Behavioral and Emotional Adjustment of Teenagers in Mainstream School Who Were Born Before 29 Weeks’ Gestation
Objectives. To investigate behavioral and emotional problems and positive adjustment of 15-to 16-year-olds who were born at extremely low gestational age (ELGA), from the perspective of parents, teachers, and teenagers.
Methods. Prospective follow-up was conducted of birth cohorts, with classroom control subjects. All infants who were born before 29 weeks in 1983–1984 (mean gestational age: 27 weeks) to mothers who resided in 3 regions of the United Kingdom were studied. A total of 82% (179 of 218) of survivors were traced at age 15 to 16. The 150 in mainstream school were compared with age- and gender-matched classroom control subjects (n = 108). Behavioral and emotional problems, delinquency, peer relations, self-esteem, and hobbies, were assessed by standardized, well-validated instruments, including the Strengths and Difficulties Questionnaire, administered by mail to parents, teenagers, and teachers.
Results. Parents were more likely to rate ELGA teenagers than control subjects as in the “abnormal” range for hyperactivity (8% vs 1%; difference: 7%; (95% confidence interval [CI]: 2–12), peer relationship problems (19% vs 5%; difference: 14%; 95% CI: 6-21), and emotional problems (18% vs 7%; difference: 11%; 95% CI: 3-19), but not conduct problems (10% vs 5%; difference: 5%; 95% CI: −1 to 12)). Teachers reported a similar pattern. In contrast, compared with control subjects, ELGA teenagers did not rate themselves as having more problems with peers, hyperactivity, conduct, depression, or low self-esteem. They reported more emotional problems but less delinquency, alcohol, cannabis, and other drug use.
Conclusions. Compared with mainstream classmates, children who are born extremely early continue to have higher levels of parent- and teacher-reported emotional, attentional, and peer problems well into their teens. However, despite these problems, they do not show signs of more serious conduct disorders, delinquency, drug use, or depression.
It is widely known that infants who are born very preterm are at increased risk for physical, sensory, and learning disability1–3 and behavioral problems.4,5 However, most studies have followed these children only as far as early or middle childhood, and less is known about outcome later in the teenage years, when antisocial behavior and mood disturbances become more common in the general population. Moreover, there is evidence that childhood learning, behavioral, and attentional problems carry an increased risk for conduct disorders and delinquency in adolescence.6,7 These problems are known to be more prevalent in low gestational age children and were detected in these cohorts at earlier ages. However, it is not known whether these early problems carry the same risk for later antisocial outcomes in young people who were born very early.8 Although some studies have followed children who were born preterm (<32 weeks) to adolescence,9,10 these have not involved cohorts that were born extremely preterm.
We also know very little about positive aspects of social adjustment in young people who were born very early. The qualities of prosocial behavior, self-esteem, peer relationships, social activities, and hobbies are likely to be important as indicators of psychological well-being and as potential protective factors against disorder.11 They also provide information about the teenager’s quality of life. In adolescence, it is important to assess psychological well-being from the teenagers’ viewpoint, as their perspective may differ from that of parents and teachers. With teenagers’ increasing independence, their perspective is essential for assessing characteristics that are unlikely to be accessible to adults, including internal emotional states and antisocial behavior outside the home.
A number of studies of vulnerable neonates have focused on those with very low birth weight, also finding high rates of emotional and behavioral problems in childhood and early adolescence.8,9,12–17 Botting et al’s12 detailed interview study found higher rates of anxiety as well as attention-deficit/hyperactivity disorder (ADHD) but not depressive and conduct disorders. Saigal et al’s18 findings were similar in some respects, in that young adolescents who were born at extremely low birth weight showed raised rates of ADHD but not conduct symptoms. Conversely, they showed increased rates of depressive but not anxious symptoms. However, these cohorts defined by birth weight included both very preterm infants and those of higher gestational age but who may have had fetal growth retardation. For exploring the risk of behavioral problems associated with preterm birth, it is better to define cohorts by gestational age rather than by birth weight.
