Decreased Birth Weight, Length, and Head Circumference in Children Born by Women Years After Treatment for Hyperthyroidism

Transcription

1 ORIGINAL Endocrine ARTICLE Research Decreased Birth Weight, Length, and Head Circumference in Children Born by Women Years After Treatment for Hyperthyroidism Hans Ohrling, Ove Törring, Li Yin, Anastasia N. Iliadou, Ove Tullgren, Mirna Abraham-Nordling, Göran Wallin, Per Hall, and Stefan Lönn Division of Endocrinology (H.O., O.Tö.), Department of Internal Medicine, Södersjukhuset, 11883, Stockholm, Sweden; Institution of Clinical Research and Education (O.Tö.), Department of Medical Epidemiology and Biostatistics (L.Y., A.N.I., P.H., S.L.), and Institution of Molecular Medicine and Surgery (M.A.-N., G.W.), Karolinska Institutet, 17177, Stockholm, Sweden; Department of Oncology (O.Tu.), Karolinska University Hospital, Solna, Sweden; and Department of Research and Development (S.L.), Region Halland, Halmstad, Sweden Context: Whether hyperthyroidism influences the birth characteristics of children born several years after treatment is unknown. Objective: The objective of the study was to compare birth characteristics in singleton newborns delivered by women previously treated for Graves disease (GD), toxic nodular goiter (TNG), or nontoxic goiter (NTG). Design: This was a nested case-control design within a national cohort registry study from 1950 through Setting: The study was conducted at a university and a hospital center in collaboration. Patients: The birth characteristics of newborns (n 3421) delivered in a cohort of women treated for GD or toxic nodular goiter by radioiodine or surgery (exposed group) at least 1 year prior to pregnancy were compared with newborns (n 2914) of mothers, previously operated for NTG (unexposed group). Main Outcome: The primary outcome was birth weight, length, and head circumference. The secondary outcome was malformations, gestational age, and type of hyperthyroidism. Results: The birth weight of exposed children was g (mean SD) compared with the unexposed, g(P.001). The cumulative odds ratio (OR) for lower birth weight was 1.29 [95% confidence interval (CI) ]. The average birth length for the exposed children was cm compared with the unexposed of 50.4 cm 2.6 cm (P.01) [cumulative OR 1.25 (95% CI )]. The head circumference was cm among exposed and cm, respectively (P.001), with an OR of 1.24 (95% CI ). No differences in birth characteristics were observed between children born after maternal GD or toxic nodular goiter. Conclusions: Previous GD or TNG may influence the birth characteristics several years after radioiodine or surgical treatment. (J Clin Endocrinol Metab 99: , 2014) ISSN Print X ISSN Online Printed in U.S.A. Copyright 2014 by the Endocrine Society Received January 17, Accepted May 20, First Published Online May 30, 2014 Graves disease (GD), or toxic multi- or uninodular goiter (TNG) are common endocrine disorders with an annual incidence of between 27.6 and 100 per women in Sweden, the United Kingdom, Scandinavia, Japan, and the United States (1 3). Hyperthyroidism occurs in 0.1% 0.2% of pregnancies of which GD accounts for 90% 95% (4). Thyrotoxicosis during pregnancy or when the maternal hyperthyroidism is poorly controlled (4) is Abbreviations: CI, confidence interval; GD, Graves disease; NTG, nontoxic goiter; OR, odds ratio; TNG, toxic nodular goiter; TRab, thyroid-stimulating antibody. doi: /jc J Clin Endocrinol Metab, September 2014, 99(9): jcem.endojournals.org 3217

2 3218 Ohrling et al Hyperthyroidism and Birth Characteristics J Clin Endocrinol Metab, September 2014, 99(9): associated with an increased risk of intrauterine growth restriction and low birth weight, prematurity, miscarriages, malformations, and maternal and other fetal complications (5 7). Earlier studies have mainly focused on the fetal and maternal complications when the woman is hyperthyroid during pregnancy. Whether previous hyperthyroidism may affect the outcome of pregnancy later on in the woman s life, however, is unknown. Yet it is a frequently asked question by fertile women with GD or TNG. TNG is treated with radioiodine or surgery, and GD, in addition, also has the option of antithyroid drugs all of which sooner or later render the patient euthyroid with or without the need of thyroxine substitution. Assumingly, previous treatment for GD or TNG therefore should not affect the fetal growth and birth characteristic of a pregnancy later in life a priori. To explore this hypothesis, we have studied the birth characteristics in children born in a large cohort of women who had previously been treated for GD or TNG by radioiodine or surgery (8). When compared with children born in a cohort of women who have been operated for nontoxic goiter (NTG), we found a lower birth weight, shorter birth length, and average head circumference of children born by mothers on average 5 years after treatment of GD or TNG hyperthyroidism. Materials