OBJECTIVES: The goals were to study the association between neonatal jaundice and disorders of psychological development in a national, population-based cohort and to study whether gestational age, parity, and season of birth influenced that association.
METHODS: A population-based, follow-up study of all children born alive in Denmark between 1994 and 2004 (N = 733 826) was performed, with data collected from 4 national registers. Survival analysis was used to calculate hazard ratios (HRs).
RESULTS: Exposure to jaundice in neonates was associated with increased risk of disorders of psychological development for children born at term. The excess risk of developing a disorder in the spectrum of psychological development disorders after exposure to jaundice as a neonate was between 56% (HR: 1.56 [95% confidence interval [CI]: 1.05–2.30]) and 88% (HR: 1.88 [95% CI: 1.17–3.02]). The excess risk of infantile autism was 67% (HR: 1.67 [95% CI: 1.03–2.71]). This risk for infantile autism was higher if the child was conceived by a parous woman (HR: 2.71 [95% CI: 1.57–4.66]) or was born between October and March (HR: 2.21 [95% CI: 1.24–3.94]). The risk for infantile autism disappeared if the child was conceived by a primiparous woman (HR: 0.58 [95% CI: 0.18–1.83]) or was born between April and September (HR: 1.02 [95% CI: 0.41–2.50]). Similar risk patterns were found for the whole spectrum of autistic disorders.
CONCLUSIONS: Neonatal jaundice in children born at term is associated with disorders of psychological development. Parity and season of birth seem to play important roles.
WHAT'S KNOWN ON THIS SUBJECT:
Little and inconclusive evidence has been published regarding the association between neonatal jaundice and autistic disorders.
WHAT THIS STUDY ADDS:
With this study, the authors confirm a positive association between neonatal jaundice and autistic disorders, as well as possibly disorders of psychological development in general. Gestational age, parity, and season of birth play important roles in this association.
Neonatal jaundice usually is a result of elevated bilirubin production caused by increased breakdown of fetal erythrocytes and a low hepatic excretory capacity resulting from general immaturity of the liver. Visual jaundice is seen for 60% of term neonates and 80% of preterm neonates.1,2 For most neonates, this is a normal transitional phenomenon that resolves itself within the first week of life.2 However, it is the most common condition requiring medical attention in newborns and is the main reason for readmission in the neonatal period.3,–,5
Exposure to high serum bilirubin levels, hyperbilirubinemia, is of concern because unconjugated bilirubin is neurotoxic and can cause death in newborns or lifelong sequelae, such as mental retardation, cerebral palsy, and kernicterus, in those who survive.6 It was suggested recently that exposure to moderate serum bilirubin levels is associated with impaired child development, especially autistic disorders classified as pervasive developmental disorders.7 We reported previously on a positive association between neonatal jaundice and autism,8 as did Jangaard et al9 and Buchmayer et al.10 Croen et al11 found no significant association, but they included relatively few cases with moderate/high levels of bilirubin in their study. Moreover, the study was conducted in a restricted geographical area.11
The cause of autistic disorders is not known. Genetic factors play a role but prenatal exposure to thalidomide, valproic acid, or viral infections also may play a role.12,13 Season of birth and parity are environmental factors that often are studied in autism research and also are relevant in studies of neonatal jaundice. Exposure to daylight is a contributing factor in decreasing bilirubin levels, and children born to primiparous women experience different exposure to antibodies in fetal life than do children born to multiparous women. In a previous case-control study, we found an almost fourfold higher risk for infantile autism (suspected to be limited to term-born children) if the child developed hyperbilirubinemia in the neonatal period.8 In the present study, we investigated (1) the association between neonatal jaundice and autistic disorders, and other disorders of psychological development, in a national cohort and (2) the role of gestational age, parity, and season of birth in autistic disorders.
This study was designed as a national, population-based, follow-up study of all children born in Denmark between January 1, 1994, and December 31, 2004. The cohort was established from the Danish Medical Birth Register (DMBR),14 which includes information on >99% of all deliveries in Denmark. The data stored in the register have been validated and are considered reliable.15 From the DMBR, we obtained information on parent's age, mother's citizenship, maternal smoking in early pregnancy, birth weight (in grams), gestational age, irregular fetal presentation, congenital malformations, and Apgar score. Additional data on exposures and outcomes were obtained from the Danish National Hospital Register (DNHR),16 the Danish Psychiatric Central Register (DPCR)17 (through the DNHR), and the Cause of Death Register.