Thus, the aims of the present study were to examine behavioral and emotional difficulties and positive adjustment in teenagers in mainstream school who were born at extremely low gestational age (ELGA; before 29 weeks’ gestation), compared with classroom control subjects of higher gestational age. We used standardized questionnaires, administered by mail to parents, teachers, and teenagers. We combined in a collaborative study 3 cohorts of infants who were born before 29 weeks’ gestation in 1983–1984 in different parts of the United Kingdom. In this way, larger numbers were available to get more precise and generalizable estimates of the risks of later adverse outcomes than would be possible with a study based on a single area. Survivors were followed up at different ages,1,14,19,20 and we planned to retrace these at 16 years of age.21 Although some children with severe disabilities are in special schools, the majority of children in the cohort are in mainstream schools, and we wanted to be able to study their adjustment in comparison with peers in the same mainstream classrooms.
The methods for tracing and recruitment of participants have been described in detail elsewhere21 and are summarized here. The original 3 cohorts were defined by maternal residence at the time of birth and included all infants who were born before 29 weeks’ gestation in 1983 in the former Northern region and in 1984 in Scotland and the former Oxford Region. Gestational age had been checked at the time of assembly of the cohorts using a combination of menstrual dates and ultrasound before 20 weeks when available. The scan date was preferred when the menstrual date was uncertain and when there was a discrepancy of >14 days between the menstrual date and scan estimate.
Tracing was conducted through the child’s general practitioner (GP) or, when necessary, through National Health Service (NHS) Central Registers, which established whether the child was still alive and gave details of their new GP. Written permission was requested from both parents and teenagers to send a questionnaire to each of them and to the teenager’s teacher. With permission, a letter was sent to the head teacher enclosing a questionnaire for the year tutor of the teenager. Thus, the questionnaire was filled in by the teacher with overall responsibility for the young person’s pastoral and academic care. The school was asked to identify as a control the teenager in the same year group of the same gender and nearest in birth date to the ELGA teenager. When the family did not wish to participate, the school then approached the teenager second nearest (or third if needed) in birth date to the ELGA teenager. Parents, teachers, and control teenagers were asked to complete the same questionnaires as for the ELGA teenagers. In 2 of the 3 geographic areas of the study, questionnaires were returned to the local coordinators. Two reminders were sent to nonresponders. In the Northern region, the questionnaires that were completed by parents and teenagers were retrieved at the time of a visit to the family by a developmental pediatrician. Teenagers’ and parents’ questionnaires were sent out and, except in the Northern Region, returned in separate envelopes.
We estimated that 170 ELGA children in mainstream schools would take part, with 170 control subjects. Given a prevalence rate of 5% or 10% in adverse behavioral outcome22 in the control subjects, this sample size would have enabled us to detect (2α = .05; β = .20) an increase in prevalence to 14% or 21%, respectively, in the ELGA children. The smaller numbers in fact enrolled enabled detection of a prevalence rate of 17% or 24%, respectively, in the ELGA children.
We used brief, standardized, well-validated questionnaires, completed by parents, teachers, and teenagers. For assessing emotional and behavioral problems, we used Goodman’s 25-item Strengths and Difficulties Questionnaire (SDQ).22 The SDQ was derived from the widely known Rutter Scales and is well-validated as a tool for screening for psychiatric disorder.22–25 It has 5 subscales, 4 that assess psychiatric symptoms including conduct (eg, tempers, stealing, fighting), hyperactive/attentional (eg, restless, fidgets, distractible), emotional (eg, fears, worries, unhappiness), and peer problems (eg, solitary, unpopular, bullied) and 1 that assesses prosocial (helpful, caring) behavior. In a study of >10 000 children and adolescents in the United Kingdom, internal consistency, stability, and cross-informant correlation were satisfactory, as was the ability of the SDQ to predict independently diagnosed psychiatric disorder.23 The SDQ has been translated into >50 languages, and its factor structure, reliability, and validity were found to hold up in a Dutch validation study.24 In a comparison study with the 118-item Child Behavior Checklist,25 scores on the 2 instruments were highly correlated, and the SDQ was as good or somewhat better at detecting psychiatric disorder, despite being much more brief.
