Author Affiliations: Department of Pediatrics, Case Western Reserve University and Rainbow Babies & Children's Hospital, University Hospitals Case Medical Center (Drs Taylor and Hack and Ms Minich), and Department of Education, Cleveland State University (Dr Klein), Cleveland, Ohio; and Office of Research and Department of Psychology, University of Nebraska–Lincoln (Dr Espy). Ms Anselmo is an educational consultant in Gates Mills, Ohio.
Objectives To assess learning problems among kindergarten students with extremely preterm birth and to identify risk factors.
Design Cohort study.
Setting Children's hospital.
Participants A cohort of 148 children born between January 1, 2001, and December 31, 2003, with extremely preterm birth, defined as less than 28 weeks' gestation or having a birth weight of less than 1000 g, and 111 classmate control individuals born at term with normal birth weight.
Interventions The children were enrolled in the study during their first year in kindergarten and were assessed on measures of learning progress.
Main Outcome Measures Achievement testing, teacher ratings of learning progress, and individual educational assistance.
Results Children with extremely preterm birth had lower mean standard scores than controls on achievement tests of spelling (8.52; 95% confidence interval, 4.58-12.46) and applied mathematics (11.02; 6.76-15.28). They had higher rates of substandard learning progress by teacher report in written language (odds ratio, 4.23; 95% CI, 2.32-7.73) and mathematics (7.08; 2.79-17.95). Group differences in mathematics achievement and in teacher ratings of learning progress were statistically significant even in children without neurosensory deficits or low global cognitive ability. Neonatal risk factors, early childhood neurodevelopmental impairment, and socioeconomic status predicted learning problems in children with extremely preterm birth; however, many children with problems were not enrolled in a special education program.
Conclusions Learning problems in children with extremely preterm birth are evident in kindergarten and are associated with neonatal and early childhood risk factors. Our findings support efforts to provide more extensive monitoring and interventions before and during the first year of school.
Children with extremely preterm birth (hereafter referred to as EP children) born before 28 weeks' gestation or with a birth weight of less than 1000 g are at high risk for neurodevelopmental disorders.1 These disorders range from severe conditions, such as cerebral palsy, sensory deficits, and overall mental deficiency, to more subtle impairments in cognition, behavior, and academic achievement, with many EP children requiring special educational assistance during their school-aged years.2- 12 Academic difficulties are of particular concern once EP children begin kindergarten because failure to recognize these difficulties can delay special educational interventions.13
We know little about the educational progress of EP children during this critical period of schooling. Studies11,12,14- 24 have documented academic difficulties at early school age per test results or per teacher ratings of achievement. However, this previous research did not examine the nature and predictors of learning problems in kindergarten. Knowledge pertaining to early academic outcomes is vital in appraising the educational needs of these children and their risks for future learning problems.25- 29 Information regarding neonatal and early childhood factors related to poor achievement in kindergarten also would be useful to identify children most in need of careful monitoring.
Our primary objectives were to assess early learning progress among EP children vs that of control individuals born at term with normal birth weight (hereafter referred to as NBW controls) and to examine associations of educational outcomes in the EP group with neonatal risk factors, early childhood neurodevelopmental impairment, and socioeconomic status (SES). The primary hypothesis was that rates of academic problems and learning assistance would be higher for EP children than for NBW controls.3,12,21,23,24,29 Secondary hypotheses were that group differences would remain even when excluding children with more severe deficits,20,30 that many EP children with academic problems would not be receiving special assistance in school,10,31,32 and that higher rates of learning problems in the EP group would be associated with neonatal risk factors, early childhood neurodevelopmental impairment, and lower SES.3,18,19,25,33- 43
Excluding 8 infants with congenital infections or abnormalities, 246 EP children born at less than 28 weeks' gestation or with a birth weight of less than 1000 g were admitted to the neonatal intensive care unit of Rainbow Babies & Children's Hospital, Cleveland, Ohio, between January 1, 2001, and December 31, 2003. Of 198 (80.5%) who survived, 148 (74.7%) were enrolled in the study. Reasons for nonparticipation included failure to locate families (21 children), relocation out of the region (16 children), and refusals or no-shows (5 children). Eight children could not be assessed because of custody issues or because they were non–English speaking. Comparison of participants with nonparticipants failed to reveal statistically significant differences in sex, race/ethnicity, or neonatal risk factors. The children were followed up longitudinally and were contacted during the initial year in which they attended kindergarten. One hundred twenty-nine EP children were in regular classrooms, 16 were in full-time special education programs, and 3 were homeschooled.
