Neurobehavioral Outcomes of School-age Children Born Extremely Low Birth Weight or Very Preterm in the 1990s FREE

Cognitive deficits, school difficulties, and behavioral problems are often reported for children born with extremely low birth weight (ELBW; birth weight <1000 g) or very preterm (<28 weeks' gestation).^1- 3 A recent meta-analysis of case-control studies reported from 1980 to November 2001 that examined cognitive and behavioral outcomes found that the mean IQ for school-aged children born very preterm was approximately two-thirds SD below that of healthy controls.⁴ In addition, the meta-analysis found that preterm children exhibited more externalizing, internalizing, and attentional behavior problems than healthy controls, although these difficulties were not present in all studies.⁴

These findings, however, were based on cohorts born before the 1990s, before a number of advances in perinatal intensive care, including exogenous surfactant therapy. In the 1990s, survival rates increased significantly, especially for children born before 26 weeks' gestation⁵ or with a birth weight of less than 750 g.⁶ It is imperative to establish the rate of cognitive and behavioral deficits for children born in the 1990s, because the improving survival of the most preterm and smallest infants may have increased the risk for adverse long-term outcomes. The aim of this study was to determine the cognitive, educational, and behavioral outcome at school-age of ELBW or very preterm infants born in the 1990s compared with normal birth weight (NBW) controls.

Of 568 consecutive live births of neonates with birth weights less than 1000 g or with gestational ages younger than 28 completed weeks in Victoria, Australia, between January 1, 1991, and December 31, 1992, 298 infants (52.5%) survived to 2 years of age; these survivors comprised the ELBW or very preterm cohort. The NBW cohort, also enrolled at birth, was composed of 262 survivors from 265 randomly selected children with birth weights of more than 2499 g born in 1 of the 3 level III perinatal centers in Victoria. Each control was born on the expected date of birth for each child with a birth weight of less than 1000 g and was matched for sex, the mother's country of birth (English speaking or not), and health insurance status (private health insurance or not). Social class was determined by the occupation of the main income earner in the family; low social class identified families where the main income earner was unskilled or unemployed. Children were considered to be small for gestational age (SGA) if their birth weight was more than 2 SDs below that expected for gestational age and sex.

All children were enrolled in the newborn period in a prospective, longitudinal study that included medical, psychological, and developmental assessments at 2, 5, and 8 years of age. Outcomes at 2 and 5 years of age for children in this cohort with gestational ages younger than 28 weeks have been reported,^5,7 as has outcome at 2 years of age for children with birth weights of less than 1000 g.⁶ Written informed consent was obtained from parents of NBW children. Follow-up was considered routine clinical care for the ELBW or very preterm children. The study was approved by the Research and Ethics Committees of the Royal Women's Hospital, Melbourne, Australia.

The psychological evaluation included standardized assessments of cognitive ability, educational progress, and behavior problems. During the assessment, psychologists were unaware of participants' perinatal details or the results of earlier assessments in childhood. Twelve children were not administered the psychological assessments due to significant neurosensory impairments. Of the remaining children, not all children could complete all tasks; the number with missing data varied with the test. We did not substitute values for those with missing data. Children were assessed for this component of the study between January 2000 and April 2002.

Cognitive ability was assessed using the Wechsler Intelligence Scale for Children–Third Edition (WISC-III).⁸ Full-scale IQ was used as a measure of general intellectual ability, and the 4-factor index scores were used to examine specific elements of cognitive functioning. The Verbal Comprehension Index (VCI) measures verbal reasoning ability, the Perceptual Organization Index (POI) assesses visual-spatial reasoning ability, the Freedom From Distractibility Index (FDI) taps attention and working memory, and the Processing Speed Index (PSI) evaluates the speed and accuracy of information processing.⁹ Each scale or index is age standardized with a mean of 100 (SD, 15). The criteria for mild intellectual impairment was a full-scale IQ between 70 and 84 (from –2 to less than –1 SD), whereas the criteria for major intellectual impairment was an IQ below 70 (less than –2 SDs).

