Associations Between Maternal Thyroid Function in Pregnancy and Obstetric and Perinatal Outcomes

Abstract

Context

The effects of maternal subclinical hypothyroidism on pregnancy outcomes are not clear.

Objective

We aimed to assess potential associations between maternal thyrotropin (thyroid-stimulating hormone [TSH]) levels in pregnancy and obstetric and perinatal outcomes.

Design

Retrospective cohort study.

Setting

Tertiary academic medical center.

Patients

Women aged ≥18 years with a singleton gestation and no known thyroid disease seen for prenatal care at Boston Medical Center from January 1, 2003 through May 22, 2014, and their fetuses and infants were included.

Main Outcome Measures

Risk ratios of adverse obstetric and perinatal outcomes.

Results

A total of 8,413 pregnant women (mean age 29.1 years, 15% white, 60% black, 13% Hispanic) and their fetuses and infants (mean gestational age at birth 38.5 weeks, 52% male, mean birth weight 3.2 kg) were included in the analyses. The median (interquartile range) TSH level was 1.06(0.62–1.60) mIU/L, and 130 women (1.6%) had TSH > 4 mIU/L. Maternal TSH levels > 4 mIU/L were associated with increased risks of prematurity (risk ratio [RR] 2.17 [95% confidence interval 1.15–4.07] P = .016) and neonatal respiratory distress syndrome (RDS) (RR 2.83 [95% confidence interval 1.02–7.86] P = .046) compared to TSH levels ≤ 4 mIU/L. Although not statistically significant, TSH levels > 4 mIU/L were also associated with increased RRs for fetal loss, preeclampsia/eclampsia, and low birth weight. TSH levels > 4 mIU/L were not associated with preterm labor, placental abruption, cesarean section, gestational hypertension or diabetes, or neonatal intensive care unit admission.

Conclusion

Maternal serum TSH concentration > 4 mIU/L in pregnancy was associated with approximately 2-fold increased risks of prematurity and RDS in offspring. Elevated TSH was also associated with statistically non-significant increases in the risk of fetal loss, preeclampsia/eclampsia, and low birth weight.

Adequate thyroid hormone is essential for normal fetal development (1). The fetal thyroid gland forms by 10 to 12 weeks’ gestation and starts to produce thyroid hormone by 18 to 20 weeks’ gestation (2,3). The fetal serum thyroid hormone levels do not reach adult levels until 36 weeks’ gestation (4). Therefore, the fetus depends on maternal thyroid hormone crossing the placenta during the critical period of development in the first trimester. During early pregnancy, maternal serum thyrotropin (thyroid stimulating hormone [TSH]) decreases, largely due to the weak stimulatory effect of human chorionic gonadotropin on the thyroid gland with subsequent increase in free T4 (FT4) levels (5). Thus, the American Thyroid Association (ATA) recommends using population-based, trimester specific TSH reference ranges or, if the assay-specific ranges are not available, using an upper reference limit of 4.0 mIU/L (6). Untreated overt maternal hypothyroidism during pregnancy is known to have adverse effects on pregnancy outcomes such as increased risk of fetal loss, preeclampsia, abruption placenta, and postpartum hemorrhage (7,8) and on perinatal outcomes such as increased risks of premature birth, low birth weight (LBW), and neonatal respiratory distress syndrome (RDS) (9,10).

Although there is consensus on the adverse effects of untreated overt maternal hypothyroidism in pregnancy, the effects of maternal subclinical hypothyroidism (SCH) on obstetric and perinatal outcomes are less clear. Several studies have reported increased risks of miscarriage (7,11–15), preterm birth (15–17), or placental abruption (17) associated with maternal SCH. A recent meta-analysis of 18 cohort studies using the 2011 ATA guidelines for trimester-specific cutoffs for TSH (18) showed increasing risk of pregnancy loss (relative risk [RR] 2.01), placental abruption (RR 2.14), premature rupture of membranes (RR 1.43), and neonatal death (RR 2.58) in women with SCH (19). Another recent meta-analysis of individual data of 47 045 women from 19 cohort studies also showed 1.29-fold increased risk of preterm birth in women with SCH, defined as TSH above 97.5th percentile (20). The risk was higher in women with positive thyroid peroxidase (TPO) antibody and serum TSH level > 4 mIU/L (20).