The EP group was composed of 56 children (37.8%) with less than 750 g birth weight, 32 children (21.6%) with less than 25 weeks' gestation, and 37 children (25.0%) with birth weight less than the 10th percentile for gestational age.44 Neonatal complications included infection or necrotizing enterocolitis in 60 children (40.5%), bronchopulmonary dysplasia (defined as supplemental oxygen at 36 weeks' corrected age) in 77 children (52.0%), severe retinopathy of prematurity (defined as stage 4 or 5 retinopathy or receipt of cryotherapy or laser therapy45) in 27 children (18.2%), and abnormal ultrasonographic findings (defined as grade III or IV intraventricular hemorrhage, periventricular leukomalacia, or ventricular dilatation at discharge) in 15 children (10.1%). At 20 months' corrected age, 60 EP children (40.5%) had neurodevelopmental impairment, defined as cerebral palsy (13 children [8.8%]), blindness or deafness (6 children [4.1%]), or a Mental Development Index of less than 70 on the second edition of the Bayley Scales of Infant Development46 (57 children [38.5%]).
The control group was composed of 111 term-born NBW children (>36 weeks' gestational age and >2500 g birth weight) from the same classrooms as the EP children (n = 93) or from other kindergarten classrooms (n = 18). To recruit controls, flyers were sent home with the students in each classroom. From the pool of students whose families responded, we recruited the classmate who most closely matched each EP child based on sex, race/ethnicity, and age at assessment. Controls were not recruited for children who were homeschooled or attending full-time self-contained special education programs (n = 19). Matches for the remaining 18 EP children in regular classrooms were not recruited because of school refusals, school locations too distant from the medical center, or difficulties in finding matches.
No group differences were observed in sex, race/ethnicity, or SES as defined by the mean of the sample's z scores for maternal education in years, caregiver occupation (averaged for 2-parent households),47 and median income for the neighborhood of residence based on data from the 2000 US Census (Table 1).48 The EP group was older than the NBW group in chronological age at school entry. However, the groups did not differ in age corrected for prematurity because proportionally more EP children had been held back a year before kindergarten entry (21.6% vs 2.7%, P < .001). Group comparisons also failed to reveal differences in age at testing, months in school when the teacher ratings were completed, or percentage of children in full-day kindergarten (83.7% vs 80.2%). Although a larger proportion of the EP group received early intervention services before school entry (83.1% vs 12.6%, P < .001), the groups showed little difference in rates of enrollment in preschool (83.8% vs 84.7%) or day care (49.3% vs 50.5%).
Children were administered tests of achievement and cognitive ability by examiners who were masked to birth status; during that time, their parents completed interviews and questionnaires to obtain information regarding family characteristics and children's health status. Children's tests were administered in a fixed order during a single session. Six EP children were untestable on most measures because of cerebral palsy and severe cognitive impairment. Scores on achievement and intelligence measures for these children and others whose functioning was too low to be tested were assumed to be less than 85. The research was approved by the University Hospitals Case Medical Center Institutional Review Board; written informed consent from parents and teachers was obtained before participation.
