Background and aims
Using data from a randomised controlled trial (RCT) comparing two policies of prostaglandin (PGE2) vaginal gel induction of labour (IOL) at term, this study aimed to determine: (i) demographic/clinical factors that predict IOL outcomes; and (ii) clinical characteristic(s) of women who would benefit from a policy of amniotomy once technically possible as opposed to giving more PGE2.

Material and methods
Following an initial PGE2 dose, women were randomised to amniotomy or repeat‐PGE2. Using RCT data, two multivariate models were developed, assessing the relationship between demographic/clinical characteristics and the outcomes of caesarean section (CS), and vaginal delivery within 24 h (VD < 24 h). Regression‐equations were used to predict the likelihood of CS and VD < 24 h, varying independent predictors from the multivariate analyses. Results Of 245 term women undergoing IOL, 90 had a CS, 155 delivered vaginally and 79 had a VD < 24 h. Controlling for confounders, nulliparity [adjusted odds ratio (aOR) = 3.71 (1.55, 8.88)] and modified Bishop's score (MBS) at first review [aOR = 0.78 (0.66, 0.92)] were independently associated with CS. Nulliparity [aOR = 0.06 (0.02, 0.15)], MBS at first review [aOR = 1.66 (1.35, 2.05)], and a policy of early amniotomy [aOR = 2.28 (1.04, 5.00)] were associated with VD < 24 h. Modelling using regression equations, and varying both MBS at first review and parity, there was no scenario where repeat PGE2 was predicted to be superior to an earlier amniotomy. Conclusions Following IOL using PGE2 vaginal gel at term, both parity and cervical favourability at first review are associated with CS and VD < 24 h. All combinations of parity and MBS at first review predicted fewer CS and greater likelihood of VD < 24 h with a policy of amniotomy once technically possible.