Hypothyroidism and pregnancy loss: comparison with hyperthyroidism and diabetes in a Danish population-based study

Transcription

1 Clinical Endocrinology (2016) 85, doi: /cen ORIGINAL ARTICLE Hypothyroidism and pregnancy loss: comparison with hyperthyroidism and diabetes in a Danish population-based study Stine Linding Andersen*,, Jørn Olsen and Peter Laurberg*, *Department of Endocrinology, Aalborg University Hospital, Department of Clinical Biochemistry, Aalborg University Hospital, Aalborg, Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, and Department of Clinical Medicine, Aalborg University, Aalborg, Denmark Summary Background Hypothyroidism is a common endocrine disease. The frequency of pregnancy loss in women with known hypothyroidism as opposed to women with a later diagnosis of hypothyroidism has not been evaluated and compared with other common endocrine diseases. Design Population-based cohort study using Danish nationwide registers. Participants All pregnancies in Denmark, , resulting in live birth (n = ), spontaneous abortion (n = ) or stillbirth (n = 2937) were identified together with information on maternal hypothyroidism, hyperthyroidism and diabetes. Methods Cox model was used to estimate adjusted hazard ratio (ahr) with 95% confidence interval (95%CI) for spontaneous abortion and stillbirth, reference: no hypo- or hyperthyroidism or diabetes (n = ). Results We identified 4951 pregnancies where maternal hypothyroidism was diagnosed before the pregnancy (group 1) and 2464 pregnancies where maternal hypothyroidism was diagnosed in the 2-year period after the pregnancy (group 2). In group 1, 825 pregnancies (167%) resulted in spontaneous abortion which was more frequent than in nonexposed (132%), (ahr 119 (95%CI )), and of the same magnitude as in hyperthyroidism (172%, P = 05) and diabetes (175%, P = 02) diagnosed before the pregnancy. In group 2, the frequency was 122% (ahr 092 ( )). In group 2, 16 pregnancies (065%) resulted in stillbirth which was more frequent than in nonexposed (036%), (ahr 181 ( )), of the same magnitude as in hyperthyroidism (082%, P = 05) and less frequent than in diabetes (29%, P < 0001) diagnosed after the pregnancy. In group 1, the frequency was 040% (ahr 111 ( )). Correspondence: Stine L. Andersen, Departments of Endocrinology and Clinical Biochemistry, Aalborg University Hospital, Sdr. Skovvej 15, 9000 Aalborg, Denmark. Tel.: ; stine.a@rn.dk Conclusions Hypothyroidism increased the risk of both early and late pregnancy loss as did hyperthyroidism and in particular diabetes. We hypothesize that undetected or insufficiently treated maternal disease in the pregnancy may be of causal importance. (Received 11 April 2016; returned for revision 22 May 2016; finally revised 14 June 2016; accepted 14 June 2016) Introduction Unintentional loss of a desired pregnancy is a feared and common obstetric complication which affects at least 30% of women who attempt to be pregnant and often remains undetected. 1 Pregnancy loss before entering the 23rd week of pregnancy is in Denmark referred to as spontaneous abortion, whereas in utero death in or after the 23rd week of pregnancy is classified as stillbirth. 2 Many factors related to either the pregnant woman and/ or the foetus increase the risk of pregnancy loss. Much attention has been given to the identification of maternal risk factors that can be reduced or eliminated, because this may potentially prevent pregnancy loss. 3 Hypothyroidism in women who are or may in the future become pregnant is predominantly of autoimmune origin. 4 Comprehensive literature indicates that women suffering from this disease overall have an increased risk of experiencing pregnancy complications including an increased risk of pregnancy loss. 5 7 It is, however, still uncertain which factors are the main determinants of such increased risk. 8 One proposed mechanism is that low levels of maternal thyroid hormone in pregnancy could lead to pregnancy loss or that thyroid autoimmunity per se is the involved mechanism which is supported by studies on pregnancy loss in euthyroid women with thyroid autoantibodies. 9 We used the Danish nationwide registers to evaluate the association between maternal hypothyroidism and pregnancy loss (spontaneous abortion and stillbirth). Newly diagnosed overt hypothyroidism in a pregnancy is clinically considered a highrisk situation, and these women should be carefully managed 962