In contrast, a cohort study of 10,990 US women reported no associations between maternal SCH, defined as the TSH level above the 97.5th percentile (4.29 mIU/L in the first trimester and 3.94 mIU/L in the second trimester), assessed in the first and second trimester of pregnancy and adverse pregnancy outcomes (21). Other investigators similarly reported no association between maternal SCH and adverse pregnancy outcomes in a large retrospective cohort in northern Finland (22,23). A recent meta-analysis of 14 cohort studies and one case control study showed no significant increase in preterm delivery risk in women with SCH, even though the risk was increased 19% in women with overt hypothyroidism and 24% with overt hyperthyroidism (24).

Given the conflicting data regarding the effects of SCH on pregnancy and perinatal outcomes, there is continued debate regarding the need for universal screening for maternal hypothyroidism during pregnancy. Currently, most guidelines recommend a case-finding approach, testing only pregnant women with clinical suspicion of thyroid disease or those with risk factors for developing thyroid dysfunction (6,25–27).

To assess potential associations between maternal thyroid status, as measured by serum TSH levels, during pregnancy and obstetric and perinatal outcomes, we conducted a retrospective cohort study of pregnant women and their offspring seen at Boston Medical Center (BMC) from 2003 to 2014.

Materials and Methods

A retrospective cohort study assessing maternal thyroid function in pregnancy and obstetric and perinatal outcomes in 8,413 pregnant women and their offspring seen at BMC from 2003 through 2014 was conducted using electronic medical records. BMC is a tertiary academic medical center and a safety-net hospital, serving largely underserved populations. Institutional review board approval for the study was obtained from the Boston University Medical Campus and Boston Medical Center, Boston, Massachusetts, US.

Study subjects

Subjects were identified from electronic medical records at BMC from January 1, 2003 to May 22, 2014 using the following inclusion and exclusion criteria. The inclusion criteria for subjects were (i) pregnant women at least 18 years of age seen in outpatient clinics for prenatal care at BMC between January 1, 2003 and May 22, 2014; (ii) pregnant women with at least one serum TSH measurement during the prenatal period; and (iii) children born or pregnancy outcomes recorded at BMC between January 1, 2003 and May 22, 2014. Exclusion criteria were (i) pregnant women less than 18 years of age; (ii) women with twin or higher order gestations; (iii) women for whom pregnancy outcomes were unknown, such as those who did not deliver at BMC or women who received care for miscarriage or neonatal demise at another institution; and (iv) women with a previous diagnosis of thyroid disease or those who were taking thyroid medications or other medications known to affect thyroid function, such as steroids or lithium, during pregnancy. Woman were included more than once if they had more than 1 pregnancy meeting inclusion criteria during the study period. Live births, miscarriages, stillbirths, and neonatal deaths (as recorded during the hospitalization immediately following birth) were identified from the medical records and linked with maternal medical records of pregnant women in the study cohort.

Study measurements

Maternal data and information on obstetric and perinatal outcomes and complications were collected from existing electronic medical records using International Classification of Disease (ICD)-9 codes.

Exposure variables assessed included maternal serum thyroid function during pregnancy including TSH, FT4, total triiodothyronine, and TPO antibody or thyroid stimulating immunoglobulin status as available. Measurement of thyroid function in pregnancy is done at discretion of treating physician at BMC. If a pregnant woman had multiple TSH values measured throughout pregnancy, the first TSH measurement of the study pregnancy was used. Based on the 2017 ATA pregnancy guidelines (6), serum TSH > 4 mIU/L was considered elevated. Normal thyroid function was defined as serum TSH between 0.1 mIU/L and 4 mIU/L. Obstetric outcomes assessed included miscarriage/fetal loss, preterm labor (based on ICD-9 classification), cesarean section, placental abruption, preeclampsia/eclampsia, gestational hypertension, and gestational diabetes. Perinatal outcomes assessed included prematurity (defined as birth at <37 weeks gestation), LBW (defined as ≤2500 g), admission to neonatal intensive care unit (NICU), APGAR score at 1 and 5 min, neonatal death (based on ICD-9 classification), and respiratory distress syndrome (RDS). Gestational age was determined based on the report on the delivery records, which were defined based on dating ultrasound.

