Cesarean delivery after non-medically indicated induction of labor: A population-based study using different definitions of expectant management.

Abstract

Most observational studies found that non-medically indicated induction of labor (IOL) is not associated with an increased risk of cesarean delivery compared with expectant management, defined as all births at a later gestation. However, given the higher rate of cesarean delivery at late term, this definition of the expectant management group might bias the results of observational studies in favor of IOL at early or full term when estimating the risk of short-term (eg up to 1week) expectant management. We conducted a retrospective cohort study including 447066 singleton term and post-term hospital births that occurred in Austria between 2008 and 2016. Multivariate logistic regression was used to test the association of IOL and cesarean delivery at each week of gestation from 37-41. Expectant management was either defined as all births at "next week or beyond" or "at next week". Non-medically indicated IOL was associated with increased odds for cesarean delivery at 37 and 38weeks, and reduced odds at 40 and 41weeks. At 39weeks, IOL resulted in comparable cesarean rates compared with expectant management defined as "next week or beyond" (17.2% vs 16.2%; adjusted odds ratio [OR] 0.93; 95% confidence interval [CI] 0.86-1.00; P=.059). However, when defined as births "at the next week", expectant management was associated with significantly reduced odds for cesarean delivery (13.6%; adjusted OR 0.76; 95% CI 0.70-0.82; P<.001). Comparison of the cesarean delivery rates for the two definitions of expectant management showed that the "next week and beyond" model underestimates the benefit of short-term expectant management by up to 1week, particularly for IOL at weeks 38 and 39. Our findings demonstrate that the definition of the expectant management group has a significant impact when analyzing the outcome of IOL in retrospective cohort studies. Non-medically indicated IOL is not an all-or-none choice between "elective" induction and indefinite expectant management. Thus, to define the control group as all births at the next week could be useful for clinical decision-making, as it allows to estimate the risks of expectant management until the next appointment compared with immediate IOL.