We used established cutoffs on the SDQ to identify those in the “abnormal” range on each subscale and thus at elevated risk for psychiatric disorder.22 Parents and teachers also rated “impact”; that is, the extent to which behavioral and emotional difficulties, when present, interfered with the teenager’s everyday life, including learning and friendships.22 Teenagers rated depressive symptoms (eg, misery, loneliness) on the Mood and Feelings Questionnaire,26 delinquency items (eg, fighting, truancy) from the Youth Self Report,27 and self-esteem.28 They were asked about frequency of use of alcohol, cannabis, and 6 other illegal drugs.21 Parents rated the amount and quality of the teenagers’ engagement in hobbies and activities outside school, using the Competence scales of the Child Behavior Checklist.27 Social class was recorded, based on occupation of the father, or mother when father was absent from the family. We asked other questions about health and educational outcomes, which are reported elsewhere.21 The project was approved by the Oxford NHS Multicenter Research Ethics Committee and by all relevant local ethics committees.
Data were double-keyed and analyzed using SPSS v9.0 and StatXact3 (Cytel Software Corp, Cambridge, MA). Comparisons between proportions were made using the χ2 test or an exact test when numbers were small. The t test was used for comparisons between means.
Of the 535 infants in the original cohorts, data from the school-age follow-ups suggested that 218 were alive at 16 years of age, and this was confirmed by the NHS central register. Of these, at least 1 completed questionnaire was available for 179 (82%) of the ELGA teenagers. Nonresponders were comparable to responders in birth weight and gestational age but (in the 2 areas for which data were available) were of lower social class (Table 1). Of the 179 ELGA teenagers on whom some information was available, 150 were in a mainstream school. The remaining 29 were in special schools and are not the subject of this article. They had moderate to severe, often multiple disabilities, including severe cerebral palsy, deafness or blindness, and severe learning and/or behavioral difficulties. The disabilities in teenagers in special and mainstream schools are described in more detail in a companion article.21 The present article refers only to the 150 in mainstream school and to the 108 control teenagers who agreed to participate. Ten of the 150 ELGA teenagers in mainstream school had moderate to severe disabilities, including 6 with cerebral palsy with very limited or no mobility and 4 with very poor vision.
Questionnaires were completed by both teenagers and parents for 142 of the ELGA group. Of the remaining 8, 5 returned only the teenager questionnaire and 3 returned only the parent questionnaire. All 108 of the control teenagers completed questionnaires, as did 100 of their parents. Because of the design of the study, questionnaires for control teenagers were completed later (mean age: 16.0 years; standard deviation: 0.46) than the ELGA group (mean age: 15.6 years; standard deviation: 0.54). Of the 145 questionnaires completed by ELGA parents, 124 were completed by mothers, 16 by fathers, and 5 by another caregiver. Of the 100 questionnaires completed by control parents, 96 were completed by mothers, 1 by a father, and 3 by another caregiver. Teachers completed questionnaires for 120 ELGA teenagers and 92 control subjects. ELGA and control groups were comparable on demographic variables (Table 2).
Behavioral and Emotional Problems
Parents of ELGA teenagers, compared with control group parents, were more likely to rate their child as scoring in the abnormal range on hyperactivity and emotional and peer problems but not on conduct problems (Table 3). The pattern of findings was closely similar for teachers. Of those who perceived their teenager to have behavioral or emotional problems, parents of ELGA teenagers were more likely to rate the problems as having an adverse impact on the teenagers’ functioning. There were no group differences in teachers’ ratings of the impact of these problems. A series of logistic regression analyses based on parent-reported outcomes found that adjusting for demographic variables (social class, maternal education, and marital status) made no material difference to the results (data not shown).
There were no group differences in teenagers’ self-ratings of hyperactivity, conduct or peer problems, or depressive symptoms. Compared with the control group, ELGA teenagers were more likely to rate themselves as having emotional problems, such as worries and lack of confidence. Teenagers in the control group rated themselves as more delinquent and as more likely to use alcohol, cannabis, or other drugs than those in the ELGA group.
