Nine RCTs were included (505 participants). Studies generally had adequate randomisation methods but only three studies had low risk of bias for blinding of participants, study personnel, and outcome assessors. All studies scored between 11 and 13 out of 15, except for one study which scored 6.
There were no statistically significant differences between treatments in the pain scores at rest after 24 hours (eight RCTs) or 48 hours (seven RCTs) after surgery, nor for pain on movement at 24 hours (seven RCTs) or 48 hours (six RCTs) after surgery; all analyses were subject to statistically significant heterogeneity.
A non-significant trend towards reduced opiate requirements was found favouring use of epidural analgesia, though significant heterogeneity was evident (WMD 10.0mg, 95% CI -1.0 to 21.1; five RCTs; Ι²=87%). There were no statistically significant differences between groups for treatment failure (six RCTs) nausea and vomiting (eight RCTs), and for local catheter complications and wound infections (seven RCTs). There was a significantly reduced incidence of urinary retention in the wound catheter group (OR 0.14, 95% CI 0.04 to 0.47, three RCTs; Ι²=53%), though significant heterogeneity was present.
Subgroup analyses showed no statistically significant differences in effect estimates according to incision type or protocol of administration (continuous versus bolus).