Association between the development of sepsis and the triglyceride-glucose index in acute pancreatitis patients: a retrospective investigation utilizing the MIMIC-IV database.

Abstract

Background: Sepsis is a serious consequence of acute pancreatitis (AP) that requires immediate detection and treatment. Triglyceride-glucose (TyG) index demonstrated predictive ability for a number of diseases. In an effort to enhance clinical care and early warning systems, this study examined the association between the TyG index and sepsis risk with the aim of improving clinical care and early warning systems.

Methods: Patients who were first admitted and satisfied the diagnostic criteria for acute pancreatitis (ICD-9: 5770; ICD-10: K85) were chosen from the MIMIC-IV database, excluding those lacking essential demographic or laboratory data. Using the Sepsis-3.0 criteria. Depending on whether they had sepsis or not, patients were divided into sepsis group and non-sepsis group. Utilizing the formula ln[(triglycerides mg/dl) × (glucose mg/dl)/2], the TyG index was calculated. The Boruta algorithm and Xgboost model were used for feature selection in order to pinpoint the important variables affecting results. Logistic regression with univariate and multivariate factors were used to assess the association between the TyG index and the start of sepsis after admission.

Results: Twenty-eight thousand AP patients were screened in all, among which 661 patients were ultimately included in the study. Of these, 228 patients (34.5%) developed sepsis. The TyG index was shown to have a significant correlation (OR = 1.891, 95% CI: 1.408-2.555) with sepsis, and an increased risk of sepsis was observed with an increase in the TyG index (all P values for trend < 0.001). Subgroup analysis showed that among patients of various ages, sexes, and with hypertension and diabetes, there was a positive association between the TyG index and the probability of sepsis (all P values for trend < 0.05). The combination of the TyG index with clinical indicators had an area under the curve (AUC) of 0.828 (0.794-0.862), which was significantly greater than that of the TyG index alone (0.657 [0.613-0.701]), with a statistically significant difference (Z= -7.362, P < 0.001).

Conclusion: In patients who have AP, the TyG index is substantially linked to a higher risk of sepsis, and when combined with clinical markers, its predictive power for sepsis is enhanced. The findings imply that the TyG index might be a helpful detection for determining which AP patients are at a higher risk of developing sepsis.