and socioeconomic status with fetal and neonatal mortality, for each gestational age at term. The association of single determinants was examined using descriptive statistics and univariable and multivariable logistics regression analyses. The associations of multiple determinants were examined using inter-categorical analyses with the inclusion of interaction terms in the multivariable logistic regression analyses. The study showed that the probability of perinatal mortality depends on the number of determinants taken into account. Importantly, we learned that decision making about the use of induction of labour to prevent fetal and neonatal mortality based on a single determinant may lead to the overuse or underuse of induction of labour. A value based health care strategy, addressing social inequity, and investing in better screening and diagnostic methods by employing an individualised and multi-determinant approach may be more effective at preventing fetal and neonatal mortality. In chapter 6 we present the results of our nationwide cross-sectional study into variation in the content of Dutch hospital protocols used in risk selection for neonatal referral to the paediatrician. Studies indicate that there is unwarranted variation in the risk selection process used in the referral of neonates. This may be caused by variation in the protocols used for neonatal referral by local hospitals. We compared the protocols of obstetric and neonatal departments in all hospitals in the Netherlands ― between regions, between neonatal and obstetrics departments and within neonatal and obstetrics departments ― for the six most common indications for neonatal referral. We found considerable interhospital, inter-department, and intra-department variation in recommendations for type of referral, admission, screening/diagnostic tests, treatment, and discharge. Furthermore, our results generally showed lower referral thresholds in neonatal departments compared to obstetric departments, and higher referral thresholds in the eastern region of the Netherlands. Our recommendations to reduce unwarranted variation in local protocols include: developing evidence-based, multidisciplinary guidelines to support local protocols, basing agreements onmultidisciplinary consensus only when evidence is lacking or remains inconclusive, paying attention to implementation of guidelines, and describing deviation from evidence-based guidelines because of specific local circumstances in the protocol. Chapter 7 is the general discussion. It reviews the findings of our research, showing how the subjectivity of, and blind spots in, risk selection limit its effectiveness. The advancements in knowledge and technology in MNC have contributed to the lowest maternal and perinatal mortality rates in history. And yet, MNC is plagued by the continued existence of unwarranted variation in care, the underuse and overuse of medical interventions, care disparities, and disrespectful care point. The contemporary understanding of risk selection is based on the presumption that risk is objective and is centred around risks generated from within the body because (1) risk is understood predominantly in terms of pathology, (2) risk is influenced by professionals’ beliefs and interests, and (3) risk is 12 SUMMARY