Maternal information considered as potential confounders included age at delivery, race and ethnicity as reported in medical records, highest education level, insurance type (used as a surrogate marker for socioeconomic status), smoking status assessed at initial and last prenatal visits prior to delivery, alcohol use assessed at initial and last prenatal visit prior to delivery, height at initial prenatal visit, maternal weight at initial and last prenatal visit prior to delivery, and medical history including history of fetal loss or preterm delivery prior to the pregnancy of interest, history of hypertension or diabetes prior to pregnancy of interest, gestational hypertension, and gestational diabetes. Body mass index (BMI) at initial and last prenatal visits was calculated by the following formula = weight in kilograms/(height in meters)². BMI was considered underweight if <18, normal if ≥18 and <25, overweight if ≥25 and <30, obese if ≥30 and <35, and morbidly obese if ≥35. All pregnant women with diabetes in our cohort had type 2 diabetes, and none had type 1 diabetes. Offspring gender was also assessed as a potential confounder.

Statistical analyses

Descriptive statistics are presented as mean (standard deviation [SD]) or median (range) as appropriate. Two sample t-tests for continuous variables and chi-square test or Fisher’s exact tests for categorical variables were used to compare the distribution of demographics between those with TSH measurements and those without TSH measurement.

Log-binomial generalized estimating equation (GEE) models were used to assess crude associations between maternal TSH > 4 mIU/L and each obstetric and perinatal outcome to account for multiple pregnancies in the same women. Based on the effect size from the crude analyses, prematurity and RDS were 2 primary outcomes of interest in our study. Log-binomial GEE models were also used for multivariable analyses to assess the impact of potential confounders. Each adverse obstetric and perinatal outcome was assessed separately. Maternal age, maternal race and ethnicity, maternal insurance category (surrogate marker for socioeconomic status), and highest maternal education level were considered “core” confounders and included in all multivariable analyses. Additional confounders were identified from the remaining covariates by adding them to the core multivariate model one at a time. Confounders that resulted in at least a 10% additional change in the adjusted RRs were retained for the final multivariate model. The covariates included in the multivariable analyses are listed in Supplementary Material in Reference (28). For statistically significant multivariate association, potential effect modification was assessed for each statistically significant covariate.

Although serum FT4, total triiodothyronine, TPO antibody, and stimulating immunoglobulin levels were collected, these were not included in the analyses because of the small number of subjects with these measurements.

The following secondary multivariable analyses for each obstetric and perinatal outcome were performed using the same covariates identified in the primary analyses and the GEE model including (i) only those with TSH measurements done in the first trimester, (ii) using the 2011 ATA guidelines for trimester-specific cutoff for serum TSH levels during pregnancy (18), and (iii) using the TSH cutoff of 10 mIU/L.

All statistical analyses were performed using SAS version 9.3. A two-tailed P-value < .05 was considered statistically significant.

Results

Of the 27,697 pregnant women seen for prenatal care at BMC during the study period, a total of 8,413 women (mean age 29.1 years, 15% white, 60% black, and 13% Hispanic) with maternal serum TSH levels measured during pregnancy were included in the analyses. Maternal demographic information for women with and without TSH measurements during pregnancy are presented in Table 1. The mean age of women with TSH measured during pregnancy was slightly higher compared to those without TSH measured during pregnancy (29.1 vs 28.2 years). Black women, non-Hispanics, women with commercial or private insurance, smoking or alcohol use during pregnancy, history of hypertension, and history of fetal loss were more likely to have TSH measured during pregnancy. The median (interquartile range) serum TSH value was 1.06 (0.62–1.60) mIU/L. A total of 130 women (1.6%) had serum TSH levels > 4 mIU/L during pregnancy, and 16 women (0.2%) has serum TSH levels > 10 mIU/L.

