Keywords
attention-deficit hyperactivity disorder - ADHD - hypothyroidism neurodevelopment
- pregnancy - preterm birth - thyroid
Attention-deficit hyperactivity disorder (ADHD), a neurodevelopmental disability affecting
9.4% of children 2 to 17 years of age, is characterized by difficulties in executive
function, inattention, hyperactivity, and impulsivity.[1] It is the most widespread behavior disorder of children and it increases the risk
of other life-long problems such as substance abuse,[2] relationship problems,[3]
[4] poorer job[5] and academic performance,[6] riskier driving,[5] criminality,[7] and suicide rates.[8] It is estimated to cost $143 to 266 billion in lost wages, increased health, and
educational costs.[9]
Many of the genes implicated in childhood ADHD including: dopamine receptor 4,[10] monoamine oxidase,[11] dopamine β hydroxylase,[12] DOPA decarboxylase,[12] adiponectin,[13] and brain-derived neurotrophic factor[14] are regulated by the thyroid hormones that are critical for various processes in
neurodevelopment such as stem cell differentiation, neuron growth, synapse formation,
and myelination. Therefore, several researchers have explored the possibility that
thyroid dysfunction during pregnancy may be a cause of the condition. However, the
results have been conflicting.[15]
[16]
[17]
[18] Thyroid peroxidase autoantibodies in the maternal plasma increased the risk of ADHD-type
problems by 1.77-fold[15] and severe hypothyroxinemia (<5th percentile) at 12.9 weeks increased the risk of
ADHD by 1.70-fold in a Dutch population.[16] Others, however, found no association of maternal thyroid peroxidase antibody or
fT4 concentrations with risk of ADHD but did report 1.39-fold increased risk of ADHD
symptoms in female offspring of women with high thyroid-stimulating hormone (TSH)
levels.[17] A Danish study revealed that maternal hyperthyroidism, not hypothyroidism was associated
with a 1.23-fold increased risk of ADHD in the children.[18] These studies were conducted in regions of Europe where there are nationalized health
care systems and ethnically homogeneous populations and the findings may not be generalizable
to children born in the United States where the health care is largely privately managed,
preterm birth rates are higher, and there is significant racial and ethnic diversity.
Different countries have different determinants of disease for both hypothyroidism
and ADHD and access to health care may amplify the association. Therefore, effect
estimates of a potential relationship between hypothyroidism and ADHD may differ in
the United States. Although a previous study, reported an association between maternal
hypothyroidism and intellectual disability/inattention in the United States,[19] it relied on subscores of the Wechsler Intelligence Scale for Children to diagnose
the inattention rather than formal diagnosis of ADHD. Furthermore, no potential interactions
of hypothyroidism with maternal race/ethnicity, timing of diagnosis, gestational age
at birth, or fetal sex were explored. Therefore, we evaluated the risk of ADHD in
children born to women with and without hypothyroidism in a large diverse Southern
California population and how this potential risk factor is impacted by timing of
exposure to the hypothyroidism, gestational age at delivery, child's sex, and race-ethnicity.
Materials and Methods
Study Population
This study was approved by the Kaiser Permanente Southern California (KPSC) Institutional
Review Board with exempt status. The study population for this retrospective cohort
study was drawn from a total population of 571,674 births that occurred between January
1, 2000 and December 31, 2016 ([Fig. 1]). The information used in this study was obtained from the KPSC electronic health
records including: perinatal service (obstetrical conditions and procedures, fetal/neonatal
outcomes), maternal and child inpatient/outpatient medical care, laboratory and pharmacy
records, maternal sociodemographic and behavioral characteristics, child race ethnicity,
age, and sex. Parental demographic and maternal medical and obstetric health records
were linked to the child medical records using medical record numbers unique for each
pregnancy. Inclusion criteria included: (1) children born to KPSC members in all KPSC
hospitals between January 01, 2000 and December 31, 2016, and (2) delivered between
280/7 and 426/7-weeks' gestation. A total of 130,556 births were excluded due to: multiple births,
nonlife births, gestation length <28 or ≥43 weeks, leaving KPSC membership prior to
age three, noncontinuous enrollment for at least 3 months between 3 and 17 years of
age, autism diagnosis, ADHD diagnosed but not treated with medication or undiagnosed
ADHD with ADHD medication, or ADHD diagnosis prior to age three but not confirmed
at later age. The latter exclusion was important to maintain a cleaner cohort regarding
exposure and outcome ascertainment. Impact of these exclusions on patient demographics
is shown in [Supplementary Table S1] (available in the online version).
Fig. 1 Flowchart of the decision rules to develop the cohort used for these studies.
Exposures and Outcomes
International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)
codes and/or ICD-10-CM: 243, 244.0, 244.1, 244.2, 244.3, 244.8, 244.9, 245.0, E06.0,
E89.0, E03.0, E03.2, E03.3, E03.8, and E.03.9 and laboratory test values, TSH >4.0 μIU/mL
or fT4 ≤0.80 ng/dL, were used to identify exposures to hypothyroidism (this is consistent
with the 2017 ATA guidelines). Distributions of the hypothyroid diagnostic codes and
thyroid laboratory values are summarized in [Supplementary Tables S2] and [S3] (available in the online version). Where possible, we further divided the population
of women exposed to hypothyroidism into several categories and subcategories based
on timing of diagnosis. This included before and during pregnancy. The “pre-pregnancy”
category was further subdivided into 120 to 61 and 60 to 1 days prior to pregnancy
and the “in pregnancy” category was subdivided between women diagnosed in first, second,
and third trimesters.
