Prognostic Factors and Nomogram for Klebsiella pneumoniae Infections in Intensive Care Unit.

Abstract

Purpose: Klebsiella pneumoniae infections pose a significant threat to public health with high morbidity and mortality rates. The early identification of risk factors for mortality and accurate prognostic evaluation are important. Therefore, we aimed to identify the risk factors for mortality in patients with K. pneumoniae infections and develop a nomogram model for prognosis.

Methods: Patients diagnosed with K. pneumoniae infection were recruited from the intensive care unit of Peking University Third Hospital. The enrolled patients were categorized into survivor and non-survivor groups. Univariate and multivariate regression analyses were performed to identify independent risk factors for 30-day mortality, and a nomogram was constructed and validated.

Results: A total of 408 patients infected with K. pneumoniae at different sites were included in this study. PO₂, lactate, respiratory failure, urinary tract infection, heart rate, 24h-urineoutput, neutrophil count, alkaline phosphatase, and vasoactive drug use were significant risk factors and were integrated into a nomogram to predict the risk of 7-day, 14-day, 21-day, and 28-day mortality. The nomogram demonstrated superior prognostic ability, achieving higher area under the receiver operating characteristic curve (AUC) (>0.8) and concordance index (C-index) (>0.8) values than the Pitt bacteremia, sequential organ failure assessment (SOFA), and acute physiology and chronic health evaluation (APACHE) II scores (all AUC and C-index < 0.75). Cross-validation of the nomogram confirmed its consistent performance, with both AUC and C-index values exceeding 0.75. The nomogram demonstrated a strong Hosmer-Leme-show goodness-of-fit and good calibration (p > 0.05). Additionally, decision curve analysis revealed that the nomogram provided significant clinical utility for prognostic prediction.

Conclusion: The 30-day mortality risk factors for K. pneumoniae infections were identified, and a predictive nomogram model was developed. The nomogram demonstrated good accuracy and predictive efficiency, providing a practical tool for short-term risk assessment and potentially improving clinical outcomes by providing early intervention and personalized patient management.