119 5 MATERNAL CHARACTERISTICS AS INDICATIONS FOR ROUTINE IOL | 11 GOODARZI et al. The transition to VBHC would benefit from investing in the prevention of known causes of TM. For example, the majority of TM at term is associated with fetal growth restriction.2,37,38 Current screening and diagnostic methods fail to accurately differentiate between fetuses and newborns with unreached growth potential and those constitutionally small but healthy.38 More importantly, current methods do not address structural social causes of growth restriction, which have been identified as the primary drivers of fetal growth restriction.37 Primary prevention by addressing social drivers of TM is the most sustainable approach toward reducing PM. Consistent with previous studies, our study showed an association between TM and non-Dutch ethnicity and low SES.12,13 This finding may convey the impression that non-Dutch and low-SES women are more likely to benefit from routine IOL. However, studies show that the associations between ethnicity and SES with TM is based on social drivers. Conditions in which people are born, grow, live, work, and age, shape health in powerful ways.39,40 It has been argued that ethnicity and SES act as proxies for complex societal processes,41 and that the association with TM is mediated by discrimination and inequity.29 (p247),42 Thus, the extent to which medical interventions, such as IOL, can intervene upon these social processes is limited.40 Nevertheless, often, social determinants of health are used as biological or genetic determinants, and differences between subgroups are medicalized. Consequently, efforts to reduce TM predominantly focus on improving and using medical interventions instead of addressing underlying societal processes. A recent example is the revised National Institute for Health and Care Excellence IOL guideline's recommendation to induce labor at 39 weeks of women with otherwise uncomplicated pregnancies and a black, Asian and minority ethnic background, because they are two to four times as likely to die during pregnancy and birth.43 This strategy adds the additional risks of IOL without addressing root causes of the adverse outcomes.44 A value-based approach is primary prevention of TM by systematically addressing institutional discrimination and inequity in society.33,41 Further research into the underlying causes of the differences in TM between the ethnic and SES categories is necessary.34 Like previous studies, our study indicates that a multi- determinant approach contributes to a better identification of women that would benefit from IOL.28,29 A recent review identified over 60 determinants of stillbirth, including maternal and medical characteristics and biomarkers.29 A multi-determinant approach can offer the possibility to study the dynamic interaction between mutually constituting social and biomedical drivers doing more justice to the complexity of pregnancy and birth.33,41 This approach includes considering multiple determinants and measuring effect modification.20,28,29,41 4.4 | Strengths and limitations To our knowledge, this study is the first to examine the association between multiple maternal characteristics and TM in different gestational weeks at term in a healthy population. The data set used in this study is unique in size and population, which provided us with the power necessary to study TM in a healthy population. However, the use of registration data also had disadvantages. First, despite the size of the database, in some subgroups, the incidence of TM was very low or zero, resulting in fluctuating outcomes. Also, they were not able to specify the analyses for all subcategories. This would be possible by combining data sets internationally, requiring high-quality data registration and collection, and comparable registration systems. Second, we were unable to include other relevant maternal characteristics associated with TM because they are not registered routinely, such as BMI and cigarette smoking, which are also associated with ethnicity and SES.2,28,29 This may have impacted the outcomes of this study. Third, we were unable to study TM in different ethnic categories because ethnicity is registered imprecisely and inconsistently in Perined. The Perined categories for ethnicity consisted of countries, continents, and racial groups. Therefore, we used a binary construction of ethnicity. This may have conveyed the impression that these groups are homogeneous. However, ethnic categories are diverse and dynamic because they are socially constructed.45 For example, in Perined, ethnicity is assigned by women's care provider, usually based on appearance, name, and/or information provided by women. However, ethnicity depends on an individual's and their family's country of birth, migration history, genealogy,46 and whether ethnicity is self-assigned or socially assigned.47 This makes the categorization of individuals into ethnic categories difficult.48Furthermore, the definition of ethnic categories changes over time and differs between settings.45 Like ethnicity, SES categories are also socially constructed.49 Caution should be taken when using ethnic or SES categories in research and practice as biogenetic determinants of health outcomes.34,50 To be able to use ethnicity and SES in a more meaningful way as a social determinant in future research, underlying categories should be registered.34 Fourth, because the moment of fetal and neonatal death was not registered, we used moment of birth as inclusion criterion, assuming fetal death and birth occurred, at most, a few days apart. Last, this study included data