Evaluate and manage thyroid and other endocrine disorders in pregnancy

In one line

Thyroid disease in pregnancy is won by anchoring every decision to trimester-specific TSH and physiology: drive overt hypothyroidism to a TSH below 2.5 mU/L from the moment a period is missed, treat hyperthyroidism with the least-teratogenic drug at the lowest effective dose while watching the fetus through the maternal TRAb, and resist treating biochemical-only subclinical disease for which the randomised evidence shows no neurodevelopmental benefit.

This chapter assumes the gland physiology, the booking-bloods groundwork and the first-line drug choices taught at Intermediate level — the hCG–TSH cross-talk, TBG-driven shifts in total hormone, the basic levothyroxine-vs-antithyroid-drug distinction. It spends its words on the consultant layer: telling the subtypes apart when the biochemistry is ambiguous, the exact thresholds and how the named treatment regimens differ, appraising the randomised trials that moved the guidelines, and the judgement calls in the refractory and atypical case.

Assessment

Thyroid biochemistry must be interpreted against the pregnancy-shifted reference range, not the non-pregnant one. hCG cross-stimulates the TSH receptor, so TSH falls (often below the non-pregnant lower limit) in the first trimester while free T4 rises; oestrogen raises thyroxine-binding globulin, inflating total T4/T3 and making free hormone and TSH the only reliable analytes. (That physiology is Intermediate groundwork; the harder question is what to do when the numbers are internally inconsistent.)

• Reference ranges. Use assay-/population-specific trimester ranges where available. When they are not — the SA public-sector reality — the ATA 2017 default is a TSH upper limit of 4.0 mU/L (the older 2.5/3.0 cut-offs over-diagnosed). A booking TSH should be read against this, not a lab printout flagged on adult norms.
• Hypothyroidism. Overt = raised TSH with low free T4 (or any TSH >10 mU/L). Subclinical = raised TSH (above the trimester limit, i.e. >4.0) with normal free T4. Isolated hypothyroxinaemia = normal TSH, low free T4. Check TPO antibodies: positivity raises miscarriage and progression risk and changes who you treat.
• Hyperthyroidism. Separate Graves disease (TRAb-positive, persists/worsens, may have orbitopathy) from gestational transient thyrotoxicosis (GTT) — an hCG-driven, self-limiting biochemical thyrotoxicosis of early pregnancy, classically with hyperemesis gravidarum. GTT is TRAb-negative, has no goitre or eye signs, and resolves by ~14–18 weeks; it needs no antithyroid drug.
• The single most important test in Graves disease is maternal TRAb, because IgG crosses the placenta and can cause fetal/neonatal thyrotoxicosis even in a euthyroid or post-ablation mother. Measure it in the first trimester and, if positive, again at 18–22 weeks.

Telling the subtypes apart — the consultant discriminators

The biochemistry only sorts patients if you read it against the mechanism generating it. Four discriminations carry most of the management consequences.

• GTT vs first-presentation Graves. Both give a suppressed TSH and a high free T4 in early pregnancy, and both can occur in a vomiting woman. The mechanism splits them: GTT is receptor cross-stimulation by hCG — a transient, dose-dependent agonism that tracks the hCG curve, peaks at 9–11 weeks and fades as hCG falls. Graves is autonomous TRAb-driven stimulation that does not care about gestation. The bedside discriminators that flow from this: GTT has no goitre, no orbitopathy, no pre-pregnancy thyroid history, a free T3 that is usually only mildly raised (hCG drives T4 more than T3), and it is TRAb-negative; the hyperthyroidism is proportionate to the vomiting and improves as the woman stops vomiting. A T3-predominant picture, a goitre with a bruit, eye signs, or a positive TRAb pulls you firmly toward Graves. The trap is starting carbimazole on a hyperemetic woman with a suppressed TSH — you are then treating a self-limiting hCG effect with a teratogen.
• Subclinical hypothyroidism vs assay artefact vs early overt disease. A "raised" TSH above 4.0 with a normal free T4 is only subclinical hypothyroidism if it is real and persistent. Reasons it may be neither: a non-pregnant reference range applied by the lab, an isolated transient rise, biotin interference (high-dose biotin supplements spuriously lower TSH and raise free T4/T3 on immunoassays — a recognised cause of a "thyrotoxic" panel in a clinically euthyroid woman), heterophile antibodies, or a free-T4 immunoassay that under-reads in the protein-shifted third trimester. The consultant move is to repeat with the gestation noted, ask about supplements, and — where the free T4 looks discordant late in pregnancy — recognise that total T4 (×1.5 the non-pregnant range) or the free T4 index can be more reliable than a third-trimester free-T4 immunoassay.
• Isolated hypothyroxinaemia. Normal TSH, low free T4. This is the subtype that looks like it should be treated and is not — the randomised evidence (below) shows no neurodevelopmental gain, and a low third-trimester free T4 is frequently an assay phenomenon rather than true maternal hypothyroxinaemia. Naming it correctly stops a reflex levothyroxine start.
• TPO-antibody status as a modifier, not a diagnosis. TPO positivity does not itself define disease; it shifts the trajectory and the threshold. A TPO-positive euthyroid woman has a higher rate of progression to overt hypothyroidism across pregnancy (so she needs repeat TSH even if the booking value is normal), a higher background miscarriage and preterm rate, and a higher risk of postpartum thyroiditis. It is the variable that tips a TSH-of-3.5 woman from "observe" toward "treat".