Educational progress was assessed using the Wide Range Achievement Test (WRAT3)¹⁰ and the Comprehensive Scales of Student Abilities (CSSA).¹¹ The WRAT3 includes 3 subtests that assess reading (word recognition and decoding), spelling, and arithmetic. Each scale is age standardized with a mean of 100 (SD, 15). Mild impairment in these educational domains was determined by scale scores from 70 to 84, whereas scale scores less than 70 represented major impairment. The CSSA is a teacher-completed questionnaire that examines 9 aspects of educational ability, including verbal thinking, speech, reading, writing, handwriting, mathematics, general facts, basic motor generalizations, and social behavior. Each scale is age standardized with a mean of 100 (SD, 15).

The Behavior Assessment System for Children (BASC)¹² includes parent and teacher rating scales. The BASC parent rating scale provides a comprehensive assessment of a child's adaptive and problem behaviors in home and community settings, whereas the teacher rating scale assesses these behaviors at school. Both scales provide composite indexes for externalizing problems, internalizing problems, adaptive skills, and overall behavioral symptoms. All indexes are age-sex standardized, providing T scores with a mean of 50 (SD, 10). For the behavioral indices, T scores of 70 and above are considered clinically significant, whereas T scores between 60 and 69 represent the at-risk range. For the adaptive index, a T score of 30 or below is clinically significant, whereas T scores between 31 and 40 represent the at-risk range.

Initial comparisons of interest were between the ELBW or very preterm cohort and the NBW controls. Within the ELBW or very preterm cohort there was also interest in the contrast between the smallest (birth weight, 500-749 g) and most immature (23-25 weeks' gestation) survivors and their slightly heavier (birth weight, 750-999 g) or more mature (26-27 weeks' gestation) counterparts.

Between-group differences were analyzed by χ² analysis for dichotomous outcome data and by t test for continuous data, since data were normally distributed. Analyses were repeated excluding children with neurosensory impairments (cerebral palsy, blindness, or deafness). Data were also analyzed by multiple linear regression to adjust for potential confounding variables, including sex, SGA status, ethnicity, language spoken at home, family structure, mother's marital status, social class, and mother's and father's education (Table 1). P<.05 was considered statistically significant. Because this study was primarily hypothesis generating rather than hypothesis proving and because each of the domains assessed was of individual interest, we did not reduce P values for multiple tests. With the exception of children with missing data, and as otherwise stated, the analyses included all children.

The initial sample size for the ELBW or very preterm cohort was computed on the basis of detecting a significant increase in survival rate relative to an earlier Victorian cohort born in 1985-1987. With the numbers in each cohort assessed, we had 80% power to detect differences in means between groups as small as 0.27 SD.

The follow-up rate at 8 years of age for the ELBW or very preterm cohort was 92.3% (275/298), significantly higher than the 85.1% (223/262) for the NBW cohort (χ²₁ = 6.6, P = .01). Of the 23 ELBW or very preterm children not assessed at 8 years of age, 8 were lost to follow-up, 11 refused to participate, and 4 were inaccessible, usually living in another country. Of the 23 ELBW or very preterm children, 18 had been assessed at 2 years of age and 9 at 5 years of age. Of the 39 NBW children not assessed at 8 years of age, 22 were lost to follow-up, 13 refused to participate, and 4 were inaccessible. Of the 39 NBW children, 22 had been assessed at 2 years of age and 12 at 5 years of age.

At 2 and 5 years of age, psychometric tests scores, where available, were lower for children ultimately not assessed at 8 years of age than those who were assessed at 8 years of age (age, 2 years: Mental Developmental Index on the Bayley Scales of Infant Development; mean difference, –4.4; 95% CI, –11.7 to 3.3; P = .27; age, 5 years: full-scale IQ on the Wechsler Preschool and Primary Scale of Intelligence–Revised; mean difference, –6.8; 95% CI, –13.6 to –0.1; P = .046). Of the children not assessed at 8 years of age, 30 (48%) of 63 were from lower social class, which was not significantly different from the rate of 227 (46%) of 497 of lower social class in those assessed at 8 years of age.

