Incidence and risk factor of sepsis in patients with severe community-acquired pneumonia: a Chinese, single-center, retrospective study

Sepsis represents a high-risk mortality cohort among patients with severe community-acquired pneumonia (SCAP). Rapid and precise identification along with prompt decision-making, serves as a practical approach to improve patient prognosis. This retrospective observational study enrolled adult patients with severe community-acquired pneumonia (SCAP) who were continuously hospitalized in the intensive care unit (ICU) of West China Hospital, Sichuan University, from September 2011 to September 2019. Univariate and multivariate logistic regression analyses were employed to identify independent risk factors for co-sepsis, followed by the utilization of LASSO regression to filter features to establish a nomogram. Model robustness was evaluated via the C index, receiver operating characteristic (ROC) analysis, and calculation of the area under the curve (AUC). Furthermore, its predictive accuracy was assessed via decision curve analysis (DCA). In total, 5855 SCAP patients were included in the present study, of whom 654 developed sepsis. Patients with sepsis exhibited a prolonged length of stay in the ICU and higher mortality rates, indicating a worse prognosis than those without sepsis. We identified 15 independent risk factors associated with the development of sepsis in SCAP patients. Further analysis incorporating 9 of these features to construct a nomogram demonstrated a C index of 0.722 (95%CI 0.702-0.742), including lactate, D-dimer, respiratory rate, heart rate, albumin, hemoglobin, activated partial thromboplastin time (APTT), glucose, and C-reactive protein (CRP) levels. The AUC values and DCA curves demonstrated that the model exhibited superior accuracy and overall net benefit in predicting co-sepsis development compared with the qSOFA, CURB-65, SOFA, and APACHE II scores. Additionally, the calibration curve confirmed good concordance between the predicted probabilities of the model. This study investigated the risk factors for co-sepsis in SCAP patients and constructed an expedited, cost-effective and personalized model for predicting the probability of co-sepsis.