School-Age Outcomes in Children Who Were Extremely Low Birth Weight
From Four International Population-Based Cohorts
Saroj Saigal, Lya den Ouden, Dieter Wolke, Lorraine Hoult, Nigel Paneth, David L.
Streiner, Agnes Whitaker and Jennifer Pinto-Martin
Pediatrics 2003;112;943-950
DOI: 10.1542/peds.112.4.943
The online version of this article, along with updated information and services, is
located on the World Wide Web at:
http://www.pediatrics.org/cgi/content/full/112/4/943
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly
publication, it has been published continuously since 1948. PEDIATRICS is owned, published,
and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk
Grove Village, Illinois, 60007. Copyright © 2003 by the American Academy of Pediatrics. All
rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
School-Age Outcomes in Children Who Were Extremely Low Birth
Weight From Four International Population-Based Cohorts
Saroj Saigal, MD*; Lya den Ouden, MD‡; Dieter Wolke, PhD§; Lorraine Hoult, BA*;
Nigel Paneth, MD, MPH储; David L. Streiner, PhD¶; Agnes Whitaker, MD#; and
Jennifer Pinto-Martin, PhD**
ABSTRACT. Objective. The aim of this study was to
determine whether learning and school problems in extremely low birth weight (ELBW) and reference children
differ between cohorts in different countries.
Methods. Participants were 4 international population-based cohorts of ELBW survivors who were 500 to
1000 g birth weight from New Jersey, central-west Ontario, Bavaria, and Holland (n ⴝ 532) and were followed
longitudinally from birth. Psychometric data were collected independently and prospectively and included at
least 1 measure of cognitive status and 1 measure of
achievement administered to the children between the
ages of 8 and 11 years. Adjustments were made for comparison of all measures based on reference norms within
each country. Information on special educational assistance and grade repetition was obtained from the parents.
Results. The overall follow-up rate was 84% (range:
74%–90%; n ⴝ 436). The proportion of children who
performed within the normal range (>85) were as follows: IQ between 44% and 62%; reading between 46%
and 81%; arithmetic between 31% and 76%; and spelling
between 39% and 65%. Children from New Jersey had
the lowest rates of cognitive and achievement deficits,
and Bavarian children did less well in achievement
scores relative to their peers and other cohorts. Despite
these differences, more than half of all cohorts required
special educational assistance and/or repeated a grade.
Conclusions. School difficulties were found to be a
serious sequelae of ELBW in all 4 countries, an observation that has social and economic implications. Pediatrics
2003;112:943–950; extremely low birth weight, school-age
outcomes, international cohorts, population-based study.
ABBREVIATIONS. ELBW, extremely low birth weight; SD, standard deviation; WISC-R, Wechsler Intelligence Scale for ChildrenRevised; ANOVA, analysis of variance.
From the *Department of Pediatrics, McMaster University, Hamilton, Ontario, Canada; ‡TNO Prevention and Health, Child Health Division, Leiden,
Netherlands; §Division of Child Health, University of Bristol, Bristol,
United Kingdom; 储Department of Epidemiology, Michigan State University,
East Lansing, Michigan; ¶Department of Psychiatry, University of Toronto,
Toronto, Ontario, Canada; #Division of Child & Adolescent Psychiatry,
Columbia University and New York State Psychiatric Institute, New York,
New York; and **Department of Pediatrics, University of Pennsylvania
School of Medicine, Philadelphia, Pennsylvania.
Received for publication Dec 17, 2002; accepted Mar 3, 2003.
Reprint requests to (S.S.) Department of Pediatrics McMaster University
1200 Main St West, Rm 4G40, Hamilton, Ontario L8S 4J9, Canada. E-mail:
saigal@mcmaster.ca
PEDIATRICS (ISSN 0031 4005). Copyright © 2003 by the American Academy of Pediatrics.
A
dvances in neonatal care have resulted in
improved survival of extremely low birth
weight (ELBW) infants.1–3 However, concerns have been expressed that these improvements
may produce an increase in neurodevelopmental
morbidity among survivors. A number of studies
now document the prevalence of a broad range of
neurologic, behavioral, and intellectual impairments
at school-age in children who were born in the early
era of neonatal intensive care.4 –9 Without exception,
these studies suggest that children who were ELBW
are more likely to have lower IQ and academic
achievement scores, experience greater difficulties at
school in mid-childhood, and require significantly
more educational assistance than children who were
born at term.
Few of these studies are population-based, and
therefore generalizability of the published data originating from single centers with high expertise may
be limited.10,11 Moreover, few studies extend beyond
the early school years, and it is difficult to assess
accurately intellectual abilities until later in childhood. Also, differences in the inception cohorts and
lack of a reference group preclude precise comparisons between studies.11 Differences in the philosophy of resuscitation and in the aggressiveness of
neonatal interventions, for example, are not often
addressed, and these practice differences may affect
survival and long-term outcome. In addition, socioeconomic factors, racial and ethnic differences, the
nature of funding of health care, and variability in
the methodologic rigor at follow-up may further contribute to differences in the reported outcomes.
Differences between studies may be even more
pronounced in international comparisons. The risk of
mortality and early neurodevelopmental morbidity
for infants ⬍26 weeks’ gestation in relation to approaches to clinical management of obstetric and
neonatal care have been investigated by Lorenz et
al12 in 2 settings in Holland and New Jersey. They
have shown that the lower mortality but higher risk
of disabling cerebral palsy in New Jersey with near
universal initiation of neonatal intensive care was
accompanied by increased utilization of health care
resources. It is plausible that these variations in management styles may have significant impact even at
school age. In this study, we describe and compare
the outcomes at mid-childhood from 4 carefully conducted international population-based cohorts of
PEDIATRICS Vol. 112 No. 4 October 2003
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
943
ELBW children in terms of cognitive and achievement measures and school difficulties. The aim of
this study was to determine whether learning and
school problems in ELBW and reference children
differ between cohorts in the different countries and
to explore the reasons for the differences, if any.