2 Hypothyroidism and pregnancy loss 963 and controlled during the pregnancy. This study, however, focused on women diagnosed before the pregnancy who potentially had inadequately treated disease in the pregnancy as well as women first diagnosed in the years after the pregnancy who potentially had undetected and untreated disease in the pregnancy. Hyperthyroidism is another thyroid disease which is mainly of autoimmune origin in women of reproductive age. 10 In Denmark, no routine testing for thyroid function or thyroid autoantibodies is performed in pregnancy, but all pregnant women are tested for glycosuria at the first pregnancy visit in general practice. To set the risk into perspective, we compared the frequency of pregnancy loss in hypothyroidism with that in hyperthyroidism and in diabetes diagnosed before or after the pregnancy. Hyperthyroidism was included because it is another thyroid disease that may seriously complicate a pregnancy 11,12, and the potential benefits and harms of routine testing for both hypothyroidism and hyperthyroidism in pregnancy are a matter of discussion. 13 Diabetes was included for comparison to a nonthyroidal disease. Diabetes is a known risk factor for pregnancy complications, and the discussion on screening for diabetes in pregnancy has been even more intense and ongoing for decades Method and materials Study population and design We designed a population-based cohort study based on all clinically recognized pregnancies in Denmark from 1 January 1997 to 31 December 2008 (n = ) leading to inpatient or outpatient hospital visit (Fig. 1). The opportunity to study all diagnosed pregnancies emerged from the Danish Civil Registration System 17 and nationwide registration of health data. In Denmark, every individual is assigned a unique ten-digit personal identification number at birth which is used in all the nationwide registers in encrypted form and enables linkage between the registers. All data were linked in Statistic Denmark, and the study was approved by the Danish Data Protection Agency. Outcome of pregnancy The Danish National Hospital Register (DNHR) 18 includes diagnoses on all inpatient visits since 1977, and all inpatients and outpatients since 1995 coded according to the 8th revision of the International Classification of Disease (ICD-8) from 1977 to 1993 and ICD-10 from We identified all hospital visits with a diagnosis of spontaneous abortion (O020 O039), induced abortion (O040 O069), molar and ectopic pregnancy (O000 O019), and births ( ) including information on whether it was a singleton (Z370, Z371) or multiple (Z372 Z377) stillbirth or live birth. Information on gestational age (GA) was registered in the DNHR as completed week plus days calculated from the first day of the last menstrual period, for example is the first day in the 9th week of pregnancy. We included all pregnancies with a registered GA in the range from (first day in the third week of pregnancy) to weeks (last day in the 45th week of pregnancy) at the end of the pregnancy. The date of pregnancy start (first day of last menstrual period) was estimated by subtracting GA (in days) from the date the pregnancy had ended. We used both the registered hospital diagnosis and GA to define the outcomes of pregnancy. Spontaneous abortion was defined as diagnosis of spontaneous abortion with a registered GA in the range from 2 + 0to21+ 6 (before the 23th week of pregnancy). 2 Stillbirth was defined as a diagnosis of stillbirth with a registered GA in the range from to44+ 6 (23rd up to and including the 45th week of pregnancy). 2 Live births were included when the diagnosis was live birth and the registered GA was or more. In Denmark, the definition of stillbirth was changed from (29th week of pregnancy) to (23rd week of pregnancy) in Thus, the birth of a child with no signs of life in the 23rd to the 28th week of pregnancy would have been registered as a spontaneous abortion in the period from 1997 to These pregnancies were excluded from the main analyses because information on the type of pregnancy (singleton/multiple) and maternal smoking status in pregnancy was not registered by midwifes when the diagnosis was spontaneous abortion. Similarly, pregnancies with missing information on GA or maternal covariates were excluded as were pregnancies resulting in induced abortion, molar pregnancy or ectopic pregnancy (Fig. 1). Maternal endocrine diseases Information on maternal endocrine diseases was obtained from the DNHR. 18 Inpatient and outpatient visits in the period from 1 January 1977 and up to two years after the end of the pregnancy were included. Hypothyroidism was defined by ICD-8: and ICD-10 groups: E03 and E89; hyperthyroidism by ICD-8: and ICD-10 group: E05; and diabetes by ICD-8: and ICD-10 groups: E10-E14. The Danish National Prescription Register (DNPR) 20 holds data on all prescription drugs redeemed from Danish pharmacies since 1995 including the type of drug prescribed according to the Anatomical Therapeutic Chemical (ATC) classification system and the date of sale. Thyroid hormones (H03A), antithyroid drugs (ATD) (H03B) and antidiabetics (A10) are sold solely as prescription drugs in Denmark, and we identified all prescriptions redeemed between 1 January 1995 and up to 2 years after the end of the pregnancy. Women were classified with hypothyroidism if they had no registration of hyperthyroidism (diagnosis or redeemed prescription of ATD) and had either a diagnosis of hypothyroidism and/ or thyroid surgery and at least one prescription of thyroid hormone or minimum two redeemed prescriptions of thyroid hormone. Women with a diagnosis of hypothyroidism before 1 January 1995 and no prescription of thyroid hormone were included as treatment may have ended before the registration was initiated (n = 61). After the identification of women with hypothyroidism, the remaining women were classified with hyperthyroidism and diabetes from similar criteria (diagnosis

3 964 S. L. Andersen et al. Fig. 1 Flowchart illustrating the selection of the pregnancy outcomes under study and exposure groups. plus one redeemed prescription or minimum two redeemed prescriptions). The time of diagnosis of maternal endocrine disease was defined by the date of the first hospital contact with a diagnosis of the disease or by the date the first prescription was redeemed, whichever came first. Women identified with onset of disease in the pregnancy (in the time period between the estimated pregnancy start and the end of the pregnancy) were excluded from the study (Fig. 1). The nonexposed group was defined as pregnancies with no registration of maternal hypothyroidism, hyperthyroidism or diabetes up to 2 years after the pregnancy. Two exposure groups were defined: (1) maternal endocrine disease diagnosed before the pregnancy and (2) maternal endocrine disease diagnosed for the first time in the 2-year period after the pregnancy. Statistical analyses In the analyses of spontaneous abortion (follow-up from pregnancy start to the end of the 22nd pregnancy week), all pregnancies were included. In the analyses of stillbirth (followup from pregnancy start to the day of birth), all pregnancies resulting in spontaneous abortion and multiple births were excluded. Induced abortions, molar and ectopic pregnancies were excluded from the main analyses, but included in a sensitivity analysis and censored from follow-up at the time of the event. The Cox proportional hazard model with GA as the underlying time-scale was used to estimate crude and adjusted hazard ratio (HR) with 95% confidence interval (95% CI) for spontaneous abortion and stillbirth in pregnancies exposed to maternal hypothyroidism vs nonexposed pregnancies. The proportional hazards assumption was justified in plots of log cumulative hazards and by Schoenfeld residuals. Robust standard error was used to account for dependency between maternal multiple pregnancies in the study period. Information was available in the DNHR and from Statistic Denmark on maternal age, parity, income, cohabitation, origin, residence and the year of pregnancy which was included in the adjusted model as categorical explanatory variables (defined in Table 1). Information on

4 Hypothyroidism and pregnancy loss 965 Table 1. Maternal characteristics at the time of the pregnancy in pregnancies with maternal hypothyroidism diagnosed before or in the 2- year period after the pregnancy and in nonexposed pregnancies maternal smoking in pregnancy was only available for the analysis of stillbirth. The distributions of gestational weeks at the end of the pregnancy were compared using nonparametric statistics (Mann Whitney U-test), and the frequency of pregnancy loss in nonexposed pregnancies vs pregnancies with maternal endocrine disease was compared using v 2 -test. Statistical analyses were performed using STATA version 11 (Stata Corp., College Station, Texas, USA). A 5% level of significance was chosen. Results Hypothyroidism Nonexposed n % n % Pregnancies Age (years) < Parity* Cohabitation Married Not married Income (quartiles) 1st (lowest) nd rd th Origin Born in Denmark Not born in Denmark Residence West Denmark East Denmark *Previous pregnancies in the study period including index pregnancy. Divided by the Great Belt. Altogether pregnancies were included in the study (Fig. 1) and resulted in live birth (n = ), spontaneous abortion (n = ) or stillbirth (n = 2937). A total of 7806 pregnancies were included because maternal hypothyroidism had been diagnosed before or in the 2-year period after the pregnancy, a total of pregnancies were included because maternal hyperthyroidism or diabetes had been diagnosed before or in the 2-year period after the pregnancy and finally pregnancies were included in the nonexposed group (Fig. 1). In pregnancies where the mother had been diagnosed with hypothyroidism, the pregnant woman was older with a higher parity and was more often not born in Denmark (Table 1). Among the 7806 pregnancies classified with maternal hypothyroidism, the woman was diagnosed with hypothyroidism prior to the pregnancy in 4951 of the pregnancies and first time in the 2-year period after the pregnancy in 2855 of the pregnancies. When maternal hypothyroidism was known prior to the pregnancy, 167% of the pregnancies resulted in spontaneous abortion which was more frequent than in nonexposed and in pregnancies where maternal hypothyroidism was diagnosed after the pregnancy with similar findings in crude and adjusted follow-up analyses (Table 2). For the analyses of stillbirth, pregnancies resulting in spontaneous abortion and multiple births were excluded, leaving pregnancies in the nonexposed control group and 6475 pregnancies as exposed to maternal hypothyroidism. Among nonexposed, the frequency of stillbirth was 036% illustrating that late pregnancy loss is a rare event compared with spontaneous abortion. The frequency of stillbirth was of the same magnitude when maternal hypothyroidism was diagnosed before the pregnancy, but not in the group of pregnancies where maternal hypothyroidism was diagnosed in the 2-year period after the Table 2. Crude and adjusted hazard ratio (HR) with 95% confidence interval (95% CI) for spontaneous abortion and stillbirth in pregnancies where maternal hypothyroidism was diagnosed before or first diagnosed in the 2-year period after the pregnancy vs nonexposed pregnancies n % HR 95% CI Spontaneous abortion* Nonexposed (n = ) Reference Hypothyroidism diagnosed before the pregnancy (n = 4951) Crude Adjusted Hypothyroidism diagnosed after the pregnancy (n = 2855) Crude Adjusted Stillbirth Nonexposed (n = ) Reference Hypothyroidism diagnosed before the pregnancy (n = 4011) Crude Adjusted Hypothyroidism diagnosed after the pregnancy (n = 2464) Crude Adjusted *Analyses included pregnancies terminated with live birth, stillbirth or spontaneous abortion. Adjusted model included the following: year of pregnancy termination, maternal age, parity, origin, geographical residence, income and cohabitation. Analyses included pregnancies terminated with singleton live birth or stillbirth.

5 966 S. L. Andersen et al. pregnancy. In the latter group, altogether 16 pregnancies had resulted in stillbirth which was about twice as frequent as in the nonexposed group and associated with an increased risk of stillbirth both in the crude and in the adjusted analysis (Table 2). After additional adjustment for maternal smoking in pregnancy, an association between maternal hypothyroidism diagnosed after the pregnancy and stillbirth was observed (adjusted HR 175 (95% CI )). The frequency of spontaneous abortion and stillbirth in women with hypothyroidism was compared with the frequency of a similar pregnancy loss in women with hyperthyroidism and diabetes (Fig. 2). Overall, the frequency of foetal loss and the distribution of events in women with known and with later diagnosed disease were rather similar for hypothyroidism and hyperthyroidism. For diabetes, an increased frequency of early and late pregnancy loss was observed both when the disease was diagnosed before and after the pregnancy, and especially, the frequency of foetal loss was high when diabetes was first diagnosed in the 2-year period after the pregnancy. To further elaborate on these findings, we evaluated the gestational week at the end of the pregnancy in nonexposed pregnancies and in pregnancies exposed to maternal endocrine disease. For spontaneous abortion (Fig. 3), the median week at the end of the pregnancy was the 9th week of pregnancy in a positively skewed distribution (upper figure). The distribution was similar in pregnancies with maternal hypothyroidism diagnosed before the pregnancy, whereas for pregnancies where maternal hyperthyroidism or diabetes was diagnosed before the pregnancy, the distribution was significantly different with a tendency towards earlier ending of the pregnancy (Fig. 3). For