Copyright 2002 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2002
To provide current information on the academic achievement of small-for-gestational-age (SGA) children at age 10 years, to analyze predictors, and to evaluate the smallness of a newborn as a risk factor for school failure.
Prospective, population-based birth cohort study with 10-year follow-up.
The SGA group consisted of 106 children born in 1985 whose birth weight had been below the 2.5th percentile on the population-based fetal growth chart. The control group consisted of 105 appropriate-for-gestational-age (AGA) children born as closely as possible in time to SGA infants and matched for gestational age and mode of delivery. Both groups were evaluated at age 10 years.
Main Outcome Measure
Twenty-five percent of SGA children were school failures (vs 14% of AGA children; P = .05). There were significantly more SGA than AGA children with unfavorable backgrounds and deficient performances associated with poor academic achievement. Independent predictors of poor academic achievement in SGA and AGA children were inattention-passiveness as rated by their teachers (masked to grouping; corresponding to a 1-point increase in inattention-passiveness score: odds ratio, 5.1; 95% confidence interval, 2.1-12.1), a low verbal IQ score (psychologists masked; corresponding to a 10-point decrease in IQ score: odds ratio, 3.2; 95% confidence interval, 1.6-6.2), and restlessness at the follow-up visit (one examiner masked and the other not; odds ratio, 14.3; 95% confidence interval, 2.1-97.3).
Being SGA at birth has a clinically significant impact on the academic achievement of a 10-year-old child.
SUCCESS IN school is an excellent marker of general adolescent well-being. School failure—poor academic performance often together with conduct problems—is a powerful sign of other high-risk behaviors, with possible subsequent school dropout and social marginalization. There are several well-recognized risk factors associated with school difficulties, for instance, behavioral disorders, low self-esteem about one's academic ability, and a disadvantaged home background.1
Newborns can be small for many reasons. As long as an unequivocal criterion that is easily applicable in the clinical setting for malnourishment of a newborn is not available, the consensus is the lightness of a newborn (birth weight for gestational age >2 SDs below the mean birth percentile). On the whole, it is crucial that the intrauterine growth of a newborn is assessed because small-for-gestational-age (SGA) infants as a group are at risk for developmental disorders,2 for remaining smaller than their peers,3 and for chronic diseases in older age.4 Currently, approximately two thirds of SGA children are born without any suspicion of their potential intrauterine growth failure,5 and many SGA infants are discharged from obstetric units without any diagnosis of potential growth failure.6 Therefore, it is not reasonable to suppose that those in primary care will pay enough attention to the intrauterine growth retardation (IUGR) of an infant or child. Previous studies7 seem to support the fact that SGA children are prone to cognitive deficits. Evaluations of the success of SGA children in school beyond the preschool years are few, with different definitions of SGA and varying information on the predictors of academic achievement. Studies have focused on different populations, for example, preterm, very low-birth-weight SGA children8- 10; preterm and full-term SGA children as one group11 or as separate groups with different definitions12 or the same definition13 of SGA for preterm and full-term children; clinically intrauterine malnourished, full-term children14; and full-term SGA adolescents only.15
Our follow-up study was undertaken in 1985. The perinatal risk factors and neonatal complications have been reported previously.16,17 The aim of the present phase of the study is to investigate the academic achievement of SGA children and control appropriate-for-gestational-age (AGA) children at age 10 years and to evaluate the associations of perinatal history, sociodemographics, neurological status, head growth, several neurocognitive abilities, and behavior with academic achievement. Does the smallness of a newborn need to be recognized as a potential risk factor for poor school performance?
The SGA group consisted of all 118 SGA infants from the original birth cohort born in 1985 in the catchment area of the Turku University Central Hospital in southwestern Finland (Figure 1). The only recruitment criterion was birth weight for gestational age below the 2.5th percentile (corresponding to a birth weight >2 SD below the mean birth percentile) on the regional population-based fetal growth chart.18 The mean birth weight of SGA infants was 2452 g (range, 540-2930 g), and the mean gestational age was 38.8 weeks (range, 27.0-42.0 weeks). Eight SGA infants (7%) were born prematurely. Forty-six SGA infants (39%) were born at community hospitals, and the remainder were born at the Turku University Central Hospital, which serves as a tertiary-level perinatal center in the region concerned. Thirteen SGA infants were born from twin pregnancies. None of the infants had a history of intrauterine infection. Except one Indian father and one Portuguese mother, all the parents were ethnically homogeneous, that is, Finnish, with 4% having Swedish as their first language.