METHODS
The design of the study was longitudinal follow-up of all 4
cohorts of ELBW survivors from birth to mid-childhood (age 8 –11
years). Selection of psychometric measures and data collection
were independent and specific to each center and collected prospectively as described below. Comparisons between centers and
sharing of the collective data were made possible through a collaborative grant to establish the nature and frequency of both
short- and long-term neurodevelopmental sequelae (principal investigator, Nigel Paneth). Details of each of the 4 inception cohorts
are provided below.
Study Populations
Central New Jersey Cohort
During the period August 1984 through June 1987, 231 liveborn
infants between 500 and 1000 g birth weight were born or transferred to 3 tertiary care hospitals in 3 counties in central New
Jersey (Table 1). These infants are a subset of 1105 infants enrolled
in the Neonatal Brain Hemorrhage study, representing 85% of
regional births ⱕ2000 g birth weight; 89% of these were inborn.13
The survival rate to hospital discharge of the ⱕ1000 g cohort was
49% (n ⫽ 113); 36% of the subjects were black; 8% were Latino;
55% were non-Latino white; 1% were other; and 25% were multiple births. All infants were followed prospectively from birth to
school age. One infant died between discharge and the 9-year
assessment. Neurologic assessments were performed at 9 years of
age by trained clinicians on 79% of the eligible subjects. Psychometric assessments took place predominantly in the homes of the
subjects at the same age.14 When home visits could not be arranged, telephone or mailed questionnaires were completed by
the mother or guardian. The proportions of eligible survivors
assessed by psychometric assessments was 60% at 9 years of age;
however, the follow-up figure was higher (74%) when maternal
TABLE 1.
assessment of school performance was also included (Table 2). A
reference group was not recruited.
Central-West Ontario Cohort
During the period 1977 to 1982, 397 live births of 500 to 1000 g
birth weight were born in the region of central-west Ontario
(Table 1).15 The majority of infants (69%) were inborn at the
tertiary care center; 16.4% were transferred from community hospitals, and a small proportion (14.3%) died at the local hospitals
before they could be transferred and are included in the denominator of live births. Intensive care was not initiated and/or was
withdrawn in 16.8% of births. The survival rate to hospital discharge was 45% (n ⫽ 179); 94% of the subjects were white. All
survivors were followed prospectively from birth to 8 years of
chronologic age4; 6 ELBW children died postdischarge. At 8 years
of age, 155 (90%) of 173 ELBW survivors and 145 sociodemographically and age-matched term control children, recruited from
a random list of children provided by the local school boards,
were assessed at the hospital clinic by research psychometrists
who were blind to the group status (Table 2). Neurologic examination was performed on all subjects at the same visit by 2 pediatricians and an experienced nurse clinician.
Bavarian Cohort
During the period February 1985 to January 1986, 203 infants
who were of 500 to 1000 g birth weight were enrolled in the study
(Table 1); of these, 95 infants survived (survival rate: 47%). These
subjects were part of a larger cohort of 682 infants ⬍32 completed
weeks’ gestation, or ⬍1500 g birth weight, delivered in South
Bavaria from a catchment area of 16 children’s hospitals.16 Comparison with state perinatal survey records indicated that ⬎99% of
all births were recorded during the study period. Three ELBW
children died postdischarge; the follow-up rate was 87% (n ⫽ 80)
at 8.5 years of age (Table 2). In addition, a full-term control group
(n ⫽ 320), as well as a normative population sample representative
of the region (n ⫽ 291), was matched for age, sex, maternal
educational status, and urban versus rural place of residence and
studied prospectively at the same ages. The children were seen in
the hospital clinic with face-to-face assessments using a battery of
standardized tests, observations, and interviews from birth to 8.5
Neonatal Demographics of Live Births and Survivors, 501–1000 g Birth Weight, in Four International Cohorts
Demographic
Live births 500–1000 g (n)
Proportion ⬍750 g (n [%])
Birth weight (g; mean [SD])
Gestation (wk; mean [SD])
Survival rates* (%)
Survivors 500–1000 g (n)
Proportion ⬍750 g (n [%])
Birth weight (g; mean [SD])
Gestation (wk; mean [SD])
Ventilated (%)
Vent days (median [IQR])†
Hospital days (mean [SD])‡
Maternal education
⬍High school (%)
High school (%)
Post–high school (%)
University (%)
Study Cohorts
All Cohorts
New Jersey
(1984–1987)
Ontario
(1977–1982)
Bavaria
(1985–1986)
Holland
(1983)
231
82 (36)
788 (138)
26.5 (2.5)
49
113
20 (18)
851 (119)
27.3 (2.5)
93
16 (7, 43)
90 (34)
397
185 (47)
763 (146)
25.9 (2.5)
45
179
47 (26)
837 (123)
27.0 (2.3)
82
32 (12, 56)
100 (32)
203
60 (30)
825 (132)
27.2 (2.3)
47
95
10 (11)
884 (103)
28.4 (2.1)
95
38 (20, 57)
120 (47)
310
71 (23)¶
830 (131)§
28.1 (2.8)㛳
47
145
20 (14)#
873 (105)§
29.5 (2.4)§
53
6 (2, 11)
97 (32)§
9
35
19
37
34
31
22
13
13¶
70¶
7¶
10¶
1141
398 (35)
799 (141)
27.5 (6.8)
47
532
97 (18)
858 (116)
28.0 (2.5)
78
22 (8, 48)
101 (37)
**
**
**
**
IQR indicates interquartile range (25th, 75th percentile).