In children assessed at 8 years of age, the mean gestational age at birth for the ELBW or very preterm group was 26.7 weeks (SD, 1.9), with 73 children (27%) born earlier than 26 weeks and 138 (50%) at 26 or 27 weeks; their mean birth weight was 884 g (SD, 162), with 54 (20%) weighing less than 750 g and 172 (63%) weighing 750 to 999 g. In contrast, the NBW cohort's mean gestational age at birth was 39.3 weeks (SD, 1.4) and the mean birth weight was 3407 g (SD, 443).

In the ELBW or very preterm cohort, 253 children (92%) were born in level III perinatal centers; 202 (73%) received antenatal steroids, 91 (33%) were administered postnatal steroids, and 108 (39%) had surfactant therapy. All children in the NBW group were born in level III perinatal centers and 3 (1%) received antenatal steroids. At 8 years of age, 34 children (12%) in the ELBW or very preterm cohort exhibited significant neurosensory impairments; 3 were blind, 4 required hearing aids, and 29 had cerebral palsy (2 of whom were also blind). In contrast, only 1 child in the NBW cohort was diagnosed as having cerebral palsy.

Sociodemographic characteristics and sex distribution of the ELBW or very preterm and NBW cohorts were similar (Table 1). The mean ages of the ELBW or very preterm and NBW cohorts at the time of the assessment were 8.7 and 8.9 years, respectively. Of the ELBW or very preterm cohort, 13.8% were SGA, whereas no NBW children were SGA (χ²₁ = 33.4, P<.001). Significantly fewer mothers in the ELBW or very preterm cohort were married compared with mothers in the NBW cohort (66.4% vs 77.2%, χ²₁ = 6.7, P = .009), although the proportion of intact families was similar. The socioeconomic status of the 2 cohorts did not differ; however, the proportions with at least 12 years of formal education were significantly greater in mothers (48.0% vs 60.8%, χ²₁ = 8.0, P = .004) and fathers (41.1% vs 61.3%, χ²₁ = 18.2, P<.001) in the NBW cohort.

On the WISC-III, the mean IQ for the ELBW or very preterm children was within the average range (range, 90-109; mean, 95.5), despite being below the IQs of the NBW children (mean, 104.9) (Table 2). The mean index scores for the ELBW and very preterm children were also within the average range, despite being between one-third and two-thirds SD below the scores of the NBW children (Table 2). Compared with the ELBW or very preterm cohort, the NBW cohort exhibited superior general intellectual ability (full-scale IQ: mean difference, –9.4; t₄₇₆ = −6.8; P<.001), verbal reasoning (VCI: mean difference, −6.8; t₄₇₆ = −5.0; P<.001), visual-spatial reasoning (POI: mean difference, −9.9; t₄₇₄ = −7.1; P<.001), attention and working memory (FDI: mean difference, −8.2; t₄₇₅ = −6.1; P<.001), and processing speed (PSI: mean difference, −6.7; t₄₆₈ = −4.9; P<.001). Excluding children with neurosensory deficits resulted in small reductions in the mean differences between the groups, but no statistical conclusions were altered (Table 3). Similarly, no statistical conclusions were altered after adjusting for sociodemographic variables on the multivariate analyses (Table 3).