Table 1.

Maternal Demographics of Subjects With TSH Measured and Without TSH Measured During Pregnancy

Demographic variable	TSH measured during pregnancy	TSH not measured during pregnancy	Comparison (P-value)
Age (years): mean (SD)	29.1 (6.1)	28.2 (6.1)	<.0001
Race (%)			<.0001
White	15	13
Black	60	36
Hispanic	14	40
Other	11	11
Ethnicity (%)			<.0001
Hispanic	13	38
Non-Hispanic	87	62
Highest education level (%)			<.0001
8th grade or less	11	17
High school	56	52
College and above	33	31
Insurance category (%)			<.0001
Public	79	88
Commercial/private	21	12
Smoking during pregnancy (%)			<.0001
Yes	7	2
No	93	98
Alcohol consumption during pregnancy (%)			<.0001
Yes	2	0.2
No	98	99.8
BMI at initial prenatal visit (%)			<.0001
Underweight	2	1
Normal	38	31
Overweight	30	31
Obese	17	21
Morbidly obese	13	16
BMI at last prenatal visit (%)			.03
Underweight	0.1	0.2
Normal	11	13
Overweight	33	33
Obese	30	29
Morbidly obese	26	24
Family history of thyroid disease (%)
Yes	0.1	0.01	.006
No	99.9	99.9
History of type 2 diabetes (%)			<.0001
Yes	0.8	0.3
No	99.2	99.7
History of hypertension (%)			<.0001
Yes	2	0.6
No	98	99.4
History of fetal loss (%)			<.0001
Yes	2	0.7
No	98	99.3
History of preterm delivery (%)			<.0001
Yes	0.6	0.2
No	99.4	99.8

Abbreviations: TSH, thyroid stimulating hormone; BMI, body mass index

A total of 8,413 fetuses and newborns (52% male) of the women with TSH measurements during pregnancy were included in the final analyses. The mean (SD) gestational age (GA) of 6,100 offspring whose GA could be calculated was 38.5 (3.3) weeks. The mean (SD) birth weight was 3155.5 (869.2) g. The median (range) APGAR score at was 8 (0–10) at 1 min and 9 (0–10) at 5 min.

The incidence of adverse obstetric and perinatal outcomes assessed in the study cohort is presented in Table 2. Preterm labor, cesarean section, and gestational diabetes were frequent, as was the incidence of prematurity and LBW in offspring.

Table 2.

Occurrence of Adverse Obstetric and Perinatal Outcomes in Study Cohort

Outcome	Number (%)
Obstetric outcomes
Fetal loss	390 (4.6%)
Preterm labor	878 (10.4%)
Placental abruption	187 (2.2%)
Eclampsia/preeclampsia	531 (6.3%)
Cesarean section	2857 (34.0%)
Gestational HTN	530 (6.3%)
Gestational DM	863 (10.3%)
Perinatal outcomes
Prematurity (GA < 37 weeks)	780 (12.8%)
Neonatal death	0 (0%)
Respiratory distress syndrome	142 (1.7%)
Admission to NICU	454 (5.4%)
Low birth weight (≤2500 g)	1075 (12.8%)

Abbreviations: DM, diabetes; GA, gestational age; HTN, hypertension; NICU, neonatal intensive care unit

The results of crude and adjusted multivariable regression analyses assessing the relationship between TSH > 4 mIU/L and adverse pregnancy outcomes are presented in Table 3. Maternal serum TSH levels > 4 mIU/L compared to 0.1 mIU/L ≤ TSH ≤ 4 mIU/L were associated with a 2.17-fold increased risk of prematurity in offspring (RR 2.17; P-value .016; 95% confidence interval [CI] 1.15, 4.07). Maternal serum TSH levels > 4 mIU/L were also associated with a 2.83-fold increased risk of RDS in offspring (RR 2.83; P-value .046; 95% CI 1.02, 7.86). Although not statistically significant, there were also increased RR for fetal loss, preeclampsia/eclampsia, and low birth weight with maternal serum TSH levels > 4 mIU/L, with adjusted RRs of 1.62, 1.44, and 2.14, respectively.