* To hospital discharge.
† Ventilator days for proportion ventilated only.
‡ Total hospitalization days, including convalescent care.
§ ANOVA P ⬍ .001.
㛳 ANOVA P ⬍ .05.
¶ ␹2 P ⬍ .0001.
# ␹2 P ⬍ .005.
** Maternal education for Holland presented in years of education: ⬍12 years ⫽ 71%; 12 ⫹ years ⫽ 13%; 15 ⫹ years ⫽ 14%; university ⫽ 1%.
944
INTERNATIONAL COMPARISON OF EXTREMELY LOW BIRTH WEIGHT INFANTS AT SCHOOL AGE
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
TABLE 2.
Cohorts
Prevalence of Neurosensory Impairments Among Survivors in Four International
Assessed (n)
Follow-up rate (%)
Age (mo; mean [SD])
Controls available
Neurosensory impairments*
All impairments (%)
Cerebral palsy
New Jersey
Ontario
Bavaria
Holland
Total
83/112
74
114 (6)
N/A
155/173
90
93 (5)
Yes
80/92
87
102 (4)
Yes
118/140
84
124 (3)
N/A
436
84
22
19
27
13
25
16
11
8
* Neurosensory impairments include cerebral palsy, mental retardation (IQ ⬎3 SD below mean),
blindness, and deafness.
years of chronological age.17 Neurologic examinations were performed by trained clinicians using a structured format.
Dutch Cohort
In 1983, 1338 infants (94% of all live births) were delivered
before 32 completed weeks of gestation and/or a birth weight of
⬍1500 g.18 Of these, 310 infants weighed 500 to 1000 g at birth
(Table 1), and 145 infants survived (survival rate 47%). All children were followed prospectively from birth to between 9 and 11
years of chronological age.5 Five children died postdischarge; the
follow-up rate was 84% (118 of 140) at school age (Table 2). At 5
years of age, neurologic assessments were performed on 96% of
the subjects in their homes by a trained pediatrician. Although
information on school difficulties was obtained by questionnaires
on 84% of the cohort, psychometric tests were administered to
only 30 randomly selected children. Term controls were not recruited, and the results were compared with the Dutch psychometric norms published in 1986.
Psychometric Tests
Each center had at least 1 measure of cognitive status and 1
measure of achievement administered to the children (Table 3).
Adjustments for Reference Standards
In view of the discrepancies related to the years of standardization of the psychometric measures, adjustments were made for
comparisons for both cognitive and achievement measures based
on the reference norms and standard deviations (SD) within each
country. Because New Jersey did not have any term controls, the
means and SD of the most current national test norms for Wechsler Intelligence Scale for Children, Third Edition (mean: 100; SD:
15)19 and the Woodcock Johnson (mean: 100; SD: 15)20 were used.
As the Wechsler Intelligence Scale for Children-Revised (WISCR)21 and Wide Range Achievement Test Revised22 standardization
sample for the Ontario cohort4 was outdated, it seemed appropriate to use the means and SD of the Ontario term matched controls
as the normative sample (WISC-R: mean: 104, SD: 12; Wide Range
Achievement Test Revised-reading: mean: 95, SD: 17; spelling:
mean: 93, SD: 17; math: mean: 94, SD: 13). The Kauffman Assessment Battery for children23 and the achievement measures24 –26
used in Bavaria were considered to be current, and therefore the
reference norms of the tests were used (mean: 100; SD: 15). As a
control group was not available for the Dutch cohort, the reference
TABLE 3.
norms (mean: 100; SD: 15) of the WISC-R Dutch version27 and the
Dutch achievement measures28 –30 were used.
Children were considered to have “normal” scores when their
scores were ⱖ1 SD of the mean of the normative sample for their
country or for the matched controls, when available. Similar comparative data were used for those who were in the “abnormal”
range, ie, ⬎2 SD below the mean of the country-specific normative
sample or that for matched controls. Untestable children with
severe impairments and cognitive deficits (IQ ⬎3 SD below mean)
were assigned scores ⬍40.
Parent Report of School Performance and Criteria for
School Placement, Grade Failure, and Special
Educational Assistance
Parents of children in all cohorts completed questionnaires
about grade repetition and special educational assistance. Criteria
for providing special education in New Jersey and Ontario varied
among schools. In general, children received part- or full-time
special assistance in their own class and sometimes outside the
class but generally within the same school. Referral for special
educational assistance was at the discretion of classroom teachers,
and children were rarely sent to special schools. Grade retention
was enforced for children with significant delay in most school
boards.
In Bavaria, all children received a school entry assessment, and
decisions for placement in a special school or delayed entry were
made jointly among the school educational and medical services,
support agencies, parents, and the head of the school. Special
schools are provided for children with physical, neurosensory
(blind, deaf), and mental handicaps and for children with learning
disabilities and language disorders. Children who attended special schools were considered to have full-time special educational
assistance. Nearly all children attended public schools, as private
schooling is rare in Germany. The school system in Holland is
somewhat similar to Bavaria. The teacher/consultant determines
school placement and grade failure (either grade repetition or
delayed school entry is accepted), and extra help within the classroom for mild problems is provided; children with physical or
significant developmental problems are sent to special schools.5
For the purposes of this study, children with school difficulties
were defined as children who received special educational assistance and/or had repeated a grade.31 However, in view of the
differences between countries in the provision of remedial ser-
Description of Psychometric Tests and Reference Standards
Country
Tests
(Standardization Year)
Tests of Achievement
(Standardization Year)
New Jersey, US
WISC-III (1991)19
Woodcock-Johnson (1988)20
Ontario, Canada
WISC-R (1974)21
Bavaria, Germany
K-ABC (1991)23
Holland
WISC-R (1984)27
Woodcock-Johnson (1988)20
WRAT-R (1982)22
Standardized Reading Test (1981)24
German Spelling Test (1983)24
Mathematical Ability Test (1993)25, 26
Reading, One Minute Test (1979)28
Spelling, Praxis 35 (1984)29
Mathematics, Tempo Test (1992)30
Reference Group for Adjustments
No controls; standardized reference
norms used (mean 100, SD 15)
Means and SD of control subjects used
Bavarian standardized tests reference
norms (mean 100, SD 15)
Dutch standardized tests reference
norms (mean 100, SD 15)
K-ABC indicates Kaufman Assessment Battery for Children; WRAT-R, Wide Range Achievement Test–Revised.
