Ten RCTs were included (680 women).
Different methods of allocation were used including sealed envelopes (5 RCTs), computer-generated lists (2 RCTs) and unspecified (3 RCTs). Six studies used an ITT analysis. None of the studies were blinded.
Nifedipine versus ritodrine.
1. Delivery within 48 hours (7 RCTs).
Homogeneity could not be rejected (p=0.22). Nifedipine reduced the risk of delivery within 48 hours, although not significantly (OR 0.85, 95% CI: 0.69, 1.0). Risk difference was -2.7% (95% CI: -12, 6).
2. Postponement of delivery to at least 36 weeks gestation (7 RCTs). Homogeneity could not be rejected (p=0.86). Nifedipine significantly reduced the risk of delivery before 36 weeks (OR 0.72, 95% CI: 0.55, 0.91). Risk difference was -11% (95% CI: B20, -1.5). There appears to be a misprint for the lower 95% CI (could be - rather than B as printed).
3. Perinatal mortality (6 RCTs).
Homogeneity could not be rejected (p=0.19). There was no significant difference between nifedipine and ritodrine (OR 1.2, 95% CI: 0.69, 2.1). Risk difference was 0.7% (95% CI: -2.8, 4.3).
4. RDS (3 RCTs).
Homogeneity could not be rejected (p=0.09). Nifedipine significantly reduced the risk of RDS (OR 0.72, 95% CI: 0.54, 0.96). Risk difference was -10% (95% CI: -20, -0.2).
5. Admission to NICU (2 RCTs).
Homogeneity could not be rejected (p=0.42). Nifedipine significantly reduced the risk of admission (OR 0.60, 95% CI: 0.42, 0.86).
6. Maternal side-effects (5 RCTs).
Meta-analysis was not possible due to inconsistencies in definitions of side-effects between studies. The percentage of patients experiencing side-effects whilst on nifedipine was 16% (23 out of 147), compared with 45% (73 out of 132) of women using ritodrine. There appears to be a misprint for the latter: based on the patient numbers provided, the proportion of patients experiencing side- effects whilst on ritodrine should be 55%.