Physician diagnosis of ADHD (ICD-9-CM codes 314.×) in children 3 to 17 years old on
at least one visit during the follow-up period was used to ascertain the outcome variable.
In a preliminary analysis, 96% of children with ADHD were diagnosed by child/adolescent
psychiatrists, developmental/behavioral pediatricians, child psychologists, or neurologists.
Follow-up began at birth and ended with ADHD diagnosis or when censoring occurred
due to health plan disenrollment, 18th birthday, non-ADHD-related death, or end of
the study (December 31, 2016).
Covariates
Factors including child sex, race-ethnicity, maternal age, education, median family
household income, parity, and timing of prenatal care, smoking habits during pregnancy,
and gestational age at delivery were evaluated as potential confounders/effect modifiers.
Child race/ethnicity was based on maternal and paternal records and classified as
non-Hispanic white (White), non-Hispanic black (Black), Hispanic, Asian/Pacific Islander,
and other/mixed race/ethnicity.
Statistical Analyses
Differences between maternal and child characteristics based on children ADHD status
were compared using χ
2-tests.
Crude and adjusted hazard ratios (aHRs) and their 95% confidence intervals (CIs) were
estimated using Cox proportional hazards models. Stratified analyses were used to
determine if the effects of hypothyroidism on ADHD risk may vary due to factors selected
a priori that include: timing of hypothyroidism diagnosis, gestational age at delivery,
child sex, and race-ethnicity. Because preterm birth may lie on the causal pathway
between hypothyroidism and ADHD, we examined the associations with and without adjustment
for gestational age at delivery in the model. However, we examined the independent
effect of hypothyroidism on ADHD after stratifying the dataset by gestational age
(preterm and term births) categories. The categorization of gestational age into term
and preterm birth was based on an a priori decision. We also stratified a priori based
on sex because males are at increased risk for several neurodevelopmental disorders,
including ADHD, that may be due to biological factors. Furthermore, we stratified
by race-ethnicity a priori because there are well documented health disparities in
minorities that may be due to access to and utilization of health care as well as
underlying biological risk factors. Children from women not diagnosed with hypothyroidism
were the reference group in all analyses. It is difficult to compare HRs between different
strata due to nonlinear effects of the denominator variation on the point estimate.
Therefore, we also calculated incidence rates (IRs, defined as the number of cases
ADHD divided by the number of person-years) and incidence rate differences (IRDs)
with 95% CIs to compare effect sizes on a linear scale.[20] Analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).
Results
Details of the children with (n = 16,696) and without (n = 312,461) ADHD diagnosis and their mothers are summarized in [Table 1]. Women who gave birth to children with ADHD were younger, tended to finish high
school but were more likely to make less than $30,000 per year than mothers of non-ADHD
children. Mothers of children with ADHD were also more likely to be nulliparous, to
deliver preterm, and smoke tobacco during pregnancy.
Table 1
Maternal and child characteristics by children's ADHD status
|
Characteristics
|
No ADHD N = 312,461 (%)
|
ADHD N = 16,696 (%)
|
p-Value
|
|
Maternal age, y
|
|
< 20
|
13,783 (4.4)
|
1,020 (6.1)
|
<0.001
|
|
20–29
|
127,830 (40.9)
|
7,070 (42.3)
|
|
30–34
|
101,309 (32.4)
|
4,971 (29.8)
|
|
≥ 35
|
69,539 (22.3)
|
3,635 (21.8)
|
|
Maternal education, y
|
|
< 12
|
23,238 (7.4)
|
1,211 (7.3)
|
<0.001
|
|
12
|
82,636 (26.4)
|
4,716 (28.2)
|
|
≥ 13
|
202,974 (65.0)
|
10,620 (63.6)
|
|
Missing
|
3,613 (1.2)
|
149 (0.9)
|
|
Median family household income, USD
|
|
< $30,000
|
21,326 (6.8)
|
1,282 (7.7)
|
<0.001
|
|
$30,000–$49,999
|
94,995 (30.4)
|
5,164 (30.9)
|
|
$50,000–$69,999
|
94,909 (30.4)
|
4,936 (29.6)
|
|
$70,000–$89,999
|
55,345 (17.7)
|
3,030 (18.1)
|
|
≥ $90,000
|
44,705 (14.3)
|
2,180 (13.1)
|
|
Missing
|
1,181 (0.4)
|
104 (0.6)
|
|
Parity
|
|
Par 0
|
121,400 (38.9)
|
7,431 (44.5)
|
<0.001
|
|
Par 1
|
106,990 (34.2)
|
5,388 (32.3)
|
|
Par 2+
|
84,057 (26.9)
|
3,877 (23.2)
|
|
Missing
|
14 (<0.1)
|
0 (0.0)
|
|
Gestational age, wk
|
|
28–36
|
22,996 (7.4)
|
1,585 (9.5)
|
<0.001
|
|
37–42
|
289,465 (92.6)
|
15,111 (90.5)
|
|
Smoking during pregnancy
|
|
No
|
297,527 (95.2)
|
15,594 (93.4)
|
<0.001
|
|
Yes
|
14,934 (4.8)
|
1,102 (6.6)
|
|
Late initiation or no prenatal care
|
23,137 (7.4)
|
1,233 (7.4)
|
0.867
|
|
Child's race/ethnicity
|
|
White
|
56,899 (18.2)
|
4,322 (25.9)
|
<0.001
|
|
Black
|
21,843 (7.0)
|
1,500 (9.0)
|
|
Hispanic
|
129,556 (41.5)
|
5,717 (34.2)
|
|
Asian/Pacific islanders
|
31,452 (10.1)
|
632 (3.8)
|
|
Other/Multiple
|
72,398 (23.2)
|
4,512 (27.0)
|
|
Unknown
|
313 (0.1)
|
13 (0.1)
|
|
Child's sex
|
|
Female
|
158,212 (50.6)
|
4,761 (28.5)
|
<0.001
|
|
Male
|
154,249 (49.4)
|
11,935 (71.5)
|
Abbreviations: ADHD, attention-deficit hyperactivity disorder; USD, United States
Dollar.
