Methimazole exposure in the first trimester is associated with a characteristic pattern of anomalies that includes aplasia cutis and choanal and esophageal atresia. Propylthiouracil was long considered free of such effects, but a Danish national study found it too was associated with birth defects, which were less severe (Andersen et al., 2013). Propylthiouracil in turn carries a risk of severe liver injury, which led to advice that its use be limited largely to the first trimester (Bahn et al., 2009). The period of greatest teratogenic sensitivity appears to be gestational weeks six to ten (Laurberg and Andersen, 2014).

Guidelines therefore recommend that women with Graves' disease be counseled before conception, that those taking methimazole who become pregnant be reviewed urgently, and that propylthiouracil be preferred when treatment is needed in early pregnancy. They also allow that a woman who is euthyroid on a low dose, with features suggesting remission, may have the drug withdrawn under close supervision with frequent thyroid function tests (Alexander et al., 2017; Kahaly et al., 2018). The distinction between supervised withdrawal in a selected patient and unsupervised cessation in any patient is the point on which the present pathway turns.

No laboratory test separates storm from uncomplicated thyrotoxicosis, since hormone concentrations overlap. Two clinical systems are in use. The Burch-Wartofsky Point Scale assigns points for temperature, central nervous system effects, gastrointestinal and hepatic dysfunction, heart rate, heart failure, atrial fibrillation and the presence of a precipitant; a total of 45 or more is highly suggestive of storm, 25 to 44 suggests impending storm and a total below 25 makes it unlikely (Burch and Wartofsky, 1993). The Japan Thyroid Association criteria require biochemical thyrotoxicosis and define storm by combinations of the same organ-system features (Akamizu et al., 2012). The American Thyroid Association recommends that the diagnosis rest on clinical judgment supported by one of these tools (Ross et al., 2016). Neither system was derived or validated in pregnant women.

Normal pregnancy raises heart rate and cardiac output, produces heat intolerance and, in the first trimester, nausea and vomiting. Chorionic gonadotropin stimulates the thyroid, suppressing thyrotropin and sometimes raising free thyroxine, and estrogen increases thyroxine-binding globulin and with it total hormone concentrations (Glinoer, 1997). Gestational transient thyrotoxicosis, often accompanying hyperemesis, must be told apart from Graves' disease. Later in pregnancy, preeclampsia, sepsis and peripartum cardiomyopathy can each mimic or precipitate decompensation. A scoring system that counts tachycardia and vomiting will therefore read higher in a pregnant woman for reasons unrelated to the thyroid, and the pathway has to allow for that.

Prevention begins well before pregnancy. Every woman of reproductive age prescribed a thionamide should be told, and have it recorded, that pregnancy should be planned where possible, that the drugs carry a small risk to the embryo, that untreated hyperthyroidism carries risks too, and that on a positive pregnancy test she should contact her clinician at once and not stop treatment on her own. The option of definitive treatment before conception is part of the same conversation (Alexander et al., 2017).

The pathway proposes that a statement by a pregnant woman with Graves' disease that she has stopped her antithyroid drug be treated as a trigger, whoever hears it: midwife, pharmacist, general practitioner or booking clerk. The trigger leads to thyroid function testing and specialist review within days. Endocrine guidance already calls for close monitoring when treatment is changed in early pregnancy (De Groot et al., 2012); the proposal here is only that the same urgency apply when the change was the patient's own.

A woman who presents unwell is assessed for temperature, pulse and rhythm, mental state, signs of heart failure, vomiting, diarrhea and jaundice. Free thyroxine, triiodothyronine and thyrotropin are measured and read against pregnancy-specific expectations. A Burch-Wartofsky score is calculated and recorded with its components, so that later readers can see which features drove it. Because pregnancy inflates some items, the score informs the decision and does not make it. A search for a second precipitant, particularly infection and preeclampsia, proceeds in parallel (Chiha et al., 2015).

Suspected or impending storm is managed in a high-dependency or intensive care setting, as the Japanese guideline advises for storm in general (Satoh et al., 2016). The pathway asks that a senior endocrinologist, obstetrician and intensivist be informed together at the point of suspicion, with an anesthetist and neonatologist added once the fetus is viable. A single named clinician coordinates.

Treatment of storm is multimodal and its components are the same in pregnancy as outside it: a thionamide to block hormone synthesis, iodine to block release, given only after the thionamide has been started, a beta-adrenergic blocker to control heart rate, a glucocorticoid, cooling, fluids and treatment of the precipitant (Ross et al., 2016; Nayak and Burman, 2006). The pathway adds nothing to these recommendations. It asks only that the order of administration and the monitoring required for each agent be written into the local protocol in advance, so that they are not worked out at the bedside.

Maternal observation follows critical care practice. Fetal heart rate is monitored at a gestation where intervention would be considered; fetal tachycardia is expected while the mother is thyrotoxic and febrile and generally improves as she does. Obstetric guidance is that delivery during storm should be avoided unless there is an obstetric indication, because labor and surgery are themselves precipitants (ACOG, 2020).

Once the acute illness is controlled, the team agrees which antithyroid drug will be continued and at what gestation it might be changed, how often thyroid function will be checked, when thyrotropin receptor antibodies will be measured to assess fetal and neonatal risk, and how the newborn will be followed (Alexander et al., 2017; Cooper and Laurberg, 2013). The postpartum period, when Graves' disease commonly relapses, and the question of definitive treatment are addressed before discharge. The conversation in Step 1 is repeated, this time with the benefit of experience.