Follow up studies of very low birthweight (VLBW) children have documented increased prevalence of neurodevelopmental disabilities and cognitive deficits.^1–3 There has been a growing awareness of psychiatric problems.^4–7 The most consistent finding is an increased risk of attention deficit/hyperactivity disorder (ADHD), especially the inattentive type.^3,6,8,9 Anxiety disorders, depression, and thought problems may also occur.^6,7

For children born small for gestational age (SGA) at term, the outcome is less clear, especially when it comes to social and emotional development.^6,10–14 VLBW and term SGA infants differ in biological basis and vulnerability, hence the long term outcome may vary. Most studies are performed in childhood using questionnaires.³ Although there are some follow up reports in adolescence, our knowledge on psychiatric disorders in this age group is limited, and few have used psychiatric interview as a diagnostic tool.^{4,5,7,15–20} By using in depth psychiatric assessment, we wanted to evaluate whether the prevalence of psychiatric symptoms and disorders is higher in VLBW and SGA adolescents than in adolescents with normal birth weight, and describe the characteristics of psychiatric symptoms and disorders in these groups.

MATERIAL AND METHODS

Study design

This study is a population based follow up at 14 years of age of two groups of adolescents with low birth weight (VLBW and term SGA) compared with a control group of normal birth weight. The VLBW children were admitted to the neonatal intensive care unit at the University Hospital in Trondheim (the referral hospital) in 1986–1988. Children born in 1988 were assessed thoroughly at 1 and 6 years of age.^21–23 The SGA and control children were born to mothers living in the Trondheim region. They were enrolled before week 20 of pregnancy in a multicentre study between January 1986 and March 1988.^24,25 A 10% random sample of women (with one or two previous pregnancies) was selected for follow up during pregnancy. At birth, all the children born to mothers in the random sample and all the SGA children were included for follow up.²⁶ The present study was carried out between November 2000 and October 2002, and included a psychiatric assessment, an evaluation of cognitive and motor abilities, and a neuropaediatric examination.

Study population

VLBW adolescents

VLBW was defined as a birth weight ⩽ 1500 g. Ninety nine children were admitted to the neonatal intensive care unit in 1986–1988. Of these, 23 died, one child with trisomy 21 was excluded, and six had moved. Of the remaining 69, 13 did not consent to participate at follow up. Thus 56 (81%) were examined (30 boys, 26 girls). Twelve had a birth weight ⩽ 1000 g. Ten participants were twins, of whom six were in twin pairs.

SGA adolescents

Of 1200 eligible women, 104 (9%) gave birth to an SGA child, defined by a birth weight < 10th centile of all infants in the multicentre study, adjusted for gestational age, sex, and parity.²⁵ At follow up, 12 had moved. Of the remaining 92, 32 did not consent, leaving 60 (65%) for assessment (28 boys, 32 girls).

Control adolescents

The control group comprised 120 children with a birth weight ⩾ 10th centile for gestational age, born at term to mothers in the 10% random sample. At follow up, 10 had moved. Of the remaining 110, 27 did not consent, leaving 83 (76%) who participated (35 boys, 48 girls).

Non-participants

There were no differences in mothers’ age at childbirth, duration of pregnancy, or the infants’ birth weight, body length, and head circumference between those who participated and those who did not consent in any of the groups.

Methods

Psychiatric assessment

Psychiatric symptoms and disorders were diagnosed using the semistructured interview Schedule for affective disorders and schizophrenia for school aged children (KSADS).^27,28 This interview is performed separately with parent and adolescent. It consists of a screening part and supplementary diagnostic interviews. Conclusions are drawn according to the Diagnostic and statistical manual of mental disorders, fourth edition (DSM-IV).²⁹ The interviewer differentiates between diagnosis, symptoms ⩾ 75% level of diagnostic criteria, or neither.