Note: Statistically significant associations are indicated in boldface.
General Effects of Hypothyroidism
Of the 329,157 pregnancies followed in this study, 9,675 (2.9%) had a diagnosis of
hypothyroidism <120 prior to, or during their pregnancy (319,482 did not). Incidence
of ADHD was greater for children born to women with a hypothyroid diagnosis prior
to, or during, pregnancy ([Table 2]). Cox-proportional hazards models also demonstrated that prenatal exposure to hypothyroidism
is significantly associated with increased risk of ADHD (aHR:1.24, 95% CI: 1.14, 1.35)
([Fig. 2]; [Table 2]). Further analysis by trimester of in utero exposure, revealed that the trend was
largely confined to the first trimester (HR: 1.29, 95% CI: 1.05, 1.59) and that hazard
ratios were not affected by adjustment for covariates ([Table 2]). Children of women whose hypothyroidism was left either untreated or treated with
<50 μg/d thyroid supplementation were not at significantly increased risk (aHR: 1.08,
95% CI: 0.75, 1.57) of ADHD but children of women whose hypothyroidism was treated
with ≥50 μg/d were at significantly increased risk (aHR: 1.26, 95% CI: 1.14, 1.39).
Exposure to hypothyroidism was associated with increased risk of ADHD when diagnosed
prior to or after conception.
Fig. 2 Kaplan-Meier curves for attention-deficit hyperactivity disorder (ADHD) prevalence
amongst children born to mothers with and without hypothyroidism in pregnancy.
Table 2
Association between prenatal exposure to pre-pregnancy and gestational hypothyroidism
and ADHD incidence and risk
|
Group
|
N
|
Person-years
|
IR (‰)
|
IRD (95% CI)
|
HR (95% confidence intervals)
|
|
Crude
|
Adjusted[a]
|
|
No Hypothyroidism
|
16,171
|
2,963,448
|
5.46
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
525
|
80,260
|
6.54
|
1.08 (0.56, 1.60)
|
1.26 (1.15, 1.37)
|
1.24 (1.14, 1.35)
|
|
Untreated
|
89
|
14,455
|
6.16
|
0.70 (−0.51, 1.91)
|
1.21 (0.98, 1.49)
|
1.23 (1.00, 1.51)
|
|
Treated
|
436
|
65,805
|
6.62
|
1.17 (0.60, 1.74)
|
1.27 (1.15, 1.39)
|
1.24 (1.13, 1.37)
|
|
<50 μg
|
28
|
5,465
|
5.12
|
−0.33 (−2.29, 1.63)
|
1.11 (0.77, 1.61)
|
1.08 (0.75, 1.57)
|
|
≥50 μg
|
408
|
60,340
|
6.77
|
1.31 (0.71, 1.90)
|
1.28 (1.16, 1.41)
|
1.26 (1.14, 1.39)
|
|
Pre-Pregnancy
|
377
|
55,792
|
6.76
|
1.30 (0.68, 1.92)
|
1.30 (1.17, 1.44)
|
1.27 (1.15, 1.41)
|
|
Diagnosis timing
|
|
|
|
|
|
|
|
120 to 61 days
|
222
|
35,116
|
6.32
|
0.86 (0.09, 1.64)
|
1.22 (1.06, 1.39)
|
1.19 (1.04, 1.36)
|
|
60 to 1 days
|
155
|
20,676
|
7.50
|
2.04 (1.03, 3.05)
|
1.43 (1.22, 1.68)
|
1.40 (1.19, 1.64)
|
|
Treatment status
|
|
|
|
|
|
|
|
Untreated
|
32
|
4,778
|
6.70
|
1.24 (−0.86, 3.34)
|
1.42 (1.00, 2.00)
|
1.41 (0.99, 1.99)
|
|
Treated
|
345
|
51,014
|
6.76
|
1.31 (0.66, 1.95)
|
1.29 (1.16, 1.43)
|
1.26 (1.13, 1.40)
|
|
<50 μg
|
22
|
3,489
|
6.31
|
0.85 (−1.60, 3.30)
|
1.33 (0.87, 2.02)
|
1.32 (0.87, 2.00)
|
|
≥50 μg
|
323
|
47,525
|
6.80
|
1.34 (0.67, 2.01)
|
1.28 (1.15, 1.43)
|
1.25 (1.12, 1.40)
|
|
In-pregnancy
|
148
|
24,468
|
6.05
|
0.59 (−0.34, 1.52)
|
1.17 (0.99, 1.37)
|
1.17 (1.00, 1.38)
|
|
Diagnosis timing
|
|
|
|
|
|
|
|
First trimester
|
91
|
14,158
|
6.43
|
0.97 (−0.25, 2.19)
|
1.29 (1.05, 1.59)
|
1.28 (1.04, 1.58)
|
|
Second trimester
|
27
|
5,322
|
5.07
|
−0.38 (−2.37, 1.60)
|
0.94 (0.64, 1.37)
|
0.93 (0.64, 1.36)
|
|
Third trimester
|
30
|
4,988
|
6.01
|
0.58 (−1.49, 2.61)
|
1.10 (0.77, 1.57)
|
1.15 (0.80, 1.64)
|
|
Treatment status
|
|
|
|
|
|
|
|
Untreated
|
57
|
9,677
|
5.89
|
0.43 (−1.04, 1.91)
|
1.12 (0.87, 1.46)
|
1.15 (0.88, 1.49)
|
|
Treated
|
91
|
14,791
|
6.15
|
0.70 (−0.50, 1.89)
|
1.20 (0.98, 1.47)
|
1.19 (0.97, 1.46)
|
|
<50 μg
|
6
|
1,976
|
3.04
|
−2.42 (−5.68, 0.84)
|
0.69 (0.31, 1.54)
|
0.66 (0.30, 1.46)
|
|
≥50 μg
|
85
|
12,815
|
6.63
|
1.18 (−0.11, 2.46)
|
1.27 (1.02, 1.57)
|
1.26 (1.02, 1.56)
|
Abbreviations: IR, incidence rates per 1000 person-years; IRD, Incidence Rate Difference,
HR, hazard ratio; CI, confidence intervals.