On formal tests of academic achievement, the ELBW or very preterm children performed below the NBW children (Table 2). The reading and spelling mean scores for the ELBW or very preterm cohort were well within the average range but significantly below the mean scores achieved by the NBW cohort (reading: mean difference, −6.7; t₄₇₆ = −4.7; P<.001; spelling: mean difference, −5.6; t₄₇₄= −4.7; P<.001). These statistical conclusions were not affected when analyses were repeated excluding children with neurosensory deficits and after adjusting for sociodemographic variables (Table 3). After controlling for IQ, the group differences in reading and spelling were no longer statistically significant (reading: adjusted mean difference, −0.8; 95% CI, −3.2 to 1.6; P = .49; spelling: adjusted mean difference, −0.9; 95% CI, −3.0 to 1.1; P = .37). For arithmetic, the ELBW or very preterm cohort performed below the average range and significantly below the NBW cohort (mean difference, −8.8; t₄₇₃ = −6.8; P<.001) even after controlling for IQ (adjusted mean difference, −2.8; 95% CI, −4.8 to −0.8; P = .006).

Teachers reported that ELBW or very preterm children were not progressing as well as the NBW children (Table 2). According to teachers, children in the ELBW or very preterm cohort were lagging in the areas of verbal thinking (mean difference, −7.2; t₄₂₇ = −5.3; P<.001), speech (mean difference, −8.4; t₄₂₇ = −5.4; P<.001), reading (mean difference, −6.4; t₄₂₇ = −4.4; P<.001), writing (mean difference, −7.6; t₄₂₇ = −5.5; P<.001), handwriting (mean difference, −8.8; t₄₂₇ = −6.3; P<.001), mathematics (mean difference, −10.0; t₄₂₇ = −6.9; P<.001), general facts (mean difference, −7.4; t₄₂₇ = −5.3; P<.001), basic motor generalizations (mean difference, −12.5; t₄₂₇ = −7.9; P<.001), and social behavior (mean difference, −6.3; t₄₂₈ = −4.7; P<.001). These statistical conclusions were not affected when analyses were repeated excluding children with neurosensory deficits and after adjusting for sociodemographic variables (Table 3). Significantly more children in the ELBW or very preterm group had repeated a grade level at school (ELBW or very preterm: n = 54, 20.2%; NBW: n = 16, 7.2%; χ²₁ = 16.9; P<.001) and required additional educational assistance (ELBW or very preterm: n = 103, 38.7%; NBW: n = 48, 21.5%; χ²₁ = 16.8; P<.001).

As shown in Table 4, parents of children in the ELBW or very preterm cohort reported more behavior problems compared with parents of children in the NBW cohort (mean difference, 2.5; Behavioral Symptoms Index [BSI]: t₄₆₇ = 2.3; P = .02). According to the parents, children in the ELBW or very preterm cohort exhibited more internalizing problems (mean difference, 2.9; t₄₆₈ = 2.9; P = .004) and fewer adaptive skills (mean difference, −2.7; t₄₆₆ = −2.7; P = .006) compared with children in the NBW cohort. More specifically, the ELBW or very preterm children displayed more attention problems (mean difference, 3.4; 95% CI, 1.5-5.3; P<.001), hyperactivity (mean difference, 2.5; 95% CI, 0.2-4.8; P = .003), somatic complaints (mean difference, 4.5; 95% CI, 2.5-6.5; P<.001), and atypical behaviors (mean difference, 2.2; 95% CI, 0.1-4.3; P = .004), whereas their social (mean difference, −2.4; 95% CI, −4.3 to −0.5; P = .01) and leadership (mean difference, −3.4; 95% CI, −5.2 to −1.5; P<.001) skills were less developed.

Similarly, teachers reported more behavior problems in the ELBW or very preterm cohort than the NBW cohort (mean difference, 2.7; BSI: t₄₅₈ = 3.4; P = .001), and again the areas of concern were internalizing behaviors (mean difference, 2.1; t₄₅₉ = 2.6; P = .01) and adaptive skills (mean difference, −4.0; t₄₄₇ = −5.0; P<.001). As rated by teachers, the ELBW or very preterm cohort displayed more attention problems (mean difference, 4.1; 95% CI, 2.3-6.0; P<.001), depression (mean difference, 2.4; 95% CI, 0.9-4.0; P = .002), and atypical behaviors (mean difference, 2.8; 95% CI, 1.3-4.4; P<.001) and less adaptability (mean difference, −2.4; 95% CI, −4.9 to −0.9; P = .003), leadership (mean difference, −4.8; 95% CI, −6.3 to −3.2; P<.001), and social skills (mean difference, −2.7; 95% CI, −4.4 to −1.0; P = .002).