Of the 199 adolescents, the first author assessed 171, and another child psychiatrist carried out 28 interviews. The first 27 interviews with adolescents were videotaped and scored to test interrater reliability. The intraclass correlation coefficient was 0.91 for all symptoms, 0.86 for anxiety symptoms, and 0.86 for ADHD symptoms. A third senior child psychiatrist, blinded to the first evaluation, performed a second diagnostic assessment based on the written interviews. Subsequently, diagnostic consensus was obtained. The child psychiatrists were blinded to birth weight status.

The interviewer scored the adolescents’ overall functioning on the children’s global assessment scale (CGAS).³⁰ This scale ranges from 1 to 100, divided into 10 main categories. A score above 80 denotes good functioning.

The psychiatric interview was supplemented with the autism spectrum screening questionnaire (ASSQ),³¹ as part of the parents’ interview, and the teachers’ report on ADHD rating scale IV.^32,33 Parents stated their own mental health on symptom checklist-90-R (SCL-90-R), and we used the global severity index as a single summary measure of psychological distress.³⁴

Evaluation of cognitive abilities

An estimate of intelligence quotient (IQ_est) was calculated using four of 10 subscales of Wechsler intelligence scales (WISC-III)³⁵: vocabulary, arithmetic, block design and picture arrangement.³⁶ We defined “low IQ_est” as below 2 SDs of the control group mean value. Data were analysed both including and excluding children with low IQ_est.

Socioeconomic status

Socioeconomic status was calculated according to Hollingshead’s two factor index of social position, based on a combination of parents’ education and occupation.³⁷

Ethics

The regional committee for medical research ethics approved the study protocol. Written informed consent was obtained from adolescents and parents.

Statistical analysis

SPSS 11.5.1 was used for data analysis. Three group comparisons were made using one way analysis of variance for variables with a normal distribution and Kruskal-Wallis for ordinal data. If a difference was found between groups (p < 0.05), two group comparisons were made using Scheffe’s post hoc test for normally distributed variables and the Mann-Whitney U test for variables with a non-normal distribution.³⁸ Differences in proportions between groups were analysed by the χ² test. Odds ratios (OR) with 95% confidence interval (95%CI) were calculated as an estimate of the relative risk that a child with low birth weight had psychiatric symptoms and disorders, compared with the control group. To control for possible confounders, we used binary logistic regression³⁸ to calculate adjusted OR, first in a bivariate analysis, followed by a multivariate analysis (backward stepwise) for covariates altering the unadjusted OR > 10%.

RESULTS

Group characteristics

Table 1 shows family and child characteristics. There were no significant differences between groups with regard to mothers’ or fathers’ education, socioeconomic status, monthly income, or maternal or paternal mental health. Likewise, there were no sex differences between the groups. Of the 56 VLBW adolescents, seven (13%) were living in single parent families, compared with five (8%) in the SGA and 15 (18%) in the control group (ns).

Ten (18%) VLBW adolescents had low IQ_est compared with four (7%) SGA and three (4%) controls. Cerebral palsy was diagnosed in six (11%) VLBW (four diplegia, one hemiplegia, one quadriplegia), and one SGA adolescent (diplegia). Four of the VLBW and the one SGA with cerebral palsy had low IQ_est. Two VLBW and one control adolescent had epilepsy. There were no major visual or hearing impairments.

Psychiatric assessment by semistructured interview (KSADS)

Of the 56 VLBW adolescents, 26 (46%) had emotional or behavioural problems ⩾ 75% of diagnostic criteria, compared with 11 of 83 (13%) controls (OR 5.7, 95%CI 2.5 to 13.0) (table 2). The most common symptoms among VLBW adolescents were attention deficit (OR 6.6, 95%CI 2.0 to 21.3 v controls) and anxiety (OR 3.5, 95%CI 1.2 to 10.0 v controls). Four VLBW adolescents had symptoms of Asperger’s disorder, compared with none in the control group (p < 0.05).

Psychiatric disorders were diagnosed in 14 (25%) VLBW adolescents compared with six (7%) controls (OR 4.3, 95%CI 1.5 to 12.0) (table 2). Anxiety disorders were most prevalent (OR 4.4, 95%CI 1.1 to 17.6 v controls), whereas four (7%) had ADHD compared with one control (ns). One VLBW adolescent had Asperger’s disorder. Six of 12 (50%) with birth weight ⩽ 1000 g and eight of 44 (18%) with birth weight > 1000 g had a psychiatric diagnosis (p  =  0.054).

