley Scales of Infant Development for cognitive function of extremely low birth weight children at school age. Pediatrics.2005;116 (2):333– 341
- ↵Horwood J, Waylen A, Herrick D, Williams C, Wolke D. Common visual defects and peer victimization in children. Invest Ophthalmol Vis Sci.2005;46 (4):1177– 1181
- ↵Department for Children, Schools, and Families. Schools and Pupils in England, January 2007 (Final). London, United Kingdom: Department for Children, Schools, and Families; 2007
- ↵Johnson S, Hennessy E, Smith R, Trikic R, Wolke D, Marlow N. Academic attainment and special educational needs in extremely preterm children at 11 years of age: the EPICure Study. Arch Dis Child Fetal Neonatal Ed.2009; Published Online First: March 12, 2009. doi:10.1136/adc.2008.152793
- ↵Brothwood M, Wolke D, Gamsu H, Benson J, Cooper D. Prognosis of the very low birthweight baby in relation to gender. Arch Dis Child.1986;61 (6):559– 564
- ↵Kraemer S. The fragile male. BMJ.2000;321 (7276):1609– 1612
- ↵Marlow N, Hennessy E, Bracewell M, Wolke D. Motor and executive function at 6 years of age following extremely preterm birth. Pediatrics.2007;120 (4):793– 804
- Copyright © 2009 by the American Academy of Pediatrics