Within the ELBW or very preterm cohort, those with birth weights of 500 to 749 g had significantly lower scores for some cognitive and academic tests than those with birth weights of 750 to 999 g (full-scale IQ: mean difference, −7.1; 95% CI, −12.2 to −2.0; P = .007; VCI: mean difference, −6.6; 95% CI, −11.6 to −1.7; P = .009; POI: mean difference, −6.3; 95% CI, −11.5 to −1.1; P = .02; FDI: mean difference, −4.8; 95% CI, −9.5 to −0.1; P = .045; spelling score: mean difference, −4.4; 95% CI, −8.5 to −0.4; P = .03), but there were no significant differences in behavior scores. Those with gestational ages of 23 to 25 weeks had lower cognitive and academic scores than those with gestational ages of 26 to 27 weeks, but none of these findings were statistically significant (eg, full-scale IQ: mean difference, −2.6; 95% CI, −7.5 to 2.2; P = .29; spelling score: mean difference, −2.3; 95% CI, −6.1 to 1.5; P = .23).

The proportions of children in the ELBW or very preterm and NBW cohorts who displayed clinically important neurobehavioral impairments were compared (Table 5). In contrast with their NBW peers, significantly more ELBW or very preterm children exhibited intellectual impairments (17% vs 6% mild, 5% vs 1% major; χ²₂ = 19.8; P <.001). Children in the ELBW or very preterm cohort were also more likely to display learning disabilities in reading (12% vs 7% mild, 6% vs 1% major; χ²₂ = 12.2; P = .002), spelling (16% vs 7% mild, 3% vs 1% major; χ²₂ = 10.1; P = .007), and arithmetic (26% vs 11% mild, 7% vs 1% major; χ²₂ = 28.7; P <.001) domains. The proportion of children who exhibited significant behavioral problems, as rated by their parents, did not differ between the groups (16% vs 13% at risk, 7% vs 4% clinical; χ²₂ = 3.2; P = .20). Finally, when compared with NBW peers, the ELBW or very preterm cohort was at greater risk for developing a clinically important neurobehavioral impairment (55% vs 29%; χ²₂ = 29.3; P<.001), as defined by a mild-to-major neurosensory, intellectual, educational, or behavioral impairment.

Children born preterm before the 1990s are at increased risk of cognitive, educational, and behavioral impairments.^1- 4 During the 1990s, the survival rate of premature and ELBW infants increased due to advances in perinatal intensive care,^13- 15 which, in addition to the administration of exogenous surfactant, included an increase in providing ventilatory assistance and corticosteroids.¹⁵ Despite decreasing mortality for preterm infants born in the postsurfactant era,¹⁶ follow-up studies in early childhood indicate similar rates of neurosensory impairment to those reported in earlier cohorts,^13- 15 and there is some suggestion that the incidence of these deficits may be increasing.¹⁷ Our study found that in middle childhood, 55% of survivors born ELBW or very preterm in the 1990s were exhibiting a clinically significant neurobehavioral impairment. We are unaware of any other reports of neurobehavioral outcomes at school-age of complete geographic cohorts of ELBW or very preterm children born in the 1990s.

Although it is generally accepted that preterm children are at risk of developing learning and behavioral problems, the magnitude of the impairments varies across studies.^{1- 4,18- 23} The discrepancy among studies is at least partly related to methodologic limitations, such as small or unrepresentative samples, exclusion of selected subgroups, inappropriate control groups, short follow-up, and high sample attrition.²⁴ In the current study, the ELBW or very preterm cohort was a regional sample, enabling the outcome for all children born very preterm or with ELBW to be determined. The NBW cohort was matched to the ELBW cohort, no subgroups of children were excluded, and psychologists who conducted assessments were blind to group status. Excellent retention rate for the ELBW or very preterm cohort (92%) was achieved, whereas the retention rate for the NBW cohort (85%) was acceptable. Hence, our study overcomes some of the primary methodologic limitations common in previous research.

