Hospital-Level Variation in the Frequency of Cesarean Delivery Among Nulliparous Women Who Undergo Labor Induction

Obstet Gynecol. 2020 Dec;136(6):1179-1189. doi: 10.1097/AOG.0000000000004139.

Abstract

Objective: To evaluate the use of administrative data for identification of labor induction and to estimate the variation in cesarean delivery rates among low-risk women who underwent labor induction.

Methods: A cross-sectional study was performed examining live births in California hospitals during 2016 and 2017 using birth certificate data linked with maternal patient discharge records. Initially, eight hospitals performed medical record reviews by using reVITALize definitions on 46,916 deliveries to assess the validity of induction identification by birth certificate or discharge diagnosis records or both. Hospital-level variation in cesarean delivery rates was then assessed among all California hospitals for women with low-obstetric-risk first births before and after further risk adjustment and after the exclusion of potential medical and obstetric indications for induction. Variation in physician-level cesarean delivery rates after induction at four large hospitals also was examined. The relationships between cesarean delivery rates among women with induced labors compared with noninduced labors and with the hospital rate of induction also were explored.

Results: Identifying induction by a combination of discharge diagnosis codes and birth certificate data had the highest accuracy (92.9%, 95% CI 92.7-93.2). Among 917,225 births at 238 birthing hospitals, there were 99,441 nulliparous women with term, singleton, vertex pregnancies who were induced. The median cesarean delivery rate after labor induction for nulliparous women with term, singleton, vertex pregnancies was 32.2%, with a range of 18.5-84.6%. This wide variation was not reduced after risk adjustment or after exclusion of all women with induction indications. A similar wide variation was noted within geographic regions, neonatal intensive care levels, and among individual physicians in the same facility. Only very weak associations were found for the cesarean delivery rate after labor induction and either the rate after noninduced labor (R<0.08) or the rate of nulliparous labor induction (R<0.12).

Conclusion: The large variation of cesarean delivery rates after induction of labor suggests that clinical management plays an important role in achieving induction success.