Severity and the atypical presentations

• Thyroid storm is the severity endpoint of hyperthyroidism and is a clinical, not biochemical, diagnosis — the degree of free-T4 elevation does not distinguish storm from uncomplicated thyrotoxicosis. The Burch–Wartofsky point scale (temperature, CNS dysfunction, GI–hepatic dysfunction, tachycardia, congestive failure, atrial fibrillation, and a precipitant) formalises it: a score ≥45 is highly suggestive, 25–44 impending. In pregnancy the precipitants are labour, caesarean, infection, pre-eclampsia and abruption — so storm characteristically declares itself intrapartum or immediately postpartum in a woman whose Graves was under-treated.
• Fetal thyrotoxicosis is the atypical presentation that has no maternal biochemical signature at all — the mother can be euthyroid, hypothyroid on replacement, or years post-ablation, yet her persisting TRAb crosses the placenta and drives a fetal tachycardia, goitre, growth restriction, advanced bone age or hydrops. The only way to anticipate it is to have measured TRAb; the only way to detect it is fetal surveillance.
• The decompensating myxoedematous picture is rare but is the hypothyroid severity endpoint — profound untreated hypothyroidism presenting with hypothermia, bradycardia and obtundation, usually in a woman lost to follow-up. It is mentioned because the management (cautious IV levothyroxine, steroid cover for possible co-existing adrenal insufficiency, treat the precipitant) diverges sharply from routine oral replacement.

Management

Frame immediate → ongoing → long-term, keeping the maternal and fetal threads explicit. The first-line drug choices are Intermediate groundwork; the consultant content here is the exact dosing, the regimen comparisons, the monitoring thresholds and the SA-specific substitution.

Hypothyroidism — immediate. A woman on levothyroxine who confirms pregnancy must increase her dose by ~25–30% at once — the pragmatic instruction is "take two extra tablets per week" (i.e. nine days' worth over seven), without waiting for a clinic TSH. Requirements rise because of TBG, transplacental transfer and increased volume of distribution. Newly diagnosed overt hypothyroidism is started on full replacement (~1.6 µg/kg/day) immediately.

Hypothyroidism — ongoing. Target TSH ≤2.5 mU/L (and within the trimester-specific range); recheck TSH every 4 weeks until mid-pregnancy, then at least once in the third trimester, adjusting in 25 µg steps. Take levothyroxine fasting and separate it from iron/calcium by ≥4 hours (both block absorption — a common cause of a "non-compliant" rising TSH).

Hyperthyroidism — immediate and ongoing. Treat true Graves thyrotoxicosis with the lowest effective antithyroid drug dose, titrating to keep free T4 in the upper third of the normal range (over-treatment causes fetal hypothyroidism/goitre). The teratogenicity trade-off drives the drug choice: propylthiouracil (PTU) in the first trimester (carbimazole/methimazole embryopathy — aplasia cutis, choanal/oesophageal atresia, omphalocele — clusters in weeks 6–10), then switch to carbimazole after the first trimester to avoid PTU's rare but serious hepatotoxicity. In the SA EML/SAMF setting carbimazole is the routinely stocked agent and PTU must be specifically sourced for the first trimester. Do not use block-and-replace in pregnancy: the antithyroid drug crosses the placenta but the levothyroxine does not, so the fetus is rendered hypothyroid. β-blockade (propranolol) controls symptoms short-term. Radioiodine is absolutely contraindicated; thyroidectomy (ideally second trimester) is reserved for drug failure/intolerance or airway-compromising goitre — and note thyroidectomy/ablation removes the gland but not the TRAb, which still threatens the fetus.