ARTICLES
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
945
vices, we also report the proportion who received both special
education and repeated a grade.
Informed Consent
Informed consent was obtained from the parents of all subjects
at the time of the assessments within each country and per the
guidelines of the local Ethics Review Board.
Neurosensory Impairments
Statistical Analysis
Analysis of variance (ANOVA) was used to compare mean
birth weight, number of days of mechanical ventilation, gestational age, and total number of neonatal hospitalization days
among survivors of the 4 cohorts. The days of ventilation are
reported as median and interquartile range (25th and 75th percentiles). However, because of the positively skewed distribution, we
also used a natural log transformation before we ran the ANOVA.
Bonferroni correction was used to adjust for multiple testing.
Significant findings were followed by Bonferroni post hoc tests to
determine which groups were significantly different. Independent
t tests were used to compare the mean psychometric scores for sex
(male vs female) and birth weight groups (⬍750 g vs 750-1000 g)
within each individual cohort. ␹2 tests for categorical data were
used to determine whether significant differences were found
between countries in the proportion of children who needed special education, had repeated a grade, and had school problems.
Regressions were run independently for each cohort to evaluate
the contribution to the variance in IQ scores made by the 3 birth
variables: birth weight, number of days of mechanical ventilation,
and sex. Holland was not included in the regression analysis or in
the independent t tests as the number of children on whom IQ
scores were available was small.
RESULTS
Demographic Data and Neonatal Illness
Table 1 shows the combined data for 1141 live
births ⱕ1000 g birth weight for the 4 cohorts. The live
births in the North American cohorts were more
immature in gestational age and had lower mean
birth weights compared with the European cohorts.
The proportion of infants ⬍750 g birth weight was
also higher for the North American cohorts, yet a
similar proportion, between 45% and 49%, survived.
Although approximately 23% of the survivors in the
North American cohorts were ⬍750 g, only 12% of
the European cohorts were this small. Three variables of neonatal morbidity collected prospectively
were selected—proportion ventilated, days of ventilation, and days of neonatal hospitalization—to measure the severity of neonatal illness among survivors.
The ANOVAs showed significant mean differences
(P ⬍ .001) between cohorts in birth weight, gestational age, number of days ventilated, and days of
neonatal hospitalization among survivors. Mean
birth weight and especially mean gestational age of
the survivors were also significantly lower in the
North American cohorts compared with the European cohorts (P ⬍ .001). The proportion of survivors
ventilated was lowest in Holland (53%) and highest
in Bavaria (95%). Even more striking is the median
number of days of ventilation, which was only 6
days for Holland, compared with 16 days for New
Jersey, 32 days for Ontario, and 38 days for Bavaria.
The mean number of days of neonatal hospitalization, including convalescent care, was highest for
Bavaria at 120 days, followed by 100 days for Ontario. The differences in mean number of days to
discharge between New Jersey and Bavaria, Holland
and Bavaria, and Bavaria and Ontario all were sig946
nificant at P ⬍ .001. From these variables, it seems
that neonatal intensive care was most aggressive in
Bavaria, followed by Ontario and New Jersey, and
considerably more selective in Holland, which offered treatment preferentially to the more mature
infants.
Table 2 shows that the total number of survivors
assessed in all cohorts was 436. The mean age at
assessments ranged between 93 months and 124
months. Control subjects were available only for Ontario and Bavaria.
Overall, the prevalence of total neurosensory impairments (defined as cerebral palsy, deafness and
blindness, and mental retardation IQ ⬎3 SD below
mean) was similar for New Jersey, Ontario, and Bavaria (22%, 27%, and 25%, respectively) and lowest
for Holland (11%). The rates of cerebral palsy in the
4 cohorts were 19% for New Jersey, 13% for Ontario,
16% for Bavaria, and 8% for Holland.
Psychometric Scores
Table 4 shows the proportions of children in the 4
cohorts with mean psychometric scores within and
below the normal range for the appropriate reference
norms. The proportion within the normal range
(ⱖ85) were as follows: for IQ, between 44% and 62%;
for reading, between 46% and 81%; for math, between 31% and 76%; and for spelling, between 39%
and 65%. A higher proportion of children in the New
Jersey cohort scored in the normal range in both IQ
and achievement measures (spelling was not obtained) relative to the other cohorts. Although the
distribution of IQ scores was similar in the Ontario
and European cohorts, Bavarian children did less
well on all achievement measures compared with
TABLE 4.
Proportion of ELBW Children Performing Within
and Below the Normal Range* in Four International Cohorts
Psychometric
Measures
Tests of intelligence (IQ)
⬍70
70–84
ⱖ85
Achievement tests
Reading
⬍70
70–84
ⱖ85
Arithmetic
⬍70
70–84
ⱖ85
Spelling
⬍70
70–84
ⱖ85
New Jersey Ontario Bavaria Holland
(%)
(%)
(%)
(%)
15
23
62
25
23
52
27
28
45
26
30
44
8
11
81
11
29
60
0†
54
46
4
26
70
15
9
76
25
15
60
47
22
31
30
13
57
—
—
—
15
24
61
40
21
39
21
13
65
* Normal range ⱖ85 according to within-country norms; borderline ⬍1 SD below mean ⫽ 70 – 84; abnormal ⱕ2 SD below mean ⫽
70.
