Development and Validation of the Prediction Model of Sepsis in Patients After Percutaneous Nephrolithotomy and Sepsis Progresses to Septic Shock.

Abstract

Background: To study the predictors of sepsis and the progression of sepsis to septic shock in patients after percutaneous nephrolithotomy (PCNL) and to establish and validate predictive models. Methods: The patients were assigned to either the development cohort or the validation cohort depending on their hospital. In the development cohort, univariate and multivariate logistic regression analyses were used to screen independent risk factors for sepsis after PCNL and sepsis progression to septic shock. Nomogram prediction models were established according to the related independent risk factors. Areas under the receiver operating characteristic curves, calibration plots, and decision curve analysis (DCA) were used to estimate the discrimination, calibration, and clinical usefulness of the prediction models, respectively. The two sets of models were further validated on the validation cohort. Results: In the development cohort, the risk factors for sepsis after PCNL were diabetes, urine nitrite, staghorn calculi, HU value, albumin-globulin ratio, and high-sensitivity C-reactive protein/albumin ratio. The pre- and postoperative white blood cell counts were risk factors for the progression of sepsis to septic shock. The area under the ROC curve value for predicting sepsis risk was 0.891 and that for predicting septic shock risk was 0.981 in the development cohort; in the validation cohort, these values were 0.893 and 0.996, respectively. In the development cohort, the calibration test p values in the sepsis and septic shock cohorts were 0.946 and 0.634, respectively; in the validation cohort, these values were 0.739 and 0.208, respectively. DCA of the model in the sepsis and septic shock cohorts showed threshold probabilities of 10%-90% in the development cohort; in the validation cohort, these values were 10%-90%. Conclusion: The individualized nomogram prediction models can help improve the early identification of patients who are at higher risk of developing sepsis after PCNL and the progression of sepsis to septic shock to avoid further damage.