To examine the association between maternal hyper- and hypothyroidism and the risk of attention deficit hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) in the child.
To examine the association between maternal hyper- and hypothyroidism and the risk of attention deficit hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) in the child.
A population-based cohort study.
Singletons liveborn in Denmark between 1991 and 2004.
A total of 857 014 singletons alive and living in Denmark at the age of 3 years.
Information on the diagnosis and/or treatment of maternal thyroid disease and the neurodevelopmental disorders ADHD and ASD in the child was obtained from Danish nationwide registers. The Cox proportional hazards model was used to estimate the hazard ratio (HR) with 95% confidence interval (95% CI) for risk of ADHD and ASD in children born to mothers with thyroid dysfunction, adjusting for potential confounding factors.
ADHD and ASD in the child.
Altogether, 30 295 singletons (3.5%) were born to mothers with thyroid dysfunction. Maternal hyperthyroidism diagnosed and treated for the first time after the birth of the child increased the risk of ADHD in the child (adjusted HR 1.23; 95% CI 1.05–1.44), whereas hypothyroidism increased the risk of ASD (adjusted HR 1.34; 95% CI 1.14–1.59). No significant association was seen for maternal diagnosis and treatment prior to the birth of the child.
Children born to mothers diagnosed and treated for the first time for thyroid dysfunction after their birth may have been exposed to abnormal levels of maternal thyroid hormone already present during the pregnancy, and this untreated condition could increase the risk of specific neurodevelopmental disorders in the child.
Neurodevelopmental disorders constitute a broad group of frequent disorders of brain function, with cognitive, mental, behavioural, and physical effects. Developmental defects may occur in up to 10% of children and the aetiology is only partly understood, but both genetics and environment play a role, making these disorders at least partly preventable.
Injury to the developing brain may occur prenatally or in early postnatal life. Brain development begins in utero and involves a complex sequence of events, with some particularly vulnerable periods. It is well established that thyroid hormones are essential during early brain development, and that maternal thyroid hormones are required before the onset of thyroid hormone production by the fetus.[4, 5] Thus, abnormal levels of maternal thyroid hormone during pregnancy may disrupt brain development in the fetus.
Thyroid disorders are common in women of reproductive age.[6, 7] Several studies have examined the association between maternal thyroid dysfunction and neurocognitive development of the child,[8-10] but the results are not consistent, and the number of studies evaluating the risk of specific neurodevelopmental disorders in the child is limited.
Using Danish nationwide registers, we studied the association between maternal hyper- and hypothyroidism and the risk of the neurodevelopmental disorders attention deficit hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) in the child. ADHD is the most common neurobehavioural disorder in childhood: it is characterised by hyperactivity, inattention, and impulsiveness. ASD is characterised by difficulties in social interaction, communication, and/or stereotyped interests/repetitive behaviour.
As thyroid hormones are involved in a number of events during the early development of the brain, we hypothesised that maternal thyroid dysfunction, either hyper- or hypothyroidism, present during the pregnancy could increase the risk of ADHD and ASD in the child via subtle changes in brain structure during early brain development. The diagnosis of thyroid disease before or during a pregnancy is expected to be followed by treatment of the disorder, but several studies have shown a high frequency of inadequate treatment in subsequent pregnancies.[13, 14] On the other hand, thyroid dysfunction may be present for a period of time before a diagnosis is made. Thus, abnormal levels of maternal thyroid hormone might already have been present during the pregnancy in cases when maternal thyroid dysfunction is diagnosed and treated for the first time after the birth. We therefore also studied any association with a diagnosis of thyroid disease in the mother after the birth of the child. To evaluate whether a genetic component exists, we also looked for an association with paternal thyroid disease.
We conducted a population-based cohort study using Danish nationwide registers. All Danish citizens are assigned a unique ten-digit personal identification number, which enables accurate linkage between the nationwide registers. Using the Danish Civil Registration System we identified all liveborn singletons in Denmark delivered between 1 January 1991 and 31 December 2004.