Note: Statistically significant associations are indicated in boldface.
a Adjustments were made for maternal age, education, median family household income,
smoking during pregnancy, prenatal care, parity, child's sex and race/ethnicity.
Stratification by Gestational Age at Delivery
Although no statistical interaction was detected between hypothyroidism and preterm
birth (p = 0.439), when stratified by the timing of first diagnosis, maternal hypothyroidism
was associated with increased incidence and risk of ADHD in preterm born infants with
the strongest associations for women first diagnosed during the pre-pregnancy period
(aHR: 1.43, 95% CI: 1.09, 1.88; [Table 3]). When further stratified, the association was significantly higher for those diagnosed
at 61 to 120 day during pre-pregnancy (aHR: 1.42, 95% CI: 1.01, 1.99), but not for
those diagnosed at 1 to 60 days prior to pregnancy. The IRDs in ADHD for infants born
to hypothyroid versus nonhypothyroid mothers were much higher for preterm infants
than it was for term infants ([Table 3]) and the incidence of ADHD in children born to hypothyroid mothers delivering at
term was strongest in the first trimester ([Table 3]), similar to what was observed in the unstratified analysis ([Table 2]). Adding gestational age of delivery to the model had no effect on the association
(aHR = 1.23, 95% CI: 1.13, 1.35).
Table 3
Association between maternal hypothyroidism and ADHD risk in the offspring based on
gestational age at delivery and timing of first hypothyroidism diagnosis
|
Birth
|
Group
|
N
|
Person-Years
|
IR (‰)
|
IRD (95% CI)
|
HR (95% Confidence intervals)
|
|
Crude
|
Adjusted[a]
|
|
Preterm
|
No hypothyroidism
|
1,515
|
218,583
|
6.93
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
70
|
7,640
|
9.16
|
2.23 (0.32, 4.14)
|
1.38 (1.09,1.76)
|
1.33 (1.04,1.69)
|
|
Pre-pregnancy
|
54
|
5,407
|
9.99
|
3.06 (0.80, 5.31)
|
1.50 (1.15,1.97)
|
1.43 (1.09, 1.88)
|
|
120 to 161 days
|
35
|
3,418
|
10.2
|
3.31 (0.49, 6.13)
|
1.52 (1.09, 2.13)
|
1.42 (1.01, 1.99)
|
|
60 to 1 days
|
19
|
1,989
|
9.55
|
2.62 (−1.06, 6.30)
|
1.47 (0.93, 2.31)
|
1.44 (0.92, 2.27)
|
|
In-pregnancy
|
16
|
2,233
|
7.17
|
0.23 (−3.23, 3.70)
|
1.09 (0.66,1.78)
|
1.07 (0.66, 1.76)
|
|
First trimester
|
11
|
1,225
|
8.98
|
2.05 (−2.63, 6.73)
|
1.39 (0.77,2.52)
|
1.40 (0.77, 2.54)
|
|
Second trimester
|
5
|
549
|
9.11
|
2.18 (−4.80, 9.15)
|
1.35 (0.56,3.24)
|
1.31 (0.55,3.16)
|
|
Third trimester
|
0
|
0
|
0.00
|
ND
|
ND
|
ND
|
|
Term
|
No hypothyroidism
|
14,656
|
2,744,865
|
5.34
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
455
|
72,620
|
6.27
|
0.93 (0.39, 1.47)
|
1.23 (1.12, 1.35)
|
1.22 (1.11, 1.34)
|
|
Pre-pregnancy
|
323
|
50,385
|
6.41
|
1.07 (0.43, 1.72)
|
1.26 (1.13, 1.40)
|
1.24 (1.11, 1.38)
|
|
120 to 161 days
|
187
|
31,698
|
5.90
|
0.56 (−0.25, 1.37)
|
1.16 (1.01, 1.34)
|
1.15 (0.99, 1.33)
|
|
60 to 1 days
|
136
|
18,687
|
7.28
|
1.94 (0.89, 2.99)
|
1.42 (1.20, 1.68)
|
1.38 (1.17, 1.64)
|
|
In-pregnancy
|
132
|
22,235
|
5.94
|
0.60 (−0.37, 1.56)
|
1.17 (0.99, 1.39)
|
1.18 (0.99, 1.40)
|
|
First trimester
|
80
|
12,933
|
6.19
|
0.85 (−0.42, 2.11)
|
1.27 (1.02, 1.58)
|
1.26 (1.01, 1.57)
|
|
Second trimester
|
22
|
4,773
|
4.62
|
−0.73 (−2.80, 1.34)
|
0.87 (0.57,1.32)
|
0.87 (0.57, 1.32)
|
|
Third trimester
|
30
|
4,529
|
5.90
|
1.28 (−0.85, 3.42)
|
1.23 (0.86, 1.76)
|
1.30 (0.91, 1.87)
|
Abbreviations: ADHD, attention-deficit hyperactivity disorder; CI, confidence intervals;
HR, hazard ratio; IR, incidence rate per 1,000 person-years; IRD, incidence rates
difference; ND, not done due to small sample.
