The role of selection bias in comparing cesarean birth rates between physician and midwifery management

Abstract

Objective: The midwifery service at our hospital has been observed to have a 2% cesarean birth rate consistently over a 10-year period. There are substantial differences in labor management style between the midwives and physicians. We sought to test the hypothesis that the low cesarean birth rate on the midwifery service was the result of patient selection bias.

Methods: A randomized blinded clinical trial was conducted in which 492 low-risk patients were assigned to either physician or midwifery management. The provider responsible for labor management was unable to determine group assignment. Patients in the midwifery group were managed by previously established protocols, and outcome was attributed to the midwives even if the patients subsequently required transfer to physician management. Route of delivery was the primary outcome measurement. Continuous variables were analyzed using Student t test and discrete variables using chi 2.

Results: There were no demographic differences between the groups, and the admission pelvic examinations were the same. The patients assigned to the midwifery group had a 2.1% cesarean birth rate, whereas those assigned to physician management had a 0.4% rate. The higher rate of operative vaginal deliveries in the physician group was statistically significant. There were no differences in neonatal outcomes. The physician-managed group had significantly more episiotomies and third- and fourth-degree extensions.

Conclusions: The 2% cesarean birth rate observed on the midwifery service appeared to be the result of patient selection bias. A low cesarean birth rate can be achieved by either physician or midwifery management in a selected low-risk population.