Achievement measures included the letter and word identification, spelling, calculation, and applied problems subtests of the Woodcock-Johnson III Tests of Achievement.49 To take into account time in kindergarten, the scores on these tests were based on grade rather than on age standards. Test reliability for these measures is good, and validity has been documented relative to other achievement tests. Estimates of IQ were obtained using the age-standardized brief intelligence assessment (BIA) score from the Woodcock-Johnson III Tests of Cognitive Abilities.50
Teachers provided information regarding special education programs as defined by an individual educational plan (IEP). The IEPs were identified by parent report when the teacher report was unavailable. Teachers of children attending regular classrooms reported on accommodations for learning difficulties within the classroom and completed ratings of student learning progress in each of several curricular objectives based on State of Ohio–mandated goals.51 Using a previously validated method,52 0 was assigned for progress toward a given objective that was behind expectations by at least a full grading period, 1 for progress that was below expectations by less than a full grading period, and 2 for progress that was at or above expectations. An advantage of these ratings compared with formal testing is that they provided information regarding children's learning progress in specific areas of competency. Teachers were not informed of children's birth status.
Analysis of covariance was used to compare the EP and NBW groups on the results of achievement testing and on the sum of teacher ratings of learning progress in written language and mathematics. Sex, race/ethnicity, and SES were included as covariates. Child race/ethnicity was defined by caregivers and was included in analyses to control for demographic background. Preliminary analyses failed to indicate any covariate × group interactions, justifying exclusion of interaction terms from the models.
Logistic regression analysis was conducted to examine group differences in learning problems as defined by standard scores of less than 85 on the achievement tests, substandard teacher ratings of learning progress (ie, 0 or 1), special education (as defined by an IEP or related services), and any classroom accommodation. Covariates in these analyses were the same as those included in the corresponding analysis of covariance. The variable of months in school was controlled for in analyses of IEPs because these placements increased during the school year.
To determine if learning deficits in the EP children could be attributed to the subset of children with cerebral palsy, blindness, deafness, or low global cognitive ability (as defined by a BIA of less than 85), group comparisons were repeated with these children excluded from the analysis. To determine if these deficits were identifiable based on having an IEP, group comparisons also were rerun after excluding this subset of children. Associations of poor achievement test performance and substandard learning progress with IEPs and classroom accommodations were investigated using χ2 analysis.
Associations of learning problems with neonatal risk factors, neurodevelopmental impairment at 20 months, and SES were examined using logistic regression analysis. Each predictor was examined in a separate analysis. Sex, race/ethnicity, and SES were covariates in analyses of the neonatal and neurodevelopmental factors.
Bonferroni-corrected α levels were used to determine statistical significance, with adjustments on a familywise basis within the domains of written language and mathematics. Therefore, α level was .05 divided by 3 (ie, .02) for each of 3 primary measures within each domain (ie, 2 test scores and the sum of teacher ratings for the analyses of covariance; deficits on the results of 2 tests and any indication of substandard learning progress for the logistic regression analyses). For the sample size, the power was 0.80 with 2-sided α = .05 for detecting effect sizes of 0.33 and differences in event rates of 15.0%. Differences of at least this magnitude were anticipated based on previous research with similar samples.3
The EP children had lower mean scores on achievement tests of spelling (8.52; 95% confidence interval, 4.58-12.46) and applied problems (11.02; 6.76-15.28) and lower sum teacher ratings of learning progress in written language and mathematics than NBW controls (Table 2). A larger proportion of the EP group compared with the NBW group was untestable or failed to reach basal levels of performance in mathematics calculation (52.1% vs 20.7%, P < .001). Even when excluding children with neurosensory disorders or a BIA of less than 85 (ie, 65 EP children and 12 NBW children), the EP group had lower scores on applied problems and lower sum teacher ratings of learning progress than the NBW group. Several of the group differences also remained statistically significant when excluding children with IEPs.
Consistent with these results, the EP group had higher rates of IEPs and classroom accommodations (Table 3). Several group differences remained statistically significant in analyses that excluded children with neurosensory deficits or a BIA of less than 85 or those with IEPs.