Information on the diagnosis of thyroid disease was obtained from the Danish National Hospital Register (DNHR), which holds data on all admissions to any Danish hospital since 1977, and on all hospital outpatient visits since 1995. For every admission, the register holds the date of admission and discharge and diagnoses classified according to the eighth revision of the International Classification of Diseases (ICD–8) from 1977 to 1993, or the tenth revision (ICD–10) from 1994 onwards. We included all in- and outpatient visits with a main or additional first-time diagnosis of thyroid dysfunction after 1 January 1977 and before 1 January 2009. Danish studies of the age-dependent epidemiology of various types of hyper- and hypothyroidism have shown that women of reproductive age predominantly suffer from autoimmune hypothyroidism and Graves' disease.[6, 7] According to the ICD classification, hyperthyroidism was defined as 242.00–242.29 (ICD–8) and E05.0–E05.9 (ICD–10, excluding thyrotoxicosis factitia [E05.4], the overproduction of thyroid-stimulating hormone [E05.8A], and thyrotoxic heart disease [E05.9A]). Hypothyroidism was defined as 243.99 and 244.00–244.09 (ICD–8, excluding secondary hypothyroidism [244.02]), and E03.0–E03.9 and E89.0 (ICD–10, excluding unspecified congenital goitre [E03.0A] and atrophy of the thyroid [congenital E03.1B; acquired E03.4]).
Information on the prescription of thyroid medication was obtained from the Danish National Prescription Registry (DNPR), which holds data on all prescription drugs dispensed from Danish pharmacies since 1995, including the patient's personal identification number, the type of drug prescribed according to the Anatomical Therapeutic Chemical (ATC) classification system, and the date of sale. Thyroid hormones (ATC H03A) and anti-thyroid medication (ATC H03B) are sold solely as prescription drugs in Denmark, and we identified prescriptions of thyroid medication dispensed between 1 January 1995 and 31 December 2008.
The time of the identification of maternal thyroid dysfunction was defined as the day the first prescription for thyroid medication was dispensed, and was categorised as before or after the birth of the child. If a hospital diagnosis was registered before 1 January 1996, the day of first admission to hospital defined the time of diagnosis of disease. Hyperthyroidism was defined by: (1) a first hospital diagnosis of hyperthyroidism and at least one prescription of antithyroid medication; or (2) if no hospital diagnosis was registered, at least two prescriptions of antithyroid medication. Hypothyroidism was defined by: (1) a first hospital diagnosis of hypothyroidism, at least one prescription of thyroid hormones, and no prescription of antithyroid medication; or (2) if no hospital diagnosis was registered, at least two prescriptions of thyroid hormones and no prescription of antithyroid medication. Singletons born to mothers with inconsistent registration of thyroid dysfunction were excluded from the analyses (Figure 1).
We used three sources to ascertain whether a child developed ADHD during follow-up: hospital diagnosis in the DNHR, diagnosis in the Danish Psychiatric Central Register (DPCR), and dispensed prescription of ADHD medication in the DNPR. ADHD was defined by F90 (ICD–10) and ADHD medication was defined by ATC ‘N06BA’. If no diagnosis was registered, ADHD was defined by at least two dispensed prescriptions of ADHD medication. Time of first diagnosis or first dispensed prescription defined the onset of disease, whichever came first. Singletons with inconsistent registration of ADHD, including diagnosis and/or dispensed prescription before the age of 3 years only, were excluded from the analyses (Figure 1).
We used two sources to ascertain if a child developed ASD during follow-up: hospital diagnosis in the DNHR and a diagnosis in the DPCR. ASD was defined by F84 (ICD–10). Time of first diagnosis defined the onset of disease. Singletons with a diagnosis of ASD before the age of 3 years only were excluded from analyses (Figure 1).
From the Danish Medical Birth Registry we obtained information on maternal age and parity at the time of a child's birth, gender and birthweight of the child, 5-min Apgar score, and gestational age at delivery.
From Statistic Denmark we obtained information on maternal cohabitation, income, origin, and geographical residence at the time of the child's birth. For maternal marriage and origin we replaced missing values by available information in the preceding or following 5 years (origin) or 3 years (marriage), whichever came first. Iodine is essential for the synthesis of thyroid hormone, and in Denmark regional differences in iodine intake exist because of different levels of iodine in the drinking water: levels are higher in east Denmark than in west Denmark (divided by the Great Belt).
