Induction of labour

We thank Hehir and colleagues1 for their interest in our systematic review.2 In response to their comment that not all inductions are equal and that the rate of cesarean delivery following induction compared with expectant management would be expected to be different for different indications for induction, baseline characteristics, methods of induction, parity and institutional cesarean delivery rates, we would like to draw their attention to the extensive subgroup analyses reported in Table 3. The relative risks for cesarean delivery depend on a wide variety of characteristics, including method of induction, indication for induction, gestational age, definition of induction, cervical status, pregnancy risk and parity. For some of these characteristics, we have precise evidence, because a relatively large number of trials reported results by the characteristics, but for others we do not have sufficient information. The results of meta-regression exploring the impact of factors such as patient’s characteristics, induction methods and definition of induction are provided in Appendix 6, available at: www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.130925/-/DC1. Unfortunately, the included trials did not report the institutional or prelabour cesarean delivery rates, so we were unable to evaluate these.

We also thank Slome Cohain3 for her interest in our systematic review. There were 2568 cesarean deliveries in the expectant management groups from a total of 15 119 participants in the 157 trials, and the rates varied from 0% to 60%, with a mean of 19.4% and a standard deviation of 13.03%. There were 2384 cesarean deliveries in the induction groups from a total of 15 966 participants, and the rates varied from 0% to 50%, with a mean of 16.5% and a standard deviation of 11.05%.

We agree that the most serious outcome of cesarean delivery is death, which is why we specifically looked for it as one of the outcomes. This analysis was limited by the data published. The 20 trials that reported maternal deaths evaluated results for 4689 women, 2387 in the induction group and 2302 in the expectant management group. There was one death in each group, giving a crude overall mortality rate of 1 in 2345. The most we can say is that there was no difference in the maternal mortality rates in the two groups, and the rates were very low. An extremely large trial or meta-analysis of individual patient data from current trials (obtaining data on maternal mortality where it is not published) could aim to address the question of whether induction is associated with mortality.

Evidence-based medicine encourages a combination of the current best evidence with clinical acumen and the preferences of the patient. Ignoring current evidence would deprive women of the knowledge needed in decision-making concerning labour induction.