and Methods Women treated for hyperthyroidism, by surgery or radioiodine, between 1950 and 2006 in Sweden were identified from an established cohort comprising hyperthyroid patients treated with radioiodine therapy or surgery. The radioiodine cohort includes women treated with radioiodine between 1950 and 2000 and has been described in detail previously (9, 10). The patients were recruited from seven hospital radiotherapy and oncology departments in Sweden. Most of the patients were included after 1975 and consisted exclusively of patients from the Stockholm region. Women who had hyperthyroidism (defined as International Classification of Diseases, seventh revision, codes , , , and ; International Classification of Diseases, eighth revision, codes , , , and ; International Classification of Diseases, ninth revision, codes 242A, 242B, 242C, 242D, 242E, 242W, and 242X; and International Classification of Diseases, tenth revision, codes E05.0, E05.1, E05.2, E05.3, E05.4, E05.5, E05.8, and E05.9), which had been surgically treated (defined as procedure codes 0810, 0811, 0814, 0815, 0820, 0825, 0826, 0831, and 0849 for years and BAA 20, BAA 25, BAA 30, BAA 40, BAA 50, BAA 60, and BAA 99 for years 1997 to the present) between 1965 and 2006 were identified through the Swedish National Inpatient Registry. A reference group of women without a history of hyperthyroidism but who had been surgically treated (same procedure codes as described above) for NTG (nontoxic multinodular goiter and nontoxic goiter simplex) between 1965 and 2006 (International Classification of Diseases, seventh revision, codes , , , and ; International Classification of Diseases, eighth revision, codes , , , , , and ; International Classification of Diseases, ninth revision, and International Classification of Diseases, tenth revision, codes E04.0, E04.1, E04.2, E04.8, and E04.9) were also identified through the Swedish National Inpatient Registry. The combined radioiodine and surgically treated GD or TNG hyperthyroidism cohort (n ) as well as the NTG group (n ) has been described earlier (8). The Swedish National Medical Birth Registry includes information prospectively collected during pregnancy, delivery, and the neonatal period on virtually all births in Sweden since Data on sociodemographic characteristics, reproductive history, and complications during pregnancy, delivery, and the neonatal period are prospectively collected in this database beginning with the first antenatal visit. Each Swedish woman is assigned a unique personal identification number given to her at birth or at immigration, which enables a linkage between all available national registries for this individual. In our study, singleton children born to women with a history of thyroid disease were linked through the Swedish Medical Birth Register through this unique personal registration number. Only children of women who became pregnant after the treatment of hyperthyroidism by radioiodine or surgery is included in this study because no registry data are available on medically treated patients from the study period. Consequently, only children born between 1973 and 2006 are included in the study. To ensure that no woman was pregnant during the treatment, we included only children born at least 1 year after treatment (n 3421). Two thousand nine hundred fourteen children were identified in the unexposed group. Children born by women with a history of hyperthyroidism were defined as exposed and children born by women with a history of NTG were defined as unexposed. The recorded outcomes included birth weight, birth length, the duration of pregnancy, gestational age based on last menstrual period, head circumference, and malformations. Additional potential influent factors were collected from the Medical Birth Registry and included maternal smoking status (from first antenatal visit), child s gender, birth order, place of birth, and time since exposure (date of treatment). The recruitment of patients in this study followed the general principles of the Helsinki Declaration and Swedish regulations for handling of patient and register data. Statistical methods Birth weight and length, head circumference, and length of pregnancy were continuous variables and categorized prior to analysis. Birth weight was divided into the five groups: less than 2000, , (reference category), , and more than 4999 g. Birth length was divided into five groups: less than 44, 44 46, (reference category), 50 51, and greater than 51 cm. Gestational age was categorized into four groups: less than 36, 36 37, (reference category), and longer than 42 weeks. Head circumference was divided into five groups: less than 33, 33 34, 35 (reference category), 36 37, and greater than 37 cm. Malformations were defined as present or not present (reference category). We measured relative risk of the exposure on multicategorical outcome by the odds ratio (OR) of the nonreference category to the reference category. The multinomial logistic model was used

