Soares 2015.

Methods	Trial design: RCT Follow up: 7 days
Participants	Country: Brazil Setting: single centre, Federal University of Ceará Inclusion criteria: ‐ People using warfarin ‐ Requiring extractions of at least 2 teeth ‐ Minimum age of 18 years Exclusion criteria: ‐ Use of associated anticoagulating medications (other than warfarin) ‐ Smokers, pregnant or lactating women ‐ Reported pre‐existing blood dyscrasias ‐ Individuals who did not return for reassessment were removed from the trial ‐ Surgical sites subjected to procedures requiring mucoperiosteal flap for dental extractions or that lasted longer than 30 min were not included in the study sample Sample size: TXA group: ‐ Total number of surgical procedures: 28 ‐ Total number of participants: 12 Fibrin sponge group: ‐ Total number of surgical procedures: 28 ‐ Total number of participants: 13 Control group (dry gauze): ‐ Total number of surgical procedures: 28 ‐ Total number of participants: 13 Number of individuals analysed/randomised: ‐ Total number of surgical procedures analysed/randomised: 84/93 ‐ TXA group: 28/33 ‐ Fibrin sponge group: 28/32 ‐ Control group (dry gauze): 28/28 Participant characteristics: ‐ Age in years (mean (SD)) TXA group: 51.0 (2.9); fibrin sponge group: 51.1 (3.3); control group (dry gauze): 51.1 (2.8) ‐ Sex (M/F): N = 22/16 ‐ Mean (SD) INR: TXA group: 2.52 (0.16); fibrin sponge group: 2.51 (0.16); control group (dry gauze): 2.51 (0.13)
Interventions	Intervention group 1: Gauze pad soaked in 4.8% TXA was applied to the surgical alveolus for 8 min under biting pressure Intervention group 2: Fibrin sponge was packed into the surgical alveolus, and the individual was asked to bite down on the gauze for 8 min for compression Control group: Dry gauze compression was performed under biting pressure on the surgical alveolus for 8 min without the use of local haemostatic agents All groups: Surgical sites were occluded with suture in "X" fashion (silk 4.0) Only 1 tooth was extracted per appointment
Outcomes	Number of postoperative bleeding episodes requiring intervention Postoperative bleeding was defined as bleeding on the extraction site that could not be controlled by gauze compression during 20 min and/or requiring medical intervention
Notes	Intervention and control groups eligible for the review: ‐ Intervention group 1: gauze pad soaked in 4.8% TXA ‐ Control group: dry gauze compression Intervention groups not eligible for the review: ‐ Intervention group 2: fibrin sponge
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Extraction sites were considered as sampling units and, according to the local haemostatic procedure to be performed, were randomly allocated into one of the three study groups by using a computer‐generated code:…" (Patients and methods, page 210)
Allocation concealment (selection bias)	Unclear risk	There is no indication of adequate allocation concealment in the text
Blinding of participants and personnel (performance bias) All outcomes	High risk	The performance of the interventions was not blinded (Patients and methods, page 211)
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"The investigator who contacted the study participants for the purpose of evaluating bleeding complications was blinded to the haemostatic protocol allocation throughout the study period” (Patients and methods, page 212)
Incomplete outcome data (attrition bias) All outcomes	Low risk	Loss to follow‐up of three participants (N = 9 surgical procedures), however the percentage of participants that were lost to follow‐up was less than 10%
Selective reporting (reporting bias)	Low risk	All pre‐specified outcome measures are reported for all trial participants
Other bias	High risk	Participants were randomised more than once. Considered as high risk of bias, as participants were not blinded

INR: international normalised ratio
 RCT: randomised controlled trial
 SD: standard deviation
 TXA: tranexamic acid