Postpartum anemia is a global health issue linked to maternal mortality and sepsis. While iron deficiency plays a role, its contribution is limited, highlighting the need for better biomarkers. Although some studies suggest potential links between certain B vitamins and anemia, their relationship with immediate postpartum anemia (IPPA) remains unclear. We aimed to explore the longitudinal trajectory patterns of B vitamins (B1, B2, B3, B5, B6, and B9) during pregnancy and their potential associations with IPPA, and to develop a preliminary predictive model based on identified factors. We recruited 200 women in second trimester, measuring B vitamin levels longitudinally through pregnancy and postpartum. IPPA was defined as hemoglobin < 11 g/dL within 24 h postpartum. Latent class trajectory models (LCTM) identified B vitamin trajectories, and logistic regression analyzed associations with IPPA. A predictive model for IPPA risk was developed using the Kolmogorov-Arnold network (KAN), with performance evaluated via the area under the curve (AUC), the net reclassification index (NRI), integrated discrimination improvement (IDI), and decision curve analysis (DCA). IPPA was diagnosed in 52.5% of participants. Unique trajectories of vitamins B1 (OR 0.25, 95% CI 0.10–0.59) and B3 (OR 0.21, 95% CI 0.08–0.57) were associated with reduced IPPA risk. Single-time-point analyses indicated that second-trimester levels of B1 (OR 0.82, 95% CI 0.67–0.98) and B3 (OR 0.84, 95% CI 0.73–0.97) may be associated with lower IPPA risk. The predictive model, integrating second-trimester levels of B1, B3, and iron, achieved an AUC of 0.78 (95% CI 0.76–0.79). The KAN-based predictive model showed an NRI of 0.32 (95% CI 0.27–0.37) and an IDI of 0.08 (95% CI 0.07–0.09), indicating superior predictive performance compared to the base model. Our exploratory findings suggest that B vitamins, particularly the previously unrecognized vitamin B3, may play a protective role against IPPA, with the second trimester emerging as a potential critical window for monitoring. The developed predictive model shows promise as a screening tool. These results require validation in larger prospective studies but point to modifiable nutritional factors that could inform future prevention strategies.
Wu et al. (Thu,) studied this question.