According to parent report, there were no group differences in teenagers’ prosocial behavior (eg, sharing, helping), although by teacher and teenager report, teenagers in the control group were seen as more prosocial. Parents rated the quality and quantity of social life and activities outside school as lower in the ELGA group, compared with the control group. There were no group differences in the teenagers’ sense of self-esteem (Table 4).
Our findings concern the group of teenagers who were born extremely preterm and are educated in mainstream school. On the positive side, these young people represent a majority, >80%, of the survivors whom we were able to trace. Moreover, only a minority of teenagers who were born extremely preterm show adjustment problems. However, compared with their mainstream classroom peers, ELGA teenagers have higher rates of psychological problems, including hyperactivity and emotional and peer problems but not conduct problems, as reported by parents and teachers. According to parents, these problems are more likely than similar problems in control subjects to have an adverse impact on the child’s functioning in areas such as learning, friendships, home life, and leisure. Teachers of ELGA teenagers, conversely, did not believe that behavioral and emotional difficulties, when they occurred, had a greater impact than those of comparison teenagers. It should be noted that this is a conservative index of the impact of behavioral and emotional problems, as it asks about impact only in the minority of children who show problems. In terms of absolute numbers, ∼2 to 3 times as many ELGA teenagers as control subjects show problems that have a significant impact on their functioning, according to teachers. The comparable figure for parent-reported impact on different areas of functioning ranges from 2 to 12 times as many ELGA teenagers as control subjects.
When teenagers reported on their own adjustment, only emotional problems (mainly nervousness and worries, rather than depression) were more common in ELGA teenagers compared with control subjects. ELGA teenagers did not have a lower sense of self-esteem, which is consistent with some recent studies of teenagers who were born at extremely low birth weight29 and moderately low gestational age10 but not others.15 It is particularly intriguing that these young people, despite having several major risk factors for antisocial behavior (attention deficit, peer problems, and, as found in the companion study, learning difficulties21), do not, according to all 3 informants, show higher rates of conduct problems, and indeed they showed lower rates of self-reported delinquency, as well as alcohol and drug use than their peers. This finding of high levels of hyperactivity but not conduct problems is consistent with studies of younger low birth weight children8,12,30 and extremely low birth weight adolescents,18 and with a recent follow-up to age 20 of a group of young people with very low birth weight,31 which found lower rates of criminal convictions and drug and alcohol use compared with control subjects. We discuss possible explanations for these findings later.
As well as showing higher rates of attentional, peer, and emotional problems, ELGA teenagers were seen by their parents as less well-adjusted in terms of their involvement in extracurricular activities and, according to teachers and to their own self-rating, as being less helpful and considerate to others. These findings cannot easily be explained in terms of the more disabled children simply being unable to participate in activities, as when the 10 most severely disabled children in the mainstream sample were excluded, we found the same pattern of significant results for all variables reported here (data not shown).
In contrast to adult reports, teenagers themselves presented a more positive picture, reporting similar levels of self-esteem and psychological adjustment as their peers in all areas apart from emotional problems. In some respects, this is in keeping with a recent study18 of adolescents who were born at extremely low birth weight, which found differences between index and control groups in ADHD and depressive symptoms according to parents but not according to adolescents. Possible explanations for these differing views could include teenage optimism about their problems in low birth weight adolescents, relative to their peers, or parental overconcern. By including data on teacher reports, our study may be able to shed some light on this. Our finding that teachers and parents perceived a similar proportion of children to have problems does not support an interpretation in terms of parental overconcern.