Note: Statistically significant associations are indicated in boldface.
a Adjustments were made for maternal age, education, smoking during pregnancy, prenatal
care, parity, median household income, child's sex, and race/ethnicity.
Stratification by Sex
As expected, there were 2.5 times more boys than girls in this study with ADHD ([Table 4]). Although a diagnosis of maternal hypothyroidism overall tended to increase the
hazard of ADHD in girls, the increased incidence was marginal (aHR: 1.19, 95% CI:
1.01, 1.40). For male infants, however, maternal hypothyroidism diagnosed prior to
pregnancy (aHR: 1.30, 95% CI: 1.15, 1.46) or within the first trimester (aHR: 1.28,
95% CI: 1.00, 1.64) was significantly associated with increased incidence and hazard
of ADHD. Evaluation of the IRs stratified by sex, suggests that hypothyroidism may
have a stronger effect on boys than it does on girls because differences between IRD
in children born of hypothyroid mothers and those whose mothers did not have hypothyroidism
were four times larger for boys than it was for girls (IRD 1.84 vs. 0.48; [Table 4]). Tests for a statistical interaction between sex and hypothyroidism did not reach
statistical significance (p = 0.563), however.
Table 4
Effect of maternal hypothyroidism on incidence and hazard of childhood ADHD by child
sex and timing of first hypothyroidism diagnosis
|
|
|
|
|
|
HR (95% Confidence intervals)
|
|
Sex
|
Group
|
N
|
Person-Years
|
IR (‰)
|
IRD (95% CI)
|
Crude
|
Adjusted[a]
|
|
Female
|
No hypothyroidism
|
4,613
|
1,482,030
|
3.11
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
148
|
41,159
|
3.60
|
0.48 (−0.06, 1.03)
|
1.22 (1.04, 1.44)
|
1.19 (1.01, 1.40)
|
|
Pre-pregnancy
|
104
|
28,345
|
3.67
|
0.56 (−0.10,1.21)
|
1.24 (1.02, 1.51)
|
1.20 (0.99, 1.46)
|
|
120 to 161 days
|
62
|
17,952
|
3.45
|
0.34 (−0.48, 1.16)
|
1.17 (0.91, 1.51)
|
1.13 (0.88, 1.45)
|
|
60 to 1 days
|
42
|
10,393
|
4.04
|
0.93 (−0.15, 2.00)
|
1.36 (1.01, 1.85)
|
1.32 (0.97, 1.44)
|
|
In-pregnancy
|
44
|
12,814
|
3.43
|
0.32 (−0.65,1.29)
|
1.17 (0.87, 1.58)
|
1.16 (0.86, 1.56)
|
|
First trimester
|
27
|
7,364
|
3.67
|
0.55 (−0.72,1.83)
|
1.31 (0.90, 1.92)
|
1.30 (0.89, 1.90)
|
|
Second trimester
|
7
|
2,709
|
2.58
|
−0.53 (−2.63, 1.58)
|
0.85 (0.41, 1.79)
|
0.84 (0.40, 1.76)
|
|
Third trimester
|
10
|
2,741
|
3.65
|
0.54 (−1.55, 2.62)
|
1.15 (0.62, 2.13)
|
1.11 (0.60, 2.07)
|
|
Male
|
No hypothyroidism
|
11,558
|
1,481,418
|
7.80
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
377
|
39,101
|
9.64
|
1.84 (0.95, 2.73)
|
1.30 (1.17, 1.44)
|
1.26 (1.14, 1.40)
|
|
Pre-pregnancy
|
273
|
27,447
|
9.95
|
2.14 (1.09, 3.20)
|
1.34 (1.18, 1.51)
|
1.30 (1.15, 1.46)
|
|
120 to 161 days
|
160
|
17,164
|
9.32
|
1.52 (0.19, 2.85)
|
1.26 (1.07, 1.47)
|
1.22 (1.04, 1.43)
|
|
60 to 1 days
|
113
|
10,283
|
10.99
|
3.19 (1.47, 4.90)
|
1.47 (1.22, 1.77)
|
1.43 (1.19, 1.72)
|
|
In-pregnancy
|
104
|
11,654
|
8.92
|
1.12 (−0.49, 2.73)
|
1.20 (0.99, 1.46)
|
1.18 (0.97, 1.44)
|
|
First trimester
|
64
|
6,794
|
9.42
|
1.62 (−0.49, 3.72)
|
1.32 (1.03, 1.68)
|
1.28 (1.00, 1.64)
|
|
Second trimester
|
20
|
2,613
|
7.65
|
−0.15 (−3.54, 3.24)
|
0.97 (0.63, 1.51)
|
0.97 (0.63, 1.50)
|
|
Third trimester
|
20
|
2,247
|
8.90
|
1.10 (−2.56, 4.76)
|
1.16 (0.75, 1.80)
|
1.16 (0.75, 1.81)
|
Abbreviations: CI, confidence interval; HR, hazard ratio; IR, incidence rate per 1,000
person-years; IRD, incidence rate difference.
