Decision-Making System for Extended Bacteremia Treatment in Patients with Hematologic Malignancy

Patients with hematologic malignancy (HM) experiencing bloodstream infections by Acinetobacter baumannii (AB) encounter considerable mortality risks, despite 14 days of standard antibiotic therapy. This research addresses a critical clinical challenge by developing a decision-making system to identify patients with hematologic malignancy and Acinetobacter baumannii bloodstream infections who would benefit from extended antibiotic therapy. Retrospective cohort research was conducted on patients with hematologic malignancies (HM) who developed bloodstream infections and were treated with a 14-day course of antibiotics within the specified period from January 2019 to April 2024. The odds ratio (OR) and risk ratio (RR) were calculated to examine the relationships among clinical and demographic data. A multivariable logistic regression model has been applied and adjusted to account for various predictors. The predictive model and the “Ex-CSEPA” decision-making system were developed using logistic regression. The performance metrics, including sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristic curve (AUC-ROC), were evaluated. The developed model demonstrated exceptional performance, achieving an accuracy of 98.7%. It exhibited a sensitivity of 99.07% and a specificity of 98.33% in predicting mortality, setting a cut-off point of 0.5 or higher as indicative of high risk for mortality after 14-day treatment. The system's ability to identify patients who would benefit from antibiotic (ATB) treatment beyond the standard 14-day period was particularly significant. The application of this predictive model in clinical practice has pushed up the potential to enhance decisions for extended-ATB duration and decrease 30-day mortality for patients with HM who are morbid with AB bloodstream infections.