Nine of the 10 VLBW adolescents with low IQ_est had psychiatric symptoms, and four had a diagnosis (ADHD, Asperger’s disorder, anxiety disorders). Of the two VLBW adolescents with cerebral palsy without low IQ_est, one had ADHD.

Although 14 (23%) SGA adolescents had psychiatric symptoms, and six (10%) had a psychiatric disorder, this did not differ significantly from controls (table 2).

There were no sex differences in the VLBW and SGA group, whereas in the control group more boys than girls had ADHD symptoms (p < 0.05).

Few had experienced drinking ⩾ 3 units of alcohol (two VLBW adolescents, three SGA, five controls) (ns). One adolescent in each group admitted experimenting with other substances (cannabis, amphetamine).

ADHD rating scale IV

VLBW adolescents had higher mean attention score and total score than controls, without significant difference in hyperactivity score (table 3). Although the SGA group had higher total score than controls, the difference was not significant. Boys had a higher total score than girls in the control (p < 0.01) and SGA groups (p < 0.05), but not in the VLBW group (p  =  0.30) (data not shown).

Autism spectrum screening questionnaire (ASSQ)

The VLBW group had higher mean sum score than controls (table 3), whereas the SGA group showed a trend towards a higher sum score compared with controls (p  =  0.08). There were no sex differences. VLBW adolescents had high scores on the items: “Lives somewhat in a world of his/her own with restricted idiosyncratic intellectual interests”, “Has a literal understanding of ambiguous and metaphorical language”, “Uses language freely but fails to make adjustment to fit social contexts or the needs of different listeners”, “Lacks empathy”, “Wishes to be sociable but fails to make relationships with peers”.

Children’s global assessment scale (CGAS)

VLBW adolescents had lower mean CGAS score than controls (table 3). Scores below 80 were found in 37 of 56 (66%) VLBW, 17 of 83 (21%) control (OR 7.6, 95%CI 3.5 to 16.3), and 24 of 60 (40%) SGA (OR 2.6, 95%CI 1.2 to 5.4 v controls) adolescents. There were no sex differences in any of the groups.

Excluding adolescents with low IQ_est

When adolescents with low IQ_est were excluded, the results were essentially unchanged, except that the VLBW group had a reduced prevalence of anxiety symptoms (p  =  0.09) and disorders (p  =  0.07) compared with controls.

Multivariate analyses

With the outcome variable symptoms of any psychiatric disorder (⩾ 75% of diagnostic criteria), logistic regression analysis was run to evaluate the confounding effect of psychosocial factors (table 4). Paternal education and maternal and paternal mental health were the only factors that influenced the increased risk of psychiatric symptoms in the VLBW group more than 10%. When these variables were included in multivariate analysis, the increased risk associated with VLBW was 5.8 (95%CI 1.9 to 17.4), with maternal mental health as the only variable causing a significant change. Analogously, after multivariate analysis, the risk associated with SGA was 2.9 (95%CI 1.0 to 8.2), with only paternal mental health causing a significant change.

DISCUSSION

In this study we found that nearly every other VLBW adolescent had psychiatric symptoms. One in four had a psychiatric diagnosis, anxiety disorders being the most prevalent. Attention deficit problems were acknowledged in every fourth VLBW adolescent, but only a minority fulfilled the diagnostic criteria. Moreover, relational problems and deficits in social skills were common among VLBW adolescents, and some had autistic spectrum symptoms. Although every fifth SGA adolescent had psychiatric symptoms, they did not differ significantly from controls.

The strength of this study is the thoroughness of the psychiatric assessment, using both questionnaires and a semistructured interview, and different sources of information. Even though the VLBW adolescents had widespread emotional or behavioural symptoms, many did not reach the diagnostic level of a psychiatric disorder. However, the CGAS results indicate that the psychiatric problems were affecting the overall functioning in both the VLBW and the SGA group, causing strain for the adolescents and their families.

