**Background:** Thyroid storm is a life-threatening endocrine emergency with reported mortality of 8%–25%. The 2016 American Thyroid Association (ATA) guidelines recommend propylthiouracil as first-line therapy based on its additional property of inhibiting peripheral conversion of T4 to T3, leading to more rapid decline in serum T3 levels. However, the Japan Thyroid Association guidelines allow either propylthiouracil or methimazole, citing observational data showing similar outcomes. Large-scale direct comparisons have been lacking, and a randomized trial is unlikely given the rarity of the condition.
**Methods:** This study used the Premier Healthcare Database (January 2016–December 2020), which captures approximately 25% of US hospitalizations. Included patients were adults (≥18 years) admitted to an ICU or intermediate care unit on hospital day 1 or 2, with an ICD-10 code for thyroid storm designated as present on admission, who received either propylthiouracil or methimazole (but not both) on hospital day 1 or 2, and who also received corticosteroids on the first day of thionamide use. The primary exposure was initiation of propylthiouracil vs methimazole, assigned akin to intention-to-treat. The primary outcome was the composite of in-hospital death or discharge to hospice. Secondary outcomes included in-hospital death alone, hospice discharge alone, organ support–free days by day 21, thionamide costs, and total hospital costs. Covariates included demographics, organ dysfunction, comorbidities, concurrent medications (β-blockers, vasopressors, corticosteroids, potassium iodide), and hospital-level factors. The primary analysis used targeted maximum likelihood estimation, a doubly robust semiparametric approach with ensemble machine learning.
**Key Results:** Of 2845 patients identified, 1383 met full inclusion criteria: 656 (47.4%) initiated propylthiouracil and 727 (52.6%) initiated methimazole. Mean age was 45 years in both groups; 72.1% of propylthiouracil and 71.5% of methimazole patients were female. The propylthiouracil group had higher rates of ICU admission (73.2% vs 60.0%), vasopressor use (15.1% vs 9.9%), invasive mechanical ventilation (18.9% vs 14.0%), and potassium iodide use (41.3% vs 26.0%). The primary composite outcome occurred in 7.4% overall (102/1383): 8.5% (56/656) in the propylthiouracil group and 6.3% (46/727) in the methimazole group. The adjusted risk difference was 0.6% (95% CI, −1.8% to 3.0%; P = .64). No significant differences were found for in-hospital death alone (adjusted RD 1.1%; 95% CI, −1.2% to 3.5%; P = .34), hospice discharge alone (adjusted RD −0.6%; 95% CI, −1.6% to 0.3%; P = .20), organ support–free days (adjusted mean difference −0.2 days; 95% CI, −0.7 to 0.3; P = .40), or total hospitalization costs (adjusted mean difference $1097; 95% CI, −$972 to $3166; P = .30). Thionamide costs were higher for propylthiouracil (adjusted mean difference $72; 95% CI, $48 to $97; P < .001). Adverse events were rare in both groups: thyroidectomy (2 vs 1), acute hepatic failure (1 vs 0), agranulocytosis (1 vs 1), and acute pancreatitis (0 vs 0). Subgroup and sensitivity analyses (restricting to one admission per patient, excluding glucocorticoid recipients, and using full optimal matching) were consistent with the primary analysis.
**Clinical Implications:** This study provides robust evidence that propylthiouracil and methimazole are associated with comparable clinical outcomes in the management of thyroid storm, challenging the current ATA recommendation favoring propylthiouracil as first-line therapy. The findings support the position of the Japan Thyroid Association that either agent can be used interchangeably. The lower-than-recommended median methimazole dose (40 mg vs the ATA-recommended 60–80 mg) may reflect real-world practice or potential undertreatment, but did not appear to worsen outcomes. Limitations include the use of claims-based diagnosis (not validated against clinical scoring systems), inability to ascertain pre-hospital medications, and lack of data on subsequent thionamide titration. Nonetheless, given the rarity of thyroid storm and the impracticality of a randomized trial, this large comparative effectiveness study offers the best available evidence to guide thionamide selection.