Exposure and Outcomes
We determined jaundice exposure status and diagnoses of disorders of psychological development on the basis of the International Classification of Diseases, 10th Revision (ICD-10).7 Exposure status (neonatal jaundice) was defined on the basis of ICD-10 codes P57.0 to P59.9, and data were retrieved from the DNHR. Neonatal jaundice caused by hemolytic diseases was not included (ICD-10 codes P55–P56). No information on phototherapy or serum bilirubin levels is available in the register. The outcome diagnoses (disorders of psychological development) were defined on the basis of ICD-10 codes F80.0 to F89.9, and data were retrieved from the DPCR through the DNHR.
The outcome diagnoses were divided into the following subgroups: specific developmental disorders of speech and language (ICD-10 codes F80–F80.9), specific developmental disorders of scholastic skills (ICD-10 codes F81–F81.9), specific developmental disorders of motor function (ICD-10 codes F82–F82.9), mixed specific developmental disorders (ICD-10 codes F83–F83.9), pervasive developmental disorders (also referred to as autistic disorders; ICD-10 codes F84–F84.9), and other and unspecified disorders of psychological development (ICD-10 codes F88–F89.9). The subgroup of pervasive developmental disorders was further restricted to infantile autism (ICD-10 code F84.0). Information on mental retardation (ICD-10 codes F70–F79.9) also was retrieved. The outcome diagnoses were obtained from the DPCR, which was established in 1938 and computerized in 1969 to include all inpatient admissions to psychiatric hospitals and wards. Since January 1, 1995, information from all contacts with psychiatric outpatient clinics has been included. In Denmark, children are referred to a specialist in child psychiatry by general practitioners, schools, or psychologists if mental impairment problems are suspected. Only child psychiatrists assign a diagnostic code, and all diagnoses are recorded in the DPCR. Denmark provides universal tax-paid health care coverage, and no private psychiatric hospitals exist; the inpatient and outpatient contacts recorded in the DPCR are thought to capture the vast majority of children with severely impaired development. Information on children's vital status and time of death was obtained from the Cause of Death Register.
The information from the different registers was linked through the unique identification numbers that are assigned to all live-born children and new residents in Denmark and are used whenever a person is in contact with the Danish health care system. This number allows for the accurate linkage of information between registries at the individual level.
We estimated the relative risks of neonatal jaundice and disorders of psychological development as hazard ratios (HRs) with 95% confidence intervals (CIs), with Cox regression models with right censoring. For all children, the follow-up time began at birth and continued until the first occurrence of 1 of the following events: date of relevant diagnosis, time of death, or end of the follow-up period on April 30, 2008. No attempts were made to adjust for emigration, which has a low rate of <1% in Denmark.18 Both crude and adjusted HRs were computed. The potential confounders included in the analysis were selected from the already known risk factors for jaundice or disorders of psychological development and included mother's smoking status, irregular fetal presentation, gender, birth weight, gestational age, Apgar score, parents' ages, mother's citizenship, and congenital malformations.
In the analysis, we considered all disorders of psychological development (ICD-10 codes F80–F89.9), followed by the analysis of 6 diagnostic subgroups (ICD-10 codes F80–F80.9, F81–F81.9, F82–F82.9, F83–F83.9, F84–F84.9, and F88–F88.9). In all of these analyses, we obtained separate estimates of the associations between neonatal jaundice and the outcomes for term and preterm births.
For term-born children, further analyses (crude and adjusted) were made for autistic disorders and the restricted subgroup of infantile autism. These analyses included stratification according to the gender of the child, the season of birth (April to September versus October to March), parity (primiparous versus parous women), and mental retardation (children with or without a diagnosis of mental retardation). To clarify the importance of gestational age, we also performed a series of supplementary analyses with 10 gestational age categories. In these analyses, the outcome was the main group of psychological development disorders and the 2 subgroups of autistic disorders and infantile autism.