† Thirty percent of Bavarian children had reading scores at exactly
70.
Untestable children: New Jersey ⫽ 5; Ontario ⫽ 8; Bavaria ⫽ 8;
Holland ⫽ 3.
INTERNATIONAL COMPARISON OF EXTREMELY LOW BIRTH WEIGHT INFANTS AT SCHOOL AGE
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
their peers in their own country, as well as in comparison with the other 3 cohorts.
Also shown in this table are the proportion of
children who performed 1 SD (⬍85) or 2 SD below
the mean (⬍70), based on the age-appropriate norms
within each country. Between 15% and 27% of ELBW
children in all cohorts performed in the abnormal
range for IQ. For achievement measures, in New
Jersey, 11% scored ⬎1 SD and 8% scored ⬎2 SD
below the mean for reading, whereas 9% scored ⬎1
SD and 15% scored ⬎2 SD below the mean for math.
In Ontario, 11%, 25%, and 15% scored below 2 SD in
reading, math, and spelling, respectively. However,
in Bavaria, a much higher proportion, 47% and 40%,
performed in the abnormal range (⬎2 SD below the
mean) in math and spelling, respectively, whereas no
children scored ⬍70 on the reading test (30% of the
children scored exactly at 70). The proportion of
Dutch children who performed ⬎2 SD below the
mean for reading, arithmetic, and spelling was 4%,
30%, and 21%, respectively.
In Ontario, no significant differences were found
in the scores between the smaller (⬍750 g) and the
larger (750 –1000 g) birth weight groups (not shown);
however, in New Jersey, the ⬍750 g birth weight
group had significantly lower mean scores in reading
(87 [SD: 26] vs 102 [SD: 26]; P ⬍ .05) and math (79
[SD: 35] vs 102 [SD: 26]; P ⬍ .05); and in Bavaria, the
smaller birth weight group had lower mean IQ (57
[SD: 14] vs 80 [SD: 20]; P ⬍ .005) and math scores (49
[SD: 24] vs 70 [SD: 24]; P ⬍ .05) than the larger
infants. Only 2 children ⬍750 g birth weight were
tested in Holland and are therefore not reported. Sex
differences were seen only in the Ontario cohort, in
which girls had higher IQs than boys (92 [SD: 19] vs
85 [SD: 20]; P ⬍ .05).
tries and ranged from 51% in Holland to a high of
68% in Bavaria (␹2 ⫽ 12.5, df ⫽ 3, P ⬍ .01). When the
proportion of children who received special education and had also repeated a grade were considered
(ie, the most learning compromised children), 14% to
17% of children in New Jersey, Ontario, and Dutch
cohorts but no Bavarian children were in this category (␹2 ⫽ 14.63, df ⫽ 3, P ⬍ .005). Clearly, Bavarian
school policies provided one or the other form of
remediation, which was mutually exclusive, in contrast to the other cohorts.
Perinatal Variables and Psychometric Scores
In New Jersey, birth weight, number of days on
mechanical ventilation, and sex were not associated
with mean IQ. In Ontario, the same variables accounted for 9% of the variance in IQ scores among
the ELBW. The overall ANOVA was significant (P ⬍
.01), with the number of days on mechanical ventilation being the only significant variable in the regression model (P ⬍ .05). In Bavaria, 26% of the
variance in IQ scores was accounted for by the same
variables (P ⬍ .001), with the number of days on
mechanical ventilation being the only significant
variable in the model (P ⬍ .005). Thus, in the Ontario
and the Bavarian cohorts, the number of ventilation
days was significantly associated with IQ scores
among ELBW.
DISCUSSION
This large, international, population-based study
indicates that ELBW children at school age experience a substantial burden of childhood disability,
school-related difficulties, and increased utilization
of special educational resources. School difficulties
thus are serious sequelae of ELBW in all 4 countries,
an observation that has social and economic implications. These difficulties, which cannot be identified
at younger ages, become steadily more apparent as
the children face increasingly complex tasks in
school. Remarkable similarities across all 4 cohorts
are found both in survival rates and in the proportion of ELBW children requiring special educational
assistance. More than half of ELBW children, wherever they are born in Western countries, experience
serious educational problems. Thus, the impact on
society, on families, and on the long-term future of
the ELBW children is similar in Western countries
providing neonatal intensive care to infants ⱕ1000 g.
In a separate report by this group, we found that
Parent Reports on School Performance
Table 5 provides data on school performance as
reported by parents with respect to special educational assistance and grade repetition. A fairly high
proportion of ELBW children from all cohorts (between 39% and 62%) were receiving either part- or
full-time special educational assistance; there were
no significant differences between countries (␹2 ⫽
4.55, df ⫽ 3, P ⫽ .2).
Grade repetition ranged between 19% and 34%,
and, again, the proportions were not significantly
different between the cohorts (␹2 ⫽ 7.65, df ⫽ 3, P ⫽
.05). However, the proportion of children who had
school difficulties (special education and/or grade
repetition) was significantly different between counTABLE 5.
