Sixty-five RCTs (3,338 patients) were included in the review. Sample sizes ranged from 20 to 188 patients. Quality scores ranged from 1 to 7 points out of 7, with a median quality score of 3 points. Double-blinding was reported in 48 trials. Randomisation procedures were described in 24 trials. Ten trials reported concealment of treatment allocation.
Efficacy
There were significant benefits observed with morphine added to intrathecally administered bupivacaine for the reduction in cumulative 24-hour morphine consumption (WMD -12 mg, 95% CI -18 to -5; 0.05 to 1mg morphine; seven trials), reduction in the proportion of patients requiring postoperative opioids (OR 0.05, 95% CI 0.02 to 0.14; six trials; NNT 2.9, 95% CI 1.5 to 25), and an increased duration of postoperative analgesia (WMD 503 minutes, 95% CI 315 to 641 minutes; 0.05 to 2mg of morphine; 13 trials). There were no differences between the intervention and control groups in postoperative pain intensity at 24 hours post-surgery (six trials). Statistically significant heterogeneity was reported for all outcomes. There was no evidence of a dose-response relationship for these outcomes.
The use of fentanyl (0.01 to 0.05mg) added to intrathecally-administered bupivacaine was associated with increased duration of analgesia (WMD 114 minutes, 95% CI 60 to 16; eight trials, with statistical heterogeneity present), but there was insufficient data on the effect of fentanyl on 24-hour morphine consumption, post-operative pain intensity, and the numbers of patients requiring administration of post-operative opioids.
Safety
The use of morphine with intrathecally-administered bupivacaine was associated with higher rates of pruritus (OR 6.92, 95% CI 4.51 to 10.60;17 trials; NNH 4), nausea (OR 1.66, 95% CI 1.05 to 2.64; 10 trials; NNH 9.8), vomiting (OR 1.88, 95% CI 1.20 to 2.9;,13 trials; NNH 10), and urinary retention (OR 3.90, 95% CI 1.94 to 7.86; seven trials; NNH 6.5). The risk of respiratory depression was not statistically significantly different between groups. There was no statistically significant heterogeneity observed for these outcomes.
Fentanyl administered with intrathecal bupivacaine was associated with a statistically significant increase in the risk of pruritus (OR 10.8, 95% CI 7.09 to 16.5; 13 trials; NNH 3.3), but there were no significant differences observed with fentanyl on nausea, urinary retention or respiratory depression.
Data on other opioids buprenorphine, diamorphine, hydromorphone, meperidine, methadone, pentazocine, sufentanil, and tramadol were presented in fewer than five trials and were not analysed further in the review.