Children with IEPs and classroom accommodations had higher rates of poor achievement test performance and substandard learning progress in written language and mathematics (P < .05 for all). However, 26 of 70 EP children (37.1%) with a score of less than 85 on 1 or more achievement tests did not have an IEP. Low-achieving EP children were less likely to have an IEP if they had normal neurosensory status and a BIA of 85 or higher (n = 14) vs a neurosensory deficit or a BIA of less than 85 (n = 12) (P = .01). Low-achieving EP children without an IEP had statistically significantly higher scores on spelling and applied problems and lower rates of less than 25 weeks' gestation, abnormal ultrasonographic findings, and neurodevelopmental impairment at 20 months compared with those with an IEP.
Within the EP group, neonatal risk factors, neurodevelopmental impairment at 20 months, and SES were associated with learning problems (Table 4). Less than 25 weeks' gestation and bronchopulmonary dysplasia were associated with higher rates of IEPs and classroom accommodations even when children with neurodevelopmental impairment at 20 months were excluded from analysis.
Survival of EP children has reached a previously unsurpassed level but is offset by high rates of neonatal morbidity.53 As increasing numbers of these children are reaching school age, it is critical that we identify their problems at school entry and factors related to early educational difficulties. Our results reveal that the EP children performed worse on achievement tests, were making less learning progress, and had higher rates of IEPs and classroom accommodations than NBW controls. Academic problems were evident based on the results of tests of written expression and mathematics and in teacher ratings of learning progress in these areas. A 2009 study20 of academic outcomes among 6-year-olds born in the 1990s with gestation of 33 or fewer weeks indicated higher rates of academic problems compared with NBW controls even for the subset of EP children without neurosensory deficits or low global cognitive ability. Our results suggest similar outcomes for children born more recently. Although other studies12,18,20,21,23,54,55 have reported deficiencies in academic readiness or initial achievement skills among preterm children, ours is the first study to date, to our knowledge, that demonstrates the pervasiveness of learning deficiencies during the first year in school among EP children born since 2000. The EP and NBW groups were similar in sociodemographic characteristics, indicating that neonatal brain insults and subsequent abnormalities in neural development may have contributed largely to the higher rates of educational difficulties for the former group.56,57
Also, this study is among the first to examine educational interventions for EP children in kindergarten and the correspondence of these interventions with learning problems. More EP children than NBW controls were receiving special assistance at school; however, many EP children with low scores on an achievement test did not have an IEP. Low-achieving EP children without an IEP had higher achievement scores and lower rates of neurosensory deficit, poor cognitive ability, and neonatal complications than those with an IEP, suggesting that children with less severe impairments are more likely to be underidentified. It is also possible that IEPs had not been completed for some children despite parent or teacher awareness of learning difficulties. Our results imply that the educational needs of EP children are not met fully.
Achievement deficits in the EP group were evident in spelling and applied mathematics but not in word recognition. When children with neurosensory deficits or low global cognitive ability were excluded from analysis, only the deficit in mathematics in applied problems remained statistically significant. These findings are consistent with other data indicating that among preterm children, difficulties in mathematics are more prominent and selective than problems in reading.1,6,20,35,58- 60
Neonatal risk factors associated with the learning problems of EP children in kindergarten included less than 25 weeks' gestation, abnormal cranial ultrasonographic findings, bronchopulmonary dysplasia, and infection. Neurodevelopmental impairment at 20 months and lower SES also predicted learning problems. Previous studies3,18,19,31,32,34,36,37,39- 43,61,62 have reported similar risks for deficits in school readiness skills and special education interventions. Therefore, these risk factors may be useful in identifying the children most in need of developmental monitoring and early childhood interventions before school entry.