Parental Report of Special Education, Grade Repetition, and School Difficulties in Four International Cohorts
New Jersey
Special education
Grade repetition
School difficulties
Special education and/or
grade repetition
Special education and grade
repetition
Ontario
Bavaria
␹2
Holland
n
%
n
%
n
%
n
%
51/82
16/84
(62)
(19)
79/154
31/143
(51)
(22)
37/80
23/80
(39)
(29)
60/118
40/118
(51)
(34)
4.55, df ⫽ 3; P ⫽ .20
7.65, df ⫽ 3; P ⫽ .05
52/82
(63)
88/154
(57)
60/80
(68)
60/118
(51)
12.53, df ⫽ 3; P ⫽ ⬍.01
13/82
(16)
22/154
(14)
0/80
(0)
20/118
(17)
14.63, df ⫽ 3; P ⫽ ⬍.005
ARTICLES
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
947
similar behavioral difficulties are also of major concern at school age in all 4 cohorts in children of this
birth weight.8 Whether and to what extent these
findings are applicable to the current survivors of
neonatal intensive care remains to be investigated. It
is likely that these children will also experience significant school difficulties because of the high disability rates as a result of the increased survival of
the sicker and more immature infants.2,32
There are also some significant differences between cohorts, and these include the less aggressive
neonatal management and lower rates of neurosensory impairments in the Dutch cohort compared
with the other cohorts.12 The New Jersey sample had
the lowest rates of IQ deficits and reading problems
compared with the Ontario and Bavarian cohorts.
Nevertheless, despite better performance on standardized tests, the New Jersey cohort had the highest
number of children in special education. There are
several factors that are likely to explain this apparent
paradox. Although 87% and 90% of the Bavarian and
Ontario samples, respectively, were followed up
with standardized tests, the ascertainment rate for
administration of psychometric tests was substantially lower (60%) in New Jersey (although 74% of
maternal respondents provided information on educational difficulties). Thus, a substantial number of
children who were not assessed with the achievement tests were included in reporting educational
difficulties and may represent a group who performed less well. It has been shown previously that
nonparticipants have more intellectual and behavioral problems than participants.33 Another issue is
that of control subjects. The Ontario and Bavarian
cohorts had same-aged control subjects, and thus
concurrent norms were used, whereas the New Jersey cohort relied on test norms standardized several
years earlier. Wolke et al34 showed that using outdated normative standards leads to significant overestimation of the mean intellectual performance of
premature children compared with concurrent representative samples. Secular trends amounting to an
average yearly increase of 0.2 to 1.2 IQ points have
been reported for both adult and child populations.35
Finally, that the New Jersey children were assessed
in their own homes rather than in a hospital setting,
as in Bavaria and Ontario, may have led to less
consistency in the assessments.
A difference between the Ontario and Bavarian
samples is that despite similar distribution of IQ
scores, the Bavarian children did more poorly in
arithmetic and spelling compared with the Canadian
children, ie, they performed more poorly than might
be expected on the basis of their intellectual level.
One explanation may be that one quarter of the
Bavarian ELBW children had delayed school entry
by at least 1 year beyond the usual average age of
school entry of 6.6 years. Thus, at the time of testing
(8.5 years), they did more poorly than control subjects, who generally already had an extra year of
schooling. It is not entirely clear why the proportion
of children with achievement scores below 2 SD of
norms is so much higher within the Bavarian cohort
948
than in the North American cohorts. We can only
speculate on some of the reasons, such as differences
in teaching methods in school, different educational
standards, cultural differences in the tests used, and
differences in the policies of providing remedial education in the school systems. In the 2 European
samples, there is a clear destreaming system, ie, it is
geared toward average to above-average performance in regular schools; children with disabilities
and below-average performance receive their education in special schools. Despite these differences, it
seems that approximately half of the ELBW survivors within each cohort encountered school difficulties in mid-childhood. Of even greater concern is the
finding in the Ontario cohort that these difficulties
persist to adolescence.31
The strengths of this study are the populationbased nature of the data that were collected prospectively by experienced clinicians who followed methodologically rigorous standards. These include the
considerable effort expended in reporting accurately
the number of live births and deaths in all cohorts,
the provision of details on subject losses, prevalence
of neurosensory impairments, inclusion and exclusion criteria for psychometric tests, the availability of
comparison or reference standards, and relevant information on standardization data for clinical and
psychometric measures. The analysis of the grouped
data was performed jointly by the principal investigators who had assembled the cohorts, an approach
that has clear advantages over meta-analyses in
which differences in the inception cohorts, modes
of assessment, and analyses are not always
known.10,36,37 At the same time, we recognize the
problems that ensue from the independent assembly
of these cohorts retrospectively, without previous
consensus as to the study design or measures. We
accept that some factors that may have an impact on
the neonatal characteristics of the populations were
difficult to control, such as varying criteria for reporting live births and deaths; differences in the
philosophy of instituting resuscitation and/or withdrawal of intensive care; and aggressiveness of neonatal intervention and sociobiological factors such as
social class, parental education, family configuration,
source of health care funding, and available resources. For example, despite careful ascertainment,
it is possible that some very immature live births
were not registered in Holland as they were considered nonviable at the time.1 This fact and a selective
policy of offering neonatal intensive care to the larger
infants may have contributed to the more mature
mean gestational age of the Dutch cohort relative to
the other 3 cohorts. The mortality rate of the cohorts,
however, did not differ, but the proportion ventilated and the mean days of ventilation in the Dutch
cohort were remarkably lower, suggesting a less aggressive approach to the care of ELBW infants relative to other cohorts. These differences in management styles could have had an impact on subsequent
morbidity and may explain the low prevalence of
impairments in the Dutch children compared with
the New Jersey cohort.12
INTERNATIONAL COMPARISON OF EXTREMELY LOW BIRTH WEIGHT INFANTS AT SCHOOL AGE
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
This is the first international collaboration to study
and carefully compare the long-term outcomes of
ELBW children. We have shown that comparison of
national or international data on school outcomes is
indeed a complex task. Although it is increasingly
recognized that hospital-based studies are not generalizable and population-based studies are preferable,11,31,38 it is difficult for many centers to maintain
their cohorts to school age without a high attrition
rate.33 For conducting future comparative studies to
determine whether management practices influence
the long-term outcomes, it is necessary to develop a
consensus nationally and internationally for the minimum data set necessary for reporting school problems. There needs to be consistency in the protocol
for assessing neurodevelopment and in the inclusion
and exclusion criteria used when reporting on children with impairments. It is also important for investigators to use current and within-country reference norms for tests of ability and achievement. The
policies for special educational assistance, grade repetition, and programs for children with disabilities
should be outlined clearly in the reports for meaningful interpretation and generalizability. Thus, we
must define more clearly our own practices when
reporting outcomes. The importance of methodologically rigorous population-based studies cannot be
overemphasized. As we continue to save infants of
borderline viability, it is imperative that the current
survivors be followed to school age to provide a
more complete picture of their overall outcome. Such
information is necessary for parents, providers of
health care and education, and the health care funding agencies.
