Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomly assigned (from Geigy tables) and stratified for age, sex and type of operation
Allocation concealment (selection bias)	Unclear risk	Unclear; not mentioned
Blinding of participants and personnel (performance bias)	High risk	Not blinded
Blinding of outcome assessment (detection bias)	High risk	Not blinded
Incomplete outcome data (attrition bias)	High risk	47 dropouts not kept for the analysis: nine in the PCA group and 38 in the epidural groups (operation changed or not done, participant chose other anaesthetic method, epidural impossible or contraindicated)
Selective reporting (reporting bias)	High risk	Not in intention‐to‐treat. Complications of epidurals not mentioned
Other bias	Low risk	Groups well balanced