PhD student: Emmanuel Rault
Title: Vaginal birth in France: evolution, characteristics of women, medical interventions and prediction of the success of the vaginal birth attempt
Supervisors: Camille Le Ray, Aude Girault
Doctoral school: Paris-Cité
Promotion: 2023
Thesis abstract:
Since the 17th century, childbirth has undergone a gradual medicalization that has transformed the way it is approached. This medicalization has significantly improved maternal and neonatal health. Nevertheless, as the current demand from women favors a childbirth experience described as “physiological,” healthcare providers aim to accompany the process with a primary principle of “first, do no harm”. However, in France, the rate of physiological childbirth, meaning childbirth without any medical interventions, is not known, and there is no classification system to describe the contribution of different groups of women to vaginal deliveries. Developing such a classification would enable the assessment of the distribution of these populations over time through national perinatal surveys and the study of their determinants.
In the face of the potential iatrogenic effects of medical interventions and the strong societal demand for the demedicalization of childbirth, any medical intervention during labor should be based on a formal medical indication and shared decision-making. Therefore, studying the evolution of medical interventions during labor holds particular importance. Previous studies using data from the National Perinatal Survey (NPS) have revealed a reduction in the use of oxytocin during labor, decreasing from 58.3% in 2010 to 42.6% in 2016. The NPS in 2021 has shown a continued reduction in these interventions, with an oxytocin utilization rate of 30.0% and artificial rupture of membranes (ARM) in 33.2% of cases. The determinants of oxytocin and ARM utilization in 2021 have not yet been studied, and these data need to be compared to previous NPS data.
Supporting women also requires the ability to assess the risk of labor in order to provide clear information for informed decision-making. Multiple factors are at play, making this evaluation complex. Predictive models based on conventional statistical tools have been developed to refine individual risk by considering maternal and pregnancy characteristics. The development of “machine learning” in the field of healthcare now offers new tools to refine and individualize patient care. The NPS data could be utilized to develop a predictive model for the success of vaginal delivery upon admission to the delivery room based on machine learning models.