In the DNHR and the DPCR we ascertained whether the mother had a diagnosis of psychiatric disease (ICD–8, 290–315; ICD–10, F00–F99), and from the DNHR we obtained information on maternal diagnosis of pre-eclampsia/eclampsia (ICD–8, 637.03–637.19; ICD–10, O14.0–O15.0) and diabetes (ICD–8, 249.00–250.09; ICD–10, E10.0–E14.9 and O24.0–O24.9). Information on maternal smoking during pregnancy was available in the DNHR from 1996.
The children were followed from the age of 3 years to the onset of disease, emigration, death, or 31 December 2008, whichever came first. The Cox proportional hazards model was used to estimate the hazard ratio (HR) with 95% confidence interval (95% CI) for risk of ADHD and ASD in children born to mothers with hyper- or hypothyroidism, compared with unexposed children. Univariate and multivariate analyses were performed, and significant associations in the main analyses were further stratified by time of maternal diagnosis before or after the birth of the child. The Cox proportional hazards assumption was tested in plots of log cumulative hazards, which showed parallel curves. Robust standard errors were used to adjust for dependence between multiple pregnancies.
Parallel analyses were performed for singletons exposed to paternal thyroid disease. Birthweight, gestational age, and Apgar score, as well as maternal psychiatric disease, pre-eclampsia/eclampsia, and diabetes were considered possible intermediates, and hence were not included in the main model. In sensitivity analyses these variables were investigated, and the impact of maternal smoking was ascertained in the group of singletons born between 1996 and 2004. Finally, analyses were restricted to first-born children and to singletons whose mother was born in Denmark.
Statistical analyses were performed using stata 11 (Stata Corp., College Station, TX, USA) and a 5% level of significance was chosen.
Among singletons born in Denmark between 1991 and 2004, and included in the final study population, 3.5% were born to mothers with a diagnosis of thyroid dysfunction and/or a prescription for thyroid medication before 1 January 2009. Table 1 presents characteristics of the mothers and the children at the time of the child's birth.
|No thyroid dysfunctiona||Hyperthyroidismb||Hypothyroidismb|
|Singletons||826 719||96.5||12 284||1.4||18 011||2.1|
|<25 years||139 387||16.8||1709||13.9||2264||12.6|
|25–29 years||317 364||38.4||4382||35.7||6256||34.7|
|30–34 years||262 757||31.8||4081||33.2||6258||34.7|
|≥35 years||107 211||13.0||2112||17.2||3233||18.0|
|Married||483 778||58.5||7405||60.3||11 586||64.3|
|Not married||342 941||41.5||4879||39.7||6425||35.7|
|First quartile (lowest)||34 149||4.1||555||4.5||872||4.8|
|Second quartile||239 869||29.0||3618||29.4||5002||27.8|
|Third quartile||421 800||51.0||6274||51.1||9262||51.4|
|Fourth quartile||130 901||15.9||1837||15.0||2875||16.0|
|Born in Denmark||741 726||89.7||10 740||87.4||15 467||85.9|
|Not born in Denmark||84 993||10.3||1544||12.6||2544||14.1|
|West Denmark||461 026||55.8||7172||58.4||9114||50.6|
|East Denmark||365 693||44.2||5112||41.6||8897||49.4|
|7–10||811 346||99.1||12 072||99.1||17 663||99.0|
|<37 weeks||37 439||4.6||676||5.5||936||5.2|
|37–41 weeks||712 240||86.9||10 690||87.0||15 583||86.5|
|≥42 weeks||70 004||8.5||917||7.5||1489||8.3|
|Mean (SD)||3540 (562)||3490 (578)||3580 (581)|
Figure 2 illustrates the percentage of children in the study population diagnosed with ADHD and ASD according to the year of the birth of the child. The occurrence was only illustrated for part of the study period and for the onset of disease before or at the age of 10 years. This restriction was made to obtain similar follow-up periods and comparable frequencies of disease by birth year. Notably, the occurrence of ADHD increased considerably during the study period.
