If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Thyroid disorders have been associated with recurrent miscarriage. Little evidence is available on the influence of subclinical hypothyroidism on live birth rates. In this cohort study, women who had experienced miscarriage and subclinical hypothyroidism (defined as thyroid-stimulating hormone >97.5th percentile mU/l with a normal thyroxine level) were investigated; the control group included women who had experienced recurrent miscarriage and normal thyroid function. Multivariable logistic regression was used to investigate the association of subclinical hypothyroidism. Data were available for 848 women; 20 (2.4%) had subclinical hypothyroidism; 818 women (96%) had euthyroidism; and 10 (1.2%) had overt hypothyroidism. The live birth rate was 45% in women with subclinical hypothyroidism and 52% in euthyroid women (OR 0.69, 95% CI 0.28 to 1.71). The ongoing pregnancy rate was 65% versus 69% (OR 0.82, 95% CI 0.32 to 2.10) and the miscarriage rate was 35% versus 28% (OR 1.43, 95% CI 0.56 to 3.68), respectively. No differences were found when thyroid stimulating hormone 2.5 mU/l was used as cut-off level to define subclinical hypothyroidism. In women with unexplained miscarriage, no differences were found in live birth, ongoing pregnancy and miscarriage rates between women with subclinical hypothyroidism and euthyroid women.
). Couples with parental chromosome abnormalities and women with uterine anomalies, endocrine disturbances, hyperhomocysteinemia and antiphospholipid syndrome have a higher risk for RM. Despite comprehensive investigations, an underlying risk factor for RM is identified in less than 50% of couples (
). At present, no effective treatment has been established to improve the live birth rates for women with unexplained RM. For couples who experience RM, a reliable prognosis for the chance of a live birth is of utmost importance in their decision whether or not to conceive again, as RM has often distressing physical and emotional consequences (
Thyroid disorders, especially hypothyroidism, have been associated with miscarriage. Overt hypothyroidism is associated with an increased risk for miscarriage (OR 5.78, 95% CI 2.4 to 14), but also with other pregnancy complications, such as low birth weight, premature delivery, placental abruption and pregnancy-induced hypertension (
). It often presents with clinical symptoms and, therefore, most women already receive treatment before conception.
Subclinical hypothyroidism is a more common thyroid disorder among women of fertile age. It is defined as a raised serum thyroid-stimulating hormone (TSH) level above the upper limit of normal with a normal level of serum free thyroxine (T4). The prevalence of subclinical hypothyroidism has been estimated to be between 4.0% and 8.5% in the normal population and between 1.5% and 4% in pregnancy (
). To date, studies investigating an association between subclinical hypothyroidism and RM have been conflicting. Also, the effect of subclinical hypothyroidism on live birth rates in women who experience RM is unclear and limited to one published study. In this observational cohort study that compared 55 patients with RM and subclinical hypothyroidism (TSH ≥2.5 mU/l) (19% of the total cohort) with euthyroid women who experienced RM, no significant difference in the subsequent live birth rates was found (
Evidence of an association between subclinical hypothyroidism and a single miscarriage is conflicting. Two cohort studies found an increased risk of miscarriage in women with subclinical hypothyroidism compared with euthyroid women (
Interpreting the data on studies of subclinical hypothyroidism is complicated by the use of differential cut-off levels for TSH and by inter-laboratory differences using various different analysers.
No consensus has been reached on the optimal TSH cut-off level to define subclinical hypothyroidism. The American Thyroid Association recommends TSH levels to be trimester specific, with 2.5 mU/l as upper limit, and the Endocrine Society advises a TSH level of 0.1–2.5 mU/l in the first trimester (
). Recommendations on cut-off levels for diagnosing subclinical hypothyroidism at pre-conception are not available. Data cannot automatically be extrapolated to all ethnicities as TSH levels are known to be population specific with intrinsic ethnic variation (
Although subclinical hypothyroidism is associated with pregnancy complications, its association with RM remains unclear. Therefore, we evaluated the effect of subclinical hypothyroidism on live birth rates in a cohort of women with unexplained RM.