Table 3.

Crude and Adjusted Risk Ratios of Adverse Obstetrics and Perinatal Outcomes with Maternal TSH > 4 mIU/L in Pregnancy from Multivariable Regression Models

		Multivariable adjusted analyses
Outcomes	Crude risk ratio	Estimate	Risk ratio	95% Confidence Interval	P-value
Obstetric outcomes
Fetal loss	1.44	0.481	1.62	0.62, 3.97	.34
Preterm labor	1.00	0.265	1.30	0.51, 3.33	.58
Placental abruption	1.30	0.034	1.03	0.24, 4.41	.96
Eclampsia/preeclampsia	1.10	0.368	1.44	0.70, 2.98	.32
Cesarean Section	1.26	0.020	1.02	0.60, 1.74	.94
Gestational HTN	0.64	–0.604	0.55	0.20, 1.51	.24
Gestational DM	1.11	–0.313	0.73	0.32, 1.67	.46
Perinatal Outcomes
Prematurity (GA < 37 weeks)	1.72	0.774	2.17	1.15, 4.07	.016
Respiratory distress syndrome	1.89	1.039	2.83	1.02, 7.76	.046
Admission to NICU	1.65	0.263	1.30	0.37, 4.52	.68
Low birth weight (≤2500 g)	1.80	0.759	2.14	0.78, 6.07	.15
APGAR score at 1 min	0.51	–0.285	0.75	0.46, 1.24	.26
APGAR score at 5 min	0.89	–0.152	0.86	0.60, 1.23	.40

Maternal age, maternal race and ethnicity, maternal insurance category (surrogate marker for socioeconomic status), and highest maternal education level were considered “core” confounders and included in all multivariable analyses. For complete list of covariates in each multivariable analyses, please refer to Supplementary Material in Reference (28).

Abbreviations: SE, standard error; HTN, hypertension; DM, diabetes; GA, gestational age; NICU, neonatal intensive care unit.

^aStatistically significant at α < .05

There were more than 30% change in the RRs from crude analyses to adjusted analyses for eclampsia/preeclampsia (31%), gestational diabetes (34%), RDS (50%), and APGAR score at 1 min (47%). After multivariable analyses with only “core” variables as described in the method section, the contribution of each additional confounder to the multivariable analyses for these outcomes was examined. For eclampsia/preeclampsia, the last BMI was the greatest contributor, resulting in a 32.9% change in the RR. For gestational diabetes, the addition of the last BMI to the model resulted in 624.5% change in RR, and addition of a history of hypertension resulted in an 815% change in the RR. For RDS, addition of the last BMI resulted in a 56.8% change in the RR. For APGAR score at 1 min, addition of prematurity resulted in a 69.5% change in the RR.

The full results of multivariable regression analyses for prematurity are presented in Table 4. Increasing maternal age was statistically significantly associated with increased prematurity risk when controlled for other variables, in addition to TSH > 4 mIU/L. “Other” race was statistically significantly associated with decreased prematurity risk when controlled for other variables. There was a trend for an association between Latino race and increased prematurity risk when controlled for other variables. There was no effect modification by maternal age or maternal race. Table 5 shows the full results of multivariable regression analyses assessing RDS. When controlled for other variables, increased BMI at last visit was statistically significantly associated with RDS. The presence of preeclampsia or eclampsia during pregnancy, and cesarean section was statistically significantly associated with increased risk of RDS. There was no effect modification by BMI at the last visit, presence of eclampsia or preeclampsia, or cesarean section.

Table 4.