In the follow-up studies, the children were followed up to the age of 18 years for onset of ADHD and ASD. Altogether, 11 351 children developed ADHD and 5311 children were diagnosed with ASD. Table 2 presents the number of children developing ADHD and ASD during follow-up among exposed children. In comparison with unexposed children, adjusted analyses showed a significantly increased risk of ADHD in children born to mothers with hyperthyroidism, and a significantly increased risk of ASD in children born to mothers with hypothyroidism. There were also trends suggesting an association between hyperthyroidism and ASD and hypothyroidism and ADHD, but these did not reach statistical significance.
|Singletons (n)||Crude HR (95% CI)||Adjusted HR (95% CI)a|
|No hyper- or hypothyroidismc||10 893||1.00 (reference)||1.00 (reference)|
|Hyper- or hypothyroidismd||458||1.04 (0.94–1.14)||1.14 (1.03–1.25)|
|Hyperthyroidism||194||1.08 (0.94–1.25)||1.18 (1.03–1.36)|
|Hypothyroidism||264||1.01 (0.89–1.14)||1.10 (0.98–1.25)|
|No hyper- or hypothyroidismc||5065||1.00 (reference)||1.00 (reference)|
|Hyper- or hypothyroidismd||246||1.24 (1.09–1.41)||1.25 (1.10–1.42)|
|Hyperthyroidism||91||1.13 (0.91–1.39)||1.18 (0.96–1.45)|
|Hypothyroidism||155||1.31 (1.12–1.54)||1.30 (1.11–1.53)|
The majority of exposed children were born to mothers diagnosed with thyroid dysfunction and treated for the first time after the birth of the child (78.2%). The associations between maternal hyperthyroidism and ADHD and maternal hypothyroidism and ASD found in the main analyses were subsequently evaluated when stratified by time of maternal diagnosis in relation to the birth of the child. In the stratified analyses, only maternal diagnosis after the birth of the child revealed an increased risk (Figures 3 and 4).
The median time from the birth of the child to the diagnosis of and treatment for maternal thyroid dysfunction was 6.5 years (hyperthyroidism 6.1 years; hypothyroidism 6.8 years). The number of exposed cases was relatively small when analyses were further stratified by time interval from the birth of the child to maternal diagnosis (≤2 years or >2 years); however, the risk of ADHD in the child seemed to be particularly high when the mother was diagnosed with hyperthyroidism within 2 years of the birth (Figure 3). For ASD, no particular difference was observed according to time of maternal diagnosis of hypothyroidism after the birth (Figure 4), but the number of exposed cases was smaller and the confidence intervals were consequently wider.
The association with maternal thyroid dysfunction diagnosed and treated after the birth of the child could have a genetic component. If so, we would also expect an association for children who had a father with thyroid dysfunction, and in our study population we identified 2338 children whose father had hyperthyroidism and 3072 children whose father had hypothyroidism. In parallel analyses, we found no significant association between paternal thyroid dysfunction and the risk of ADHD (hyperthyroidism, adjusted HR 1.21 [95% CI 0.87–1.68]; hypothyroidism, adjusted HR 1.16 [95% CI 0.87–1.55]) or ASD (hyperthyroidism, adjusted HR 1.25 [95% CI 0.78–1.98]; hypothyroidism, adjusted HR 0.90 [95% CI 0.57–1.44]) in the child. Additional adjustment for paternal age at the time of the child's birth and paternal diagnosis of psychiatric disease did not change the results.
As the results could be confounded by maternal smoking, we additionally adjusted for maternal smoking during pregnancy for the time period when smoking was registered. After such adjustment, the association between maternal hypothyroidism and ASD was unaltered, whereas for hyperthyroidism and ADHD the association weakened, but still showed an increased risk in children born to mothers diagnosed and treated within 2 years of the birth of the child (adjusted HR 1.72 [95% CI 1.13–2.63]).
In a number of sensitivity analyses, the impact of possible intermediates not included in the main model was investigated. Overall, results were consistent, irrespective of additional adjustments for maternal diagnosis of psychiatric disease and/or prescription of ADHD medication, maternal diabetes, maternal pre-eclampsia/eclampsia, birthweight of the child (<2500 or ≥2500 g), gestational age at birth (<37, 37–41, or ≥42 weeks), and 5-min Apgar score (<7 or ≥7). A small group of children had a diagnosis of thyroid disease (n = 777), and the results were unaltered after the exclusion of these children. Restricting analyses to the first-born child and to children whose mothers were born in Denmark also revealed similar associations (data not shown).