ACKNOWLEDGMENTS
This study was supported by funding from NIH Grant HS09493-01. Germany was supported by DFG Grant SCHN 315/15-1.
We thank Madeleine Lenski (Michigan), Barbara Stoskopf and
Peter Rosenbaum (Ontario), Renate Meyer (Bavaria), and Elysee
Hille and Elske Sigmond-de Bruin (Holland) for assistance in
providing data from individual centers and for helpful comments.
In addition, we thank the children and their parents for cooperation and Diane Turcotte for typing the manuscript.
REFERENCES
1. Lorenz JM, Wooliever DE, Jetton JR, Paneth N. A quantitative review of
mortality and developmental disability in extremely premature newborns. Arch Pediatr Adolesc Med. 1998;152:425– 435
2. Wood NS, Marlow N, Costeloe K, Gibson AT, Wilkinson AR, for
the EPICure Study Group. Neurologic and developmental disability after extremely preterm birth. N Engl J Med. 2000;343:378 –
384
3. Vohr BR, Wright LL, Dusick AM, et al. Neurodevelopmental
and functional outcomes of extremely low birth weight infants in
the National Institute of Child Health and Human Development
Neonatal Research Network, 1993–1994. Pediatrics. 2000;105:
1216 –1226
4. Saigal S, Szatmari P, Rosenbaum P, Campbell D, King S. Cognitive
abilities and school performance of extremely low birth weight children
and matched term control children at age 8 years: a regional study.
J Pediatr. 1991;118:751–760
5. Hille ETM, Den Ouden AL, Bauer L, van den Oudenrijn C, Brand R,
Verloove-Vanhorick SP. School performance at nine years of age in very
premature and very low birth weight infants: perinatal risk factors and
predictors at five years of age. J Pediatr. 1994;125:426 – 434
6. Hack M, Taylor HG, Klein N, Eiben R, Schatschneider C, MercuriMinich N. School-age outcomes in children with birth weights under
750 g. N Engl J Med. 1994;331:753–759
7. Szatmari P, Saigal S, Rosenbaum P, Campbell D. Psychopathology and
adaptive functioning among extremely low birthweight children at
eight years of age. Dev Psychopathol. 1993;5:345–357
8. Hille ETM, den Ouden AL, Saigal S, et al. Behavioural problems in
children who weigh 1000g or less at birth in four countries. Lancet.
2001;357:1641–1643
9. Saigal S, Pinelli J, Hoult L, Kim M, Boyle M. Psychopathology and social
competencies of adolescents who were extremely low birth weight.
Pediatrics. 2003;111:969 –975
10. Escobar GJ, Littenberg B, Petitti DB. Outcome among surviving very
low birthweight infants: a meta-analysis. Arch Dis Child. 1991;66:
204 –211
11. Wolke D. Psychological development of prematurely born children.
Arch Dis Child. 1998;78:567–570
12. Lorenz JM, Paneth N, Jetton JR, den Ouden L, Tyson JE. Comparison of
management strategies for extreme prematurity in New Jersey and the
Netherlands: outcomes and resource expenditure. Pediatrics. 2001;108:
1269 –1274
13. Pinto-Martin J, Paneth N, Witomski T, et al. The central New Jersey
neonatal brain hemorrhage study: design of the study and reliability of ultrasound diagnosis. Paediatr Perinat Epidemiol. 1992;6:
273–284
14. Pinto-Martin J, Whitaker AH, Feldman JF, Van Rossem R, Paneth N.
Relation of cranial ultrasound abnormalities in low-birthweight infants
to motor or cognitive performance at ages 2, 6 and 9 years. Dev Med
Child Neurol. 1999;41:826 – 833
15. Saigal S, Rosenbaum P, Hattersley B, Milner R. Decreased disability rate among 3-year-old survivors weighing 501 to 1000 gram
at birth and born to residents of a geographically defined region
from 1981 to 1984 compared with 1977 to 1980. J Pediatr. 1989;114:
839 – 846
16. Wolke D, Meyer R. Cognitive status, language attainment and prereading skills of 6 year-old very preterm children and their peers:
The Bavarian Longitudinal Study. Dev Med Child Neurol. 1999;41:
94 –109
17. Wolke D, Schulz J, Meyer R. Entwicklungslangzeitfolgen bei ehemaligen, sehr unreifen Fruehgeborenen. Monatsschr Kinderheil. 2001;149;
S53–S61
18. Verloove-Vanhorick SP, Verwey RA, Brand R, Bennebroek Gravenhorst J, Keirse MJCN, Ruys JH. Neonatal mortality risk in relation to
gestational age and birthweight. Results of a national survey of preterm
and very low birthweight infants in the Netherlands. Lancet. 1986;1:
