The association between triglyceride glucose-body mass index and mortality in intensive care unit patients: a propensity score matching analysis

Abstract

Purpose: Triglyceride glucose-body mass index (TyG-BMI) represents a combined measure to evaluate insulin resistance and predict cerebral and cardiovascular disease risk and the resulting negative consequences. Nevertheless, the prognostic value of TyG-BMI for predicting outcomes, such as mortality, among critically ill patients in the intensive care unit (ICU-CIP) remains understudied. Our study seeks to ascertain the relation between all-cause mortality (ACM) and TyG-BMI among ICU-CIP, regardless of specific diseases, to recognize individuals at high risk and enhance prediction strategies.

Methods: The data were acquired from the Medical Information Mart for Intensive Care (MIMIC)-IV database, version 3.2, and estimated the TyG-BMI, incorporating fasting blood glucose, fasting triglycerides, and BMI. The formula used was ln{[fasting triglycerides (mg/dL) × fasting blood glucose (mg/dL)]/2} × BMI. Herein, we included all first-time admitted adult patients, evaluated their TyG-BMI., and conducted a 1:1 propensity score matching (PSM) approach to address possible confounding variables. The critical TyG-BMI level influencing patient survival was determined utilizing maximally selected rank statistics. Kaplan-Meier survival analysis along with multivariate Cox proportional hazards (PH) regression models were utilized to estimate the impact on short- and long-term ACM. Furthermore, restricted cubic spline (RCS) methods explored the linear or non-linear relation between TyG-BMI and ACM, with additional knowledge acquired from interactions and analyses of subgroups.

Results: A total of 9,175 ICU-CIP was included; after PSM, the analysis involved 3,642 matched participant pairs. Cox PH fully adjusted regression models demonstrated a significant correlation between higher TyG-BMI (≥ 239.54) and decreased 90 day ACM, both before (hazard ratio [HR] 0.77; 95% confidence interval [CI] 0.69-0.85) and after PSM (HR 0.76; 95% CI 0.68-0.85). Comparable associations were observed for 30 day, 180 day, and 365 day ACM. Post-PSM, RCS analysis revealed a negative L-shaped non-linear relation between both short- and long-term ACM and TyG-BMI. Notably, significant interaction effects were noticed in age, race/ethnicity, and hypertension subgroups, while no interaction effects were found in diabetes and gender subgroups.

Conclusion: TyG-BMI is a novel, non-invasive predictor of mortality in ICU-CIP. These findings may inform risk stratification and public health strategies, although validation in diverse populations is warranted.

Keywords: Biomarker; Critical ill; Intensive care unit; MIMIC-IV; Mortality; Triglyceride-glucose body mass index.

Fig. 1

Study flow chart

Fig. 2

Evaluation of the optimal cut-off point of TyG-BMI by utilizing maximally selected rank statistics

Fig. 3

The Kaplan–Meier survival curves of (A) 90 day, (B) 180 day, and (C) 365 day of ACM were categorized by binary and tertiles of TyG-BMI before PSM

Fig. 4

The absolute standardized variances for the matching variables between the groups

Fig. 5

The Kaplan–Meier survival curves of (A) 90 day, (B) 180 day, and (C) 365 day of ACM were categorized by binary of TyG-BMI after PSM

Fig. 6

Forest plots for (A) 90 day, (B) 180 day, and (C) 365 day of ACM in different subgroups

Fig. 7

Restricted cubic splines at different time intervals for ACM at (A) 90 day, (B) 180 day, and (C) 365 day before and after propensity score matching