Follow-up profile of small-for-gestational-age (SGA) and appropriate-for-gestational-age (AGA) children.
After the exclusions (Figure 1), the total number of eligible survivors in the SGA group was 112; 106 (95%) of them were available for study at age 10 years. Their gestational ages ranged from 27 to 42 weeks (mean, 38.8 weeks). The attrition analysis showed that neither the age (P = .28) nor the basic education (P = .85) of the mothers or the degree of domicile urbanization (P = .22) in participants differed significantly from those in the 6 nonparticipants, whereas the birth weight (P = .03) and the SGA index (P = .002) were significantly higher in nonparticipants, indicating less severe IUGR. The SGA index was calculated by dividing actual birth weight by a 50th percentile reference birth weight, and it is presented as a percentage.
The control AGA group was randomly selected and consisted of 118 singletons matched for gestational age (mean, 38.8 weeks; range, 27-42 weeks) and mode of delivery and born at the Turku University Central Hospital as closely as possible in time to the study group SGA infants (Figure 1). The birth weight was appropriate for gestational age, that is, between the 10th and 90th percentiles on the same fetal growth chart (mean birth weight, 3378 g; range, 1210-4100 g). Eight AGA children were born prematurely. None of the infants had evidence of chromosomal or congenital abnormality at birth.
The total of 105 AGA children were evaluated at age 10 years. Their mean gestational age was 38.9 weeks (range, 27.0-42.0 weeks). Participants and nonparticipants did not differ significantly regarding maternal age (P = .58), maternal basic education (P = .09), or degree of domicile urbanization (P = .25).
The perinatal history of SGA and AGA children was evaluated in 1995 by means of hospital records (Table 1). Although the birth weight of SGA children was less than the 2.5th percentile, the degree of the severity of IUGR was further evaluated as an SGA index (mean, 70.7%; range, 46.1%-78.4%).
All the children were examined at the mean (SD) age of 10.1 (0.2) years (range, 9.4-10.8 years) during outpatient clinic visits at the Turku University Central Hospital. Table 1 gives the assessment battery and the amount of data available for analysis. In addition to routine pediatric examination, neurological examination was performed by one of us (O.H.). Head circumference was measured using a flexible metallic measuring tape as the maximum circumference between the supraorbital ridge and the occiput. The classification of subnormal head circumference (>2 SDs below the reference percentile) was based on the reference data of the Finnish growth charts.26 To obtain sociodemographic information, the parent(s) was interviewed by one of us (O.H.).
For an overall measure of intellectual ability, the Finnish or Swedish version of the revised Wechsler Intelligence Scale for Children–Revised (WISC-R) was used.27 The tests were conducted by 2 psychologists who were masked to grouping. The WISC-R subtests were also grouped into 3 domains according to Kaufman.28 A comprehensive battery of tests was administered to test visuomotor perception, fine motor abilities, memory, and attention (Table 1).
The classroom teachers were asked to rate the child's behavior at school by completing the Conners' Abbreviated Teacher Rating Scale34 (4-point scale: "not at all true" to "very true" of the child). The teachers were masked to grouping. The child's behavior at home was assessed by the mother, the father, or both by means of the Conners' Abbreviated Parent Rating Scale.34 When both maternal and paternal ratings were available, the mean score was taken as representative. The child's behavior was also evaluated during the follow-up visit by the pediatrician (O.H.) and the psychologists (behavior scale: timid, restless, or appropriate).
In Finland, there were no standardized achievement tests available in 1995. Therefore, to rate the academic achievement of a child, the type of education (special vs mainstream) was first taken into account. All the children in special education were intellectually disabled, that is, they had an intellectual impairment instead of physical, sensory, or behavioral impairments as a reason for not attending mainstream school. The children in mainstream education were divided according to whether they were in an appropriate grade level or not. The academic achievement of the children in an appropriate grade level was assessed on the basis of their school reports. The classroom teacher was asked to post a copy of a child's school report on which mathematics and reading and writing were assessed. The school report assessments were rescored as good (knowledge or skill attained) or poor (knowledge or skill not attained). Among school failures were children (1) who were in special education; (2) who were in an inappropriate grade level, that is, they had either repeated a grade or started basic education later than their peers; or (3) whose achievements in mathematics, reading and writing, or both were rescored as poor.
