Eighteen RCTs were included (n=1,239 patients). Only two RCTs were graded level I. Ten RCTs scored 2 or less out of 5 points on the Jadad score. In 13 RCTs, allocation concealment was unclear. Follow-up was adequate in all trials. The follow-up period for mortality ranged from 48 hours to 16 days.
There was insufficient evidence of the effects of general anaesthesia versus regional anaesthesia on the following: mortality (two RCTs reported no difference); cardiovascular morbidity (six RCTs reported effects of different interventions on different outcomes); or deep vein thrombosis and pulmonary embolism when thromboprophylaxis was used (two RCTs reported no difference in proximal deep vein thrombosis).
Four of 10 RCTs reported reduced blood loss in regional anaesthesia compared to general anaesthesia groups; the other six RCTs reported no significant difference.
There was no difference between general anaesthesia and regional anaesthesia for duration of surgery (12 RCTs).
Compared to systemic analgesia, regional analgesia was associated with a reduction in post-operative pain or morphine consumption (in 10 out of 11 RCTs) and nausea and vomiting (four of nine RCTs, the other five reported no significant difference).
There was no difference between regional anaesthesia, regional analgesia or both versus general anaesthesia plus systemic analgesia for length of stay in hospital (four RCTs) or rehabilitation outcomes (three RCTs).
Some results were reported in supplementary online tables (see URL for Additional Data).