Hypothyroidism in pregnancy: pre-pregnancy thyroid status influences gestational thyroxine requirements


Dr A Kothari, 23 Barmouth Avenue, Perivale, Greenford, Middlesex UB6 8JS, UK. Email akothari@doctors.org.uk


There is considerable uncertainty about the management of hypothyroidism in pregnancy. Our aim was to establish the pattern of thyroxine dose adjustment needed and to determine the clinical reasons for these changes and the contributory factors. Of 89 pregnancies, thyroxine dose was unchanged in 50, increased (by a mean of 38 micrograms) in 34, and decreased in 5. Twenty-three percent of women who were tested in the first trimester needed an immediate increase in thyroxine. One-quarter (26%) of the women who needed a gestational increase in thyroxine dose had had a recent pre-pregnancy increase in thyroxine requirement (compared with 0% in women on static dose in pregnancy, P < 0.001). Furthermore, they did not require a decrease in thyroxine dose postpartum, suggesting a long-term need for more thyroxine rather than a transient gestational effect. None of the women who had stable doses of thyroxine during pregnancy had required recent pre-pregnancy changes in dose or needed postnatal changes. Inadequate pre-pregnancy control of thyroid function is associated with a need to increase thyroxine dosage during pregnancy.


Hypothyroidism is one of the most common medical disorders in pregnancy. However, there is a lack of agreement regarding the correct reference ranges to use, whether thyroid-stimulating hormone (TSH) or free thyroxine (fT4) levels are the most important determinants of normality, what the impact of inadequate1 or excess2 treatment might be (on the mother or the baby), and whether correct treatment can improve neonatal outcome. These difficulties are often further compounded by practical issues such as suboptimal pre-pregnancy control, late gestation at booking, poor compliance, and malabsorption related to pregnancy-induced vomiting or to the use of iron or calcium supplements. Recommendations for management are generally based on expert opinion, and even protocols from leading groups are not based on strong evidence and do not concur.

It is generally accepted that a proportion of women with hypothyroidism need to increase their dose of thyroxine during pregnancy, but it is unclear how these dose changes should be decided. While some studies suggest that decisions should be based on thyroid function tests (TFTs) at the booking visit and during pregnancy3 and have reported that many women would not need an increment,3 others have proposed a global increase in the thyroxine dose as soon as pregnancy is confirmed.4 There are concerns about this as not only may it be unnecessary for many women but also fetal exposure to excess T4 and T3 may be associated with miscarriage and low birthweight.2

We sought to establish the pattern of thyroxine dose adjustment in our population and to establish the clinical reasons for the changes and the contributory factors.


From January 1999 to December 2003, 167 pregnancies in women taking replacement doses of thyroxine for treatment of primary hypothyroidism were identified from the prospectively completed database of the antenatal endocrine clinic at the West Middlesex University Hospital. Sixty-seven were excluded from the study (clinical records unavailable for 47 pregnancies, 3 had had thyroid cancer and were on suppressive doses of thyroxine, 3 had early pregnancy losses, and on review, 14 were actually euthyroid or had incomplete data). This left 100 pregnancies in 91 women (9 women had 2 pregnancies) requiring treatment for hypothyroidism to be included in the study. Hundred sets of clinical and computer records and 335 TFTs were reviewed. Data were collected on demographic characteristics, pre-pregnancy management of hypothyroidism, gestation at booking, timing of TFTs, the timing and basis of alteration of the thyroxine doses, compliance with treatment, and perinatal outcomes. If more than one set of tests was performed in any one trimester, the highest TSH value together with the corresponding fT4 values were used to minimise any bias towards avoidance of dose change. The pre-pregnancy and first set of postpartum TFTs were also recorded when available. Hormone measurements were made in the pathology laboratory at West Middlesex University Hospital. fT4 was tested with an Advia Centaur competitive immunoassay and TSH with an Advia Centaur two-site sandwich immunoassay, both of which use direct chemiluminometric technology. Nonpregnant TSH and fT4 reference ranges were used to interpret the TFTs (Table 1), with an allowance made for third-trimester changes.5

Table 1.  Mean TSH (mIU/l), mean fT4 (pmol/l) and median thyroxine dose (T4, micrograms) pre-pregnancy, in each trimester, and postpartum in pregnancies of women with hypothyroidism according to whether dose alteration was required or not
Pregnancy groupPre-pregnancyFirst trimesterSecond trimesterThird trimesterPostpartum
  1. The reference range for TSH was 0.4–5.5 mIU/l between beginning 1999 and mid 2002 and 0.3–4.2 mIU/l between mid 2002 and end of 2003, and for fT4 was 10.3–23.2 and 9.0–26.0 pmol/l for the same time periods.