stillbirth (Fig. 4), the median week at the end of the pregnancy was the 37th week of pregnancy. For each of the endocrine diseases, the individual cases of stillbirth are depicted in Fig. 4 according to the week the pregnancy had ended and stratified by whether maternal disease was diagnosed before or after the pregnancy. For hyperthyroidism, no difference was observed in the distribution of pregnancy weeks, whereas for hypothyroidism and diabetes, the week the pregnancy had ended tended to be later when maternal disease was first diagnosed after the pregnancy. Sensitivity analyses (data not shown) revealed consistent findings when restricted to the woman s first pregnancy in the study period, to women who were only pregnant once during the study period and to pregnancies in the years , although data on stillbirth were limited. Similar associations were observed when induced abortions were included in the analyses and censored from follow-up at the time of the event. Results for hypothyroidism did not change when women with a diagnosis of iatrogenic hypothyroidism were excluded or when analyses were restricted to women who redeemed prescriptions for a period of more than 1 year. Similarly, the findings for diabetes were the same in the group of women who only redeemed prescriptions of insulin and in the group who redeemed prescriptions of antidiabetic drugs +/ insulin. Discussion Principle findings Fig. 2 Frequency of spontaneous abortion (upper figure) and stillbirth (lower figure) in pregnancies with maternal hypothyroidism, hyperthyroidism or diabetes diagnosed before or in the 2-year period after the pregnancy. The dotted horizontal line illustrates the frequency of spontaneous abortion (upper figure) and stillbirth (lower figure) in nonexposed pregnancies. Frequencies significantly different from nonexposed pregnancies (P < 0001 in v 2 -test) are marked with *. Notice the different scaling on the y-axis. In a Danish population-based study including all clinically recognized pregnancies in Denmark and indicators of exposure from maternal diagnosis and treatment of hypothyroidism, an increased risk of spontaneous abortion was seen in women with known hypothyroidism at the time of pregnancy and women first diagnosed with hypothyroidism in the years following the pregnancy had an increased risk that the pregnancy resulted in a stillbirth. The findings were similar in hyperthyroidism, but even more pronounced in diabetes. Thyroid hormones and pregnancy loss Thyroid hormones are essential developmental factors and perform their regulatory action via binding to nuclear thyroid

6 Hypothyroidism and pregnancy loss 967 Fig. 3 Histograms illustrating the distribution of gestational week at the end of pregnancy in pregnancies resulting in spontaneous abortion (nonexposed and with maternal hypothyroidism, hyperthyroidism or diabetes diagnosed before the pregnancy). Gestational week was calculated from the first day of the last menstrual period. Gestational week 3 is the third week of pregnancy ranging from to 2 + 6, and so on up to gestational week 22 which is the 22nd week of pregnancy ranging from to P-values are result of the Mann Whitney U-test for comparison of each exposures to nonexposed pregnancies. Notice the different scaling on the y-axis. Fig. 4 Histograms illustrating the distribution of gestational week at the end of the pregnancy in pregnancies resulting in stillbirth. The upper figure illustrates the distribution in nonexposed pregnancies, the lower figures in pregnancies with maternal hypothyroidism, hyperthyroidism or diabetes diagnosed before (upper part of figures) or in the 2-year period after the pregnancy (lower part of the figures). Gestational week was calculated from the first day of the last menstrual period. Gestational week 23 is the 23rd week of pregnancy ranging from to , and so on up to gestational week 45 which is the 45th week of pregnancy ranging from to P-values are result of the Mann Whitney U-test for comparison between the group diagnosed before and the group diagnosed in the 2-year period after the pregnancy. Notice the different scaling on the y-axis. hormone receptors. Comprehensive evidence from both experimental and human studies describes the role of thyroid hormones in female reproduction and points towards a regulatory role of thyroid hormones before, during and also after implantation of the fertilized egg in the uterus. 21 Considering the role of

