Beaussier 2005.

Methods	Randomized, double‐blind, double‐dummy, active‐controlled Medications administered at the beginning of wound closure
Participants	Type of surgery: inguinal hernia repair Propacetamol group Entered/completing: 90/90 Age (mean, SD): 46 ± 14 Sex (male, %): 94 Parecoxib group Entered/completing: 92/90 Age (mean, SD): 42 ± 14 Sex (male, %): 91
Interventions	Intervention: 2 g propacetamol over 15 min Control: parecoxib 40 mg IV
Outcomes	Primary: opioid consumption (morphine) Pain intensity at rest and while coughing (VRS, VAS) and derived summary measures
Source of funding	Pfizer SA
Were treatment groups comparable at baseline?	Yes: intraoperative opioid consumption and time to tracheal extubation did not differ between groups; demographics (age, sex, weight, height) similar
Details of preoperative pain	Patients with chronic pain excluded
Notes	—
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomization performed centrally as blocks of 4 and with a 2:2 treatment ratio. In each center, treatment allocation was performed on the basis of one complete treatment block.
Allocation concealment (selection bias)	Low risk	Central allocation
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐dummy design employed. Propacetamol was administered by slow infusion over 15 min whereas parecoxib was injected by rapid bolus. Each participant received both an active product and the placebo of the other product.
Incomplete outcome data (attrition bias) All outcomes	Low risk	For primary outcome, data reported for both modified‐ITT population and per‐protocol population. Secondary outcomes reported data for ITT population.
Selective reporting (reporting bias)	Low risk	All outcomes reported with mean data or raw numbers
Size	Unclear risk	50 to 199 participants per arm of the study (90 propacetamol, 92 parecoxib)