Limitations of the study should be noted. As far as response rates of ELGA children are concerned, nonresponders were similar in birth weight and previously reported levels of impairment. However, as found in other studies,32 they were of lower social class, which may lead to slight underestimation of the level of problems in the ELGA teenagers. As far as control subjects are concerned, there was a variable level of compliance within both schools and families, and we have no knowledge of the characteristics of the potential control subjects who failed to respond. Although demographic characteristics were comparable between ELGA and control groups, there may have been some subtle group differences. However, a series of logistic regression analyses showed that demographic variables did not account for the findings. Furthermore, there was no evidence that control subjects (in the 2 English cohorts) came from more advantaged backgrounds than average, as their scores on national tests at age 14 were average for England.21 It is not possible to rule out the potential source of bias that parents of problem children might be more likely to refuse consent. However, because behavioral and academic problems tend to co-occur, any such bias would tend to lead to higher-than-average scores on national tests in the control group, which was not the case. Thus, it is unlikely that bias in the control subjects could be the whole explanation, especially given the possible underestimation of problems in the ELGA group. Furthermore, whereas any bias toward more advantaged families in the control group might explain their greater involvement in hobbies and social activities or their lower rates of ADHD symptoms, it could not explain their higher rates of delinquency and substance use, compared with ELGA teenagers.
Overall return rates were high for a postal questionnaire, especially given that many years had elapsed since the last contact with the families. Adjustment was measured using questionnaires with well-validated clinical cutoffs, rather than in-depth clinical assessment. However, our measure of behavioral and emotional problems (the SDQ) was able to capture an important aspect of clinical diagnostic assessment, namely the functional impact of problems, as perceived by parents and teachers. In common with other studies,12 we adopted a strategy of studying only those ELGA teenagers who were in mainstream schools. This has the advantage of allowing us to compare all children with classroom peers but brings the drawback that the sample is defined by a criterion (placement in mainstream school) that is influenced by level of disability but also by a variety of other factors, such as parental preference or local education department policies and resources.
The finding that rates of conduct problems are similar and rates of delinquency and alcohol and drug use are lower in the ELGA teenagers compared with the control group is surprising and merits some additional discussion. It seems to call into question fairly well-accepted theory6 that low cognitive ability and symptoms of ADHD, caused by genetic and adverse perinatal factors, are the major biological risk factors for later antisocial behavior. However, this group of children, born under circumstances of very high perinatal risk and showing high levels of both risk factors, seem to be somewhat less antisocial that their peers. Hack et al31 attempted to explain their similar finding in terms of parents’ nurturing and monitoring a preterm child better than they would a full-term child. Parenting is clearly a very important factor in the development of antisocial behavior33,34; however, Hack’s explanation would require that parents of ELGA children, as a group, show better-than-normal parenting skills in the face of considerable biological risk. This seems implausible, as it is not easy to parent with greater skill than usual a child who shows early and persistent ADHD and other behavioral symptoms. As in the general population, many parents of ELGA teenagers lived in circumstances of social disadvantage, factors that make parenting more difficult as well as increase the risk of delinquency. If better-than-average parenting in this group were not the explanation, then an alternative might be to postulate that early ADHD and perinatal risk factors are not, on their own, of major importance in the causation of antisocial behavior. Others have suggested that the interaction between perinatal and social risk may be critical.35 Additional studies are needed to elucidate how child outcomes are influenced by the interaction between parenting and perinatal risk.
This study represents the first detailed follow-up to age 16 of a group of children who are defined by extremely preterm birth and are in mainstream school, compared with classroom control subjects, using multiple informants, including the young people themselves, to identify positive and problematic areas of adjustment. A sizable majority of children who survive extremely preterm birth are functioning in mainstream school and do not show major adjustment problems. Nevertheless, it is clear that these young people at age 15 to 16, although apparently at lower risk of delinquency and drug use, continue to be at increased risk for other kinds of psychological problems, including hyperactivity and emotional and peer problems. It is not clear what the implications of these difficulties will be for adult life, although we might speculate that in the absence of antisocial behavior, the prognosis is relatively good. Findings from earlier follow-up studies of this and other cohorts suggest that when behavioral and emotional problems do occur, they are likely to be longstanding. Thus, when there are signs of management difficulties, many families might welcome and benefit from early identification and preventive intervention, particularly evidence-based interventions to help parents manage early behavioral and attentional problems.36–38 Many of the most well-developed and replicated programs38 are aimed at preventing conduct problems; nevertheless, there is increasing evidence for their beneficial effects on symptoms of hyperactivity.37,39,40 Moreover, even if conduct problems do not persist to adolescence in those who are born very preterm, they are nevertheless common in early and middle childhood, and it therefore may be important to help parents and children to manage these distressing problems. Furthermore, it may be appropriate to offer early interventions that target problems that seem to persist into adolescence, including parenting interventions for hyperactivity36 and child-based programs for improving peer relationships and emotional difficulties.41
This study was supported by an Action Research project grant. Core funding for National Perinatal Epidemiology Unit is provided by the UK Department of Health.
