Table 1.

Characteristics of included studies.

Study	Study characteristics	Patient characteristics	Outcomes
Chan 2015 (12)	Retrospective cohort study in Hong Kong. Follow-up duration: 24 months. Hyperthyroidism was defined as serum FT4 > 23 pmol/L with concomitant TSH < 0.03 mIU/L.Definition of AF was not stated.	Six hundred forty-two patients with new-onset non-valvular TAF. Mean age of patients with TAF = 71.9 ± 14.6 years.One hundred thirty-six patients with TAF were on warfarin, 243 on aspirin, and 263 had no treatment. Mean CHA2DS2-VASc score was 3.3 ± 1.9 for all patients, 2.8 ± 1.7 for those on no treatment, 3.8 ± 1.9 for those on aspirin, and 3.2 ± 2.0 for those on warfarin. Mean HAS-BLED score was 1.8 ± 1.0 for all patients, 1.6 ± 0.9 for those on no treatment, 2.0 ± 1.0 for those on aspirin, and 1.7 ± 1.1 for those on warfarin.Baseline characteristics of patients on warfarin and those on antiplatelet therapy were compared. There were statistically significant differences in mean age, hypertension, CAD, prior stroke or TIA, mean CHA2DS2-VASc score, and mean HAS-BLED score between groups.	Stroke and ICH based on medical records and discharge summaries during the follow-up period. Stroke was defined as a sudden neurologic deficit persisting > 24 hours corresponding to a vascular territory in the absence of primary haemorrhage or other causes and confirmed by CT or MRI of the brain.ICH was determined based on new-onset neurological symptoms with radiological confirmation.
Chan 2020 (20)	Retrospective cohort study in Taiwan. Follow-up from date of first prescription of anticoagulant until occurrence of outcome or end of study. Hyperthyroidism and AF were determined by ICD-9-CM codes recorded in the Taiwan National Health Insurance Program.	Four thousand ninety-four patients with non-valvular TAF. Onset of AF was not mentioned. Mean age of patients with TAF = 69.5 ± 13.3 years. 1181 patients with TAF were on warfarin and 3213 were on DOACs. Mean CHA2DS2-VASc score was 2.5 ± 2.0 for TAF patients on warfarin and 3.2 ± 1.8 for TAF patients on DOACs. HAS-BLED score was not stated.Baseline characteristics of TAF patients with and without DOACs were compared. P-values were not stated.	IS/SE, ICH, hospitalized GIB, and all major bleeding based on ICD-9-CM codes in the National Health Insurance Research Database.
Chen 2014 (7)	Prospective cohort study in China. Follow-up duration: 34.1 ± 0.6 months.Hyperthyroidism was defined as serum FT4 > 23 pmol/L with concomitant TSH < 0.03 mIU/L.AF was confirmed by ≥ two ECGs recorded ≥ four hours apart.	Sixty-two patients with TAF. Patients with RHD were excluded. Onset of AF was not stated but eight patients had persistent AF. Mean age of patients with TAF = 55.1 ± 1.7 years. four patients with TAF were on warfarin, 12 were on aspirin, and 58 had no treatment. Mean CHA2DS2-VASc score was 1.6 ± 0.2 for TAF patients. HAS-BLED score was not stated.Baseline characteristics of TAF and non-thyroid AF patients were compared. There was statistically significant difference in the mean age between groups.	IS, SE, and TIA. IS was defined as a sudden neurologic deficit persisting > 24 hours corresponding to a vascular territory in the absence of primary haemorrhage or other causes and confirmed by CT or MRI of the brain.Definition of SE was not stated.
