- Top of page
- What this paper adds
Aim The aim of this study was to evaluate the impact of pre- and perinatal factors on the risk of developing attention-deficit–hyperactivity disorder (ADHD).
Method We investigated the medical history of 237 children (206 male; 31 female) from Malmö, Sweden born between 1986 and 1996 and in whom a diagnosis of ADHD (Diagnostic and Statistical Manual of Mental Disorders-IIIR or IV) was subsequently made at the Department of Child and Adolescent Psychiatry, Lund University, and a reference group of 31 775 typically developing children from Malmö using data from the Swedish Medical Birth Register.
Results The results of multiple logistic regression analysis revealed that ADHD was significantly associated with a young maternal age (odds ratio [OR] for 5y increase 0.87; 95% confidence interval [CI] 0.76–0.99), maternal smoking (OR 1.35; 95% CI 1.14–1.60), maternal birthplace in Sweden (OR 2.04; 95% CI 1.45–2.94), and preterm birth <32 weeks (OR 3.05; 95% CI 1.39–6.71), and a male predominance (OR 6.38; 95% CI 4.37–9.32). Apgar scores at 5 minutes below 7 were significantly associated with ADHD in the univariable analysis (OR 2.60; 95% CI 1.15–5.90). The population-attributable fraction of ADHD caused by the perinatal factors studied was estimated to be 2.8%.
Interpretation The results indicate that the studied factors constitute weak risk factors for developing ADHD.
Attention-deficit–hyperactivity disorder (ADHD) is a neurodevelopmental disorder that affects about 4 to 7% of school-age children.1 Although genetic factors are considered to be most important in the aetiology of ADHD, shared and non-shared environmental factors have been estimated to account for between 10% and 25% of the statistical variance in twin ADHD scores.2 For example, many studies have shown that maternal smoking during pregnancy has a significant association with ADHD.3 A case–control study by Mick et al.4 using perinatal data obtained from parent interviews found that low birthweight was associated with a diagnosis of ADHD. Case–control studies by Botting et al.5 using data collected at birth supported this finding. However, in their case–control study with perinatal data collected at birth, O’Callaghan et al.6 could not find any association between low birthweight and ADHD. Similarly, Milberger et al.7 used perinatal data from parent interviews in their case–control study and failed to show an association between low birthweight and ADHD. Whitaker et al.8 performed a follow-up study of children aged 6 years who had a low birthweight and who had undergone cranial ultrasound. Neither low birthweight nor low gestational age increased the risk of ADHD independent of ultrasound findings.
Other possible risk factors for ADHD that have emerged from different studies are toxaemia, eclampsia, poor maternal health, maternal age, duration of labour, fetal stress, and antepartum haemorrhage.9 In a case–control study in 1990 utilizing retrospective data, Barkley et al.10 failed to find an association between ADHD and any pregnancy or birth complications, including low birthweight.
The findings in different studies have been contradictory, and study methodologies vary in controlling for confounding variables, sample sizes, and data gathering.7 Few studies have been made on population-based samples using data gathered at the time of delivery. A study involving all clinical participants in a geographic area and a population-based reference group without a diagnosis may provide information that makes it possible to estimate the relative strength of both different risk and protective factors.
In our study, we examined the Swedish Medical Birth Register (MBR), which contains obstetric data on most children born in Sweden. Obstetric data are recorded in the Swedish MBR by physicians and nurses at the time of delivery.
The aims of this study were to identify relevant pre- and perinatal risk and protective factors for ADHD and to estimate the extent to which pre- and perinatal factors affect the risk of developing ADHD.
- Top of page
- What this paper adds
The basic demographic characteristics of the children with ADHD and reference children are shown in Table I. As evident from the table, fewer children in the group with ADHD than in the reference group were born either early or late within the study period. This fact makes it necessary to consider year of birth when designing the final analysis. The mothers of children with ADHD were significantly younger and were more often smokers than the mothers of children in the reference group. A significant negative association was found between ADHD and maternal age (entered as a linear continuous variable), and a significant positive association was found between ADHD and maternal smoking (entered as a semicontinuous variable; see Method). These estimates did not change more than marginally in the multivariable analysis (Table II). Other factors that were found to be significant independent risk factors for ADHD were gestational age below 32 weeks, Apgar score at 5 minutes below 7, and male sex (Table II). Furthermore, a negative association between ADHD and maternal birth outside Sweden was found. No associations were found between ADHD and twinning, moderate preterm birth (between 32 and 36 gestational weeks), and the index child’s number among siblings.