Materials and methods
The study population consisted of women 18–40 years of age with RM who presented to the Recurrent Miscarriage Clinic at the Liverpool Women's Hospital, Liverpool, UK, between 2004 and 2011. In accordance with the Special Interest Group for Early Pregnancy (European Society of Human Reproduction and Endocrinology) consensus statement, RM was defined as two or more, not necessarily consecutive, miscarriages before 20 weeks of gestation, verified by a pregnancy test, ultrasonography, or both (
Biochemical pregnancies were not included as pregnancy losses, and were categorized as ‘other pregnancies’. Unexplained RM was defined when an underlying risk factor for RM was not present. Diagnostic workup for RM included testing for antiphospholipid syndrome (lupus anticoagulant, immunoglobulin G and immunoglobulin M anti-cardiolipin antibodies (according to the Sapporo Classification Criteria 1999 for the antiphospholipid antibody syndrome (
), uterine abnormalities, thrombophilia (Factor V Leiden mutation, prothrombin gene mutation, protein C deficiency, protein S deficiency, antithrombin deficiency) and hyperhomocysteinaemia.
Women with pre-existent thyroid disease or women who were using thyroid drugs were excluded. Women were not included for analysis if their evaluation did not include thyroid function tests (TSH, T4, or both) and when no data for the outcome measure were available. The diagnostic workup of RM did not include testing for the presence of TPO-Ab according to recent guidelines, where TPO-Ab screening is not advised in routine RM workup (
). During the study period, women who had experienced RM were given no other intervention than routine prescription of folic acid. Women who became pregnant had an ultrasound scan every 2 weeks in the first trimester.
The study group was defined as women with unexplained RM and subclinical hypothyroidism (defined as a serum TSH level above the 97.5th percentile with a normal serum thyroxine (T4) level between 2.5th and 97.5th percentile). The control group consisted of euthyroid women (TSH between 2.5th and 97.5th percentile) with unexplained RM.
Data on RM diagnostic workup and subsequent pregnancy outcomes were collected. Data were anonymized before analysis. The present study was defined retrospectively.
Since 2003, thyroid function tests were measured with immunometric assay conducted on the e602 analyzer (Roche Diagnostics) with a detection limit of 0.01 mU/l and total assay variation of 2–4%. The reference ranges for TSH were 0.3–6.0 mU/l and 60–150 nmol/l for total T4.
The primary outcome was live birth rate, defined as a live birth after 24 weeks of gestation. Secondary outcome measures were ongoing pregnancy rates, defined as a pregnancy of more than 12 weeks' gestational age and miscarriage rates, defined as pregnancy loss before 20 weeks of gestational age.
For descriptive statistics, mean with SD was used. To calculate differences in outcome measures, independent sample t-tests (two tailed) were used for continuous variables with a normal distribution. For continuous variables without a normal distribution, Mann–Whitney U test was applied. To investigate the association between subclinical hypothyroidism, live birth rates, ongoing pregnancy rates and miscarriage rates, multivariable logistic regression was conducted. The covariates maternal age and number of previous miscarriages were selected a priori and added in the multivariate model. A two-tailed P < 0.05 was judged statistically significant. Statistical analyses were conducted using SPSS 20.0 (IBM Corp., USA). In a sensitivity analysis, the above analyses were repeated with a serum TSH above 2.5 mU/l with a normal serum T4 level.
Approval for this study was obtained from the Medical Regional Ethics Committee of the Liverpool Women's Hospital (Number LWH0914) on 15 March 2013.
The selection process is presented in Figure 1. Between 2004 and 2011, a total of 1956 women visited the Recurrent Miscarriage Clinic at the Liverpool Women's Hospital. One hundred and fifty-six women did not undergo full diagnostic workup for recurrent miscarriage. A total of 1800 women (92%) underwent investigations to assess associated factors for RM. In 384 women (21%), thyroid function tests were not assessed. Women with associated factors for RM were excluded: 79 women (4.4%) with antiphospholipid syndrome, 27 women (1.5%) with pre-existing thyroid disease, 25 women (1.4%) with thrombophilia and 11 women (0.6%) with uterine abnormalities. The remainder of the patients were 1274 women with unexplained RM and thyroid function tests. Data on the primary outcome were available for 848 patients (67%).
Median TSH was 1.7 mU/l (range 0.05–13.9 mU/L). The 2.5th and 97.5th percentiles for TSH were 0.5 and 4.6 mU/l, respectively. The median TT4 value was 98 nmol/L, the 2.5th and 97.5th percentiles for TT4 were 71 and 140 nmol/l, respectively. Subclinical hypothyroidism was defined as a serum TSH above 4.6 mU/L, calculated from the 97.5th percentile with normal TT4 levels. According to this definition, 20 (2.4%) women met the criteria of subclinical hypothyroidism; 818 (96%) were euthyroid. Ten (1.2%) patients had overt hypothyroidism and were excluded from further analysis.