Dose conversion and the switch mechanics

The instruction "switch PTU to carbimazole after the first trimester" only protects the fetus if the doses are right, and dose errors at the switch are a classic source of either a relapse or an iatrogenic fetal goitre.

• Equivalence. The conventional clinical conversion is carbimazole 5 mg ≈ methimazole 5 mg ≈ PTU 50 mg (PTU is roughly one-tenth as potent on a milligram basis, and carbimazole is a prodrug almost wholly converted to methimazole). So a woman on PTU 150 mg/day converts to roughly carbimazole 15 mg/day — then titrate, do not assume, rechecking free T4 in 2–4 weeks because the equivalence is approximate.
• The narrow target. Aim free T4 in the upper third of the trimester reference range on the smallest dose that holds it there. The mechanism that makes over-treatment dangerous is that antithyroid drugs cross the placenta far more readily than maternal thyroid hormone does, so a maternal free T4 in the low-normal range usually means the fetus is already hypothyroid and may be growing a goitre. A maternal free T4 deliberately kept high-normal protects the fetus; chasing a "perfectly normal" maternal TSH does not.
• When to stop. Graves frequently remits across the second half of pregnancy as the immune milieu shifts; many women can have the antithyroid drug reduced or stopped in the third trimester. Stopping is reasonable when the dose is already minimal and free T4 is stable, with the explicit warning of a postpartum flare.
• The first-trimester counselling reality. Both drugs are teratogenic (below), so a woman already pregnant and stable on a low carbimazole dose at booking presents a genuine judgement call: a within-first-trimester switch to PTU is the guideline default, but if she presents after organogenesis (past ~10 weeks) the embryopathy window has largely passed and switching may add little. The reasoning matters more than the rule.

Thyroid storm — the named regimen and its sequence

Storm is the one thyroid emergency where the order of drugs is decisive, and the sequence is mechanistically driven:

1. Antithyroid drug first — PTU is preferred in storm (it both blocks new synthesis and inhibits peripheral T4→T3 conversion): PTU loading then high-dose maintenance via NG/oral.
2. Iodine (Lugol's or potassium iodide) at least one hour AFTER the antithyroid drug — given first, iodine would supply substrate for more hormone synthesis (the Jod-Basedow phenomenon); given after thionamide blockade it shuts down hormone release (the Wolff–Chaikoff effect).
3. β-blockade (propranolol) to control the adrenergic storm and blunt peripheral conversion.
4. Corticosteroid (hydrocortisone or dexamethasone) — reduces T4→T3 conversion and covers possible relative adrenal insufficiency.
5. Supportive care — active cooling, fluids, treat the precipitant (commonly infection, labour or caesarean), fetal monitoring. Delivery does not treat maternal storm — stabilise the mother first; an unstabilised storm is a worse environment for both than a controlled delay.

Fetal surveillance in Graves disease. If TRAb is >3× the upper limit of normal at 18–22 weeks, arrange serial growth and fetal-thyroid ultrasound looking for fetal goitre, tachycardia (>160 bpm), growth restriction, advanced bone age or hydrops — the signature of fetal thyrotoxicosis — and warn the neonatal team, as neonatal thyrotoxicosis can emerge 7–10 days postnatally once maternal antithyroid drug clears but TRAb persists. The 3× threshold is the ATA 2017 trigger: below it, the transplacental antibody load is unlikely to produce fetal disease and routine surveillance suffices; above it, fetal-thyroid imaging is the discriminator, because a fetal goitre with tachycardia and accelerated bone maturation signals fetal thyrotoxicosis (the mother is under-treated, or post-ablation with high TRAb), whereas a fetal goitre with bradycardia/normal heart rate and a maternal antithyroid drug on board signals fetal hypothyroidism (the mother is over-treated). The two look similar on a single scan and the management is opposite — reduce the maternal drug for fetal hyperthyroidism, reduce it likewise (or, rarely, intra-amniotic levothyroxine in expert hands) for the drug-induced hypothyroid goitre. This is a maternal–fetal-medicine decision, not a district-hospital one.