Likelihood ratio tests were used to test for effect measure modification by gestational age (term versus preterm) in the entire cohort with a full range of diagnoses and within the diagnostic subgroups. Similarly, in the analysis of term-born children, we assessed whether the associations between neonatal jaundice and outcomes depended on the gender of the child, the season of birth, parity, and mental retardation. We used Stata 9 (Stata Corp, College Station, TX) for data analysis. The study was approved by the Danish Data Protection Agency.
A total of 733 826 children were included in the cohort. There were 35 766 children (4.9%) with a diagnosis of neonatal jaundice in the register. Of those cases, 0.09% were ascribed to hemolytic disease and the remaining were unspecified. A total of 1721 children received a diagnosis in the spectrum of disorders of psychological development during the follow-up period, and 4257 children died.
Crude and Adjusted Analyses
Table 1 shows the distribution of the cohort according to exposure status, gender, birth weight, gestational age, parents' ages, and malformations. Larger proportions of children exposed to jaundice in the neonatal period were seen for boys, children born preterm, children with low birth weights, and children with malformations. Table 2 presents the association between exposure to neonatal jaundice and the risk of receiving a diagnosis within the spectrum of disorders of psychological development. Data were stratified for term (≥37 weeks of gestation) versus preterm (<37 weeks of gestation) birth. Overall, children exposed to jaundice had an almost 90% greater risk of diagnosis of a psychological development disorder, compared with unexposed children. With adjustment, the association between neonatal jaundice and disorders of psychological development was attenuated but remained statistically significant (HR: 1.29 [95% CI: 1.06–1.56]). With stratification for gestational age, the association was found only for term-born children (tests for no interaction were statistically significant only for the main group [ICD-10 codes F80–F89.9]; P = .001). Term-born children exposed to jaundice as neonates had 56% to 88% greater risks of a disorder within the spectrum of psychological development disorders, compared with children not exposed to jaundice as neonates (Table 2).
To elucidate further the role of gestational age, a stratified analysis was made for each week from 33 to 42 weeks of gestation. The risk of receiving a diagnosis within the spectrum of psychological development disorders (ICD-10 codes F80–F89.9) was higher for children born at 40 to 42 weeks of gestation. Similar patterns were found for the 2 autistic subgroups (ICD-10 codes F84–F84.9 and F84.0) (Fig 1).
Stratified Analyses for Term-Born Children
Table 3 shows the association between neonatal jaundice among term-born children and the 2 autistic subgroups, presented in separate categories for gender, parity, season of birth, and diagnosis of mental retardation. No significantly elevated risks for autistic disorders were found for children with or without a diagnosis of mental retardation. The risk of an autism diagnosis increased if the child was born between October and March and disappeared if the child was born between April and September. However, we found no statistically significant interaction. Similarly, we found an association between jaundice and autistic disorders if the child was born of a parous woman, and the risk disappeared if the child was born of a primiparous woman. For parity, the interaction was statistically significant for pervasive developmental disorders (P = .02) and infantile autism (P = .01).
In this study, we included all children born in Denmark in 1994–2004; therefore, the study is unlikely to be affected by selection bias. According to Statistics Denmark, the Danish population is characterized as being stable, with little migration. Data for the cases were obtained from the DPCR through the DNHR. The completeness and validity of the diagnostic data for autistic disorders obtained from the DPCR seems to be high,19 but we did not validate the other diagnostic codes in the spectrum of psychological development disorders. Exposure data and information on confounders were obtained from the DNHR and the DMBR, respectively. The DMBR contains data on >99% of all births in Denmark, and the information stored in the register is considered reliable.15 Data on exposure status with respect to neonatal jaundice were obtained from the DNHR, which means that we captured only the most-severe cases. Our data on exposure are most likely affected by nondifferential exposure misclassification, which may make our result underestimated. Potential confounders were selected on the basis of the literature for already known confounders for jaundice, autistic disorders, and developmental disorders.
An increased risk of psychological development disorders, including autistic disorders, was observed for children exposed to jaundice as neonates. In the stratified analysis, the association between neonatal jaundice and disorders of psychological development was limited to term-born infants. This is in accordance with observations made in our previous study of hyperbilirubinemia and infantile autism8 and in a recently published Swedish study.10 The difference in risk for term and preterm children might suggest that brain development undergoes a sensitive period with special vulnerability to bilirubin exposure at ∼40 weeks of gestation. The difference also might be explained by confounding by treatment; preterm infants are hospitalized and receive routine treatment as soon as they are exposed to bilirubin. Unfortunately, in the Danish national registers, there is no information on treatment with phototherapy or on measured serum bilirubin levels. Therefore, we were unable to calculate hyperbilirubinemia risk estimates and dose-response estimates in this study.