In Finland, basic education starts in August for children turning 7 years of age during the year concerned. The comprehensive school education has been a 9-year course since the early 1970s. The school system aims to keep every child in mainstream education, if necessary with remedial teaching or school assistants. If children still cannot keep up with the others in class, they are referred to a class or school for special education. The school performance rating system is based on assessments by a teacher expressed either numerically (scale: fail = 4 to excellent pass = 10) or in words (indicating whether a pupil has attained a certain skill or knowledge or not).
To test the differences between SGA and AGA children, the 2-sample t test (parental age, head growth, auditory memory, visual memory, and fine motor abilities), Fisher exact test (perinatal history, neurological status, intelligence, and academic achievement), Mann-Whitney test (attention, memory and verbal learning, visuomotor integration, and questionnaires on behavior), and Pearson χ2 test (socioeconomic status, parental basic education, and behavior at the follow-up visit) were applied. Evaluations of the univariate association between academic achievement and predicting variables were performed using cross tabulation and the χ2 test. This univariate analysis was followed by multivariate analysis with stepwise logistic regression. The variables were classified into subgroups, and the analyses were initially carried out for variables within one subgroup at a time and afterward for all statistically significant variables irrespective of grouping. P<.05 was used as the cutoff point of significance. Odds ratios and 95% confidence intervals were calculated from the logistic regression models.
The study was approved by the Joint Commission on Ethics of the University of Turku and the Turku University Central Hospital. Informed written consent was obtained from the parent(s) of the children.
The frequency of the perinatal risk factors was 18% in the SGA group and 12% in the AGA group (P = .27). Two SGA children had 5-minute Apgar scores of 0 to 3, 2 SGA children had umbilical artery pH less than 7.0, 10 SGA and 13 AGA full-term children had bilirubin levels greater than 12 mg/dL (>205 µmol/L) and 1 AGA preterm child had a bilirubin level greater than 15 mg/dL (>255 µmol/L), and 8 SGA children had blood glucose levels less than 30 mg/dL (<1.7 mmol/L). Blood glucose and bilirubin levels were determined only if clinically indicated.
The socioeconomic status of the family and parental basic education were significantly better in AGA than in SGA children (Table 2).
Eleven SGA children and 3 AGA children had major neurological or sensory impairments (P = .049): 3 had cerebral palsy (2 SGA and 1 AGA), 1 (SGA) was blind without any known cause, 2 had severe hearing loss needing hearing aid (1 SGA and 1 AGA), and 8 were intellectually disabled (7 SGA and 1 AGA). Three neurological assessment domains by Touwen25 showed significant differences between SGA and AGA children, namely, coordination of the extremities (right fingertip nose test, P = .003; left fingertip nose test, P = .02; and right diadochokinesis, P = .03), balance of trunk (walking along a straight line, P = .003), and gross motor functions (standing on left leg, P = .007; standing on right leg, P = .04), with respect to which SGA children performed poorer.
The mean head circumference (presented as an SD from reference circumference) of SGA children (–0.98) was significantly smaller than that of AGA children (+0.09; P<.001). The percentage of children with subnormal head growth at age 10 years was significantly greater in the SGA group than in the AGA group (14% vs 2%; P = .001). Among the mothers of these 15 children with subnormal head circumference, 3 (2 of SGA children and 1 of AGA children) had similarly subnormal head circumferences. The head circumferences of the fathers were within the reference range (measurement was missing for 5 fathers). The SGA index, the degree of the severity of IUGR, indicated a significant correlation with 10-year head circumference (r = 0.27; P = .008).
The mean WISC-R verbal, performance, and full-scale IQ scores of SGA and AGA children are presented in Table 3. According to the results on the WISC-R, 5% of SGA children (vs 1% of AGA children) were mentally retarded (full-scale IQ score <70), 15% (vs 12%) had borderline IQ scores (full-scale IQ score of 70-84), and 80% (vs 87%) had normal intelligence (full-scale IQ score ≥85); the differences were not statistically significant. Although SGA children showed significantly lower overall intellectual performance, the profiles of the WISC-R subtests were similar for SGA and AGA children (Figure 2). The SGA children had significantly poorer short-term auditory memory and perceptual organization ability and a higher incidence of inattention than AGA children (Table 3).
