Ojima 2017.

Methods	RCT, parallel design
Participants	Total number of randomized participants: 100 Inclusion criteria: pathologically confirmed thoracic oesophageal carcinoma; planned right transthoracic oesophagectomy with 2‐ or 3‐field lymph node dissection; clinical stage I, II, III or IV according to the UICC TNM classification; Eastern Cooperative Oncology Group performance status 0‐1; 20‐85 years of age; normal major organ function Exclusion criteria: resting HR < 50 bpm, 2nd‐ or 3rd‐degree atrioventricular block; history of AF. In addition, at randomization, participants were excluded if they were receiving dopamine, SBP < 80 mmHg or > 160 mmHg, HR < 50 bpm, or with arrhythmias or requiring ventilator assistance Type of surgery: oesophagectomy Baseline characteristics Intervention group (landiolol) Age, median (range): 68 (31‐85) years Gender, M/F: 36/14 History of MI, %: 4 Ejection fraction, median (range), %: 60.7 (57.5‐70.2) History of hypertension, %: 44 History of COPD, %: 20 Control group (placebo) Age, median (range): 69 (45‐83) years Gender, M/F: 41/9 History of MI, %: 2 Ejection fraction, median (range), %: 60.6 (44.8‐83.0) History of hypertension, %: 48 History of COPD, %: 14 Country: Japan Setting: hospital; single centre
Interventions	Intervention group (landiolol) Randomized, n = 50; losses = 0; analysed, n = 50 (use of ITT analysis not reported) Details: started on first postoperative day, landiolol hydrochloride as an infusion starting at 3 µg/kg/min, and discontinuing after 72 h Control group (placebo) Randomized, n = 50; losses = 0; analysed, n = 50 Details: 5% glucose solution at 3 mL/h over the same period as the intervention
Outcomes	Outcomes measured/reported by study authors: AF (between days 1‐7); rate of occurrence of AF in hospital; postoperative complications; haemodynamic parameters; change in inflammatory markers; mobility scores; side effects (to include hypotension and bradycardia); mortality Outcomes relevant to the review: AF, hypotension, bradycardia; mortality
Notes	Funding/declarations of interest: funding not reported. Study authors declare no conflicts of interest Study dates: March 2013‐January 2016 Notes: if AF occurred during the study period, then treatment was unblinded, and were treated with landiolol hydrochloride at 3‐5 µg/kg/min
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomization at a central registry
Allocation concealment (selection bias)	Low risk	Organized at a central registry
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Patients, care providers and investigators were blinded to the treatments." Quote: "The appearance and labelling of the dosage were indistinguishable between landiolol and placebo"
Blinding of outcome assessors (detection bias) All outcomes	Low risk	As above
Incomplete outcome data (attrition bias) All outcomes	Low risk	No losses
Selective reporting (reporting bias)	Low risk	Prospective clinical trial registration (UMIN000010648). Outcomes are reported according to clinical trial documents
Other bias	High risk	Participants in either group who had AF were treated with landiolol