The meta-analysis by Bhutta et al⁴ identified 15 studies with cognitive data and 16 studies with behavioral data published between 1980 and November 2001 considered worthy of review. The studies in their review involved cohorts born before 1989, before surfactant was routinely available. Also, the preterm groups reported in the review were predominantly composed of infants who were more mature at birth or who had greater birth weights than in our study. In the few studies in the review of Bhutta et al⁴ that had cohorts of infants with birth weights less than 1000 g, the mean difference in IQ between ELBW children and controls was mostly larger than the overall mean difference reported. Moreover, most studies in their review excluded children with neurosensory impairments. In our study, the weighted mean difference in IQ for the ELBW or very preterm and NBW cohorts (9.4) was similar to that reported by the meta-analysis⁴ (10.9); however, our study was composed of more survivors at higher biological risk, especially those of birth weight less than 750 g (20%) or gestational age of younger than 26 weeks (27%). We found that children in these biological risk groups had lower cognitive scores than those slightly heavier or more mature at birth. Our study also included children with neurosensory impairment, and when we excluded such children from our analyses, the mean IQ differences between ELBW or very preterm and NBW cohorts decreased to −8.8. Considering these subtle differences, our ELBW or very preterm children born in the 1990s were certainly no worse relative to NBW controls than in reports for children born before 1990; indeed, the overall neurobehavioral outcome may have improved slightly relative to earlier reports given that more very small and immature infants are surviving.

A potential concern for our study was a differential follow-up rate between the ELBW children and NBW controls. Because the children not assessed at the age of 8 years had lower mean developmental or IQ scores when assessed earlier in childhood, it is possible that the NBW children were relatively advantaged compared with the ELBW or very preterm children when assessed at 8 years of age. Adjusting for the imbalance in sociodemographic variables between groups caused by the differential follow-up rate resulted in only a small reduction in the mean IQ differences. Another limitation is our inability to directly compare the 8-year results with those reported at 2 and 5 years of age, since the selection criteria differed. Developmental change will be investigated in a separate report. Including ELBW children with very preterm infants increased the proportion of SGA infants in our ELBW or very preterm cohort. However, because SGA was not a significant confounder, this did not alter any statistical conclusions. Another limitation is the lack of IQ data for either or both parents. However, because the follow-up rates are relatively high for both cohorts who were recruited from birth, unless parents of all ELBW or very preterm children had lower average IQs, this would not have been the only explanation for such large differences in cognitive function between the groups in our study.

Although the mean full-scale IQ of our ELBW or very preterm cohort was within the average range, this cohort was at significantly greater risk of a mild-to-major intellectual impairment than their NBW peers. The full-scale IQ for the NBW cohort was one-third SD above the normative mean, which could reflect the differential follow-up rate, as discussed previously. It is also consistent with the well-recognized phenomenon that IQ scores gradually drift upward as norms become outdated.⁸ In comparison with the NBW controls, the ELBW or very preterm cohort displayed deficits across all cognitive domains; however, the severity of the difficulties was slightly greater for the perceptual organization and freedom from distractibility indices, which tap visual reasoning and attention and working memory, respectively. Neuropsychological studies^25- 28 with earlier preterm cohorts have reported deficits across a wide range of skills, including language, visual-motor integration, attention, memory, processing speed, and executive function.