We thank the young people and their parents and teachers who took part in the study; schools who helped with piloting; the Steering Committee and colleagues at the NPEU for helpful advice; Shona Donald for coordinating the project in Scotland and Susan Fritz for coordinating the project in Newcastle; and Jaideep Singh for data collection in the Northern Region. Members of the ELGA Steering Group were Sarah Arkle, Edmund Hey, Michael Jones, Barbara Maughan, Anne Stewart, and the authors.
- ↵Johnson A, Townshend P, Yudkin P, Bull D, Wilkinson A. Functional abilities at age 4 years of children born before 29 weeks gestation. BMJ.1993;306 :1715– 1718
- ↵Levy-Shiff R, Einat G, Mogilner M, Lerman M, Krikler R. Biological outcomes and environmental correlates of developmental outcome of prematurely born infants in early adolescence. J Pediatr Psychol.1994;19 :63– 78
- ↵Hogan AE. Cognitive functioning in children with oppositional defiant and conduct disorder. In: Quay HC, Hogan AE, eds. Handbook of Disruptive Behavior Disorders. New York, NY: Kluwer Academic/Plenum; 1999:317–339
- ↵Scottish Low Birthweight Study Group. The Scottish Low Birthweight Study: II.Language attainment, cognitive status and behavioural impairment at 4 years. Arch Dis Child.1992;67 :682– 686
- ↵Elgen I, Sommerfelt K, Markestad T. Population based controlled study of behavioural problems and psychiatric disorders in low birthweight children at 11 years of age. Arch Dis Child Fetal Neonatal Ed.2002;87 :F128– F132
- McCormick MC, Workman-Daniels K, Brooks-Gunn J. The behavioral and emotional well-being of school-age children with different birth weights. Pediatrics.1996;97 :18– 25
- ↵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
- ↵Wariyar U, Richmond S, Hey E. Pregnancy outcome at 24–31 weeks gestation: mortality. Arch Dis Child.1989;64 :670– 677
- ↵Johnson A, Yudkin P, Bowler U, et al. Health and school performance of teenagers who were born before 29 weeks gestation. Arch Dis Child.2003;88 :F190– F198
- ↵Achenbach TM. Integrative Guide for the 1991 CBCL/4–18, YSR and TRF Profiles. Burlington, VT: University of Vermont, Department of Psychiatry; 1991
- ↵Saigal S, Lambert M, Russ C, Hoult L. Self-esteem of adolescents who were born prematurely. Pediatrics.2002;109 :429– 433
- ↵Patterson GR. Coercive Family Process. Eugene, OR: Castalia; 1982
- ↵Gardner F, Ward S. Parent-child interaction and children’s well-being: reducing conduct problems and promoting conscience development. In: Buchanan A, Hudson B, eds. Promoting Children’s Emotional Well-being. Oxford, UK: Oxford University Press; 2000:95–128
- ↵Arseneault L, Tremblay R, Boulerice B, Saucier J. Obstetrical complications and violent delinquency: testing two developmental pathways. Child Dev.2002;73 :498– 508
- ↵Hartman RR, Stage SA, Webster-Stratton C. A growth curve analysis of parent training outcomes: examining the influence of child risk factors (inattention, impulsivity, and hyperactivity problems), parental and family risk factors. J Child Psychol Psychiatry.2003;44 :388– 398
- ↵Webster-Stratton C. The Incredible Years Parent, Teacher and Child Training series. Book Eleven: Blueprints for Violence Prevention. Boulder, CO: Institute of Behavioral Science, University of Colorado; 2001
- Copyright © 2004 by the American Academy of Pediatrics