3 doi: /jc jcem.endojournals.org 3219 to estimate OR and 95% confidence intervals (CIs). In case of dichotomous outcome, the multinomial logistic model was reduced to the binomial logistic model. The analysis was adjusted for smoking status, gender, birth order, the age of the mother, time between treatment of hyperthyroidism and pregnancy, and type of treatment for hyperthyroidism. In addition, we measured the overall relative risk of the exposure on multicategorical outcome by the cumulative OR. The estimated cumulative OR was an average OR of categories of lower vs higher-ordered values. The cumulative logistic model was used to estimate the cumulative OR and 95% CI. A small number of subjects had missing values for the multicategorical outcomes, and the results were obtained after removing these subjects (see Table 2). However, inclusion of missing values as a separate category in the analysis led to essentially identical results. Results We identified 3421 singleton children born by women who at least 1 year prior to the pregnancy had been treated for hyperthyroidism by radioiodine or surgery. Of those children, 2833 had mothers diagnosed with GD, 433 with TNG, and 155 with mixed hyperthyroidism of both forms, ie, patients registered as GD and toxic multinodular goiter. We identified 2914 singleton children, defined as unexposed, born to women treated with surgery for NTG. Basic characteristics for the study population indicated that mothers with previous hyperthyroidism had a lower mean age at pregnancy ( y; mean SD) compared with women with a history of NTG ( y) (P.001) and higher frequency of smoking (739 of 3421 vs 461 of 2914 in the nonexposed group, respectively) (P.001). In our study population, exposed children were born in earlier calendar periods (average year 1987) compared with unexposed children (average year 1989) (P.001). Fewer children were born as the third child or more in the exposed group [7.04% (241 of 3421)] compared with the unexposed group [7.68% (224 of 2914)], but the difference was not significant (P.33). For the other included variables in our study, there were no statistically significant differences between the exposed and unexposed children (Table 1). The average birth weight for exposed children was g (mean SD), significantly lower compared with unexposed children ( g) (P.001). The adjusted OR was higher (OR 1.28; 95% CI ) in the weight class g and lower (OR 0.79; 95% CI ) for g (Table 2). The cumulative OR for birth weight was 1.29 (95% CI ). The average birth length for the exposed children was shorter ( cm) compared with unexposed children (50.4 cm 2.6 cm) (P.01), with a cumulative OR of 1.25 (95% CI ). The adjusted OR for birth length shifted toward lower values for children greater than 49 cm of birth length and was statistically significant greater than 51 cm [OR 0.75 (95% CI )] (Table 2). The average size of the head circumference was slightly smaller for the exposed ( cm) compared with the unexposed children ( cm) (P.001), with a cumulative OR for a smaller head circumference of 1.24 (95% CI ). The length of pregnancy was overall similar between the groups, and the cumulative OR was 1.00 (95% CI ). No significant increased risk of overall malformations or influence on gestational age among children born by women treated for hyperthyroidism was observed (Table 2). Neither was any indication of increased risk of specific malformations detected, but numbers were too small for a meaningful analysis. The mean time that had elapsed between the treatment of hyperthyroidism (surgery or radiation) and pregnancy was 5.26 years, and the mean time between surgery for NTG and pregnancy was comparable (5.49 y). The time from treatment of hyperthyroidism to delivery of the child was analyzed for all outcomes, but no variation of the ORs by time period since treatment was done was observed (results not shown). When all analyses were stratified by the type of hyperthyroidism (GD and TNG), the results were similar and not statistically different between the two treatment modes (results not shown). The average birth weight for GD, TNG, and NTG for each 5-year interval during the study period is shown in Figure 1. The impact of radioiodine treatment was also evaluated, but no difference compared with surgical treatment could be detected, but the numbers of observations were small in many categories (results not shown). In addition, stratifying the analysis by calendar year (as groups from Table 1), gestational age, and the mothers age at delivery did not have any essential impact on the results of weight, length, and head circumference of the children. Excluding the last calendar period (born later then 1994) did not have any essential impact on the results. The results overall were similar between the crude statistical model and the adjusted models (Table 2). The results were also adjusted for the age of the mother and the child s birth order, but no differences in the results were found. Most women in both groups gave birth during the period The mean birth body weight of children born by women years of age in the Swedish background population, which among available data from the public birth registry best represents our study group, increased by g during this period, g between and , and for the

