Predictors of vaginal delivery in medically indicated early preterm induction of labor

When delivery is indicated prior to 34 weeks, many providers perform a cesarean delivery rather than induce labor based on perceptions of a high failure rate. Given the morbidity of cesarean delivery, an accurate estimate of the success rate and factors associated with success in preterm induction of labor is important in management decisions.

We sought to develop a prediction model for successful induction of labor in preterm patients using factors known at the time the decision is made to deliver.

A retrospective cohort study of all live singletons undergoing an indicated induction of labor between 23 and 34 0/7 weeks from 2011 through 2015. Pregnancies with major fetal anomalies or no intrapartum fetal monitoring were excluded. Successful induction of labor was defined as vaginal delivery. The cohort was randomly split into a training cohort to develop a prediction model for vaginal delivery and a validation cohort to test the model. Factors significantly associated with vaginal delivery were identified using univariate analyses, and candidate factors were used in the multivariate logistic regression model. Only factors known at the start of the induction of labor were used in the model. Receiver-operating characteristic curves were created to estimate the predictive value of the model. Sensitivity and specificity of the model were assessed.

Of 331 patients who underwent induction of labor, 208 (62.8%) delivered vaginally and 123 (37.1%) by cesarean delivery. Of the factors significantly associated with cesarean delivery, the final model included gestational age, simplified Bishop score, suspected intrauterine growth retardation, chronic hypertension, and body mass index. In the training cohort, the model correctly classified 72.3% of subjects with a sensitivity (cesarean delivery predicted/cesarean delivery performed) of 56.7% and a specificity (vaginal delivery predicted/vaginal delivery performed) of 84.1%. When applied to the validation cohort, 73.9% of subjects were correctly classified, with a sensitivity of 44.6% and specificity of 89.0%. Receiver-operating characteristic curves had an area under the curve of 0.75 for the training cohort and 0.77 for the validation cohort.

More than 60% of women undergoing induction of labor at <34 0/7 weeks deliver vaginally. For women undergoing induction of labor at <34 0/7 weeks, this prediction model rarely classifies individuals who can have a vaginal delivery as needing a cesarean delivery. This model may provide an accurate assessment tool to evaluate which patients will likely deliver vaginally to avoid the morbidity of cesarean delivery while conversely identifying subjects at high risk of cesarean delivery <34 0/7 weeks.