Table II. Risk or protective factors for having a diagnosis of attention-deficit–hyperactivity disorder. Only risk factors with p values below 0.20 are shown. Estimate and 95% confidence intervals (CIs) obtained by simple and multiple logistic regression analysis respectively, are shown
| ||Univariate estimate||Multiple modela|
|OR||95% CI||OR||95% CI|
|Maternal age (5y increase)||0.86||0.76–0.98||0.87||0.76–0.99|
|Mother born outside Sweden||0.47||0.33–0.67||0.49||0.34–0.69|
|Year of birth|
| 1991–1993||Reference|| ||Reference|| |
|<32 weeks of gestation||3.34||1.56–7.17||3.05||1.39–6.71|
|Apgar score at 5min, 1–6||2.60||1.15–5.90||2.17||0.93–5.06|
In the univariate analysis there was a slight indication of an association between elective caesarean sections or vacuum extraction/forceps delivery and ADHD (p=0.14 and p=0.08 respectively). In the preliminary multivariable analysis (including all variables with p<0.20 in the univariate analysis), there was no sign of any association between delivery mode and ADHD (p=0.23 for elective caesarean section and p=0.24 for vacuum extraction/forceps). Thus, the variables on delivery mode were never added to the final multiple model. A possible association between pre-eclampsia and ADHD was indicated in the univariate analysis (p=0.16). However, in all models, including information on Apgar score and gestational length, no suggestion of any independent association between pre-eclampsia and ADHD was found. Even though second-grade models were tested, it was revealed that in the current data set, the birthweight SD scores variable was best entered to the models as a linear continuous variable. In the univariate analysis, there was an indication of a negative association between increasing SD scores and ADHD (p=0.13). However, when SD scores were included in a multiple model (in which maternal smoking and maternal age were included), the p value increased to 0.5. Thus, SD scores never entered the final multiple model.
In order to evaluate the overall impact of obstetric risk factors on the aetiology of ADHD, birth before 37 weeks of pregnancy, Apgar score at 5 minutes below 7, small for gestational age, and large for gestational age were combined into a designed dichotomous variable. The OR, adjusted for sex, year of birth, maternal age, and maternal smoking, was 1.01 (0.70–1.45; p=0.97). If, instead, the two variables that were found to be independent risk factors for ADHD in the current study (Apgar score below 7 and very preterm birth before 32wks) were combined, the adjusted OR (with 95% CI) was 2.68 (95% CI 1.45–4.97). However, only 11 out of the 237 children (4.6%) with ADHD had any of these risk factors. With the overall risk of 0.7% for ADHD in the current study, the OR of 2.68 corresponds to a number needed to harm of 84, and the population-attributable fraction caused by perinatal factors was estimated to be 2.8%.
- Top of page
- What this paper adds
The following factors were found to have significant associations with a diagnosis of ADHD: young maternal age, maternal smoking during pregnancy, maternal birthplace in Sweden, child born before week 32 of pregnancy, male sex (as expected), and Apgar score below 7 at 5 minutes after delivery.
A population-based study in Malmö found a frequency of ADHD among children aged between 5 years 6 months and 10 years of about 4%.15 In the present study, the frequency of ADHD was about 0.7%. As the problem has been that of underdiagnosing, and as assessment methods have been good, it may be that the children who have been diagnosed have been accurately assessed but that they represent only one-fifth of all the children with ADHD. Children with ADHD not identified should constitute no more than 4 to 5% of the children in the reference group. We had a sex bias, typical of clinical samples, with six times more males than females. In Sweden, the diagnosis of deficits in attention, motor control, and perception has been used; this diagnosis can be translated in DSM terms as indicating children who have both ADHD and developmental coordination disorder.16 About half of the children with ADHD in our study were also diagnosed as having deficits in attention, motor control, and perception.
Both young and old mothers are considered to belong to an obstetric risk group according to clinical experience and research.17 We found a significant statistical association between a clinical childhood diagnosis of ADHD and young age of the mother. We could also detect a trend towards increased risk among the oldest mothers, although this association was not statistically significant.
