Predictors of mortality in sepsis patients with Acinetobacter baumannii and Klebsiella pneumoniae bacteremia

Background and aim

Sepsis with persistent bacteremia caused by Acinetobacter baumannii (A. baumannii) and Klebsiella pneumoniae (K. pneumoniae) poses a significant mortality risk in intensive care units (ICUs). The role of follow-up blood cultures (FUBCs) in predicting outcomes remains debated. This study investigates the key predictors of mortality and the importance of microbiological non-clearance (MNC) through FUBCs among these predictors in bacteremia due to K. pneumoniae and A. baumannii infections.

Materials and methods

We conducted a single-center, retrospective study at a tertiary teaching hospital in India involving 218 ICU patients with K. pneumoniae and A. baumannii bacteremia from October 2019 to December 2021. Blood cultures were analyzed using the BACT/ALERT VIRTUO system. Data were analyzed using logistic regression, receiver operating characteristic (ROC) curves, and an artificial neural network (ANN) model to determine the normalized importance of key predictors of mortality. Factors with a more than 50 % normalized importance were considered significant contributors to mortality.

Results

The overall mortality rate was 84 %, with 91 % in patients co-infected with K. pneumoniae and A. baumannii. Microbiological clearance (MC) was associated with a lower mortality rate (56.5 %) compared to MNC (97 %) or no FUBCs (84 %). Multivariable logistic regression identified the Charlson Comorbidity Index (CCI) score and MNC as independent predictors of mortality, with a predictive accuracy of 91.6 %. The ANN model confirmed the normalized importance of MNC as a key predictor, followed by CCI score, Acute Physiology and Chronic Health Evaluation II (APACHE II) scores at ICU admission, Sequential Organ Failure Assessment (SOFA) scores at ICU admission, age, SOFA score on culture-positive day, Pitts bacteremia score, and renal replacement therapy (RRT).

Conclusion

Persistent bacteremia involving K. pneumoniae and A. baumannii is associated with high mortality, particularly in co-infections. Achieving MC through FUBCs is critical for improving patient outcomes. Multivariate logistic regression analysis revealed that the CCI score, and MNC were identified as key predictors of mortality. The ANN analysis further highlighted these predictors and additional factors, including the APACHE II score at ICU admission, SOFA score at ICU admission, SOFA score on culture-positive day, Pitts bacteremia score, age, and RRT. These findings emphasize the importance of rigorous follow-up and optimal management strategies in ICU settings to improve survival outcomes.