Educational problems have been frequently reported in preterm children born in the presurfactant era, with many children requiring special educational assistance.^{1,18- 20,24,29- 30} In our study, the educational progress of the ELBW or very preterm cohort was delayed across all domains (reading, spelling, arithmetic) compared with NBW controls, although the reading and spelling problems were associated with their lower intellectual ability. Arithmetic was the educational domain in which the ELBW or very preterm cohort exhibited most problems, which is consistent with studies with earlier cohorts.^{19- 20,24,29- 30} Calculation difficulties may be related to specific neuropsychological impairments, such as attentional control, mental flexibility, and working memory. Interestingly, even though ELBW or very preterm children in our study had lower mean test scores on all 3 scales of the WRAT3 than the NBW controls, they may have improved compared with children born before the 1990s; the size of the difference in the WRAT3 between groups in our study was approximately halved compared with the differences reported by Saigal et al²⁹ for their cohorts born in 1977-1982, when tested at ages 8 and 12 to 16 years.

Teachers' qualitative ratings of educational progress supported the findings from the standardized educational assessment. Not surprisingly, given teachers' concerns across a wide range of areas, 20% of the children in the preterm cohort had repeated a grade level at school and 39% required extra educational assistance. Thus, children born ELBW or very preterm in the 1990s continue to display educational problems, as they did before that time. It should be pointed out, however, that most ELBW or very preterm children did not have educational problems (61%), had not repeated grades (80%), and did not require additional educational assistance (61%).

Attention-deficit/hyperactivity disorder,^3,21- 22 as well as internalizing and externalizing behavior problems,^21,31 has been linked with prematurity. The recent meta-analysis⁴ of 16 studies involving preterm children born before the 1990s reported a significant elevation of externalizing problems in 69% of studies, a significant elevation of internalizing problems in 75% of studies, and higher rates of attentional problems in 67% of studies. In our study, both parents and teachers of children in the ELBW or very preterm cohort reported mild internalizing problems and fewer adaptive skills compared with NBW children. Interestingly, significant externalizing behavior problems were not identified in our ELBW or very preterm cohort. Therefore, ELBW or very preterm children born in the 1990s remain at increased risk for developing attentional difficulties, internalizing behavior problems, and impaired adaptive functioning.

The impairments displayed by children born ELBW or very preterm are likely to be partly related to cerebral pathologic abnormalities and immaturity; however, these relationships are yet to be clearly defined. Neuroimaging studies^32- 36 indicate that prematurity may be linked with reduced white matter volume, white matter lesions, thinning of the corpus callosum, decrease in whole brain volume and cortical gray matter, reduced hippocampal volumes, and ventricular enlargement. Abnormalities are not restricted to those with intraventricular hemorrhage or periventricular leukomalacia,^32,37 suggesting that prematurity per se may also affect normal brain development. Extensive neuropathologic abnormalities may be observed in preterm children; white matter injury is the most prominent cerebral abnormality associated with prematurity.^36,38 Conceptually, periventricular white matter abnormalities common in preterm children may be associated with deficits such as visual-motor impairments, attentional problems, slowed processing speed, reduced working memory, and calculation difficulties.

Sociodemographic and environmental factors also greatly influence long-term outcome and quality of life of children born preterm.^39- 43 Sociodemographic variables, such as socioeconomic status, ethnicity (including language spoken at home), and parental education, account for some of the variance associated with outcome,^39- 42 also observed in our study, but it has been argued that environmental factors, such as family functioning, social climate, opportunities, and resources, are at least equally important.^39- 41,43 Case-control cohort studies can match children according to sociodemographic factors, but controlling for the impact of environmental factors is problematic. Further complicating this issue is causality. In many instances, the trauma and problems associated with having a very premature infant are sufficient to result in short- and long-term consequences on the families' environment.^44- 46

In summary, our study found that school-age children born ELBW or very preterm in the 1990s continue to display more cognitive, educational, and behavioral impairments relative to NBW controls. More than 50% of ELBW or very preterm children exhibited a clinically significant impairment, and thus it is important that medical and psychosocial interventions that aim to reduce the frequency, magnitude, and impact of these neurodevelopmental impairments are developed and evaluated.