Subtype-specific management at a glance

• GTT / hyperemesis-associated thyrotoxicosis — supportive only: rehydration, antiemetics, thiamine, treat the hyperemesis. No antithyroid drug. Recheck thyroid function at ~14–18 weeks to confirm spontaneous resolution. A short course of a β-blocker for symptomatic tachycardia is the most that is justified.
• Graves, mild and stable — lowest-dose antithyroid drug, free-T4-targeted, TRAb-stratified fetal surveillance.
• Graves, drug-intolerant (agranulocytosis, hepatotoxicity) or refractory — this is the case for thyroidectomy in the second trimester after rendering euthyroid with β-blockade and iodine; radioiodine remains absolutely contraindicated. Surgery removes the gland but the TRAb persists, so fetal surveillance continues unchanged.
• Post-ablation / post-thyroidectomy mother on levothyroxine with high TRAb — biochemically hypothyroid, fetally at risk: the trap is to relax because "her thyroid is gone". Measure TRAb and monitor the fetus exactly as for an active-Graves mother.

Long-term / postpartum. Halve or stop antithyroid drugs as disease often improves to term then flares postpartum. Postpartum thyroiditis (destructive, often TPO-positive, classically thyrotoxic then hypothyroid phase in the first year) must be distinguished from recurrent Graves — low radioiodine uptake / TRAb-negativity separates them; the thyrotoxic phase is managed with β-blockade, not antithyroid drugs. PTU, carbimazole (≤20 mg/day) and levothyroxine are all breastfeeding-compatible.

Postpartum thyroiditis vs recurrent Graves — the consultant distinction

These present identically (a thyrotoxic woman 2–6 months postpartum, often TPO-positive, often with a new tremor and weight loss) and are managed oppositely, so the discrimination is the point:

The practical consequence: giving carbimazole to the thyrotoxic phase of postpartum thyroiditis does nothing for the thyrotoxicosis and risks tipping the woman into a deeper, symptomatic hypothyroid phase. Up to a quarter of women with postpartum thyroiditis remain permanently hypothyroid, and recurrence in a future pregnancy is high — so the hypothyroid phase needs a TSH check at 12 months and pre-conception counselling, not just reassurance.

The evidence & the controversy

Does treating subclinical hypothyroidism help the child? This is the central modern controversy and the reason the thresholds moved. Observational data (Maraka's meta-analysis) link subclinical hypothyroidism to roughly doubled pregnancy loss and neonatal death, which built the case for screen-and-treat. But the two large randomised trials — CATS and the Casey/MFMU trials — found no improvement in childhood IQ or neurodevelopment from levothyroxine started in the first half of pregnancy. The honest synthesis: treatment plausibly reduces obstetric loss (especially in TPO-positive women) but the neurodevelopmental rationale for universal screening collapsed under RCT scrutiny — so guidelines (ATA, RCOG) endorse targeted case-finding, not universal screening, and reserve treatment for overt disease, TSH >10, or TPO-positive subclinical disease.

The CATS story did not end at 3 years. The objection to CATS was always that 12–13 weeks might be "too late" and that 3-year IQ is an insensitive endpoint. CATS-II re-tested the same children at age 9.5 years and found the same null: mean IQ 101.76 (treated) vs 102.31 (untreated), with IQ <85 in 7.56% vs 11.22% — no significant treatment effect (p=0.576). The interesting positive finding was that suboptimal-function mothers had a similar proportion of low-IQ children to normal-function mothers, suggesting the maternal thyroid signal on cognition is weaker than the observational literature implied. A longer follow-up with a more sensitive test still did not rescue the neurodevelopmental hypothesis.

Why not just screen everyone? Because CATS and Casey show the downstream "benefit" (a smarter child) does not materialise, and treatment is not free of harm (over-replacement, anxiety, medicalisation). The principle is that a strong association (Maraka) does not license a screening programme once randomised treatment fails to move the outcome the programme was justified by.

The thyroid-antibody-and-miscarriage question is now answered — and the answer is "no". A separate, very common clinical situation is the euthyroid TPO-antibody-positive woman with recurrent miscarriage, in whom levothyroxine was widely prescribed on the observational association between antibodies and loss. Two randomised trials closed this: TABLET (NEJM 2019, n=952) gave levothyroxine 50 µg pre-conception in TPO-positive euthyroid women with miscarriage/infertility and found live-birth rates of 37% (176/470) vs 38% (178/470), RR 0.97 (95% CI 0.83–1.14), p=0.74 — no benefit. T4-LIFE (Lancet Diab Endocrinol 2022) asked the same in recurrent pregnancy loss and found live birth 50% vs 48%, RR 1.03 (0.77–1.38) — again no benefit (the trial stopped early for recruitment). The consultant conclusion: do not start levothyroxine in a euthyroid woman just because she is TPO-positive and has lost pregnancies — treat the thyroid only if the TSH is genuinely raised. This is a distinct point from the subclinical-hypothyroidism debate.

