Articles Included	RCTs (n = 3)
Study 1: [26]
Methods Participants Interventions Comparator Outcomes Comments Risk of Bias Selection bias (low risk) Blinding performance and detection bias (low risk) Baseline comparability of treatment and control (low risk) Conflict of Interest (unclear risk)	Randomised Controlled Trials All patients undergoing clean surgery – 660 patients. Inclusion criteria: not stated. Exclusion criteria: existing airborne infection where face mask was specifically required. Participants had baseline similarity for age, gender, weight, prophylactic antibiotics and ASA classification. Group 1: Mask (n = 313) Group 2: No mask (n = 340) National Nosocomial Infection Surveillance System definition of wound infection used: superficial, deep and organ space. Group 1: Mean follow up 33.4 days (SD 22.1) Group 2: Mean follow up 33.4 days (SD 22.8) Missing data for 7 clean cases created a unit of analysis error. Only non-scrubbed staff was asked to comply with randomisation. Scrubbed staff was not included in the trial. Computer-generated randomisation of operation lists into two groups, mask and no mask and allocation was by phone call immediately before the session commenced, to a person who was unaware of the type of list in the theatre reducing the risk of selection bias. Patients unaware of allocation. Data of any post-operative wound infection was obtained by routine surveillance of infection control staff that was unaware of the treatment protocol. The two groups were comparable for baseline characteristics of type of surgery, wound and ASA classification as well as age, gender, pro-operative hospitalisation, BMI and prophylactic antibiotics. Source of funding not stated.
Article 2: [17]
Methods Participants Interventions Comparator Outcomes Comments Risk of Bias Random sequence/Allocation concealment (selection bias high-risk) Blinding (performance and detection bias) Baseline comparability of treatment	Randomised Controlled Trials. 1429 clean surgery performed in the operating room. Inclusion criteria: surgical incision through intact skin with primary wound closure. Exclusion criteria: patients not informed or consent not given for all category of surgery. Baseline comparability: similar for age, acute and cold surgery. Group 1: Mask (n = 706). Group 2: No mask (n = 723). Wound infection defined by visible pus and/or cellulitis requiring debridement, drainage with antibiotic treatment. Follow up was until after discharge but no post discharge findings stated. Patients had 2 to 3 body washes pre-operatively with 4% Chlorhexidine before elective surgery, at least 1 body wash in most acute cases, creating a unit of analysis error. A random list was set up for 1 year, denoting weeks as “masked” or “unmasked”. The list was inversed for the 2^nd and 3^rd part of the year to avoid seasonal differences between the groups, making selection bias a high risk. Concealment inadequate as investigators enrolling participants could possibly foresee allocation and thus introduce selection bias. Patient blinding not described (unclear bias). Notification of the trial was issued with each wound swab (high risk). Baseline comparability stated for age and type of surgery, wound, ASA classification, preoperative hospitalisation, BMI and prophylactic antibiotics.
Study 3: [3]
Methods Participants Interventions Comparator Outcome Comments Risk of Bias	Randomised controlled trials 2113 patients undergoing clean surgery in the operating room. Inclusion criteria: clean surgery in the operating room. Exclusion criteria: unclean surgery not done in operating room. Group 1: Mask (n = 1036). Group 2: No mask (n = 1077). Wound infection without standard follow up time of included studies. Limited by size and number of included studies. Publication bias could not be assessed due to the small number of studies included. Heterogeneity of included studies prevented pooling of data and meta-analysis. Risk of bias assessed. Stated no conflict of interest.