Note: Statistically significant associations are indicated in boldface.
a Adjustments were made for maternal age, education, smoking during pregnancy, prenatal
care, parity, median household income, and child's race/ethnicity.
Stratification by Race–Ethnicity
Like child sex, race-ethnicity is a significant risk factor for adverse neurodevelopmental
outcomes. Incidence of ADHD was highest in White children, followed by Blacks and
Hispanics with Asian/Pacific Islander children having the lowest rates of ADHD. Hypothyroidism
increased the incidence of ADHD in Whites, Blacks, and Hispanics to varying degrees
but not for Asian/Pacific Islander and children of Other/Multiple ethnicities, possibly
because of limited numbers of women experiencing hypothyroidism in these groups ([Table 5]). Formal tests for a statistical interaction with the Cox-proportional hazards models
were marginally nonsignificant (p = 0.055).
Table 5
Effect of maternal hypothyroidism on incidence rate and risk of ADHD by child race-ethnicity
and timing of first hypothyroidism diagnosis
|
|
|
|
|
|
HR (95% confidence intervals)
|
|
Race-Ethnicity
|
Group
|
N
|
Patient-Years
|
IR (‰)
|
IRD
|
Crude
|
Adjusted[a]
|
|
NH-White
|
No hypothyroidism
|
4,131
|
531,076
|
7.78
|
0.000 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
191
|
21,162
|
9.03
|
1.25 (0.03, 2.46)
|
1.21 (1.05, 1.40)
|
1.22 (1.05, 1.41)
|
|
Pre-pregnancy
|
151
|
15,776
|
9.57
|
1.79 (0.39, 3.19)
|
1.28 (1.09, 1.51)
|
1.29 (1.09, 1.52)
|
|
120 to 161 days
|
83
|
9,978
|
8.32
|
0.54 (−1.21, 2.28)
|
1.11 (0.89,1.38)
|
1.12 (0.90, 1.39)
|
|
60 to 1 days
|
68
|
5,798
|
11.73
|
3.95 (1.66, 6.24)
|
1.57 (1.23, 1.99)
|
1.57 (1.24, 2.00)
|
|
In-pregnancy
|
40
|
5,386
|
7.43
|
−0.35 (−2.72, 2.02)
|
1.00 (0.73, 1.37)
|
1.02 (0.74, 1.39)
|
|
First trimester
|
28
|
3,316
|
8.44
|
0.66 (−2.34, 3.68)
|
1.19 (0.82, 1.72)
|
1.29 (1.09, 1.52)
|
|
Second trimester
|
4
|
1,002
|
3.99
|
ND[c]
|
ND
|
ND
|
|
Third trimester
|
8
|
1,068
|
7.49
|
ND
|
ND
|
ND
|
|
Hispanic
|
No hypothyroidism
|
5,535
|
1,246,455
|
4.44
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
182
|
30,118
|
6.04
|
1.60 (0.84, 2.37)
|
1.43 (1.23, 1.66)
|
1.45 (1.25, 1.68)
|
|
Pre-pregnancy
|
117
|
20,027
|
5.84
|
1.40 (0.47, 2.33)
|
1.38 (1.15, 1.65)
|
1.39 (1.15, 1.67)
|
|
120 to 161 days
|
72
|
12,708
|
5.67
|
1.22 (0.06, 2.39)
|
1.34 (1.06, 1.69)
|
1.36 (1.08, 1.72)
|
|
60 to 1 days
|
45
|
7,319
|
6.15
|
1.71 (0.18, 3.24)
|
1.44 (1.07, 1.93)
|
1.42 (1.06, 1.92)
|
|
In-pregnancy
|
65
|
10,091
|
6.44
|
2.00 (0.69, 3.31)
|
1.53 (1.20, 1.95)
|
1.59 (1.24, 2.03)
|
|
First trimester
|
40
|
6,045
|
6.62
|
2.18 (0.49, 3.86)
|
1.61 (1.18, 2.20)
|
1.71 (1.25, 2.33)
|
|
Second trimester
|
14
|
2,135
|
6.56
|
2.12 (−0.71, 4.94)
|
1.48 (0.88, 2.50
|
1.49 (0.88, 2.52)
|
|
Third trimester
|
11
|
1,911
|
5.75
|
1.32 (−1.67, 4.30)
|
1.33 (0.74, 2.40)
|
1.35 (0.75, 2.44)
|
|
NH-Black[b]
|
No hypothyroidism
|
1,481
|
232,596
|
6.37
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
19
|
2,623
|
7.24
|
0.88 (−2.20, 3.95)
|
1.16 (0.74, 1.83)
|
1.25 (0.80, 1.97)
|
|
Pre-pregnancy
|
16
|
985
|
9.90
|
3.53 (−0.38, 7.45)
|
1.57 (0.96, 2.58)
|
1.64 (1.00, 2.68)
|
|
120 to 161 days
|
11
|
1,001
|
10.99
|
4.26 (−0.34, 9.58)
|
1.74 (0.96, 3.15)
|
1.79 (0.99, 3.25)
|
|
60 to 1 days
|
5
|
615
|
8.13
|
ND
|
ND
|
ND
|
|
In-pregnancy
|
3
|
1,007
|
2.98
|
ND
|
ND
|
ND
|
|
Asian/PI[b]
|
No hypothyroidism
|
612
|
289,003
|
2.12
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
20
|
7,844
|
2.55
|
0.43 (−0.60, 1.47)
|
1.28 (0.82, 2.00)
|
1.32 (0.35, 2.07)
|
|
Pre-pregnancy
|
16
|
5,544
|
2.89
|
0.77 (−0.46, 1.99)
|
1.44 (0.88, 2.37)
|
1.48 (0.90, 2.43)
|
|
120 to 161 days
|
8
|
3,477
|
2.30
|
ND
|
ND
|
ND
|
|
60 to 1 days
|
8
|
2,067
|
3.87
|
ND
|
ND
|
ND
|
|
In-pregnancy
|
4
|
2,300
|
1.74
|
ND
|
ND
|
ND
|
|
Other/Multiple[b]
|
No hypothyroidism
|
4,399
|
454,344
|
6.65
|
0.00 (Reference)
|
1.00 (Reference)
|
1.00 (Reference)