Multivariable Regression Model Assessing Prematurity

Covariates	Estimate (SE)	Risk ratio	95% confidence interval	P-value
Maternal agea	0.043 (0.008)	1.04	1.03, 1.06	<0.0001
Race—black	0.132 (0.136)	1.14	0.88, 1.49	0.33
Race—Latino	0.534 (0.336)	1.71	0.88, 3.29	0.11
Race—othera	–0.456 (0.216)	0.63	0.45, 0.97	0.035
Ethnicity—Hispanic	–0.160 (0.322)	0.85	0.45, 1.60	0.62
Insurance—commercial	0.033 (0.120)	1.03	0.82, 1.31	0.78
Education—high school	0.079 (0.156)	1.08	0.80, 1.47	0.61
Education—college and above	–0.217 (0.170)	0.81	0.58, 1.12	0.20
TSH > 4mIU/La	0.774 (0.322)	2.17	1.15, 4.07	0.016
TSH < 0.1 mIU/L	0.261 (0.214)	1.30	0.85,1.97	0.22
BMI at last visit	–0.016 (0.008)	0.98	0.97, 1.00	0.064

White race, non-Hispanic ethnicity, public insurance, the highest education level below 8th grades, and 0.1 mIU/L ≤ TSH ≤ 4 mIU/L were used as the reference groups.

Abbreviationss: SE, standard error; TSH, thyroid stimulation hormone; BMI, body mass index.

^aStatistically significant at α < .05.

Table 5.

Multivariable Analysis Predicting Respiratory Distress Syndrome

Covariates	Estimate (SE)	Risk ratio	95% confidence interval	P-value
Maternal age	0.018 (0.016)	1.02	0.99, 1.05	.27
Race—black	0.035 (0.276)	1.04	0.60, 1.78	.91
Race—Latino	–0.309 (0.323)	0.73	0.39, 1.38	.34
Race—other	–0.452 (0.432)	0.64	0.27, 1.48	.34
Ethnicity—Hispanic	–0.065 (0.240)	0.94	0.59, 1.50	.79
Insurance—commercial	0.049 (0.257)	1.05	0.63, 1.74	.85
Education—high school	0.345 (0.360)	1.41	0.70, 2.86	.34
Education—college and above	–0.167 (0.401)	0.85	0.39, 1.86	.68
TSH > 4 mIU/La	1.039 (0.521)	2.83	1.02, 7.86	.046
TSH < 0.1 mIU/L	0.181 (0.418)	1.20	0.53, 2.	.66
BMI at last visita	–0.052 (0.019)	0.95	0.91, 0.99	.007
Presence of eclampsia/ preeclampsiaa	1.395 (0.230)	4.03	2.57, 6.33	<.0001
Cesarean sectiona	1.177 (0.209)	3.24	2.15, 4.88	<.0001

White race, non-Hispanic ethnicity, public insurance, the highest education level below 8th grades, and 0.1 mIU/L ≤ TSH ≤ 4 mIU/L were used as the reference groups.

Abbreviations: SE,standard error; TSH, thyroid stimulation hormone; BMI, body mass index.

^aStatistically significant at α < .05.

When associations were assessed in the subset of women with serum TSH measurements done in the first trimester of pregnancy (n = 3,844), the overall results were similar, but the risk ratios (RR) for prematurity and RDS were greater (RR 2.95; P-value .006; 95% CI 1.36, 6.41 for prematurity, and RR 8.73; P-value < .001; 95% CI 2.93, 26.02 for RDS).

Secondary analyses using previous ATA recommendations for trimester-specific cutoffs for serum TSH levels during pregnancy (18) were also performed (n = 6,100). These reference ranges for the trimester-specific TSH levels were as follows: first trimester 0.1 to 2.5mIU/L, second trimester 0.2 to 3.0 mIU/L, and third trimester 0.3 to 3.0 mIU/L. By these criteria, 383 (6.3%) women had trimester-specific serum TSH values above the reference range. The results of multivariable regression analyses for TSH above these trimester-specific reference ranges were similar to the primary analyses (data not shown). Elevated maternal serum TSH levels compared to normal maternal serum TSH levels in pregnancy were associated with a 1.52-fold increased risk of prematurity in offspring (P-value .021; 95% CI 1.06, 2.16). There was also an increase in the risk of fetal loss, although it was not statistically significant (RR 1.55; P-value .11; 95% CI 0.91, 2.62). No other adverse obstetric or perinatal outcomes were associated (data not shown).