Dose unchanged (n = 50)1.36 (10)14.3 (8)100 (50)1.31 (19)15.3 (16)100 (50)2.26 (47)12.45 (46)100 (50)1.9 (38)11.85 (36)100 (50)3.16 (10)12.62 (10)125 (5)
Dose increased (n = 34)4.63 (9)14.09 (6)62.5 (10)7.77 (19)10.7 (20)100 (34)5.77 (34)10.76 (34)105 (34)2.82 (30)10.94 (30)125 (31)1.09 (7)15.24 (10)112.5 (10)


The mean age of the subjects (including all the 100 pregnancies) at delivery was 31 years (range 19–43 years), and the median duration of hypothyroidism before conception was 3 years (range from just pre-pregnancy to 25 years). In 11 pregnancies, treatment was only in pregnancy, and these were therefore ineligible for the analysis of treatment changes.

Of the 89 pregnancies included in the analysis, in 43 pregnancies, TFTs were performed in the first trimester, in 86 pregnancies in the second trimester, and in 71 pregnancies in the third trimester, including two women who booked at this time. The number of TFTs performed on each woman during pregnancy ranged from 1 to 7 (median 3). All changes in dose were made when TFTs fell outside the reference ranges, except for one woman with borderline elevation of TSH who was also clinically hypothyroid.

There were no significant differences between the pregnancies where the doses had to be changed (n = 39) and those where it did not (n = 50) with regard to age (mean 31.06 versus 31.29 years, respectively; P = 0.816), parity (41% nulliparous versus 32%, respectively; P = 0.537), body mass index (median 24 [interquartile range 23–29] versus 27 [24–31], respectively; P = 0.08), duration of hypothyroidism (median 3.0 years [1.5–5] versus 3.0 years [1.5–5], respectively; P = 0.806), or causes of hypothyroidism (the majority being due to autoimmune hypothyroidism and the rest to treated Grave’s disease). Not surprisingly, pregnancies where the dose had to be altered had more TFT performed than those who did not (median 3 tests [3–5] versus 3 tests [2–4], respectively; P = 0.003).

Pregnancies not requiring thyroxine dose change

In 50 pregnancies, dose changes were not made. In 47 of these, the women remained clinically and biochemically euthyroid. The mean TSH and fT4 values and median dose of thyroxine preconception, in each trimester, and postnatally are shown in Table 1. In 28 (56%) of these pregnancies, the women reported monitoring of TFTs approaching pregnancy, and none of them needed a dose alteration during that time (Table 2). In 19 pregnancies, the women (38%) were first tested before 13 weeks of gestation. In 46 pregnancies, the women claimed that they were fully compliant. Only four women (8%) were considered to have suboptimal compliance (one poor and three moderate), including one who had first-trimester vomiting that temporarily reduced compliance and one who admitted omitting one dose per week. Compliance was better in this group of pregnancies than in those in which a dose increment was needed (Table 2).

Table 2.  Effect of compliance and pre-pregnancy dose adjustment on thyroxine requirements in pregnancy
 Dose increase (n = 34) (%)No change in dose (n = 50) (%)P (chi square)
TFT pre-pregnancy4756NS
Dose change pre-pregnancy260<0.001
Suboptimal compliance248<0.05

In retrospect, women in 3 of the 50 pregnancies might have benefited from dose changes. All had normal fT4 measurements on each occasion, but one in whom there were concerns about compliance had raised TSH in all three trimesters, one had raised TSH in first and third trimesters and postnatally but not in the second or prior to conception, and one had raised TSH in the second but not in the third trimester.

Pregnancies requiring increase in thyroxine dose

In the other 50 pregnancies, thyroxine dose changes were made, including the 11 where the women were diagnosed with hypothyroidism and started on treatment just before or during pregnancy; these 11 were excluded from further analysis (see below). In 34 pregnancies, thyroxine dose was increased. Changes in TSH and fT4 values (Table 1) suggest improved control in the third trimester and postpartum. In 16 pregnancies (47%), the women recalled TFTs approaching pregnancy; 5 of the 9 available results indicated an elevated TSH. In at least 9 of these 34 pregnancies (26%), the women were having dose adjustments prior to (as well as during) pregnancy compared with none in the group not needing alteration in pregnancy (P < 0.001) (Table 2). Postpartum data were available for ten pregnancies; no dose reductions were needed (mean 7 weeks postpartum), strongly suggesting that at least some of the dose alterations during pregnancy were not specifically due to the pregnancy. In 20 (59%) of these 34 pregnancies, there were first-trimester measurements, 10 of which indicated immediate dose increases (and 6 of which had further increases in the second [n = 5] or third [n = 1] trimesters); the rest were initially normal and only required dose changes later. In 19 pregnancies, increases were needed in the second trimester alone, in 3 in the third trimester alone and in 2 in both second and third trimesters.

Thus, of these 42 dose increases, only 6 were in the third trimester. Twenty-two adjustments were in response to elevated TSH (including two pregnancies where the women were symptomatic), 11 to both elevated TSH and low fT4 (including one symptomatic woman), 5 to hypothyroxinaemia (one woman had symptoms), and 1 to symptomatic hypothyroidism (slow ankle jerks, with borderline elevation of TSH); three increases were inappropriate.

Compliance was described as suboptimal in eight pregnancies (poor in six and moderate in two), either because the women forgot the medication on multiple occasions, took a lower dose than advised, or had persistent vomiting (Table 2).