The psychiatric outcome was significantly worse for the VLBW group, with consistent results across all instruments. This association is unlikely to be due to chance. The lack of statistical significance for the SGA group may be due to the limited sample size.

Of the eligible children, 73% (199/271) took part in the study. There were no differences in key variables between participants and those who did not consent to participate. It is therefore unlikely that our main results are caused by selection bias.

The parents of VLBW children knew that their child belonged to a risk group. However, the examiners were blinded, and assessment was based on separate interviews with adolescent and parent, including clinical judgment and teachers’ report on ADHD rating scale, making it less likely that the results are caused by information bias.

The VLBW children were classified according to a birth weight ⩽ 1500 g, and some of them may have been growth retarded in utero. However, they were all born preterm, and we have chosen to treat them as a group in this study. The 10th centile definition of SGA is crude, and may include a certain proportion of normal small infants, whereas some growth retarded infants may have been classified as controls. Thus, the non-significant differences found between SGA and controls may be underestimates of the real differences.

The results were mainly unchanged when we excluded the adolescents with low IQ_est. Moreover, the increased risk of psychiatric symptoms among VLBW adolescents persisted after possible psychosocial confounders had been controlled for.

Our results on psychiatric symptoms are consistent with other studies on VLBW adolescents at ages 10–14 years.^4,5,15,18,39 Prevalence reports on psychiatric disorders vary from 24% to 32%, which is well in keeping with our results.^5,15,39

The prevalence of anxiety disorders corresponds to reports of 8–10% in VLBW and LBW (birth weight < 2500 g, ages 6–12).^5,19,40 We did not find an increased prevalence of depression, which has been reported in one study of adolescents with birth weight ⩽ 1000 g.²⁰

Our finding that 25% of VLBW adolescents had considerable attention deficit symptoms is in keeping with other studies.³ Reports on the prevalence of ADHD vary between 5% and 32%,^{5,15,18,19,39,41–43} presumably because of differences in study design, inclusion criteria, assessment age, and methods. We found that only 7% had ADHD according to DSM-IV criteria. This supports the minority of researchers who claim a low prevalence of ADHD in adolescence.^15,18,41 The male predominance described in children with ADHD⁴⁴ was not found in the VLBW group, nor did they have conduct disorders. This is consistent with several reports in which the authors argue that VLBW children have a specific (“pure”) form of ADHD with a dominating neurological cause.^5,6,45

We found autistic spectrum symptoms in the VLBW group. Studies on autism include pregnancy complications and intrauterine growth retardation as risk events, but the main conclusion is that there is a strong genetic component in the cause of autism.^46–48 Few follow up studies of VLBW children have reported on autistic spectrum symptoms,^19,49 although there are reports on deficits in social skills.^18,19,50 Sykes et al⁵⁰ suggested that VLBW children have problems adjusting to the social environment, because of a failure in self regulatory functions. This hypothesis is further elaborated by Davis and Burns.⁵¹ Hille et al⁵² argued that the attention deficit may cause difficulties in social relations through poor processing of cognitive stimuli. We speculate that the core point may be specific problems in relating to other people, with deficits in comprehending the subtle clues of social relations, comprising both emotional and cognitive components.

In this study, we found an association between VLBW and psychiatric problems that could not be explained by socioeconomic status or the parents’ mental health. A biological cause-effect relation seems plausible, as these children are born with immature brains and a high risk of damaging incidents in the neonatal period.^{5,6,8,16,40,43,45,52,53} To be born SGA at term seems to involve less biological risk of psychiatric problems.¹³ We also recognise that psychosocial experiences may interact with the biological vulnerability, as reported in other studies.^4,13,54 Regardless of the cause, our results should draw attention to the fact that VLBW children may need long term health services.

In conclusion, we found that VLBW, but not SGA adolescents, have a high risk of developing psychiatric symptoms and disorders by the age of 14, especially attention deficit, anxiety, and relational problems affecting their social skills.