Secondary analyses to compare those with TSH > 10 mIU/L with those with TSH ≤ 10 mIU/L (n = 16) were performed (data not shown). The magnitude of the associations in these secondary analyses were not different than those of the primary analyses; however, the association with prematurity and RDS were no longer statistically significant because few women had TSH levels > 10 mIU/L.

Discussion

Our study of 8,413 pregnant women and their offspring showed that elevated maternal TSH levels > 4 mIU/L during pregnancy were associated with a 2.17-fold increase in risk of prematurity and a 2.83-fold increased risk of RDS in offspring. Within the subset of pregnant women with TSH measured in the first trimester, the RR for both prematurity and RDS in offspring were higher (2.95 and 8.73, respectively), suggesting particular importance of adequate maternal thyroid hormone levels in early pregnancy. The increased risk of prematurity with elevated maternal TSH levels has been shown in other studies (15–17,20). Although it was not statistically significant, there was a trend toward an association between elevated maternal TSH levels and an increased risk of fetal loss, consistent with what has been previously reported (7,11–14,19). The lack of statistical significance may be due to the small number of losses stemming from the underestimation of early loss, a limitation of this cohort study utilizing pre-existing medical records. There were also trends towards associations between elevated maternal TSH levels and preeclampsia/eclampsia and low birth weight. The positive associations for prematurity and RDS did not persist when a cutoff of TSH > 10 mIU/L was used. However, this is likely due to small numbers, as only 16 women in our cohort had serum TSH levels > 10 mIU/L. When trimester-specific cutoffs for serum TSH were used as recommended by 2011 ATA guidelines, the increased risk of prematurity persisted, although with a smaller RR, but the increased risk of RDS was no longer seen (RR 1.18; P-value .70; 95% CI 0.52, 2.68). The results support the change in the cutoff for serum TSH concentrations from 2.5 to 3 mIU/L to 4 mIU/L for diagnosis of hypothyroidism in pregnant women recommended in the 2017 ATA guidelines for diagnosis and treatment of thyroid disease in pregnant women. The prevalence of SCH in our cohort was lower than anticipated (1.6% vs 3.5% (29)), which may have been due to exclusion of women who were treated with levothyroxine during pregnancy.

Strengths of our study include a large number of mother–offspring pairs with diverse ethnicity, as previous studies included mostly homogenous white or Chinese populations. A large proportion of women (45.7%) in our cohort had their TSH measured during the first trimester, the most critical time for fetal development. We also assessed the potential effect of maternal SCH using the most recent guideline for TSH cutoff of 4 mIU/L, whereas many previous studies used previous trimester-specific TSH cutoff of 2.5 mIU/L in the first trimester and 3mIU/L in the second and third trimesters.

Limitations of the study include potential selection bias as we could not determine why some women had their thyroid function tested while others did not. We compared demographic information of those who had TSH measured during pregnancy and those who did not to try to ascertain potential reasons for TSH measurement during pregnancy. We found that there was a higher proportion of women with history of fetal loss among those who had TSH measured compared to those who had not (2% vs 0.7%, respectively). This is likely because of the well-recognized association between overt thyroid dysfunction and risk of fetal loss. We also found that there was a much higher proportion of Latino and Hispanic women among those who did not have TSH measured compared those who had TSH measures (40% vs 14% and 38% vs 13%, respectively). Given the retrospective nature of this study, we cannot ascertain why there were such differences in 2 groups. We can speculate that language barriers or provider preference may have played a role. Another limitation is likely underestimation of the incidence of early fetal loss, as early fetal loss may occur without clinical detection and may be evaluated and treated before patients present for prenatal visits. In addition, we did not have FT4 measurements or TPO antibody status available in most of our subjects. Isolated low FT4 levels with normal TSH levels (hypothyroxinemia) has been previously associated with fetal loss (30) and preterm labor (21). Several studies have also shown an additive risk of adverse obstetric outcomes such as miscarriage and prematurity in women with both positive TPO antibody and SCH (11,16,20). Given the lack of data on FT4 levels and TPO antibody status, we were not able to assess potential effects of isolated FT4 deficiency or positive TPO antibodies on adverse obstetric or perinatal outcomes.