Baseline characteristics are presented in Table 1. Only one significant difference was found: the number of first-trimester pregnancy losses (P = 0.03). In women with subclinical hypothyroidism, a lower rate of first-trimester pregnancy losses was found compared with euthyroid women (median 2 [0 – 4] versus 3 [0 – 10]). The mean maternal age in women with RM and subclinical hypothyroidism was comparable to euthyroid women (33.8 versus 32.9 years). The median TSH was 5.4 mU/l (4.6–13.9 mU/l) in the group of women with subclinical hypothyroidism and 1.6 mU/l (0.5–4.5 mU/l) in the euthyroid group.
The live birth rates were 45% (9/20) in women with subclinical hypothyroidism and 52% (428/818) in euthyroid women. With multivariate logistic regression, adjusted for maternal age and number of previous pregnancy losses, no difference was found in live birth rates between the two groups (OR 0.69, 95% CI 0.28 to 1.71) (Table 2).
Table 2Logistic regression of live births, ongoing pregnancy and miscarriage rate in women who have experienced recurrent miscarriage for thyroid stimulating hormone greater than 4.5 mU/l adjusted for maternal age and previous pregnancy losses.
The ongoing pregnancy rates were 65% (13/20) in women with subclinical hypothyroidism and 69% (562/818) in euthyroid women. The odds ratio adjusted for maternal age and number of previous pregnancy losses suggested no significant difference in ongoing pregnancy rates between the two groups (OR 0.82, 95% CI 0.32 to 2.10) (Table 2).
The miscarriage rates were 35% (7/20) in women with subclinical hypothyroidism and 28% (230/818) in euthyroid women. The odds ratio adjusted for maternal age and number of previous pregnancy losses suggested no significant difference in miscarriage rate between the two groups (OR 1.43, 95% CI 0.56 to 3.68) (Table 2).
When subclinical hypothyroidism was defined as a serum TSH above 2.5 mU/l with a normal serum TT4 level, 176 (21%) women met the criteria for subclinical hypothyroidism; 651 (77%) were euthyroid; 21 (2%) patients had overt hypothyroidism and were excluded from analysis. The live birth rates according to this definition were 55% (97/176) in women with subclinical hypothyroidism and 52% (339/651) in euthyroid women (OR 1.11, 95% CI 0.79 to 1.57). The ongoing pregnancy rates were 70% (123/176) in women with subclinical hypothyroidism and 69% (451/651) in euthyroid women (OR 1.06, 95% CI 0.73 to 1.53). The miscarriage rates were 29% (51/176) in women with subclinical hypothyroidism and 29% (186/651) in euthyroid women (OR 1.07, 95% CI 0.73 to 1.55) (Table 3).
Table 3Logistic regression of live births, ongoing pregnancy and miscarriage rate in women who have experienced recurrent miscarriage for thyroid stimulating hormone greater than 2.5 mU/l adjusted for maternal age and previous pregnancy losses.
We found a prevalence of 2.4% of subclinical hypothyroidism in a cohort of 848 of women with unexplained RM. No evidence was found of a difference in live birth rates (OR 0.69, 95% CI 0.28 to 1.71), ongoing pregnancy rates (OR 0.82, 95% CI 0.32 to 2.10) and miscarriage rates (OR 1.43, 95% CI 0.56 to 3.68) between women with subclinical hypothyroidism and euthyroid women.
Comparison of results with existing literature
The prevalence of subclinical hypothyroidism in this cohort (2.4%) corresponds with the prevalence of 1.5–4% in pregnant women and is slightly lower compared with the prevalence in non-pregnant women (4–8.5%) (
). The incidence of subclinical hypothyroidism is affected by age, sex, race, geographic location, and varies according to the TSH level used to define subclinical hypothyroidism. When a TSH cut-off level of 2.5 mU/l was chosen, a prevalence of 21% of subclinical hypothyroidism was found. This is higher compared with recent reports with an application of TSH >2.5 mU/L as a threshold, with prevalence varying between 15.0 and 15.6% (
). When TSH of over 2.5 mU/l was used as a cut-off level to define subclinical hypothyroidism, this could have resulted in an overestimation of the prevalence of women with subclinical hypothyroidism. As the diagnosis of subclinical hypothyroidism was made before conception, women tended to have higher levels of TSH compared with during pregnancy. On the other hand, women who had experienced RM could possibly have had a higher average serum TSH and sufferered more frequently from subclinical hypothyroidism than stated in this study. As reference ranges are determined with the 2.5th and 97.5th percentiles, we choose to define subclinical hypothyroidism with serum TSH above the 97.5th percentile. When using a lower cut-off for serum TSH levels, no difference in live birth rates was seen either.