Wechsler Intelligence Scale for Children–Revised (WISC-R) scaled score means (SEMs) for small-for-gestational-age (SGA) and appropriate-for-gestational-age (AGA) children at age 10.1 years. Asterisk indicates P<.05.
Most children behaved themselves in an appropriate manner throughout the follow-up visit. However, 10% of the SGA children were restless (vs 5% of AGA children), and 15% were overtly timid (vs 5% of AGA children; P = .02). According to the parents, SGA children had significantly more learning problems than AGA children; the median scores (and interquartile ranges [IQR = Q3-Q1, where Q1 and Q3 correspond to the 25th and 75th percentiles, respectively]) were 0.50 (0.89) and 0.38 (0.53), respectively (P = .007). The teachers' reports revealed more inattention-passiveness in SGA children than in AGA children (0.63 [0.91] vs 0.44 [0.88]), which was not significant (P = .07).
The academic achievements of 103 of 106 SGA and 100 of 105 AGA children were assessed. Twenty-six SGA children (25%) and 14 AGA children (14%) were school failures (P = .05). Seven SGA children (7%) were in special education owing to intellectual disability compared with 1 AGA child (1%; P = .07). Three SGA children had repeated a grade, and another 3 had started basic education 1 year later than their peers, whereas all the AGA children except the child in special education were in mainstream education at an appropriate grade level (P = .01). There were 13 SGA and 13 AGA children in mainstream education whose assessments for mathematics, reading and writing, or both were rescored as poor. According to the behavioral questionnaires, the parents of SGA children reported significantly more learning problems than those of AGA children.
The adverse perinatal history of SGA children was in association with their school achievement (P = .049), whereas in AGA children it was not. Factors such as prematurity, birth weight less than 2000 g, maternal smoking, maternal diseases, and IUGR suspicion during pregnancy were not associated with poor academic achievement in SGA or AGA children. The numbers of SGA children with various unfavorable background factors and deficient performances were significantly greater than the numbers of AGA children (Figure 3).
Statistically significantly different characteristics of small-for-gestational-age (SGA) children compared with their appropriate-for-gestational-age (AGA) peers.
According to the univariate analysis, the school failures in both groups shared several factors significantly associated with poor academic achievement (Table 4). The subsequent multivariate logistic regression analysis that was first conducted for significant variables within every predictor subgroup, each of which having the smallness for gestational age of a newborn as 1 variable, was followed by the analysis between the predictor subgroups. The independent predictors of poor academic achievement were inattention-passiveness as rated by teachers, a low WISC-R verbal IQ score, and restlessness at the follow-up visit (Table 4).
Our study showed SGA children performing worse at school than matched AGA controls. The result was derived from a population-based, prospective follow-up study with nonclinical participants; an internationally agreed-on definition of SGA; a representative enrollment rate; a high retention rate; an excellent return rate; use of standard instruments; controls matched for gestational age; and masking of the developmental assessors except the pediatrician (O.H.). Because of the lack of standardized achievement tests, academic achievement was evaluated from 3 points of view: type of education, appropriateness of grade level, and assessments in school reports. The AGA children approximated to a normal Finnish population with respect to academic achievement, head growth, and neurological status. Some predictors of poor academic achievement concerned only SGA children, such as the adverse perinatal history, whereas the independent predictors—inattention-passiveness, a low verbal IQ score, and restlessness—were shared by SGA and AGA children.