The antithyroid teratogenicity calculus. The instinct to call PTU "safe" is wrong: Andersen's Danish cohort showed both drugs raise birth defects above background, just with different patterns — methimazole/carbimazole 9.1% vs PTU 8.0% vs 5.7% unexposed. The defensible plan is therefore PTU only for the organogenesis window, then carbimazole — accepting that neither is risk-free and that the lowest dose is the real safety lever. The arithmetic: against a 5.7% background, carbimazole's 9.1% is an absolute excess of ~3.4 percentage points (about 1 extra malformation per 30 exposed) and PTU's 8.0% an excess of ~2.3 points — so the residual risk of "switching to PTU" is real, not zero, and dose minimisation is what actually changes outcomes.

Iodine in the SA context. Pregnancy needs 250 µg iodine/day (WHO). South Africa's mandatory salt iodisation (since 1995) eradicated overt goitre, but Johannesburg data show pregnant women slip toward only borderline sufficiency as salt intake falls — a legitimate argument for iodine-containing antenatal supplements where dietary salt is low, without tipping iodine-replete women into excess. The mechanism to keep straight is that the fetal thyroid cannot autoregulate against an iodine load the way the adult gland can, so a large iodine bolus (radiology contrast, kelp supplements, amiodarone) can induce fetal hypothyroid goitre — iodine sufficiency is the goal, iodine excess a distinct harm.

Landmark trials & key evidence

These are the defining studies. The recurring lesson: a strong observational association with thyroid biochemistry did not survive randomisation for the neurodevelopmental endpoint — so treatment is justified from obstetric loss, not child IQ.

Worked viva — how to structure the answer

Take a stem such as "a 28-year-old at 9 weeks, vomiting, weight loss, TSH <0.01, free T4 raised — how do you proceed?" The answer runs:

1. Frame it as a discrimination, not a diagnosis. "The first decision is whether this is gestational transient thyrotoxicosis driven by hCG — which needs no antithyroid drug — or Graves disease, which does. I do not start carbimazole reflexively on a suppressed TSH at 9 weeks."
2. Sort the subtype on the discriminators. "I look for a goitre, orbitopathy, a pre-pregnancy thyroid history, the T3:T4 pattern, the severity of the hyperemesis, and I send TRAb. GTT is TRAb-negative, eye- and goitre-free, proportionate to the vomiting and resolves by ~14–18 weeks; Graves is TRAb-positive and gestation-independent."
3. Manage the likely subtype, name the alternative. "If this is GTT I rehydrate, give thiamine and antiemetics, control symptoms with a short β-blocker if needed, and recheck thyroid function at ~16 weeks. If TRAb is positive or there are eye/goitre signs, I treat as Graves with the lowest-dose antithyroid drug — PTU in the first trimester, switching to carbimazole after it — titrating free T4 to the upper-normal range, and I stratify fetal surveillance on the TRAb level."
4. State the SA reality. "Carbimazole is the EML-stocked agent; I would specifically source PTU for the first trimester, and the TRAb-driven fetal surveillance belongs at a unit with maternal-fetal-medicine and neonatal support."
5. Appraise from evidence. "If she were instead a euthyroid TPO-positive woman with recurrent miscarriage, I would not start levothyroxine — TABLET and T4-LIFE show no benefit. And I would not over-treat subclinical disease for neurodevelopment — CATS, CATS-II and Casey are negative."
6. Close the loop. "Postpartum I anticipate a Graves flare and screen for postpartum thyroiditis, distinguishing the two on TRAb and uptake; both PTU and low-dose carbimazole are breastfeeding-safe."