Pregnancies requiring a decrease in thyroxine dose

In five pregnancies, the women required a decrease in thyroxine dose, by 25 micrograms in three cases, 50 micrograms in one, and by 11 micrograms in one (150 micrograms decreased to 150 micrograms on 4 days alternating with 125 micrograms on 3 days).

Women started on thyroxine in pregnancy

In 11 pregnancies, the women were started on thyroxine treatment on the basis of hypothyroxinaemia (mean fT4 8.5 pmol/l) and raised TSH (mean TSH 11.6 mIU/l) in 9, hypothyroxinaemia in 1, and raised TSH in 1. In seven pregnancies, a history of thyroid disease or thyroid surgery provoked referral to the antenatal endocrinology clinic by the booking midwife, and in four, the diagnosis was made as the pregnancy was confirmed.

Pregnancy outcome

Among the 100 pregnancies in women with hypothyroidism, the mean birthweight was 3315 g (SD 577 g) and mean gestational age at delivery was 39.2 weeks (SD 2 weeks); one baby (whose mother had been euthyroid throughout pregnancy) who was born at term had a birthweight below the 5th centile. Ninety-five percent babies delivered at term; all the women who delivered preterm were on the same thyroxine dose throughout pregnancy.


The aim of thyroxine replacement treatment in early pregnancy is to provide a euthyroid environment for the developing fetal brain. It is logical but not evidence based that ensuring maternal biochemical euthyroidism in the first trimester when the fetus is dependent on maternal thyroxine might optimise fetal outcome. Such dependence is unlikely thereafter as the placenta does not allow significant thyroxine transfer beyond this stage of pregnancy.5 Adjustments in thyroxine dose in pregnancy do not usually influence maternal wellbeing.

In our study, in 56% (50/89) of the pregnancies, no change in thyroxine dose was required; if the nine pregnancies where women who were altering doses at conception are excluded, the women in 63% (50/80) of the pregnancies remained on a stable dose. Thus, in 34 of 89 (38%) pregnancies, the women increased their dose during pregnancy, and in 5 (6%), reduced their dose. In pregnancies where there were both first- and second-trimester tests, 23% of women (10/43) needed to increase their dose immediately (three of whom were having pre-pregnancy dose adjustments) and 23% later in the pregnancy, 9% had a dose reduction, and 44% remained on a static dose. Our findings are in stark contrast to Alexander et al.4 who concluded that as 9 of their 12 pregnant women on replacement doses of thyroxine needed to increase thyroxine in the first trimester, all women should increase by 30% as soon as possible after conception. Others have made similar recommendations, with 20–75% women increasing their dose in pregnancy, although the gestational ages at which testing or dose adjustment are made is not clearly specified.5,6

Our data suggest strongly that women whose dose is being adjusted prior to conception or who are poorly compliant are more likely to need dose increases. This has not been reported previously, and possibly, together with other issues not related to the pregnancy may contribute to the apparently wide range of thyroid dose changes needed after conception. This finding will allow improved identification of women most likely to need dose changes, so that they can be targeted for closer surveillance. Since the first trimester is the only time when maternal thyroxine might directly influence fetal development, all women who take thyroxine should be tested as soon as possible in pregnancy, but especially those who have poor control pre-pregnancy. Women who are euthyroid in the first trimester, on a stable dose of thyroxine and fully compliant with their regimen are less likely to need changes later in pregnancy, and as a biochemical picture of hypothyroidism is very uncommon in the third trimester (at a time when thyroxine changes would not influence fetal outcome), they do not need further testing. We acknowledge that 47 sets of notes were unavailable, which could have affected the results.

The next challenge will be to encourage the large number of women on thyroxine who did not have their thyroid function tested in the approach to pregnancy to do so in the future as this is the only way that maximal fetal outcome might be achieved, and it will allow a rational approach to their care during pregnancy. It is also important to test thyroid function in those women with a history of thyroid dysfunction or surgery who are not on treatment.


Achieving euthyroidism preconception and in the first trimester should be the goals of management in hypothyroid pregnancy as this may be important in optimising fetal neurodevelopment. Almost one-quarter of women tested in the first trimester need a dose increment. As women with pre-pregnancy dose adjustment and poor compliance are more likely to need an increase at this time, targeting them might improve care. As very few dose changes are needed in the third trimester, this is a time when hypothyroid pregnancy can be demedicalised. In our series, the majority of women did not need a dose increment and a less number needed a reduction; therefore, a global increase in thyroxine in early pregnancy is not appropriate.

Conflict of interest

No conflict of interest.

Contribution to authorship

A.K. participated in the collation, analysis, computerisation of the data, and preparation of the manuscript and has seen and approved the final version. A.K. is the corresponding author and has full access to all the data in the study and had final responsibility for the decision to submit for publication.

J.G. prospectively maintained the clinical database, conceived the original idea for the paper, and participated in analysis of the data and preparation of the manuscript; she has seen and approved the final version.

Details of ethics approval and funding

None required.


We would like to express our sincerest gratitude to Ms Michelle Wu for her help with data analysis.