Chen 2019 (21)	Retrospective cohort study in Taiwan. Follow-up from index date to death, loss to follow-up, or end of study. Hyperthyroidism and AF were determined based on ICD-9-CM codes recorded in the Taiwan National Health Insurance Program. Only patients with ≥ two consensus-diagnosed episodes of thyrotoxicosis or who had been prescribed anti-thyroid drugs were included.	One thousand eight hundred sixty-eight patients with new-onset TAF. Patients with RHD were excluded. Mean age of patients with TAF was not stated. Four hundred fifty patients with TAF were on warfarin, six on were DOACs, and 1163 were on aspirin. Number of patients with TAF on no treatment was not stated. Of the TAF patients, 19.7% had CHA2DS2-VASc score of 0, 33.9% had score of 1, 18.5% had score of 2, and 27.9% had score ≥ 3. HAS-BLED score was not stated.Baseline characteristics of TAF and non-thyroid AF patients were compared. There were statistically significant differences in ACEI/ARB, beta blocker, digoxin, and statin use between groups.	IS, mortality, major bleeding, hemorrhagic stroke, minor GIB, and bleeding during re-operation based on ICD-9-CM codes in the National Health Insurance Research Database.
Fauchier 2016 (22)	Retrospective cohort study in France. Patients were followed-up from first record of non-valvular AF to end of study, with mean follow-up duration of 944 days. Definition of hyperthyroidism was not stated. AF was diagnosed based on the replacement of consistent P waves by rapid oscillations or fibrillatory waves that vary in amplitude, shape, and timing, associated with an irregular, frequently rapid ventricular response when atrioventricular conduction was intact.	Thirty-six patients with TAF. Onset of AF was not stated. Patients with valvular AF were excluded. Mean age of patients with TAF was not stated. Number of patients with TAF on warfarin, DOACs, aspirin, and no treatment were not stated. CHA2DS2-VASc and HAS-BLED scores of patients with TAF were not stated. Baseline characteristics of patients with AF from temporary and non-temporary causes were compared. There were statistically significant differences in age, heart failure, hypertension, diabetes, vascular disease, CAD, previous myocardial infarction, previous percutaneous intervention, renal failure, chronic pulmonary disease, CHA2DS2-VASc score, HAS-BLED score, ACEI/ARB, and diuretic use between groups.	Stroke, SE, bleeding, and death based on ICD-10 codes in the discharge data from a computerized system, and mortality data from an online regional database.
Goldstein 2019 (14)	Double-blind randomized controlled trial in North America, South America, Europe, and Asian Pacific countries. Follow-up duration: 21.6 months (median).	Three hundred twenty-one patients with new-onset and existing TAF. Patients with moderate-to-severe mitral stenosis and prosthetic mechanical valves were excluded. Median age of patients with TAF = 69 years (IQR: 62–76). Number of patients with TAF on warfarin and DOACs were not stated. Eighty-six patients with TAF were on aspirin and seven were on clopidogrel. Mean CHA2DS2-VASc score of patients with TAF was 2.2 ± 1.2. HAS-BLED score was not stated.	Composite of ischemic or hemorrhagic stroke or SE, major bleeding, any stroke, IS, hemorrhagic stroke, all-cause mortality, myocardial infarction, hospitalization for heart failure, any bleeding. Outcomes were adjudicated based on pre-specified criteria established by a clinical events committee.
	Hyperthyroidism was determined based on investigator responses on the ARISTOTLE data intake form. Patients who were noted to be taking antithyroid therapy were included as hyperthyroid. AF or atrial flutter were determined by ECG ≥ two weeks apart in the 12 months before enrolment.	Baseline characteristics of patients with TAF and non-thyroid AF were compared. There were statistically significant differences in gender, ethnicity, weight, prior bleeding, and prior vitamin K antagonist use within 1 month between groups.	IS was defined as a focal neurological deficit, from a nontraumatic cause, lasting ≥ 24 hours and was categorized as ischemic, hemorrhagic, or of uncertain type.Bleeding was defined according to the ISTH criteria, as clinically overt bleeding accompanied by a decrease in the hemoglobin level of ≥2 g/dL or transfusion of ≥2 units of packed cells, occurring at a critical site, or resulting in death. Clinically relevant non-major bleeding was defined as clinically overt bleeding that did not satisfy the criteria for major bleeding and that led to hospital admission, physician-guided medical or surgical treatment, or a change in antithrombotic therapy.Definition of SE, all-cause mortality, myocardial infarction, and hospitalization for heart failure were not stated.