7 968 S. L. Andersen et al. thyroid hormones after implantation, thyroid hormone receptors and type 3 iodothyronine deiodinases are present in the uteroplacental unit, and alterations in the local activity of thyroid hormones within this unit have been proposed to play a role in pregnancy loss. 21. Spontaneous abortion Spontaneous abortion is the most common obstetric complication and potential risk factors have been extensively studied. 3 An increased frequency of early pregnancy loss in women suffering from autoimmune hypothyroidism and in euthyroid thyroid autoantibody positive women has been consistently reported, but as recently reviewed in detail, the underlying pathophysiological aspects are not clear. 8 It still remains uncertain whether low levels of thyroid hormones, a direct pathogenic effect of thyroid autoantibodies, or both, are involved. Adding to this, individuals with thyroid autoantibodies tend to have a higher TSH. 22 We observed an increased risk of spontaneous abortion in women diagnosed with hypothyroidism before the pregnancy, and we hypothesize that insufficiently treated hypothyroidism is the most likely mechanism involved. No increased risk was observed when hypothyroidism was first diagnosed after the pregnancy which we would expect if thyroid autoantibodies per se were the main determinants. The strength of our population-based design was the possibility to study more than pregnancies resulting in early pregnancy loss and to compare the risk of pregnancy loss in women diagnosed with hypothyroidism before and after the pregnancy. We had no information on results of thyroid function tests or thyroid autoantibodies, but our results are in line with previous reports including actual measurements of thyroid function in pregnancy. Several studies from different countries have shown that women with hypothyroidism may not be adequately treated by the time they become pregnant Taylor et al. observed that half of levothyroxine-treated women had a TSH above 25 mu/l in early pregnancy and an increased risk of early pregnancy loss in women with a TSH above 45 mu/l. 25 Stillbirth Stillbirth is a rare, but serious, obstetric complication late in pregnancy. We observed an increased risk of stillbirth in women first diagnosed with hypothyroidism after the pregnancy, and we hypothesize that these women had undetected low levels of thyroid hormones in the pregnancy which could have influenced the maintenance of the pregnancy. No increased risk was observed when hypothyroidism was diagnosed before the pregnancy which we would expect if thyroid autoantibodies per se were the main determinants. The similar findings for hyperthyroidism 12 and diabetes diagnosed after the pregnancy support the concern of undetected maternal disease in the pregnancy. The strength of our nationwide design was the inclusion of almost 3000 pregnancies resulting in stillbirth, and the possibility to study this rare outcome of pregnancy with sufficient power, but the analyses on stillbirth were less robust than the analyses on spontaneous abortion which is a much more frequent outcome of pregnancy. We had no information on results of thyroid function tests or thyroid autoantibodies, but our results are in line with studies including such measurements. 27,28 In a recent study by Nijkamp et al., 28 the authors studied 896 women who had a stillbirth and had TSH and ft4 measured in a blood sample drawn in the hospital after stillbirth was detected, but before delivery. Among the 875 women with no history of thyroid disease, 16% had overt hypothyroidism, 46% had subclinical hypothyroidism and 10% had hypothyroxinaemia. The authors discussed that these prevalences were higher than in the general pregnant population, but concluded that there was no specific underlying cause of stillbirth in women with thyroid dysfunction and that routine testing for thyroid dysfunction after a stillbirth is not justified. The study by Nijkamp et al. 28 emphasizes that thyroid disease is not the predominant cause of stillbirth, but one of many risk factors. Thus, maternal thyroid disease may overall show little impact on the population prevalence of stillbirth. On the other hand, the detection and control of maternal thyroid disease in pregnancy and the potential prevention of a pregnancy loss may benefit considerably at an individual level. Routine testing in pregnancy In Denmark, no routine testing of thyroid function in pregnant women is implemented, but it is recommended to perform thyroid function testing if the woman has a family history of thyroid disease or other autoimmune diseases. We observed that the risk of spontaneous abortion was higher in women with known hypothyroidism, whereas no