Dixon 2006.

Methods	Blinded, 3‐arm, RCT. Clean (class 1) surgery
Participants	778 participants, 18 years and older who were able to consent and comply with treatment, with 1801 skin lesions requiring excision. Exclusion criteria: skin contamination before surgery; surgical site not amenable to a moist, occlusive dressing; known allergy to the dressing or study agent. Partial‐thickness skin graft donor sites were not included. Participants, not wounds, were randomized to receive the intervention, but results were reported by wound. Setting: metropolitan skin cancer clinic Geelong, Australia
Interventions	Intervention: (n = 262 participants; 562 wounds) mupirocin ointment 20 mg/g Comparative intervention: (n = 269 participants; 729 wounds) paraffin ointment Control: (n = 247 participants; 510 wounds) no ointment
Outcomes	SSI rates (dichotomous, using clinical criteria); postoperative pain (self‐devised 6‐point scale, higher = greater pain); adverse outcomes/complications (dichotomous), clinically assessed until healing was complete, sutures were removed, or if a complication developed.
Notes	Definition of SSI: presence of clinical criteria for wound infection. Definition of adverse effects 1: presence of clinical criteria for complications and adverse outcomes. This included allergic contact dermatitis. Definition of adverse effects 2: postoperative pain on a 6‐point scale. Concurrent illness: None reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Using coloured discs in a barrel, quote: "Patients (not wounds) were randomized prospectively to one of 3 groups by an independent person drawing one of 150 discs (50 for each group) from a barrel; upon completing the barrel the process was repeated." Comment: this method of random sequence generation, although a little unconventional, is acceptable in principle.
Allocation concealment (selection bias)	High risk	Comment: the randomisation was performed by an 'independent person' drawing coloured discs from a barrel, but actual allocation concealment was not described. If the 'independent person' was the practice nurse, then this was not concealed from the researcher who applied the treatment.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Neither surgeon nor patient was aware of the randomisation, although patients could not be completely blinded to the application of an ointment by the nursing staff." The nurse (personnel) and participants were aware whether they applied/received an ointment or not. It was unclear whether the active component could be differentiated by colour or smell from the paraffin ointment, and no attempt was made to see if the patients were blinded successfully by asking them which group they thought they were in.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Comment: it is not clear who assessed wound outcomes. Only the surgeon was blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: complete outcome data were reported for all participants for infection and adverse effect outcomes. ITT analysis was performed.
Selective reporting (reporting bias)	Unclear risk	Comment: there were no incidences of allergic contact dermatitis. Several of the adverse event outcomes that were listed in the methods section of the study were not reported in the results, however it is likely that this is because they did not occur. Study protocol was not available to identify any unreported outcome.
Other bias	Unclear risk	Comment: wounds instead of participants were the basis for the main analysis.