|
|
Hypothyroidism
|
113
|
18,461
|
6.35
|
−0.52 (−1.72, 0.67)
|
0.96 (0.80, 1.16)
|
1.02 (0.84, 1.23)
|
|
Pre-pregnancy
|
77
|
12,790
|
6.02
|
−0.63 (−2.05, 0.80)
|
0.94 (0.75, 1.18)
|
1.03 (0.82, 1.29)
|
|
120 to 161 days
|
48
|
7,929
|
6.05
|
−0.59 (−2.40, 1.21)
|
0.95 (0.71, 1.26)
|
1.03 (0.77, 1.37)
|
|
60 to 1 days
|
29
|
4,861
|
5.97
|
−0.68 (−2.98, 1.62)
|
0.93 (0.65, 1.34)
|
1.02 (0.71, 1.48)
|
|
In-pregnancy
|
36
|
5,671
|
6.35
|
−0.30 (−2.43, 1.84)
|
1.00 (0.72, 1.39)
|
0.99 (0.71, 1.38)
|
Abbreviations: ADHD, attention-deficit hyperactivity disorder; HR, hazard ratio; IR,
incidence rate per 1,000 person-years; IRD, incidence rate difference; ND, not done;
PI, Pacific Islander.
Note: Statistically significant associations are indicated in boldface.
a Adjustments were made for maternal age, education, smoking during pregnancy, prenatal
care, parity, median household income, and child's sex.
b Trimester-specific data are not provided due to insufficient observation in the categories.
c Analysis “not done” due to insufficient observations in the categories.
Discussion
We found that maternal hypothyroidism during the preconceptual and prenatal period
significantly increases the hazard of ADHD in the children and that this association
is strongest for boys, those born prematurely, and for Hispanic children. IRs for
ADHD were generally higher for children born to hypothyroid women regardless of race-ethnicity,
sex, timing of birth, or timing of in utero hypothyroid exposure with most of the
exceptions being for children of women diagnosed in mid-late pregnancy. The lack of
association of maternal hypothyroidism diagnosed later in pregnancy could be due to
the sparseness of the data in that category. It is also possible, however, that maternal
hypothyroidism has more severe effects on fetal brain development if it occurs prior
to 18 weeks gestation when the fetus starts making its own thyroid hormones.[21]
Consistent with previous studies,[22]
[23] we found that preterm birth is associated with increased incidence of ADHD in the
offspring. Although we did not detect an interaction in the statistical models, maternal
hypothyroidism had a stronger effect on incidence of ADHD in preterm (IRD = 2.23)
than term (IRD = 0.93) infants. Disordered neurodevelopmental patterns caused by the
lack of thyroid hormones may make the brain more susceptible to injury as a result
of the preterm birth. Thyroid hormone supplementation reduced brain damage in animal
models of traumatic brain injury and stroke.[24] Maternal hypothyroidism also increases the risk of preterm birth[25]
[26] and it is possible that the correlation between these outcomes is due to the fact
that infants who are at risk for ADHD because of maternal hypothyroidism are also
at increased risk for preterm birth. Our finding is that the risk for ADHD remained
elevated when models were adjusted for gestational age at delivery with minimal effect
on the magnitude of association. This suggests that preterm birth may not be in the
causal pathway of ADHD but more likely an outcome independently associated with maternal
hypothyroidism.
A stratified analysis based on race-ethnicity categories revealed that the effect
of hypothyroidism was largest for Hispanic children. It is unlikely that this difference
is due to racial ethnic differences in screening because of the standardized screening
tools for both ADHD and maternal hypothyroidism that are in place across all KPSC
centers. Further studies are needed to determine if there are genetic, dietary, or
lifestyle factors that explain why children of Hispanic hypothyroid women develop
ADHD at greater rates than other race-ethnicities. Although our adjusted proportional
hazards models revealed some evidence for a statistical interaction, this could also
be, in part, due to under sampling of some race ethnicities. Further studies with
stratified sampling are needed to confirm our observations.