increased risk was observed in women diagnosed after the pregnancy. This finding stresses the importance of careful management and control of women with known hypothyroidism, but is not an argument to do general screening. On the other hand, the risk of stillbirth was higher in women diagnosed after the pregnancy and not increased in women with known disease. This finding may suggest a possible benefit of screening for maternal hypothyroidism in pregnancy. However, only an indicator of exposure in pregnancy was available and it is likely that only a proportion of the women identified with hypothyroidism after the pregnancy had abnormal thyroid function already in the pregnancy. For example, some of the women who developed severe postpartum thyroiditis that led to permanent hypothyroidism after the pregnancy may have been euthyroid during the pregnancy. Such women would have been misclassified as hypothyroid during the pregnancy in the present study. If the hypothesis of the present study is corroborated in studies with actual measurement of maternal thyroid function in stored biobank sera from pregnant women, such more exact classification of exposure is expected to strengthen the association further. In the present study, we compared the frequency of pregnancy loss in women suffering from hypothyroidism with that in hyperthyroidism and diabetes. Hypo- and hyperthyroidism showed very similar results, 12 whereas for diabetes, the risk of

8 Hypothyroidism and pregnancy loss 969 both early and late pregnancy loss was even more pronounced. Diabetes in pregnancy is a known risk factor for maternal and foetal complications and strategies for detection, management and control of maternal hyperglycaemia are considered worldwide. 29 In Denmark, selective screening for diabetes in pregnant women is performed with an oral glucose tolerance test in case of glycosuria or if the woman has other known risk factors. 19 Such selective screening approach will, however, not identify all cases of diabetes in pregnancy. 30 The high risk of pregnancy loss in women with diabetes observed in the present study adds to the hypothesis that insufficiently treated or undetected and untreated maternal endocrine disease was an underlying mechanism in the increased risk of pregnancy loss in diabetes as well as in hypo- and hyperthyroidism. The effect of such an intervention should, however, be documented in large randomized controlled trials. Methodological considerations We studied all clinically recognized pregnancies in Denmark leading to hospital visit. This could potentially induce bias if the likelihood of early pregnancy detection or referral to hospital is different among exposed and nonexposed. The pregnancy week distribution for spontaneous abortion was similar in hypothyroidism and in nonexposed, but modestly skewed towards earlier weeks for hyperthyroidism and diabetes, although this did not influence the median. Referral to hospital may be more likely in women with a previous pregnancy loss, but our results were consistent when analyses were restricted to first pregnancy. We considered bias from induced abortions (incomplete followup), 26 but the associations did not change when induced abortions were included in the follow-up analyses and censored at the time of the event. The validity of the Danish nationwide registers is in general considered high, 18,20 and we defined maternal thyroid disease as well as diabetes by medical treatment of the disease which further strengthens the classification of exposure. On the other hand, we did not include women with a diagnosis of disease who did not receive medical treatment, and we were not able to distinguish between subtypes of hypothyroidism, hyperthyroidism and diabetes. Conclusion Hypothyroidism was associated with an increased risk of pregnancy loss in a Danish population-based design with sufficient power to study even rare events. Indicators of exposure in pregnancy and comparison with hyperthyroidism and diabetes suggest that insufficiently treated or undetected maternal disease in pregnancy may be one of perhaps several underlying mechanisms. Studies with actual measurement of maternal thyroid function in pregnancy are needed. Declaration of interest Nothing to declare. References 1 Wilcox, A.J., Weinberg, C.R., O Connor, J.F., et al. (1988) Incidence of early loss of pregnancy. The New England Journal of Medicine, 319, Statens Serum Institut (2014) Fællesindhold for basisregistrering af sygehuspatienter. Vejledningsdel, 6, Feodor Nilsson, S., Andersen, P., Strandberg-Larsen, K., et