4 3220 Ohrling et al Hyperthyroidism and Birth Characteristics J Clin Endocrinol Metab, September 2014, 99(9): Table 1. Basic characteristics of children born to mothers previously treated for hyperthyroidism due to GD or TNG (exposed) compared with children born to mothers operated for NTG (nodular or simplex) and without a history of hyperthyroidism (unexposed) Exposed Unexposed n 3421 % n 2914 % Age of the mother, y Younger than Older than Type of hyperthyroidism Graves hyperthyroidism N/A N/A TNG N/A N/A Other types a N/A N/A Treatment Surgery Radiation N/A N/A Time since treatment, y Smoking during pregnancy Yes No Unknown Country of birth for the mother Sweden Other countries Gender of child Male Female Birth order of child First Second Third or higher Year of birth of the child Before or after a All other types of hyperthyroidism. period before 1980 compared with that after 1999, with a relative increase after 1994 (Figure 1). To evaluate the possible impact of categorization of birth weight, the categories were redefined using a 500-g interval, and a trend of a lower birth weight among children born by mothers with a previous history of hyperthyroidism was observed (Figure 2). All ORs for the weight categories greater than 3500 g were below 1.0, and ORs for categories less than 3000 g were greater than 1.0 except the lowest category (Figure 2). Consequently, the results were not depending on how the different categories of birth weight were defined. Discussion A restriction of growth has previously been seen in children born to mothers who were hyperthyroid during their pregnancy. Our results clearly indicate that mothers treated for hyperthyroidism more than 1 year, on average 5 years prior to delivery, are more likely to give birth to smaller children as reflected by lower birth weight, shorter birth length, and smaller head circumference. The conclusion is supported by results from a multinomial logistic model and a cumulative OR model. Our observations, however, should be interpreted with caution, acknowledging the relative small detected differences we found and the possibility of unknown bias. One cause for lower birth weight would have been intrauterine hyperthyroidism as illustrated by the observations of Vaidya et al (11). A fetus that has had nonautoimmune intrauterine hyperthyroidism due to an autosomal dominantly inherited activating TSH receptor mutation has a lower birth weight and an increased risk of premature delivery (11). This is a very rare cause, but in-

5 doi: /jc jcem.endojournals.org 3221 Table 2. OR for birth characteristics in children born to mothers previously treated for hyperthyroidism due to GD or TNG (exposed) compared with children born to mothers operated for NTG and without a history of hyperthyroidism (unexposed) a Outcome Exposed, n Unexposed, OR, Adjusted n OR, Crude OR, Adjusted b Except Smoking Malformations Yes ( ) 1.15 ( ) 1.13 ( ) No (reference) Length of pregnancy, wk ( ) 0.90 ( ) 0.92 ( ) ( ) 1.06 ( ) 1.07 ( ) (reference) ( ) 0.86 ( ) 0.85 ( ) Weight at birth, g ( ) 0.87 ( ) 0.90 ( ) ( ) 1.28 ( ) 1.32 ( ) (reference) ( ) 0.79 ( ) 0.77 ( ) ( ) 0.59 ( ) 0.57 ( ) Length at birth, cm ( ) 1.07 ( ) 1.07 ( ) ( ) 1.20 ( ) 1.20 ( ) (reference) ( ) 0.87 ( ) 0.84 ( ) ( ) 0.75 ( ) 0.72 ( ) Head circumference at birth, cm ( ) 1.26 ( ) 1.32 ( ) ( ) 1.12 ( ) 1.13 ( ) 35 (reference) ( ) 0.91 ( ) 0.90 ( ) ( ) 0.80 ( ) 0.77 ( ) The 95% CIs are shown in parentheses. a OR for exposed to unexposed children. b Adjusted for gender of child, age of mother, time elapsed since treatment of hyperthyroidism, birth order of child, type of treatment for hyperthyroidism, and smoking. trauterine hyperthyroidism due to transplacental passage of thyroid-stimulating antibodies (TRabs) is a well-known although unusual clinical problem, with an increased risk if the levels of maternal TRabs are elevated (12, 13). Although TRabs usually normalize within 1.5 years after subtotal thyroidectomy, they rise and may remain elevated for an extended period, sometimes more than 5 Figure 1. The average birth weight of children born to mothers who had been treated for hyperthyroidism due to GD (n 2833), TNG (n 433), or NTG (n 2914) at least 1 year before delivery. Data are shown for each 5-year interval during the study period. The average birth weight of children born by women years of age of the whole Swedish population (POP) is also shown. years after radioiodine treatment (14, 15). Restriction in fetal growth was in our study was also noted for women with toxic nodular