Forty-eight RCTs with more than 9,800 women were included.
The methodological flaws of the trials included post-randomisation exclusions and underpowered trials. There was a lack of consistency in the methods used to report the outcomes, making synthesis difficult. Intramuscular pethidine versus placebo (1 double-blind RCT involving 224 women).
Women were significantly more dissatisfied with placebo, compared with pethidine, both during labour (83% versus 71%; p=0.04) and after labour (54% versus 25%; p=0.00004).
Intramuscular opioid versus an alternative intramuscular opioid (20 RCTs involving 3,188 women, of which 17 were double-blind RCTs).
Maternal pain relief was similar among the different opioids. Meptazinol was the most common alternative. There was no convincing evidence that the alternatives were better than pethidine. Dissatisfaction with pain relief varied widely among studies (27 to 87%), as did maternal side-effects. Few trials assessed neonatal outcomes, and those that did reported no differences in substantive neonatal outcomes.
Intramuscular opioid versus an alternative dose of the same intramuscular opioid (3 double-blind RCTs involving 273 women).
Higher doses of tramadol (100 versus 50 mg) and pethidine (80 versus 40 mg) were associated with improved pain relief. There was no difference between 1 and 2 mg butorphanol. The side-effects were also similar. The data on neonatal outcomes were lacking.
Intravenous versus intramuscular pethidine (1 unblinded RCT involving 39 women). Women in the intravenous group received significantly more pethidine and reported significantly lower labour pain than the intramuscular group in this small trial. Neonatal outcomes were reported to be similar between the treatment groups, but seven women who had Caesarean sections were excluded from this analysis.
Intravenous versus a different intramuscular opioid (8 RCTs involving 772 women).
Five different opioids were compared with pethidine, resulting in limited data for each comparison. There was no convincing evidence of any differences between the different opioids.
Intravenous versus a different mode of administration of the same opioid (2 RCTs involving 141 women).
The data were limited and neonatal outcomes were not reported. There were no differences in the study outcomes between patient-controlled opioids and intermittent bolus for either pethidine or fentanyl.
Parenteral opioids versus epidural analgesia (11 RCTs involving 3,320 women).
All of the five RCTs that measured pain relief reported improved pain relief and satisfaction with pain relief for epidural, compared with intravenous butorphanol, pethidine and fentanyl. There was significant heterogeneity (p<0.0014) among the studies for the duration of first and second stages of labour and rates of oxytocin augmentation. Labour-related outcomes favoured opioids. The neonatal outcomes were similar between the treatment groups.
Intramuscular pethidine versus paracervical block (1 double-blind RCT involving 117 women).
There was a significant difference in pain relief favouring paracervical block up to 60 minutes, but not thereafter.
Parenteral opioids and co-drugs versus parenteral opioids. Benzodiazepine as a co-drug (2 RCTs involving 432 women): one RCT provided data on nausea and vomiting. The other RCT reported significantly better pain relief with intramuscular pethidine plus oral lorazepam, compared with pethidine alone (weighted mean difference -22.00, 95% CI: -37.5, -6.50), and decreased dissatisfaction with labour pain relief (0% versus 45%; p=0.005). Sedation was significantly increased in the co-drug group (45% versus 10%; p=0.04), while nausea and vomiting were similar.
Tranquillisers and anti-emetics (phenothiazines) as a co-drug (2 double-blind RCTs involving 992 women): the results from studies examining promethazine were conflicting.