A Prediction Model for the Diagnosis of Sepsis Based on the Classification of Acute Gastrointestinal Injury

<b>Aims/Background</b> The existing screening approaches for sepsis demonstrate lower sensitivity, potentially resulting in misdiagnosis of septic conditions. The gastrointestinal tract is the primary and most susceptible organ during sepsis. Therefore, this study aims to establish and evaluate a predictive model for sepsis based on the classification of acute gastrointestinal injury (AGI), to improve diagnostic sensitivity. <b>Methods</b> This retrospective study included patients with confirmed infections or suspected infections who were admitted to the general ward of Changshu Hospital Affiliated to Soochow University (Changshu First People's Hospital, China) between April 2023 and December 2023. Patients were randomly divided into a developing cohort (n = 1667) and a validation cohort (n = 712) in a 7:3 ratio. Furthermore, data were collected for various variables, including general variables, inflammatory factors, hemodynamic variables, organ dysfunction variables, and tissue perfusion variables. Univariate analysis was used to screen the risk factors associated with sepsis, and logistic regression analysis was employed to identify the independent predictive factors. The nomogram of the model was constructed based on these independent predictive factors. Additionally, the prediction significance of the model was evaluated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) across both the developing and validation cohorts. <b>Results</b> Out of the total of 2379 patients in the study, the sepsis rate was 12.5%. The incidence of AGI in septic patients was 96.0%, with 23.2% in grade I, 52.3% in grade II, 16.1% in grade III, and 4.4% in grade IV. Factors like age (Odds Ratio (OR) = 1.029, 95% Confidence Interval (CI) 1.015-1.043, <i>p</i> < 0.01), hypotension (OR = 3.863, 95% CI 2.372-6.290, <i>p</i> < 0.01), oxygen saturation (SpO₂) (OR = 0.795, 95% CI 0.751-0.840, <i>p</i> < 0.01), thrombocytopenia (OR = 5.657, 95% CI 2.835-11.289, <i>p</i> < 0.01) and AGI grade (OR = 7.151, 95% CI 5.040-10.144, <i>p</i> < 0.01) were observed as independent predictors for sepsis. Based on these five variables, a predictive model nomogram (model B) was developed. Model B achieved area under the curve (AUC) of 0.947 (95% CI 0.932-0.963) and 0.962 (95% CI 0.945-0.978) for the developing and validation cohorts, respectively, which were significantly higher than the AUC value of quick Sequential Organ Failure Assessment (qSOFA) (model A). Furthermore, the calibration curves for both the developing and validation datasets were close to the ideal model. Decision curve analysis revealed that model B exhibited a better net clinical benefit than model A. <b>Conclusion</b> This study developed and validated a novel model based on AGI that could predict sepsis patients with infections in general wards, significantly helping in clinical decision-making.