Our findings are largely consistent with previous studies linking thyroid dysfunction
with neurodevelopmental disorders. In the generation R study, low concentrations of
thyroxine (≤5th percentile) in pregnancy were associated with increased risk of teacher-reports
of hyperactivity and inattention at 5 years of age.[16] A follow-up study of this same patient population at 8 years of age also found that
severe hypothyroidism during pregnancy significantly increases the risk of ADHD behaviors
as rated by the parents. Although they adjusted for factors such as child ethnicity,
age, sex, and maternal demographics, no effects of these covariates on the association
between maternal hypothyroidism and ADHD index in the children were reported.[27] We performed stratified analyses on a much larger sample size that relied on clinical
diagnosis of both ADHD and maternal hypothyroidism.[27] As in previous studies,[28] we found that the effects of hypothyroidism on adverse neurodevelopmental outcomes
were gestational age dependent. Our findings also differ from those of Andersen et
al, who reported no association of maternal hypothyroidism with increased risk of
ADHD in Denmark.[18] This could be due to differences in timing of diagnosis or patient demographics.
In nearly 80% of the pregnancies studied by Andersen et al, the thyroid dysfunction
was diagnosed after the birth of the child; the vast majority of the Danish patient
population is Caucasian. We found no effect of hypothyroidism diagnosed late in pregnancy
on risk of childhood ADHD and that the effect was much stronger for Hispanics than
other races-ethnicities. There are large race/ethnic disparities in access to and
utilization of health care in the United States and this may also contribute to undertreatment
of hypothyroidisms and thus disparities in ADHD risk. This finding is consistent with
our previous work that examined race-ethnicity as a risk modifier for autism in children
born to women with hypothyroidism during pregnancy.[29]
ADHD may be one of several neurodevelopmental disorders that results from maternal
hypothyroidism. Psychological testing of 7- to 9-year old children born to women with
second-trimester TSH concentrations in the 98th percentile tended to have lower intelligence
quotients (IQs), some language difficulties, higher rates of school problems, and
to be more easily distracted[19] than children born to mothers with normal TSH levels.[19] Although we did not examine cognitive or motor function, we previously found higher
rates of autism in children born to hypothyroid mothers in our patient population.[29] Epilepsy, a common comorbidity in autism and ADHD, is also increased in children
born to hypothyroid mothers.[30]
Hyperthyroidism may be equally disruptive to neurodevelopment. Our recent studies
have demonstrated that hyperthyroidism increases the risk of autism.[31] Furthermore, Hales et al, found that overtreatment of hypothyroidism in pregnant
women increased the risk of ADHD in the offspring.[32] This is consistent with previous studies that demonstrated a U-shaped curve between maternal thyroxine levels in pregnancy with childhood IQ.[28] It is unclear from our study if the risk associated with higher levels of treatment
reflect a toxic effects of treatment or a more severe form of the hypothyroidism.
Strengths of our study include, use of a large, well-maintained electronic database
covering a large number of pregnancies over a long period of time, and measurements
on the children with high follow-up. Consistent standards for diagnosing ADHD in all
KPSC care settings minimize the risk for misclassification but may make it difficult
to generalize our findings to other populations where diagnosis is made by a primary
care provider. Ethnic diversity of the population studied as well as access to birth
records and demographic information has further advantages. Although difficulties
in diagnosing and treating minority children with ADHD are well-known,[33] any differences in background risk for diagnosing ADHD would tend to bias the results
of our study on hypothyroidism as a risk factor toward the null. By focusing our analysis
on children with both a diagnosis and prescribed medications for ADHD, we further
limited the possibilities of misclassification. Even with inclusion of these children
in our sensitivity analyses, the effect size remained unchanged. Although we adjusted
for several potential confounding factors, our findings are also limited by several
residual factors that are inherent to retrospective epidemiological studies. Given
the difficulty in accounting for all the comorbidities that may occur with ADHD in
early preterm births, we excluded infants born at <28 weeks gestation. This limits
our conclusions to moderately and late preterm birth. We also used strictly diagnostic
codes. It is unclear if the hypothyroidism was due to autoimmune thyroid disease,
thyroid cancers, Hashimoto's thyroiditis, environmental toxins, iodine deficiencies,
or other causes of hypothyroidism. Although our findings are biologically plausible
and the timing of the hypothyroidism precedes the development of ADHD, it is unclear
if the relationship is dose-dependent and causal. Additional studies are also needed
to determine the extent to which maternal hypothyroidism correlates with fetal hypothyroidism
in our population. The placenta is rich in enzymes and other proteins that regulate
iodine and thyroid hormone concentrations at the maternal–fetal interface.[21] It is possible that fetal hypothyroidism occurs in the absence of maternal hypothyroidism
or vice versa. We are also unable to explore what specific changes that thyroid hormone
deficiencies cause in the fetal brain to result in ADHD and how they may differ from
other neurodevelopmental disorders associated with maternal hypothyroidism.
In summary, maternal hypothyroidism during the preconceptual period and first trimester
significantly increases the risk of ADHD in children. Further studies are needed to
identify different types of hypothyroidism that is contributing to the increased risk
and to better understand the molecular mechanisms behind them.