In addition, we found that children born at term from parous women had a greater risk of autism than did children born at term from primiparous women. This may indicate that parous women have accumulated higher levels of antibodies during more pregnancies and that hemolytic disease rarely affects firstborn children.20 It also may reflect the differences of access to health care in the first days after delivery. In Denmark, parous women with healthy term newborns are normally discharged from the hospital shortly after delivery. Children born of primiparous women are discharged 3 to 4 days after delivery, which makes it more likely that firstborn children with high serum bilirubin levels would receive diagnoses before leaving the hospital. Recently, a Danish study found that the incidence of hyperbilirubinemia was constant for children in hospitals, whereas the incidence of hyperbilirubinemia among children admitted from home had increased threefold.21
The seasonal variation of birth has been studied in relation to neuropsychiatric disorders, but the results have been inconclusive.22,23 Exposure to daylight is a contributing factor in decreasing the levels of bilirubin.24 It is possible that children born in months with less exposure to daylight have prolonged exposure to bilirubin. Therefore, we stratified our analyses for autistic disorders and infantile autism in term-born children into a 6-month summer period and a 6-month winter period, and we found that the risk was higher in the winter period. These results may reflect the different exposures to daylight but also the fact that children born in the winter period are more exposed to other contributing agents, such as infections. In a previous study, we found no association between prenatal fever or infections and infantile autism.25 A recent meta-analysis reviewing data on obstetric risk factors and autism found inconclusive results for prenatal infections and autism.26 However, Rosen et al27 found a modestly elevated risk of infection in the first 30 days of life for children with autistic disorders, and recent animal studies indicated that viral infections in mice were associated with changes in brain development in the offspring that bore a resemblance to changes seen for children with schizophrenia and autism.28,29 Finally, prenatal and postnatal vitamin D deficiency has been associated with impaired brain development.30,31 It is known that prenatal vitamins containing 400 IU of vitamin D3 have little effect on 25-hydroxyvitamin D concentrations, especially during the winter months,32 and populations with high skin pigmentation and low sun exposure have a profoundly higher prevalence of vitamin D deficiency.30 Several studies of autistic disorders found higher risks among children of immigrants.25,26,33
Accumulating evidence suggests an association between exposure to neonatal jaundice and autistic disorders, as well as perhaps other disorders of psychological development. Gestational age, parity, and season of birth seem to play important roles in this association. Additional evidence to distinguish the genetic and environmental components is needed to explain the association between neonatal jaundice and autistic disorders.
This study was supported by the University of Aarhus Research Foundation and the Augustinus Foundation.
- Accepted July 22, 2010.
- Address correspondence to Rikke Damkjær Maimburg, MPH, PhD, Aarhus University, School of Public Health, Department of Epidemiology, Bartholins Allé 2, 8000 Aarhus C, Denmark. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- HR =
- hazard ratio •
- CI =
- confidence interval •
- DMBR =
- Danish Medical Birth Register •
- DNHR =
- Danish National Hospital Register •
- ICD-10 =
- International Classification of Diseases, 10th Revision •
- DPCR =
- Danish Psychiatric Central Register
- Escobar GJ,
- Greene JD,
- Hulac P,
- et al
- Oddie SJ,
- Hammal D,
- Richmond S,
- Parker L
- 7.↵World Health Organization. International Classification of Diseases. 10th ed. Geneva, Switzerland: World Health Organization; 1993
- Jangaard KA,
- Fell DB,
- Dodds L,
- Allen AC
- Buchmayer S,
- Johansson S,
- Johansson A,
- Hultman CM,
- Sparén P,
- Cnattingius S
- Croen LA,
- Yoshida CK,
- Odouli R,
- Newman TB
- Gardener H,
- Spiegelman D,
- Buka SL
- Rosen NJ,
- Yoshida CK,
- Croen LA
- Levenson CW,
- Figueiroa SM
- Copyright © 2010 by the American Academy of Pediatrics