Di Biase 2014.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Total duration of the study: January 2010 to April 2014 Duration of follow‐up: 48 hours for the stroke events Number of study centres and location: 7 centres (Texas Cardiac Arrhythmia Research Foundation, University of Kansas, California Pacific Medical Center, Stanford University, Case Western Reserve University, Southlake Regional Health Centre, Catholic University, Italy)
Participants	Inclusion criteria: aged ≥ 18 years, INR 2.0–3.0 at 3–4 weeks before ablation, and CHADS2 score ≥ 1 Exclusion criteria: known bleeding disorders or inherited thrombophilic disorder, oral contraceptives or oestrogen replacement therapy, prosthetic heart valves, and contraindications to warfarin therapy Number of randomised participants: 1584 Total number of participants: 1584 Number lost to follow‐up/withdrawn: 0 Number of analysed participants: 1584 Number of participants in each treatment group: warfarin interrupted: 790; warfarin uninterrupted: 794 Baseline characteristics Interrupted anticoagulation Age, years, mean: 61 (SD 10) Men, n (%): 602 (76) Type of AF, n (%): paroxysmal: 229 (29) persistent: 174 (22) long‐standing persistent: 387 (49) Comorbidities, n (%): CAD: 182 (23) hypertension: 640 (81) CHF: 118 (15) diabetes mellitus: 302 (38) prior stroke/TIA: 55 (7) CHADS2 score, n (%): 1 = 229 (29) 2 = 268 (34) 3 = 170 (22) 4 = 94 (12) ≥ 5 = 32 (4.1) Uninterrupted anticoagulation Age, years, mean: 62 (SD 12) Men, n (%): 590 (74) Type of AF, n (%): paroxysmal: 200 (25) persistent: 189 (24) long‐standing persistent: 405 (51) Comorbidities, n (%): CAD: 206 (26) hypertension: 660 (83) CHF: 136 (17) diabetes mellitus: 318 (40) prior stroke/TIA: 64 (8) CHADS2 score, n (%): 1 = 206 (26) 2 = 284 (36) 3 = 152 (19) 4 = 101 (13) ≥ 5 = 48 (6.0) Group differences: baseline characteristics and risk factors well balanced between groups
Interventions	Periprocedural anticoagulation Anticoagulant used: warfarin Dose: as per the participant's scheduled dose Duration of anticoagulant therapy in trial: all participants received warfarin before the procedure to achieve 3–4 weeks of therapeutic INRs Intensity of anticoagulation or dose adjustment: participants had to have a therapeutic INR. If, on the day of the procedure, participants had an INR > 3.5, they were excluded. If the INR was 3–3.5, fresh‐frozen plasma was administered a few hours before the procedure. Some participants presented on the day of the procedure with a subtherapeutic INR and were not excluded Adherence to anticoagulant treatment: not reported Time of interruption of the anticoagulants prior to procedure: warfarin was discontinued 2–3 days before the ablation Heparin bridge therapy: participants in group 1 (off‐warfarin) were bridged with enoxaparin 1 mg/kg twice daily (a low molecular weight heparin) until the evening before the ablation procedure. After the procedure, enoxaparin 0.5 mg/kg twice daily was routinely started. It was stopped when the INR was > 2. Warfarin was restarted the night of the procedure Time and strategy of resumption of OAC after procedure: warfarin was restarted the night of the procedure Duration of follow‐up: 48 hours' postprocedure TOE: performed in all participants in group 1 and when the participant presented with a subtherapeutic INR on the day of the procedure in group 2 Ablation procedure Type of ablation: pulmonary vein antrum isolation Ablation energy source: RFCA Intraprocedural anticoagulation Group 1 received 15,000 international units IV and continuous infusion of heparin 1000 units/hour was started to maintain an ACT > 350 seconds Group 2 received bolus of 10,000 IU in men and 8000 IU in women. ACT was maintained at > 300 seconds. In both groups, transseptal sheaths were pulled when the ACT was < 200 seconds
Outcomes	Primary outcome Thromboembolic events (during 48 hours after ablation), defined as stroke/TIA or systemic thromboembolism Stroke was defined as the onset of a new neurological deficit that occurred anytime during or within 48 hours of the procedure. If the duration of the deficit was < 24 hours, it was defined as a TIA Secondary outcomes Bleeding complications defined as major (requiring intervention) occurrence of cardiac tamponade or haemopericardium requiring intervention, causing symptoms, or requiring transfusion; haematoma requiring intervention; massive haemoptysis; haemothorax; and retroperitoneal bleeding minor (not requiring intervention) bleeding occurrence of haematoma or any bleeding that did not require any intervention or did not cause any symptoms Pericardial effusions were analysed separately as a secondary endpoint of the study Silent thromboembolic lesion: no definition reported
Notes	Investigators' conflicts of interest: quote: "Dr Di Biase serves as a consultant for Hansen Medical, Biosense‐Webster, and St. Jude Medical. Dr Di Biase also received speaker honoraria from Biotronik and Atricure. Dr Gallinghouse is a consultant for Hansen Medical. Dr Natale has received honoraria for serving on the speakers’ bureau for St. Jude Medical, Boston Scientific, Medtronic, and Biosense‐Webster. Dr Natale is consultant for Biosense Webster and St. Jude Medical. The other authors report no conflicts" Funding: Texas Cardiac Arrhythmia Research Foundation Country: US Setting: St David's Medical Center Comments: none Author's name: Andrea Natale Institution: Texas Cardiac Arrhythmia Institute Email: dr.natale@gmail.com Address: 3000 N I‐35, Ste 720, Austin, Texas, USA, 78705
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "block randomization was performed with study center as the blocking variable. A central randomization algorithm was used to generate the randomization code."
Allocation concealment (selection bias)	Low risk	Used a central randomisation algorithm.
Blinding of participants and personnel (performance bias) all outcomes	High risk	Quote: "operators were not blinded to the anticoagulation management, which introduced a bias in the study."
Blinding of outcome assessment (detection bias) all outcomes	Low risk	Quote: "Stroke and TIA diagnoses were performed by a neurologist who was blinded to the patient's group assignment. Diagnoses of peripheral embolic events or deep venous thrombosis were performed by other physicians blinded to the group assignment."
Incomplete outcome data (attrition bias) all outcomes	Low risk	Participant flow chart was provided, no lost to follow‐up, no missing data.
Selective reporting (reporting bias)	Low risk	The primary outcome matched the protocol based on the history of changes. Although there was no prespecified plan for silent thromboembolic lesions outcome, we did not consider this a major bias given the importance of primary outcomes.