goiter, a condition in which the TRab is absent, however. We did not have any information of maternal TRab levels, but taken together, it seems unlikely that TRab-mediated intrauterine hyperthyroidism could explain the growth restriction we have observed. Neither do the type of maternal hyperthyroidism because there was no difference in birth outcome between the two diagnoses nor were the age of the mother at pregnancy, smoking, birth order of the child, or any of the other tested relations. A potential explanation to the growth restriction, however, could be the use of maternal L-thyroxine replacement therapy after the radioiodine or surgical treatment for hyperthyroidism because it is lifelong compared with the unexposed group in which the women most likely remain euthyroid because the resection of the thyroid is partial. We know from clinical experience that the dose of L-thyroxine is usually also increased during pregnancy to maintain an adequate metabolism. We speculate that the increased dose, which for the mother is acceptable, may

6 3222 Ohrling et al Hyperthyroidism and Birth Characteristics J Clin Endocrinol Metab, September 2014, 99(9): Figure 2. ORs and 95% CIs (error bars) for birth weight in children born to mothers treated for hyperthyroidism (GD or TNG) compared with children born to mothers operated for NTG. *, P.02; **, P.01; ***, P.001. Odds ratio was determined for exposed to unexposed children, adjusted for gender of child, age of the mother, time elapsed since treatment of hyperthyroidism, birth order of the child, type of treatment for hyperthyroidism, and smoking have caused a slight increase in the metabolism of the fetus, which could have caused a slight restriction in fetal growth. Blazer et al (16) in 2003 studied the neonatal outcome in hypothyroid women with hormone replacement during pregnancy. The women on L-thyroxine thyroid hormones generally gave birth to children with lower birth weight and a smaller head circumference compared with euthyroid pregnant woman. However, because the exact extension of surgery, for instance, the degree of partial thyroidectomy for NTG as well as the thyroid function before and after surgery is unknown, oversubstitution with L-thyroxine as an explanation remains a theory. With respect to the smaller head circumference and body length, the differences of less than 0.5 cm are probably close to the limit of accuracy of anthropometric measurements taken at birth in a clinical setting. However, random errors of measurement are unlikely to fall in the same direction. We did not find that previous hyperthyroidism significantly increased the risk of overall or specific malformations or influenced the gestational length. Neither did we find that previous radioiodine treatment implied a different outcome than surgical treatment. With respect to the possible influence of time trends of the population of birth body weight in Sweden, it is difficult retrospectively to find an exact matched and representative group of women from the background population. Women years of age are most similar to our study group and showed a slight increase of birth body weight of g during the period in which most of the women in our study gave birth ( ) and g for the period before 1980 to after Because there were only 2 years of difference in mean birth year (1987 vs 1989) between our two study groups, we find the change in birth body weight in the background population unlikely to explain our results. Taken together, we have observed a lower birth weight, smaller length, and slightly smaller average head circumference in women previously treated for hyperthyroidism at least 1 year before. At present there is no clear explanation or indications of the cause(s). Our results are corroborated by a recent Danish National Hospital Register study in which a 50-g lower birth weight was observed in a subcohort consisting of 4381women who had had a diagnosis of hyperthyroidism before giving birth (17). The lower average birth weight of the exposed children may at first seem small and negligible, but to put our results in a wider perspective, we have compared these with the well-known influence of maternal tobacco exposure during pregnancy on birth characteristics (18). Although maternal smoking may cause an estimated decrement of g lower birth weight compared with children born by nonsmokers, our results are comparable with the figure for environmental or passive tobacco exposure, ie, g (18). In contrast to the effect of tobacco exposure on the other hand, we did not find that hyperthyroidism increased the risk of preterm delivery. The strength of our study is that it is based on data collected from well-established national registers such as the Swedish Inpatient Registry and the Swedish Medical Birth Registry (19). The study was population based using a linkage between national registers, minimizing bias related to self-reported events, missing data from the follow-up