A limitation of this study was that teacher ratings of learning progress were obtained only for EP children attending regular classrooms. This procedure was warranted to ensure that expectations for progress could be judged relative to the general kindergarten population but is likely to have underestimated the effect of EP on learning progress. A further limitation is that information regarding IEPs was provided by teachers and parents, so we had no means of determining the extent to which these plans were implemented or addressed the children's learning problems. Our EP cohort was hospital based, and we were unable to follow up 25.3% of the families. The sample was predominantly urban and African American and, although largely reflective of EP births within the region, was not representative of the population of the United States as a whole. Educational interventions may have varied from those provided elsewhere in the United States. Because of privacy issues, parents of NBW children were selected from among those who agreed to be contacted, introducing possible bias in the selection of controls. However, the children attended many schools throughout the region; and no differences were observed in background factors between participants and nonparticipants or between the EP and NBW groups. Our results are similar to those from previous studies20,21,30 of early educational outcomes.
In older school-aged children with very low birth weight, poor reading skills are associated with reduced cerebellar volume63 and mathematics difficulties with reduced gray matter in the left parietal lobe.64 Further studies are needed to investigate the neural basis of early academic difficulties in EP children and to examine the cognitive and behavioral problems associated with these difficulties. Deficits in specific skill areas, such as processing speed and working memory, may be related closely to academic difficulties, with different patterns of association for reading and mathematics disabilities.11,59,65 Additional study of environmental influences on academic achievement at school entry is needed to identify characteristics of families and schools that buffer or exacerbate the effects of extremely preterm birth on early learning progress.29,66
In conclusion, our findings reveal high rates of learning problems and educational interventions among EP children at the time of school entry, as well as associations of these outcomes with neonatal risk factors and neurodevelopmental impairment at 20 months. The results underscore that many EP children with learning problems in kindergarten are not receiving special interventions and that teacher ratings of learning progress hold promise as a simple and effective means to screen for these problems. Efforts to minimize the adverse effects of EP on later academic achievement may need to include more extensive developmental and educational interventions beginning in kindergarten or earlier.18- 21,25- 27 A system of follow-up care that provides ongoing monitoring of children's status and needs from birth through school age, such as a medical home,26 may be helpful in ensuring early detection and intervention.
Correspondence: H. Gerry Taylor, PhD, Department of Pediatrics, Rainbow Babies & Children's Hospital, University Hospitals Case Medical Center, 10524 Euclid Ave, Cleveland, OH 44106-6038 (firstname.lastname@example.org).
Accepted for Publication: March 15, 2011.
Author Contributions: Dr Taylor had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Taylor, Klein, Anselmo, Minich, Espy, and Hack. Acquisition of data: Taylor, Klein, and Hack. Analysis and interpretation of data: Taylor, Klein, Anselmo, Minich, Espy, and Hack. Drafting of the manuscript: Taylor and Hack. Critical revision of the manuscript for important intellectual content: Taylor, Klein, Minich, Espy, and Hack. Statistical analysis: Taylor and Minich. Obtained funding: Taylor, Minich, Espy, and Hack. Administrative, technical, and material support: Taylor, Anselmo, and Hack. Study supervision: Taylor, Klein, Minich, and Hack.
Financial Disclosure: None reported.
Funding/Support: This work was funded by grant HD050309 from the National Institutes of Health (principal investigator, Dr Taylor), which provided financial support for the design and conduct of the study.
Additional Contributions: Anne Birnbaum, AA, Elizabeth Roth, PhD, Dan Maier, MA, Andrea Barkoukis Gefteas, MS, Michelle R. Jacobs, MA, Alice Kostiuc, BA, and Ketrin Lengu, BA, assisted in recruitment, data collection, and coding.
Taylor HG, Klein N, Anselmo MG, Minich N, Espy KA, Hack M. Learning Problems in Kindergarten Students With Extremely Preterm Birth. Arch Pediatr Adolesc Med. 2011;165(9):819-825. doi:10.1001/archpediatrics.2011.137