Considering the definition of ADHD in the child, the association with maternal hyperthyroidism was very similar when ADHD was defined by either medical treatment alone or by both medical treatment and hospital diagnosis (data not shown).
Our study suggests an association between maternal thyroid dysfunction diagnosed and treated after the birth of a child and the neurodevelopmental disorders ADHD and ASD in that child. As we found no association with paternal thyroid disease, nor an association with maternal disease diagnosed before the birth of the child, and thus treated, these associations are probably not caused by a genetic factor. Symptoms are often absent or are uncharacteristic, even in the presence of biochemically overt thyroid disease, and thyroid dysfunction may remain stable for a long period of time. We previously reported an increased risk of preterm birth and birthweight not appropriate for gestational age when the mother was diagnosed with thyroid dysfunction before, during, or after the pregnancy. Our findings in the present study support the hypothesis that untreated maternal thyroid dysfunction already present during the pregnancy and subsequently diagnosed affects fetal brain development.
To study the association with undiagnosed thyroid disease we had to ‘condition upon the future’, which may cause bias because the mother must survive for a certain time period after the pregnancy for the disease to be detected. Mortality is low in this age group in Denmark, however, even among women with thyroid disease, and we consider this bias to be small.
Our study was population based with almost complete follow-up, which makes selection bias unlikely. The number of exposed cases in the stratified analyses was limited, however, and we cannot exclude that the lack of an association with paternal thyroid disease and with maternal thyroid disease diagnosed before the birth of the child resulted from a lack of power, as the majority of singletons were born to mothers diagnosed after the birth. ADHD was more common than ASD, and consequently the results were more robust for ADHD. The time-dependent dynamics in the HRs for both hyper- and hypothyroidism diagnosed before and after the birth of the child contradict type 1 error.
We did not have information on maternal thyroid hormone levels or on the presence of thyroid autoantibodies, but we were able to identify maternal thyroid disease and outcome diseases by both hospital diagnosis and medical treatment of disease. The validity of a diagnosis of thyroid disease in the DNHR only revealed misclassification in less than 2% of the cases in the general population. Also, a diagnosis of childhood ASD in the DPCR was confirmed in 94% of cases. By using the prescription data we also included children treated for ADHD outside the hospital system. Besides ADHD, this medication is only used for the treatment of the rare condition narcolepsy, and we found similar associations when ADHD was defined by medical treatment alone.
We obtained information on a number of possible confounding factors, but unmeasured or residual confounding might still exist.
Thyroid hormones play a crucial role in the development of the fetal brain, and abnormal levels of maternal thyroid hormone during pregnancy may lead to structural and functional changes in the developing brain. Maternal hypothyroidism may lead to reduced myelination, faulty differentiation and migration of neurons as well as altered dendritic structure and synaptogenesis in the developing fetal brain. But an excess of thyroid hormones can also disrupt fetal brain development, and maternal hyperthyroidism may lead to the compromised expression of neuronal cytoskeletal proteins and the inhibition of the proliferation of neural stem/progenitor cells.[27, 28]
Attention deficit hyperactivity disorder is one of the most common disorders of childhood, and is showing an increasing trend. It is characterised by difficulties with concentration and attention, in controlling behaviour, and hyperactivity. Structural imaging studies have shown a delay in brain maturation, with reduced grey and white matter and cortical thickness; functional imaging studies have focused on the neurochemical environment in the prefrontal cortex, which regulates attention, behaviour, and executive function through complex networks with other brain regions. Small changes in the levels of the catecholamines norepinephrine and dopamine, and reduced glucose metabolism in various regions of the brain, may play a role in the pathophysiology of ADHD.
In our study population, the increased risk of ADHD was predominant among children born to mothers diagnosed with hyperthyroidism for the first time within 2 years of the birth of the child. Hyperthyroidism in young women typically occurs in Graves' disease, as a result of autoimmunity to thyroid-stimulating hormone receptors. Such autoimmunity may fluctuate over time, and would often be present for a period of time before the hyperthyroidism approaches clinical detectability. Following this line of thought, we might speculate whether subtle changes in brain structure and function secondary to elevated levels of maternal thyroid hormone during the pregnancy may coincide with the neurodevelopmental abnormalities in ADHD. Notably, thyroid hormones interact with the catecholaminergic system in the brain, and positron emission tomography studies in adults have demonstrated regional decreased cerebral glucose metabolism in the hyperthyroid state.