55–57
19. Wechsler D. Manual for the Wechsler Intelligence Scale for Children. 3rd ed.
(WISC-III). New York, NY: Psychological Corporation; 1989
20. Woodcock R, Mather N. Woodcock-Johnson Tests of Achievement Revised:
Standard and Supplemental Batteries. Allen, TX: DLM Teaching Resources;
1998
21. Wechsler D. The Wechsler Intelligence Scale for Children-Revised. New
York, NY: Psychological Corporation; 1974
22. Jastak S, Wilkinson G. Wide Range Achievement Test-Revised (WRAT-R).
Wilmington, DE: Jastak Associates; 1984
23. Melchers P, Preuss U. K-ABC: Kaufman Assessment Battery for Children.
Deutschsprachige Fassung. Frankfurt, Germany: Swets & Zeitlinger; 1991
24. Müller R. Diagnostischer Rechtschreibtest DRT 2. Weinheim, Germany:
Beltz; 1983
25. Stigler JW, Lee S-Y, Stevenson HW. Mathematical Knowledge of Japanese,
Chinese, and American Elementary School Children. Reston, VA: The National Council of Teachers of Mathematics; 1990
26. Wolke D, Leon-Villagra J. Mathematiktest. Munich, Germany: University
Children’s Hospital; 1993 (available from D. Wolke at:
dieter.wolke@bris.ac.uk)
27. deBruyne EEJ, vanderSteene G, vanHaassen PP (eds). WISC-R Wechsler
Intelligence Scale for Children-Revised. Dutch edition. Lisse, the
Netherlands: Swets & Zeitlinger; 1986
28. Brus BTH, Voeten MJM. [One Minute Test. School Achievement Test for
Technical Reading for the Second to Sixth Grade of Primary Schools. 2nd rev
ed.] Nijmegen, the Netherlands: Berkhout; 1979
29. Van den Heuvel F. [Praxis 35: Spelling Tests for Primary Schools]. den
Bosch: Malmberg; 1984
30. De Vos T. [Tempo Test Mathematics. Assessment of Calculation and Automa-
ARTICLES
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
949
31.
32.
33.
34.
tisation for Primary and Secondary Schools]. Nijmegen, the Netherlands:
Berkhout; 1992
Saigal S, Hoult LA, Streiner DL, Stoskopf BL, Rosenbaum PL. School
difficulties at adolescence in a regional cohort of children who were
extremely low birth weight. Pediatrics. 2000;105:325–331
Emsley HCA, Wardle SP, Sims DG, Chiswick ML, D’Souza SW. Increased survival and deteriorating developmental outcome in 23 to 25
week old gestation infants, 1990 – 4 compared with 1984 –9. Arch Dis
Child Fetal Neonatal Ed. 1998;78:F99 –F104
Wolke D, Söhne B, Ohrt B, Riegel K. Follow-up of preterm children:
important to document dropouts. Lancet. 1995;345:447
Wolke D, Ratschinski G, Ohrt B, Riegel K. The cognitive outcome of
very preterm infants may be poorer than often reported: an empirical
35.
36.
37.
38.
investigation of how methodological issues make a big difference. Eur
J Pediatr. 1994;153:906 –915
Flynn J. Massive IQ gains in 14 nations: what IQ tests really measure.
Psychol Bull. 1987;101:171–191
Aylward FP, Pfeiffer SI, Wright A, Verhulst SJ. Outcome studies of low
birth weight infants published in the last decade: a metaanalysis. J Pediatr. 1989;115:515–520
Bhutta AT, Cleves MA, Casey PH, Cradock MM, Anand KJS. Cognitive
and behavioral outcomes of school-aged children who were born preterm. A meta-analysis. JAMA. 2002;288:728 –737
Kiely JL, Paneth N. Follow-up studies of low-birthweight infants: suggestions for design, analysis and reporting. Dev Med Child Neurol. 1981;
23:96 –100
A HERETICAL PROPOSAL
“It is time to refocus attention away from birth weight adjustment. As a measure
of perinatal risk, birth weight is an inadequate proxy. Comparisons of hospital
mortality rates would be more valid if adjustment were made for actual indicators
of high risk, such as measures of preterm delivery, fetal distress, preexisting
medical conditions, or currently unidentified biomarkers. Comparisons across
communities would be more valid (and frequently more precise) without adjustment for birth weight.”
Hertz-Picciotto I. Is it time to abandon adjustment for birth weight in studies of infant mortality?
Paediatr Perinat Epidemiol. 2003;17:114 –116
Submitted by Student
950
INTERNATIONAL COMPARISON OF EXTREMELY LOW BIRTH WEIGHT INFANTS AT SCHOOL AGE
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009
School-Age Outcomes in Children Who Were Extremely Low Birth Weight
From Four International Population-Based Cohorts
Saroj Saigal, Lya den Ouden, Dieter Wolke, Lorraine Hoult, Nigel Paneth, David L.
Streiner, Agnes Whitaker and Jennifer Pinto-Martin
Pediatrics 2003;112;943-950
DOI: 10.1542/peds.112.4.943
Updated Information
& Services
including high-resolution figures, can be found at:
http://www.pediatrics.org/cgi/content/full/112/4/943
References
This article cites 26 articles, 11 of which you can access for free
at:
http://www.pediatrics.org/cgi/content/full/112/4/943#BIBL
Citations
This article has been cited by 21 HighWire-hosted articles:
http://www.pediatrics.org/cgi/content/full/112/4/943#otherarticle
s
Subspecialty Collections
This article, along with others on similar topics, appears in the
following collection(s):
Premature & Newborn
http://www.pediatrics.org/cgi/collection/premature_and_newbor
n
Permissions & Licensing
Information about reproducing this article in parts (figures,
tables) or in its entirety can be found online at:
http://www.pediatrics.org/misc/Permissions.shtml
Reprints
Information about ordering reprints can be found online:
http://www.pediatrics.org/misc/reprints.shtml
Downloaded from www.pediatrics.org. Provided by University of Warwick on November 26, 2009