Abstract
Anemia in elderly patients presents a significant health concern, necessitating a thorough understanding of its multifaceted etiology and effective predictive strategies. This retrospective cohort study investigates the influential factors contributing to anemia in elderly patients and develops predictive models to aid in early detection and intervention. We analyzed demographic, clinical, and laboratory data from 864 elderly inpatients and developed a predictive model for anemia using multivariate logistic regression. Model performance was evaluated through discrimination and calibration. Notably, advanced age, elevated levels of C-reactive protein (CRP), and activated partial thromboplastin time (APTT), alongside decreased levels of albumin (ALB), calcium (Ca), and 25-Hydroxyvitamin D [25(OH)D], emerged as significant predictors of anemia. Furthermore, multiple drug resistance (MDR) was identified as a notable risk factor. Through meticulous modeling, incorporating demographic, comorbidity, and laboratory parameters, robust predictive frameworks were developed, achieving an area under the receiver operating characteristic curve (AUC) of 0.955, which indicates a high level of accuracy for risk prediction. This study underscores the complex interplay of factors contributing to anemia in elderly patients and provides a practical approach for its early identification and management.