Joyce Molenaar

67 Predicting population-level vulnerability among pregnant women and harm (22, 33, 35). Given the added value of self-reported data however, we suggest to explore how perceived health can be systematically included into screening guidelines and care registries for professionals, to enhance the provision of personalized care and support while further improving population-level predictions in the future. In our study, adding self-reported data led to better model performance and self-reported health indicators were found as important predictors to multidimensional vulnerability. Consistent with the psychosocial literature, several subjective measures (e.g. self-reported ‘insufficient financial resources’) outweighed objective measures (e.g. registered ‘income’ and ‘debts and payment arrears’) as predictors in our study. For example, multiple studies reported a stronger link between people’s subjective SES and wellbeing and physical health compared to objective SES based on income or education (36-39). Arguably, perceiving your circumstances through the lens of limited resources impacts decisionmaking and behaviour (e.g. favouring short-term over long-term considerations), increases uncertainties and stress, and thus exacerbates pre-existing vulnerabilities (40-43). Other studies reported how self-reported health or vulnerability correspond to, outperform or complement clinical measures in predicting physical health and mortality (18-20). However, using self-reported health also has its challenges. For instance, it provides little guidance regarding what respondents consider when reporting ‘poor health’ and whether they refer to physical pain, mental wellbeing, less vitality or other factors (21). Additionally, people can have diverse perceptions of their health influenced by cultural contexts, social positions, and personal health experiences (e.g. people suffering from the same illness for a longer time may report better levels of health due to various coping and self-management strategies) (22, 44). Nevertheless, self-reported health seems to be an important measure which can capture components of health or vulnerability that other measures alone cannot. Strengths, limitations and future research The availability of nationwide data on a wide range of risk and protective factors to vulnerability in many different domains was an important strength of this study. The outcome ‘multidimensional vulnerability’ was also based on 42 variables (12). Additionally, we conducted several sensitivity analyses, all of which yielded similar results, underscoring the robustness of our model. However, this study also had several limitations, mostly related to the data. One limitation concerns the representativeness of the dataset used to construct and evaluate the predictive models. It is possible that some factors (e.g. asylum seeker status) did not emerge as primary predictors because they were less present among the 4172 women, despite their association with vulnerability and adverse outcomes in the literature (45, 46). This may have also led to a slight underestimation of the actual percentage of multidimensional vulnerability. Additionally, we missed data on important topics that can contribute to vulnerability such as stress, health literacy, coping skills, and adverse (childhood) experiences including violence. Another limitation is that we insufficiently considered the dynamics around pregnancy in relation to vulnerability, since we merely incorporated data prior to pregnancy that can be subject to change. Future research should take into account that vulnerability can exist prior to pregnancy, but also 3

RkJQdWJsaXNoZXIy MTk4NDMw