Gundlund 2019 (15)	Retrospective cohort study in Denmark. Patients were followed-up from 4 weeks after discharge until outcome occurred, death, 5 years after discharge, emigration, or end of study. Hyperthyroidism and AF were determined by ICD-8 and ICD-10 codes obtained from the Danish Registries.	Two thousand five hundred seven patients with new-onset TAF. Patients with valvular AF were excluded. Median age of patients with TAF = 73 years (IQR: 63–81). One thousand hundred patients with TAF were on anticoagulation and 1407 had no anticoagulation. Median CHA2DS2-VASc score of patients with TAF was 3 (IQR: 2–4) and median HAS-BLED score of patients with TAF was 2 (IQR: 1–3).Baseline characteristics of TAF and non-thyroid AF patients were compared. P-values were not reported.	Composite of IS, TIA, or SE, AF re-hospitalization and all-cause mortality based on ICD-8 and ICD-10 codes in the Danish Registries.
Siu 2009 (13)	Prospective cohort study in Hong Kong. Follow-up duration: 12 months. Hyperthyroidism was defined as serum FT4 > 23 pmol/L with concomitant TSH < 0.03 mIU/L.AF was confirmed by ≥ two ECGs recorded ≥ four hours apart.	One hundred sixty patients with new-onset TAF. Patients with structural heart disease were excluded. Mean age of patients with TAF = 64.7 ± 1.3 years. 47 patients with TAF were on warfarin and 62 were on aspirin. Mean CHA2DS2-VASc score of patients with TAF was 0.7 ± 0.1. HAS-BLED score was not stated.Baseline characteristics of TAF and non-thyroid AF patients were compared. There were statistically significant differences in hypertension, CHADS2 score, beta blocker, calcium antagonist, digoxin, propafenone, sotalol, and amiodarone use between groups.	IS and sinus rhythm conversion based on data from discharge summaries and outpatient clinic records. IS was defined as a sudden neurologic deficit persisting > 24 hours corresponding to a vascular territory in the absence of primary hemorrhage or other causes and confirmed by CT or MRI of the brain.
Wong 2017 (6)	Retrospective cohort study in Hong Kong. Follow-up duration = 47 months (median) (IQR: 24 – 82).Hyperthyroidism was defined as elevated FT4 or FT3 and concomitant suppressed TSH. AF or atrial flutter was confirmed with standard 12-lead ECG.	One hundred thirty-three patients with TAF. Onset of AF was not stated. Patients with significant mitral valve disease were excluded. 16 patients had moderate-to-severe mitral regurgitation. Mean age of patients with TAF was not stated. 12 patients with TAF were on warfarin, 1 was on DOACs, and 50 were on aspirin. Mean CHA2DS2-VASc score of patients with TAF and IS was 2.9 and that of patients with TAF and no IS was 2.3. HAS-BLED score was not stated. Baseline characteristics of patients with spontaneous AF resolution and those with persistent AF were compared. There were statistically significant differences in smoking status, FT4 level, history of thyrotoxicosis, ACEI/ARB, and digitalis use between groups.	IS and sinus rhythm conversion based on data from a Clinical Data Analysis and Repository System. IS was defined as a sudden neurologic deficit persisting > 24 hours corresponding to a vascular territory in the absence of primary hemorrhage or other causes and confirmed by CT or MRI of the brain.

ACEI: angiotensin-converting enzyme inhibitor, AF: atrial fibrillation, ARB: angiotensin-2 receptor blocker, CAD: coronary artery disease, CT: CT, DOACs: direct oral anticoagulants, ECG: ECG, FT3: free thyronine, FT4: free thyroxine, GIB: gastrointestinal bleeding, ICD: International Classification of Diseases, ICD-9-CM: International Classification of Diseases 9^th Revision Clinical Modification, ICH: intracranial hemorrhage, IQR: inter-quartile range, IS/SE: ischemic stroke or systemic thromboembolism, ISTH: International Society on Thrombosis and Haemostasis, MRI: MRI, RHD: rheumatic heart disease, TAF: thyrotoxic atrial fibrillation, TIA: transient ischemic attack, TSH: thyroid stimulating hormone.