period, and selection procedures. The group of children born to mothers who have undergone thyroid surgery for NTG between 1965 and 2006 (unexposed children) were selected as a relevant comparative group because the mother also had goiter, were operated as inpatients, and had been exposed for anesthesia and other procedures in relation to hospital care. A limitation of our study is the use of data collected from historical registers because available variables are limited. For instance, neither information with respect to use of antithyroid drugs (ATD) or types of ATD nor of maternal body mass index or weight, maternal alcohol intake, socioeconomic status, diabetes, hypertension, and drugs were available in the register or can be extracted. Information on the use of L-thyroxine, which may have affected the outcome for the child, is also lacking as well as data on the severity by laboratory values and length of hyperthyroidism, thyroid-associated orbitopathy, etc. In addition, we lack data on patients who were medically treated only for their hyperthyroidism because outpatients

7 doi: /jc jcem.endojournals.org 3223 are not registered in the national in-patient registry up until In conclusion, we have observed an increased risk for a mother who on average 5 years but at least 1 year previously has been treated for GD or toxic nodular goiter to give birth to a slightly lighter and shorter child. These observations needs further studies for verification and clarification of possible underlying mechanism(s). Acknowledgments Address all correspondence and requests for reprints to: Ove Törring, MD, PhD, Associate Professor, Institution for Clinical Science and Education, Karolinska Institutet, Södersjukhuset, Stockholm, Sweden. ove.torring@ki.se. Disclosure Summary: The authors have nothing to disclose References 1. Vanderpump MP, Tunbridge WM, French JM, et al. The incidence of thyroid disorders in the community: a twenty-year follow-up of the Whickham Survey. Clin Endocrinol (Oxf). 1995;43: Lazarus JH. Hyperthyroidism. Lancet. 1997;349: Abraham-Nordling M, Bystrom K, Torring O, et al. Incidence of hyperthyroidism in Sweden. Eur J Endocrinol. 2011;165: Davis LE, Lucas MJ, Hankins GD, Roark ML, Cunningham FG. Thyrotoxicosis complicating pregnancy. Am J Obstet Gynecol. 1989;160: Millar LK, Wing DA, Leung AS, Koonings PP, Montoro MN, Mestman JH. Low birth weight and preeclampsia in pregnancies complicated by hyperthyroidism. Obstet Gynecol. 1994;84: Phoojaroenchanachai M, Sriussadaporn S, Peerapatdit T, et al. Effect of maternal hyperthyroidism during late pregnancy on the risk of neonatal low birth weight. Clin Endocrinol (Oxf). 2001;54: Bahn RS, Burch HB, Cooper DS, et al. Hyperthyroidism and other causes of thyrotoxicosis: management guidelines of the American Thyroid Association and American Association of Clinical Endocrinologists. Endocr Pract. 2011;17: Abraham-Nordling M, Lonn S, Wallin G, et al. Hyperthyroidism and suicide: a retrospective cohort study in Sweden. Eur J Endocrinol. 2009;160: Hall P, Lundell G, Holm LE. Mortality in patients treated for hyperthyroidism with iodine-131. Acta Endocrinol (Copenh). 1993; 128: Holm LE, Hall P, Wiklund K, et al. Cancer risk after iodine-131 therapy for hyperthyroidism. J Natl Cancer Inst. 1991;83: Vaidya B, Campbell V, Tripp JH, Spyer G, Hattersley AT, Ellard S. Premature birth and low birth weight associated with nonautoimmune hyperthyroidism due to an activating thyrotropin receptor gene mutation. Clin Endocrinol (Oxf). 2004;60: Mitsuda N, Tamaki H, Amino N, Hosono T, Miyai K, Tanizawa O. Risk factors for developmental disorders in infants born to women with Graves disease. Obstet Gynecol. 1992;80: Peleg D, Cada S, Peleg A, Ben-Ami M. The relationship between maternal serum thyroid-stimulating immunoglobulin and fetal and neonatal thyrotoxicosis. Obstet Gynecol. 2002;99: Torring O, Tallstedt L, Wallin G, et al. Graves hyperthyroidism: treatment with antithyroid drugs, surgery, or radioiodine a prospective, randomized study. Thyroid Study Group. J Clin Endocrinol Metab. 1996;81: Laurberg P, Wallin G, Tallstedt L, Abraham-Nordling M, Lundell G, Torring O. TSH-receptor autoimmunity in Graves disease after therapy with anti-thyroid drugs, surgery, or radioiodine: a 5-year prospective randomized study. Eur J Endocrinol. 2008;158: Blazer S, Moreh-Waterman Y, Miller-Lotan R, Tamir A, Hochberg Z. Maternal hypothyroidism may affect fetal growth and neonatal thyroid function. Obstet Gynecol. 2003;102: Linding Andersen S, Olsen J, Wu CS, Laurberg P. Low birth weight in children born to mothers with hyperthyroidism and high birth weight in hypothyroidism, whereas preterm birth is common in both conditions: a Danish national hospital register study. Eur Thyroid J. 2013;2: Rogers JM. Tobacco and pregnancy. Reprod Toxicol. 2009;28: Cnattingius S, Ericson A, Gunnarskog J, Kallen B. A quality study of a medical birth registry. Scand J Soc Med. 1990;18: