Previous studies show that children born below 28 completed weeks of gestation have an increased risk of attention–deficit hyperactivity disorder (ADHD),^1–4 and cognitive and behavioural deficits.⁵ However, most preterm children are born with higher gestational ages of 28–36 completed weeks. Whether premature delivery at 28–36 completed weeks of gestation increases the risk of clinically verified ADHD is unknown. Additionally, no previous study on birth weight and hyperkinetic disorder (HKD) or clinically verified ADHD has been conducted among children born at or above term (gestational age of 37 completed weeks or more), but previous findings indicate that intrauterine growth retardation (IUGR) at term may have long term effects on growth and development.⁶

HKD and ADHD are characterised by inattention, hyperactivity, and impulsivity.⁷ HKD measured in accordance with ICD-10⁸ is one of the most prevalent mental illnesses in child psychiatry (1–2%), and HKD is the clinical correlate to ADHD combined type.⁹ New findings show that HKD is the fourth (7.3%) most frequent discharge diagnosis in child psychiatry in Denmark; this percentage is increasing.¹⁰

Our aim was to study the association between prematurity (gestational age 26–36 completed weeks of gestation) and clinically verified HKD. We also studied the association between birth weight in term born children and the risk of clinically verified HKD.

METHODS

Subjects and data assessment

We conducted a nested case-control study based on data from four Danish longitudinal registers: the Danish Psychiatric Central Register,¹¹ the Danish Medical Birth Registry,¹² the Integrated Database for Labour Market Research (the IDA database),¹³ and the Danish Civil Registration System.¹⁴ The Danish Civil Registration System contains a specific personal identification number for all individuals residing in Denmark which also allows linkage to parents and siblings.¹⁴ The personal registration number enables linkage between information in all the national registers.

The Danish Psychiatric Central Register covers all inpatient admissions and outpatient contacts (from 1995) at Danish psychiatric departments. The register includes cumulative records and discharge data, dates, and diagnoses.¹¹ From 1994, the diagnoses are in accordance with the International Classification of Diseases (ICD), 10th edition;⁸ before 1994, the ICD, 8th edition. The Danish Medical Birth Register holds detailed information on all births in Denmark,¹² provided by the midwife present at the delivery. The IDA database contains longitudinal information on labour market affiliation for the total population and sociodemographic data. Continuous annually updated information is available and missing data occur only if the father is dead, unknown at birth, or has immigrated.¹³

This study includes all children born between 1980 and 1994 and registered in the Danish Psychiatric Central Register until the end of December 1999 with HKD as their main diagnosis (n = 834): disturbance of activity and attention (F90.0; n = 524), equivalent to ADHD combined type; hyperkinetic conduct disorder (F90.1; n = 194), equivalent to ADHD with conduct disorder; other hyperkinetic disorders (F90.8; n = 10), equivalent to ADHD inattentive type and hyperkinetic disorder; unspecified (F90.9; n = 106), equivalent to ADHD other types.

The diagnostic groups of children with HKD were categorised as mutually exclusive. If the child had been admitted to a psychiatric department or an outpatient clinic more than once and appeared in the register under different HKD diagnoses, the child was registered with the first diagnosis. To increase external validity, all children with childhood autism, Asperger’s syndrome, other pervasive developmental disorders (F84.0–F84.9; 308.0), and mental retardation (F70–F79; 312–315) were excluded. Other comorbidities were allowed; 279 (33%) had other diagnoses. Specific developmental disorders of speech, language, scholastic skills, and motor function (F80.0–F89.3) were the most commonly recorded first subsidiary diagnosis (165 cases; 20%).

By using a nested case-control design, based on the total Danish population, each of the 834 case children with HKD was matched with a random sub-sample of 25 single born children of the same gender, born at the same date, alive and undiagnosed at the particular date the case child was diagnosed.^15,16

Gestational age, measured in completed weeks (26–44), was reported to the Danish Medical Birth Register by the midwife present at the delivery on a mandatory coding sheet. Gestational age was based on either early fetal ultrasound measures or detailed information on the woman’s last menstrual period. Gestational age was categorised into the following groups: 26–33, 34–36, 37–39, 40–42, and 43–44 completed weeks. Gestational age between 40 and 42 weeks was used as the reference. Gestational age below 37 completed weeks was defined as preterm delivery; gestational age below 34 weeks was defined as very preterm delivery.

Birth weights between 690 and 5990 grams were categorised into the following groups: 690–1499, 1500–2499, 2500–2999, 3000–3999, and 4000–5990 g. Birth weights between 3000 and 3999 were used as the reference. A proxy measure of intrauterine growth retardation (IUGR) at or above term (37 completed gestational weeks or more) was defined as birth weight less than 2500 grams. The difference between the mean birth weight adjusted for gestational age among cases and controls was also calculated.

Only subjects with valid information on gestational age and birth weight were included in the final analyses. The final dataset consisted of 834 cases with HKD and 20 100 controls.

Information on hospitalisations and outpatient contacts of cases, controls, parents, and siblings was obtained from the Danish Psychiatric Central Register.¹¹ Psychiatric data were obtained at the time the cases and the controls were defined. Socioeconomic data on the parents were obtained from the IDA database¹³ at the time of birth of the children.

The national and local ethics committees and the Danish Data Protection Committee approved the study.

Statistical analysis

The association between gestational age, birth weight, and HKD (table 1) was calculated as a rate ratio (RR) with 95% confidence intervals (CI).¹⁵

Potential confounding factors such as socioeconomic factors, parental age, parity (table 2) and familial psychopathology were included one at a time in a multiple conditional logistic regression model, in keeping with Greenland’s suggestion.¹⁷ In the final model (table 3), the potential confounding factors (socioeconomic factors, parental age, and familial psychopathology) were included at the same time based on the a priori assumption that they all might confound the results.¹⁶ Information on missing values for each variable was included as a separate dummy variable in all analyses.

Next, we restricted the analysis to children with family members without a history of mental disorders. Furthermore, analyses were restricted to children with HKD without conduct disorder and with no recorded comorbidity other than comorbid disorders that are typically related to HKD (i.e. specific developmental disorders of speech, language, scholastic skills, and motor function (F80.1–F83.9)). Finally, analyses stratified on gender were performed.

RESULTS

Gestational age and HKD

Compared with children born at term, children with gestational ages between 34 and 36 completed weeks had an 80% increased risk of HKD, and children with gestational ages below 34 completed weeks had a threefold increased risk (unadjusted results) (table 1).

Birth weight and HKD

Among children born at 37 completed weeks of gestation, the mean birth weight was lower among cases compared with controls (unadjusted difference −98 g; 95% CI −134 to −61). Children born at term with birth weights between 1500 and 2499 g had more than a twofold increased risk of HKD compared with children born at term with birth weights above 2999 g, whereas children with birth weights between 2500 and 2999 g had a 70% increased risk (unadjusted results) (table 1).

Gender and age

Of the 834 children with HKD, 750 (90%) were boys. The increased risk of HKD was basically the same in both genders, but analyses on girls only yielded very wide confidence intervals as most of the cases were boys. The age of the children at the time of diagnosis varied between 2 and 18 years (median 8.8; interquartile range 3).

Social factors, previous admissions, family history of psychopathology

Single parent families, disadvantageous social factors, and young age of the parents were associated with an increased risk of HKD in the offspring; the effect of the income level of the father was generally higher than for the mother (table 2).

Previous psychiatric admissions and contact as outpatients of the cases and the controls (RR 21.7, 95% CI 17.0 to 27.5) and psychopathology in the immediate family, measured as admissions to psychiatric hospitals or departments or as outpatient contacts for the mother (RR 2.6, 95% CI 2.1 to 3.3), the father (RR 2.1, 95% CI 1.7 to 2.7), or siblings (RR 3.9, 95% CI 3.0 to 5.1) increased the risk of HKD.

Adjustments for social factors, history of psychiatric disorders in the parents and siblings, and parental age did not change the results substantially (table 3). The small changes in the risk estimates were due to a joint effect from all the variables under study and not accounted for by a single variable. When the results on birth weight in table 3 were adjusted for gestational age in weeks, the results were also unchanged (adjusted difference −94 g; 95% CI −143 to −45). For both exposures under study, a dose-response relation was present

Taking into account previous admissions and outpatient contacts and the time period since the last admission before the diagnosis, the results remained unchanged (data not shown). The results were essentially unchanged after restriction to children with parents or siblings without psychiatric admissions and contact as outpatients, thereby excluding 184 cases and their controls (data not shown).

Conduct disorder and other comorbidity

Excluding cases and their controls with conduct disorders either as main diagnosis (F90.1; 194 cases) or as subsidiary diagnoses (F90.1, 91.1, F91.3, F91.9; 13 cases) also failed to change the results. Exclusion of the 206 children with HKD (and their matched controls) with comorbid disorders other than specific developmental disorders of speech, language, scholastic skills, or motor function recorded as subsidiary diagnoses did not change the results.

Excluding children of parents with a history of mental disorders, children with hyperkinetic conduct disorder (F90.10), and children with other comorbid disorders except for specific developmental disorders (F80.0–F89.3) did not change the results (table 3). The mean birth weight remained lower among cases compared with controls.

Maternal smoking during pregnancy and gestational age

Among children born between 1991 and 1994 (n = 3935, n = 170), information on maternal smoking status was available (smoker, non-smoker).¹⁸ When the analysis of preterm delivery and HKD was performed among non-smokers (n = 2443; n = 65), the risk was still increased (RR 4.1, 95% CI 1.4 to 11.8).

DISCUSSION

This large population based study showed that preterm delivery near term and proxy measures of intrauterine growth in children born at or above term increase the risk of HKD.

Strengths and weaknesses

The longitudinal registers in Denmark provide detailed information on each individual. Our findings were based on all single born children without pervasive developmental disorders born during a 14 year period and registered with HKD as a main diagnosis. The controls were randomly selected from births during the same period. This procedure reduces problems with selection bias and optimises external validity.

Our large population based sample and prospective collection of information on gestational age and birth weight eliminate the risk of parental recall problems, which may result in differential misclassification. We were also able to adjust for several potential confounding factors that previously were poorly controlled. Adjustment for genetic predisposition to mental disorders is an important advantage over previous studies.^1–4 These adjustments reduce the possibility that genetic factors explain our findings.¹⁹ We are, however, aware that our adjustment for genetic predispositions may be incomplete because information on personality traits was unavailable.

Validation studies of the variables in the Danish National Birth Register show a high validity regarding gestational age in the time period under study.²⁰ We used birth weight less than 2500 grams at or above term as a proxy measure of intrauterine growth retardation. However, intrauterine growth is difficult to conceptualise from the birth weight and gestational age at birth and our results on intrauterine growth retardation should therefore be taken with caution.²¹

The diagnosis of HKD follows international guidelines,²² but the validity of HKD in the Psychiatric Case Registry and the prevalence of HKD in Denmark remains unknown. By excluding subjects with comorbidity unrelated to HKD, we examined whether our results could be explained by comorbid disorders. Apparently they were not. However, the ICD-10 has a diagnostic hierarchal system that may lead to an underassessment of comorbid disorders in the register. Information on psychiatric outpatient contacts of parents and siblings before 1995 was unavailable, and an underestimation of confounding by mental illness in the family is also possible. A potential overestimation of the results is also possible, if children born preterm or with low weight were more likely to have been referred to psychiatric departments owing to heightened awareness of their mental health. If the associations we found are true, undiagnosed children in the control group would tend to produce an underestimation of the reported risks.

Comparison with other studies

All previous studies on the association between gestational age and clinically verified ADHD were performed in very small samples of children with extremely low gestational ages at birth (below 28 completed weeks of gestation).^1–4 These studies also report an increased risk of ADHD compared with term controls.^1–4 A meta-analysis on cognitive and behavioural deficits in extremely premature infants support these findings.⁵ However, the majority of preterm babies are born at higher gestational ages, and our findings may therefore have more important public health perspectives.

A serious limitation of previous studies on birth weight and the risk of ADHD is the lack of information on gestational age.^23–30 Some of the studies found an increased risk of ADHD among children with birth weights below 2500 grams,^23–26 but the apparent effect of birth weight was not separated from a potential effect of gestational age. However, other studies found that intrauterine growth may have an effect on long term learning, cognition, and attention.⁶

Possible explanations for the findings

Detailed knowledge of the mechanisms by which the fetal brain is affected by preterm birth and intrauterine growth retardation does not exist.³¹

In humans, repeated incidents of hypoxia and hypotension are common in premature children below 34 completed weeks of gestation.³² Results from animal studies show that fetal hypoxia and hypotension may induce focal injury in the striatal complex of the basal ganglia,³³ with upregulation of the number of dopamine receptors.³⁴ This shared neurochemical abnormality for ADHD and injuries in preterm children is thought to be one of the explanations for an association between preterm delivery and ADHD in human,^35,36 and may also be one of several potential explanations in children with IUGR.

Morbidity and mortality are higher among preterm boys than among girls.³⁷ We were not able to detect any gender difference between the risks of the exposures under study and HKD. This could be because of the very small number of girls. Preliminary results indicate that the central dopamine system matures slower in males than in females, thereby increasing the period of vulnerability in the dopamine transmitter system.³⁸ This may partly explain the higher prevalence of boys with HKD and ADHD.³⁴

Because developing neurones are more vulnerable to cell death during the perinatal period,³⁹ biological and environmental insults associated with preterm birth⁵ may promote some of the anatomical differences found among children with ADHD.³⁷ Unfortunately, we were not able to study these factors in detail.

Animal studies show that undernutrition in critical fetal periods during brain development may have long term effects on the brain, affecting attention, learning, and memory.⁴⁰ Our results support this hypothesis.

Since we have now shown an association between gestational age and birth weight and HKD, it could be interesting to investigate some of the complications more closely related to gestational age and birth weight. This, however, was beyond the scope of this paper. Further studies of mechanisms related to the delivery, causes of preterm delivery, and conditions affecting the child in the perinatal period are therefore needed.

Conclusions

Children born preterm, also close to term, or at 37 or more completed weeks of gestation with low birth weights (1500–2499 g) had an increased risk of clinically verified HKD.