Exam traps & red flags

• Reading TSH against non-pregnant norms. A first-trimester TSH of 0.2 mU/L with normal free T4 is usually physiological hCG effect, not Graves — don't start carbimazole on it.
• Treating gestational transient thyrotoxicosis as Graves. TRAb-negative, self-limiting, hyperemesis-associated thyrotoxicosis needs supportive care, not an antithyroid drug; the discriminator is TRAb plus eye/goitre signs.
• Forgetting to up-titrate levothyroxine at conception. Waiting for a clinic TSH lets the first-trimester fetus run hypothyroid — the "two extra tablets a week" rule exists for this.
• Block-and-replace in pregnancy. The classic wrong answer: it makes the fetus hypothyroid because only the antithyroid drug crosses the placenta meaningfully.
• Starting levothyroxine for TPO antibodies alone. A euthyroid TPO-positive woman with miscarriage does not benefit (TABLET, T4-LIFE) — treat only a genuinely raised TSH.
• Targeting a "normal" maternal TSH on antithyroid drugs. Aiming for a normal maternal TSH means a low-normal free T4, which usually means a hypothyroid, goitrous fetus — keep maternal free T4 in the upper third of the range on the smallest dose.
• Misreading the fetal goitre. Fetal goitre + tachycardia + advanced bone age = fetal hyperthyroidism (mother under-treated/post-ablation); fetal goitre + normal/slow heart rate on maternal antithyroid drug = fetal hypothyroidism (mother over-treated). Opposite management — get it the wrong way round and you worsen it.
• Thinking maternal thyroidectomy/radioiodine removes the fetal risk. TRAb persists and can still cause fetal/neonatal thyrotoxicosis — keep monitoring TRAb regardless of maternal thyroid status.
• Radioiodine in pregnancy (ablation or scan) — absolutely contraindicated; exclude pregnancy before any radioiodine.
• Missing the postnatal window. Neonatal thyrotoxicosis can present 7–10 days after birth as antithyroid drug clears faster than maternal TRAb; and thyroid storm (precipitated by labour, infection, caesarean) is a maternal emergency — PTU, β-blocker, iodine (given after the antithyroid drug), corticosteroid and cooling.
• Treating the thyrotoxic phase of postpartum thyroiditis with carbimazole. It is a destructive leak, not over-synthesis — antithyroid drugs do nothing and deepen the later hypothyroid phase; use a β-blocker.
• Iodine excess as well as deficiency. A large iodine load (contrast, kelp, amiodarone) can induce fetal hypothyroid goitre — sufficiency, not loading, is the goal.
• Calling subclinical hypothyroidism a neurodevelopmental emergency. CATS, CATS-II and Casey say otherwise — over-treating biochemical-only disease is a defensible-sounding but evidence-poor answer.

Evidence anchors

• ATA 2017 — Guidelines for the Diagnosis and Management of Thyroid Disease During Pregnancy and the Postpartum (Alexander et al, Thyroid 2017;27:315–389)
• RCOG Green-top Guideline No. 76 — Management of Thyroid Disorders in Pregnancy (Chan et al, BJOG 2025;132:e130–e161)
• CATS — Antenatal Thyroid Screening and Childhood Cognitive Function (Lazarus et al, NEJM 2012;366:493–501)
• CATS-II — Effect of Treating Maternal Suboptimal Thyroid Function on Child Cognition at 9.5 years (Hales et al, JCEM 2018;103:1583–1591)
• Casey/MFMU — Treatment of Subclinical Hypothyroidism or Hypothyroxinemia in Pregnancy (NEJM 2017;376:815–825)
• TABLET — Levothyroxine in Women with Thyroid Peroxidase Antibodies before Conception (Dhillon-Smith et al, NEJM 2019;380:1316–1325)
• T4-LIFE — Levothyroxine in euthyroid TPO-antibody-positive women with recurrent pregnancy loss (van Dijk et al, Lancet Diabetes Endocrinol 2022;10:322–329)
• Maraka — Subclinical Hypothyroidism in Pregnancy: Systematic Review and Meta-Analysis (Thyroid 2016;26:580–590)
• Andersen — Birth Defects After Early Pregnancy Use of Antithyroid Drugs: a Danish Nationwide Study (JCEM 2013;98:4373–4381)
• van Dijk — Maternal TRAb Concentration and the Risk of Fetal and Neonatal Thyrotoxicosis: a Systematic Review (Thyroid 2018;28:257–264)
• WHO — Iodine supplementation during pregnancy (e-Library of Evidence for Nutrition Actions)
• Iodine status of pregnant women living in urban Johannesburg, South Africa (PMC8710104)
• National Department of Health, Guidelines for Maternity Care in South Africa and the SA EML/SAMF (carbimazole as the routinely stocked antithyroid drug; PTU sourced for first-trimester use) — no single stable canonical URL