al. (2014) Risk factors for miscarriage from a prevention perspective: A nationwide follow-up study. BJOG, 121, Carle, A., Laurberg, P., Pedersen, I.B., et al. (2006) Epidemiology of subtypes of hypothyroidism in Denmark. European Journal of Endocrinology, 154, De Groot, L., Abalovich, M., Alexander, E.K., et al. (2012) Management of Thyroid Dysfunction during Pregnancy and Postpartum: an Endocrine Society Clinical Practice Guideline. The Journal of Clinical Endocrinology and Metabolism, 97, Teng, W., Shan, Z., Patil-Sisodia, K., et al. (2013) Hypothyroidism in pregnancy. The Lancet. Diabetes & Endocrinology, 1, Lazarus, J., Brown, R.S., Daumerie, C., et al. (2014) 2014 European Thyroid Association guidelines for the management of subclinical hypothyroidism in pregnancy and in children. European Thyroid Journal, 3, Vissenberg, R., Manders, V.D., Mastenbroek, S., et al. (2015) Pathophysiological aspects of thyroid hormone disorders/thyroid peroxidase autoantibodies and reproduction. Human Reproduction Update, 21, Prummel, M.F. & Wiersinga, W.M. (2004) Thyroid autoimmunity and miscarriage. European Journal of Endocrinology, 150, Carle, A., Pedersen, I.B., Knudsen, N., et al. (2011) Epidemiology of subtypes of hyperthyroidism in Denmark: a population-based study. European Journal of Endocrinology, 164, Cooper, D.S. & Laurberg, P. (2013) Hyperthyroidism in pregnancy. The Lancet Diabetes and Endocrinology, 1, Andersen, S.L., Olsen, J., Wu, C.S., et al. (2014) Spontaneous abortion, stillbirth and hyperthyroidism: a Danish populationbased study. European Thyroid Journal, 3, Laurberg, P., Andersen, S.L., Pedersen, I.B., et al. (2013) Screening for overt thyroid disease in early pregnancy may be preferable to searching for small aberrations in thyroid function tests. Clinical Endocrinology, 79, Garner, P. (1995) Type I diabetes mellitus and pregnancy. Lancet, 346, Feig, D.S. & Palda, V.A. (2002) Type 2 diabetes in pregnancy: a growing concern. Lancet, 359, Damm, P., Houshmand-Oeregaard, A., Kelstrup, L., et al. (2016) Gestational diabetes mellitus and long-term consequences for mother and offspring: a view from Denmark. Diabetologia, 53, Pedersen, C.B., Gotzsche, H., Moller, J.O., et al. (2006) The Danish Civil Registration System. A cohort of eight million persons. Danish Medical Bulletin, 53, Andersen, T.F., Madsen, M., Jorgensen, J., et al. (1999) The Danish National Hospital Register. A valuable source of data for modern health sciences. Danish Medical Bulletin, 46, Sundhedsstyrelsen (2013) Anbefalinger for Svangreomsorgen. Sundhedsstyrelsen, 2, Kildemoes, H.W., Sorensen, H.T. & Hallas, J. (2011) The Danish National Prescription Registry. Scandinavian Journal of Public Health, 39,

9 970 S. L. Andersen et al. 21 Colicchia, M., Campagnolo, L., Baldini, E., et al. (2014) Molecular basis of thyrotropin and thyroid hormone action during implantation and early development. Human Reproduction Update, 20, Pedersen, I.B., Knudsen, N., Jorgensen, T., et al. (2003) Thyroid peroxidase and thyroglobulin autoantibodies in a large survey of populations with mild and moderate iodine deficiency. Clinical Endocrinology, 58, Hallengren, B., Lantz, M., Andreasson, B., et al. (2009) Pregnant women on thyroxine substitution are often dysregulated in early pregnancy. Thyroid, 19, Granfors, M., Akerud, H., Berglund, A., et al. (2013) Thyroid Testing and Management of Hypothyroidism During Pregnancy: a Population-based Study. The Journal of Clinical Endocrinology and Metabolism, 98, Taylor, P.N., Minassian, C., Rehman, A., et al. (2014) TSH levels and risk of miscarriage in women on long-term levothyroxine: a community-based study. The Journal of Clinical Endocrinology and Metabolism, 99, Hubaveshka, J., Michaelsson, L.F. & Nygaard, B. (2014) The dose of levothyroxine in pregnant women with hypothyroidism should be increased by 20-30% in the first trimester. Danish Medical Journal, 61, A Allan, W.C., Haddow, J.E., Palomaki, G.E., et al. (2000) Maternal thyroid deficiency and pregnancy complications: implications for population screening. Journal of Medical Screening, 7, Nijkamp, J.W., Korteweg, F.J., Groen, H., et al. (2015) Thyroid function testing in women who had a stillbirth. Clinical Endocrinology (Oxford), doi: /cen [Epub ahead of print]. 29 McCance, D.R. (2015) Diabetes in pregnancy. Best Practice & Research. Clinical Obstetrics & Gynaecology, 29, Miailhe, G., Kayem, G., Girard, G., et al. (2015) Selective rather than universal screening for gestational diabetes mellitus? European Journal of Obstetrics, Gynecology, and Reproductive Biology, 191,