The results of our study are in line with a recent study on live birth rates in women with subclinical hypothyroidism and RM (
). When the results of these two studies, which both defined subclinical hypothyroidism with a TSH of over 2.5 mU/l, were pooled in a meta-analysis, no difference in live birth rates was found (OR 1.07, 95% CI 0.79 to 1.46) (Figure 2A). A high miscarriage rate was found in our study population (28.3%). This might be explained by the finding that the most important risk factors for another miscarriage after RM are increasing maternal age and number of previous pregnancy losses (
As far as we are aware, this is the second cohort study investigating live birth rates in women who have experienced RM and subclinical hypothyroidism. We believe that our study generates more understanding of this, so far. unsolved, subject.
As RM is multifactorial, research becomes challenging in assessing the effect of a single factor in isolation. We accounted for this by excluding women with antiphospholipid syndrome, thrombophilia and uterus anomalies, which are all well-known risk factors for RM (
). Genetic analyses of the pregnancy tissues have not been conducted in our study cohort, as it was not a routine test at that time. As a result, the number of truly unexplained RM diagnoses may be lower than found in this study.
Potential limitations of the present study are the retrospective design and the risk for selection bias. The live birth rate in our study was lower than expected. This might be explained by the fact that, in 33% of our study population, data on the primary outcome measure were missing and therefore excluded from analysis. This might have introduced a selection bias as it can be assumed that women with a normal subsequent pregnancy do not always visit the tertiary RM centre again.
Despite the large database, a relatively low number of participants with subclinical hypothyroidism remained, owing to a low prevalence. Sample size calculation showed that about 24,000 women would be required to detect a 5% absolute difference in live birth rate (80% power at a two-sided alpha level of 0.05), assuming a 75% live-birth rate in the euthyroid group and 2.5% prevalence of subclinical hypothyroidism. When a TSH cut-off level of 2.5 mU/l would be used, 3600 women would be required. Therefore, this study was underpowered. In our study with 848 women, with a prevalence of 2.5% of subclinical hypothyroidism and a live birth rate of 52%, we could prove a difference in live birth rate of 28%. When a TSH cut-off level of 2.5 mU/l was applied and a live birth rate of 52%, we could prove a difference of 13% in live birth rates.
The live birth rates were 45% (9/20) in women with subclinical hypothyroidism and 52% (428/818) in euthyroid women. When multivariate logistic regression, adjusted for maternal age and previous number of pregnancy losses was conducted, however, no evidence of a difference was found in live birth rates between the two groups.
The diagnostic workup of RM did not include testing for the presence of thyroid peroxidise antibodies (TPO-Ab) according to recent guidelines, where TPO-Ab screening is not advised in routine RM workup (
). Recently more interest has been shown in thyroid auto-immunity and its association with RM; therefore, the lack of such investigations might be a limitation. A recent meta-analysis showed an increased risk of a miscarriage (OR 3.73, 95% CI 1.8 to 7.6) and recurrent miscarriage (OR 2.3, 95% CI 1.5 to 3.5) if thyroid antibodies are present (
). No evidence-based treatment advice is currently available, and therefore routine screening is still a matter of debate. Results of treatment RCTs have to be awaited.
More knowledge on pregnancy chances and prognostic factors for success in women who have experienced recurrent miscarriage is required. Although we did not find lower birth rates in women with RM and subclinical hypothyroidism in our study, such an association could not be ruled out, and further prospective studies are mandatory. It would be interesting to investigate TPO-Ab in this relationship simultaneously. Especially, as the American Thyroid Association guideline recomends defining subclinical hypothyroidism according to the TPO-Ab status (
). The results of this study support this advice. It remains to be established whether screening and subsequent treatment will improve pregnancy outcomes in women with subclinical hypothyroidism and RM. The potential benefit of any screening strategy critically depends on the relative contribution of thyroid dysfunction to adverse pregnancy outcomes and on the effect of treatment.
The authors acknowledge Linda Roberts, without whom the data collection never would have been successfully completed.
Overt and subclinical hypothyroidism complicating pregnancy.
Myrthe van Dijk attended medical school at the Academic Medical Centre in Amsterdam, the Netherlands. She is currently a resident in Obstetrics and Gynaecology in Leiden, the Netherlands. She is studying for her PhD at the Centre of Reproductive Medicine, Department of Obstetrics and Gynaecology, at the Academic Medical Centre in Amsterdam. Her research interests include early pregnancy and thyroid disorders.
Published online: September 20, 2016
Received in revised form:
Declaration: The authors report no financial or commercial conflicts of interest.