The most powerful independent predictor of academic achievement was the child's inattentive-passive behavior at school. Such internalizing or withdrawn behavioral problems have been reported in association with IUGR.35 Furthermore, the stability with regard to internalizing problems, attention problems, and social problems is high in SGA, although preterm, children.36 The teachers consistently rated SGA children as more problematic in all 3 domains by Conners. The parents' ratings varied widely and were inconsistent with those of the teachers, with the exception that it was agreed that SGA children had more learning problems (one main domain in the parent rating scale and a single item of the inattention-passiveness domain in the teacher rating scale) than AGA children. Parents may overreport as often as underreport problems with their children. It is also possible that parental attitudes change with time so that parents might be more anxious in the first years of the lives of their children with a perinatal hazard but develop a tolerance for problems as their children grow older.36
The second most powerful predictor was verbal IQ score. On average, the overall intellectual performance of SGA children was lower than that of AGA children but within the reference range.15,37 Sameroff38 showed that every environmental risk factor is able to reduce the child's IQ score by 4 points. To our knowledge, the rate of mental retardation (full-scale IQ score <70) among SGA children has not been reported; in our study, 5% of SGA children were mentally retarded (vs 1% of AGA children). Head growth can be used as an index of overall brain development.39 We confirmed the trend for children failing at school having smaller head circumferences than those performing well at school.9 Ounsted et al40 and Hack et al9 found a significant mother-child correlation for head circumference. In our study, only 3 of 15 children with small head circumference had a mother with small head circumference, indicating the possible genetic effect. Thus, the greater incidence of small head circumference in SGA children might be a manifestation of IUGR.9,41,42 Poor head growth may result from hypoxic-ischemic encephalopathy without IUGR.43 Among small-headed children, none had umbilical artery acidemia, and only 1 had a low 5-minute Apgar score, indicating potential asphyxia. Magnetic resonance imaging diagnostic for hypoxic-ischemic encephalopathy was not performed at the time the study children were born.
The third most powerful predictor was restless behavior during the outpatient clinic visit as rated by the examiners. Other studies8,44 also show similar overactive behavioral characteristics in association with IUGR. The pediatrician involved in examining the 10-year-old study children practically corresponded with a community pediatrician or school physician in his or her work. Schoolchildren showing restlessness or difficulties in concentration in a face-to-face situation should be provided with an overall evaluation of their school performance.
To prevent adverse sequelae of IUGR, we would like to emphasize not only potentially preventable risk factors such as fetal challenge to smoking45 but also diagnostic tools for IUGR46 and the best date to deliver the growth-retarded fetus.46 The SGA children are at risk for an adverse outcome. It has been reported previously12 that full-term SGA children perform more poorly at school than do preterm AGA children, indicating the significance of IUGR—and not of gestational age as such. As early as the 1970s, Neligan et al47 declared that "it is better to be born too soon than too small." However, there are studies48 concerning the academic achievement of low-birth-weight children that disadvantageously leave the relation of gestational age to birth weight unaddressed.
Characteristics such as hyperactivity, verbal deficits, and attention problems recognizable in early childhood can predict school difficulties and should arouse the attention of those involved in child health. A focus should be put on school quality and community support, too. Failure in the first years of school may result in low self-esteem and behavioral problems that might be antecedents of high-risk behaviors and school dropout. Strauss49 showed that those born SGA not only have increased academic difficulties persisting into adolescence but also deficits in professional and economic attainment in adulthood. Prenatal growth failure should be recognized as a risk factor for school failure, although not only SGA children but all children at risk need effective programs for managing learning difficulties.
In conclusion, SGA children perform poorer at school than their AGA peers. Independent predictors of poor academic achievement in SGA and AGA children are inattention-passiveness as rated by teachers, low verbal IQ score, and restlessness. There are poorer performances in neurocognitive tests and less favorable socioeconomic background among SGA children than among AGA children. The smallness of a newborn is worth recognizing as a risk factor for school failure.
Accepted for publication September 30, 2001.
This study was supported by the Local Turku Fund of the Foundation for Pediatric Research, the South-West Finnish Fund of Neonatal Research, and the Turku University Foundation, Turku.
We thank Heikki Lyytinen, PhD, from the University of Jyväskylä, Finland, who kindly made the computerized test of attention available for us; Olli Kaleva, BSc, for skillful computation of the statistical analyses; and Eija Suopajärvi for secretarial assistance.
This prospective, population-based, birth cohort, 10-year follow-up study shows that SGA children have neurocognitive deficits, behavioral problems, and less favorable sociodemographic backgrounds associated with poor academic achievement.
The number of school failures among SGA children is nearly twice as great as among their AGA peers. Being SGA at birth is worth recognizing as one of the risk factors for poor school performance.
Corresponding author and reprints: Outi Hollo, Department of Public Health, University of Turku, Lemminkäisenkatu 1, FIN-20014 Turun Yliopisto, Finland (e-mail: firstname.lastname@example.org).
Hollo O, Rautava P, Korhonen T, Helenius H, Kero P, Sillanpää M. Academic Achievement of Small-for-Gestational-Age Children at Age 10 Years. Arch Pediatr Adolesc Med. 2002;156(2):179-187. doi:10.1001/archpedi.156.2.179