Several previous studies have identified smoking in pregnancy as a risk factor for the development of ADHD.3 In studies of twins in which adjustments for genetic risk have been made, maternal smoking during pregnancy has been found to have a true environmental effect on ADHD symptoms, but a recent study by Thapar et al.18 involving children born using assisted reproductive technologies found no significant association between ADHD and smoking if the mother and child were genetically unrelated. Our study shows that maternal smoking during pregnancy is a significant risk factor. To determine if this factor represents a true environmental effect or if it is a consequence of the genetic risk factor, further research is needed – either more studies like that of Thapar et al.18 utilizing children born with the aid of assisted reproductive technologies or studies in which comparisons can be made between children with different degrees of heredity for ADHD.
According to Bhutta et al.,19 children of low birthweight have a threefold increased risk of developing ADHD, and Mick et al.4 have proposed that 13.8% of all cases of ADHD can be attributed to low birthweight. We did not find any significant association between birthweight and ADHD, but this may be because the statistical power of the study was too low. The strength of the association between birthweight and ADHD increased when smoking during pregnancy was introduced in the regression model, which is interesting as smoking during pregnancy has a known association with low birthweight.
Preterm birth has been described as a risk factor for the development of ADHD,7 and we found that the prevalence of ADHD was higher among children born before week 32 of gestational age than among term-born children. In other studies, pre-eclampsia has been shown to be a risk factor for ADHD.20 In our sample there was a non-significant tendency for an association between maternal pre-eclampsia and ADHD.
We found that having a mother born in another country was a relative protective factor (OR<1; thus, statistically, a protective factor from developing ADHD). In another study from Malmö concerning perinatal risk and protective factors for developing autism spectrum disorders,21 having a mother born in another country was found to be a risk factor for developing autistic syndrome but a protective factor for developing Asperger syndrome. A possible explanation for the finding that having a mother born in another country is a relative protective factor for Asperger syndrome and ADHD could be that immigrant families are less inclined to seek help from child psychiatry, and as these conditions are milder than autistic syndrome, it may be the case that affected individuals are less likely come into contact with child psychiatry and to be diagnosed. This hypothesis is supported by statistics from the Department of Child and Adolescent Psychiatry in Malmö, which show that the proportion of families seeking help from child psychiatry is lower in the areas of Malmö with a high proportion of immigrants than in those areas with very few immigrant families. This might be explained by cultural factors and difficulties with integration, causing some immigrant families to have less confidence in Swedish child psychiatry services. However, some professionals also believe that ADHD-like behaviour may be explained by social and cultural adjustment problems.
In our study, the population-attributable fraction caused by the perinatal factors studied was estimated to be 2.8%, which is less than estimates made by several other researchers.2
The incidence of birth complications in Europe and the United States is now very low since the dramatic improvements in obstetric care that occurred at the end of the twentieth century; however, it remains high in many developing countries.22 Earlier work, such as the studies by Knobloch and Pasamanick,23 were conducted at a time when concepts such as minimal brain damage still were in use and the incidence of birth complications such as neonatal asphyxia was high. Neonatal encephalopathy following perinatal asphyxia can affect the striatum, which hypothetically might lead to symptoms of ADHD.24 Marlow et al.25 found that moderate but not mild neonatal encephalopathy leads to a significantly increased frequency of hyperactivity in children of school age. In many developing countries the rates of child mortality are still very high, and intrapartum hypoxia and birth asphyxia are widely regarded as major causes of morbidity and mortality in these countries.26 In Europe and the United States, the survival of extremely preterm infants has continued to improve.27 Most recent studies do not indicate that improved survival is associated with increased neonatal morbidity.27 These data suggest that perinatal risk factors associated with the development of ADHD may be of less clinical urgency in countries such as Sweden.
In this study, we had to rely on clinical diagnoses of ADHD, which are not as reliable as research diagnoses. As the sample was large, smaller aberrations caused by individuals not being diagnosed with ADHD would only moderately weaken the statistical associations that we have studied because the prevalence of ADHD among the children in the reference group would be very low (certainly <5%). A limitation of our study is that data on socioeconomic status, family dysfunction, and abuse were not available for analysis. Although the statistical power seems good, it might not be high enough to detect some relevant associations.