A novel nomogram to predict the risk of requiring mechanical ventilation in patients with sepsis within 48 hours of admission: a retrospective analysis.

Abstract

Objective: To establish a model that can predict the risk of requiring mechanical ventilation within 48 h after admission in patients with sepsis.

Methods: Data for patients with sepsis admitted to Dongyang People's Hospital from October 2011 to October 2023 were collected and divided into a modeling group and a validation group. Independent risk factors in the modeling group were analyzed, and a corresponding predictive nomogram was established. The model was evaluated for discriminative power (the area under the curve of the receiver operating characteristic curve, AUC), calibration degree (Hosmer-Lemeshow test), and clinical benefit (decision curve analysis, DCA). Models based on the Sequential Organ Failure Assessment (SOFA) scores, the National Early Warning Score (NEWS) scores and multiple machine learning methods were also established.

Results: The independent factors related to the risk of requiring mechanical ventilation in patients with sepsis within 48 h included lactic acid, pro-brain natriuretic peptide (PRO-BNP), and albumin levels, as well as prothrombin time, the presence of lung infection, and D-dimer levels. The AUC values of nomogram model in the modeling group and validation group were 0.820 and 0.837, respectively. The nomogram model had a good fit and clinical value. The AUC values of the models constructed using SOFA scores and NEWSs were significantly lower than those of the nomogram (P < 0.01). The AUC value of the integrated machine-learning model for the validation group was 0.849, comparable to that of the nomogram model (P = 0.791).

Conclusion: The established nomogram could effectively predict the risk of requiring mechanical ventilation within 48 h of admission by patients with sepsis. Thus, the model can be used for the treatment and management of sepsis.