In addition to conflicting data regarding the effects of maternal SCH on pregnancy outcomes, there is also lack of concrete evidence for benefit of levothyroxine treatment for women with SCH during pregnancy. A randomized controlled trial of 4,562 women in Italy compared adverse obstetric outcomes in pregnant women whose TSH was measured via either a universal screening or a case-finding approach. In the case-finding group, serum from those categorized as “low-risk” were stored until study completion to measure thyroid function. Consequently, these “low-risk” women with hypothyroidism in the case-finding group were not treated during pregnancy. On the other hand, “low-risk” women with hypothyroidism in the universal screening group were treated with levothyroxine if TSH was ≥2.5 mIU/L and TPO antibody was positive. Although overall, the study did not show differences between universal screening and case-finding approaches, treatment of the “low-risk” women in the universal screening group was associated with lower odds of the composite adverse obstetric outcome compared to the untreated “low-risk” women (31). A cohort study of 5405 pregnant women with SCH defined using insurance data showed lower odds of pregnancy loss with levothyroxine treatment if the pretreatment TSH level was over 4 mIU/L, although there were increased odds of preterm delivery, gestational diabetes, and preeclampsia (32). A randomized interventional study of 366 pregnant women with negative TPO antibody levels and SCH showed a 62% decrease in the risk of preterm delivery with levothyroxine treatment for serum TSH level > 4 mIU/L, but not >2.5 mIU/L (33). On the other hand, a small observational study did not show any significant improvement of pregnancy loss with levothyroxine treatment of maternal SCH (34). In euthyroid women with positive TPO antibody levels, some studies showed lower rate of pregnancy loss with levothyroxine treatment (35,36), while others did not show any benefit of levothyroxine treatment for miscarriage or preterm delivery rates (37). A large randomized clinical trial of 677 women with SCH did not show significant differences in pregnancy outcomes such as preterm birth, miscarriage/ still birth, placental abruption, gestational hypertension or diabetes, and preeclampsia with levothyroxine treatment for serum TSH level > 4 mIU/L (38). However, the median gestational age when the treatment was initiated was 16.6 weeks, in the second trimester.

Given the conflicting data regarding the effects of maternal SCH on pregnancy and perinatal outcomes in the current literature and the lack of certainty regarding the effectiveness of levothyroxine treatment, the need for universal screening for thyroid disease in pregnant women or treatment of maternal SCH in pregnancy remains controversial. Thus, there is substantial regional and individual variation in the rates of maternal thyroid function testing during pregnancy. A study of national laboratory data reported a pregnancy testing rate of 23% (39), while it was 84.6% at BMC (40). The case-finding approach relies on physicians’ diligence to identify appropriate patients to assess for thyroid dysfunction and may miss a significant proportion of pregnant women with hypothyroidism, since the symptoms of mild hypothyroidism are frequently nonspecific and may be similar to those of normal pregnancy.

The ATA currently recommends treating pregnant women with levothyroxine supplementation if TSH is 4 to 10 mIU/L with a positive TPO antibody titer and considering treatment in pregnant women with TSH 4 to 10 mIU/L with negative TPO antibody titers (6). Our study finds that there is a statistically significantly 2.15-fold increased risk of prematurity and 2.80-fold increased risk of neonatal RDS in offspring of women with TSH > 4 mIU/L during pregnancy; these RRs were higher when elevated maternal TSH levels were detected in the first trimester. There were also increased risks of fetal loss, preeclampsia/eclampsia, and low birth weight with elevated maternal TSH levels > 4 mIU/L during pregnancy, even though they were not statistically significant. Our findings add to the past studies suggesting a link between maternal SCH in pregnancy and premature delivery. However, more interventional studies are needed to ascertain the benefit of treatment of maternal SCH in pregnancy.

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability

Restrictions apply to the availability of data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.