Attention deficit hyperactivity disorder (ADHD) has been strongly associated with a mutation in the thyroid receptor β–gene leading to resistance to thyroid hormone and a compensatory increase in thyroid hormone levels. The thyroid receptor α in the brain is thus exposed to excessive thyroid hormone, which is a probable cause of ADHD in these children.
A few studies specifically addressed the association between maternal thyroid dysfunction and the risk of ADHD in the child. In these studies, an association with maternal hypothyroxinaemia (induced by iodine deficiency) in early pregnancy and elevated titres of thyroid peroxidase antibodies were reported.[36, 37]
Autism spectrum disorder is a neurodevelopmental disorder characterised by difficulties with social relationships and communication, and stereotyped behaviour. A number of associated abnormalities in brain structure have been shown, including increased brain size, with disproportionately larger white matter, impaired neuronal migration, and abnormalities in synaptic structure and function. It has been proposed that maternal hypothyroxinaemia during pregnancy may increase the risk of autism in the child. Thus, similarities may exist between the changes in brain structure caused by low levels of maternal thyroid hormones during early brain development and the neurodevelopmental abnormalities in ASD. Impaired regulation of cell migration might coincide with this, and alterations in the neuronal expression of the glycoprotein Reelin have been demonstrated in both experimental maternal hypothyroidism and in ASD.[40, 41]
Hypothyroidism in women of reproductive age is often autoimmune in origin, and the increased risk of ASD could be caused by low levels of thyroid hormones, autoimmune mechanisms, or both. Autoimmunity has been proposed to play a role in the development of ASD, but studies that considered thyroid autoimmunity did not report an increased risk.[43, 44]
It has been much discussed whether pregnant women should be screened for abnormalities in thyroid function, but no consensus has been reached. A major issue is the lack of evidence for the treatment of small aberrations in thyroid function in early pregnancy, which might be secondary to placental dysfunction. The present study supports the importance of identifying and treating regular thyroid disease in pregnancy, as previously discussed.
Maternal thyroid dysfunction is a neurodevelopmental risk that clinicians should be aware of. The untreated condition may increase the risk of ADHD and ASD in the child, but additional studies are needed to corroborate the results of the present study. The study further supports the importance of carefully evaluating pregnant women for thyroid disease, but gives no indication of the possible benefit of treating small aberrations in thyroid function in early pregnancy.
The authors declare that they have no disclosures to report.
S.L.A. performed the statistical analyses and drafted the first version of the article. P.L., C.S.W., and J.O. contributed to the idea for the study, the analytical strategy, and the interpretation of the results. All authors revised and approved the final version.
The study was approved by the Danish Data Protection Agency.
C.S.W. is supported by individual postdoctoral grants from the Danish Medical Research Council (FSS: 12-32232).
Andersen SL, Laurberg P, Wu CS, Olsenb J. Attention deficit hyperactivity disorder and autism spectrum disorder in children born to mothers with thyroid dysfunction: a Danish nationwide cohort study BJOG 2014;121:1365–74.
A pregnant women with hypothyroidism is concerned after reading about the association between thyroid disease and autism. She asks: ‘Is my thyroid problem going to affect my baby?’
|Participants||Singletons born in Denmark between 1991 and 2004, and alive and living in Denmark at the age of 3 years|
|Intervention||Maternal thyroid dysfunction (hyper- or hypothyroidism), registered before or after childbirth|
|Comparison||Children born to mothers without registered diagnosis of thyroid dysfunction in pregnancy|
|Outcomes||Diagnosis of autism spectrum disorder (ASD), diagnosis of ADHD, and/or prescription of ADHD medication|
|Study design||Population-based cohort study|
Women's Health Research Unit, Queen Mary, University of London, London, UK
Join us at #BlueJC: Follow @BJOGTweets to stay updated on #BlueJC sessions or email firstname.lastname@example.org to host a journal club on Twitter. Find out more on our journal club page by visiting bjog.org