Risk-based management in obstetrics

Abstract

Nowadays, medicine is increasingly focused on risk-based or personalized medicine. This largely means that treatment, preventive and other - e.g. diagnostic or prognostic - interventions are administered based on individual risks for having or developing a certain health outcome. The prevailing thought is that combinations of predictors into a so-called risk prediction or decision model allows for better risk assessments and thus patient selection than when based on individual factors. Obviously, besides personalized risk based management, prediction models can also be useful to identify specific subgroups of patients that benefit more from a certain preventive, therapeutic or other additional (e.g. diagnostic) intervention. The studies presented in this thesis aim to study and enhance risk-based management in obstetrics. In the first section of this thesis we focused on prognosis in obstetrics, specifically on the identification of risk indicators for a referral during labour and the prediction of neonatal metabolic acidosis, operative delivery, and stillbirth and neonatal death of very preterm births. The second part of this thesis focused on individual participant data meta-analyses in obstetrics aimed at quantifying treatment effects of progestogen in twin pregnancies and ST-analysis in term singleton pregnancies in cephalic presentation, in specific patient groups. The third and last section of this thesis focused on methodological issues related to prognostic research (unexpected predictor-outcome associations)and the reliability of subgroup analysis in meta-analysis. Obstetrics would benefit from the conduct and implementation of proper prognostic research and subgroup analysis to achieve more optimal patient care