Clinical data banking in obstetrics—advantages and disadvantages of the strategy using the example of preterm birth risk modeling in multiple pregnancies

BACKGROUND: Despite technological advancements, the key resource for predicting obstetric complications remains the collection and analysis of clinical data. Risk stratification is crucial in obstetrics, enabling tailored antenatal care and preventive measures to reduce preterm birth rates. This is particularly important in multiple pregnancies, where preterm birth occurs in 40%–60% of cases, which causes a high risk of developing organic and functional disorders, leading to long-term disabilities and social challenges for children. AIM: The aim of this study was to develop preterm birth risk prediction models for multiple pregnancies using clinical data and to evaluate the advantages and disadvantages of data banking. METHODS: This retrospective single-center case-control study was conducted using a registry of 630 dichorionic twin deliveries (RU2024621911 as of May 3, 2024). All cases were characterized by 212 clinical parameters, with a new approach to identifying promising areas for collecting biological samples of twins being developed. RESULTS: The study comprised spontaneous preterm deliveries (main group, n = 204) and term deliveries (control group, n = 323). Multifactorial modeling of preterm birth risk in dichorionic twin pregnancy showed that very early preterm birth (31 weeks) is associated with type 1 diabetes mellitus and cervical insufficiency (good predictive power). Early preterm birth (31–33 weeks) is associated with type 2 diabetes mellitus, prior induced abortion, chronic pyelonephritis, and cervical insufficiency (good predictive power). Late preterm birth (33 weeks) is associated with IVF conception, cholestatic hepatosis, and cervical insufficiency (moderate predictive power). CONCLUSION: Clinical data registries are valuable for risk modeling, but standalone predictive models may have limitations. Integrating biobanks that combine clinical data with biological samples could enhance prediction accuracy and advance obstetric care.