Nonlinear association of the C-reactive protein-triglyceride-glucose index with 28-day mortality in critically ill children: a retrospective cohort study.

BACKGROUND: The C-reactive protein-triglyceride-glucose index (CTI) has shown prognostic value in adults, but its association with short-term mortality in critically ill children remains unclear. We investigated the association of CTI with 28-day mortality, including its nonlinear relationship and incremental predictive value. METHODS: This retrospective cohort study included 5,039 critically ill children from the Pediatric Intensive Care database. Associations between CTI and 28-day mortality were evaluated via multivariable Cox regression, restricted cubic splines (RCS), and segmented two-piecewise Cox models. Competing-risk analysis, subgroup analyses, and sensitivity analyses were performed to test the robustness of results. Time-dependent AUC was used to evaluate model discrimination, while continuous NRI and IDI were calculated to determine the incremental predictive value. Calibration and decision-curve analysis were used to assess model performance and clinical utility. RESULTS: During the 28-day period after ICU admission, 227 patients (4.5%) died in hospital and 220 (4.4%) died in the ICU. In the fully adjusted model, compared with the lowest CTI tertile, the highest tertile was associated with higher risks of in-hospital mortality (HR 1.65, 95% CI 1.19-2.29; P = 0.003) and ICU mortality (HR 1.57, 95% CI 1.13-2.20; P = 0.008). Significant nonlinear associations were observed for both outcomes, with an exploratory breakpoint estimated at approximately 9.58. Adding CTI to the clinical model yielded only modest incremental predictive improvement. CONCLUSIONS: Higher CTI was independently associated with 28-day mortality in critically ill children, with nonlinear associations and an exploratory breakpoint near 9.58, however the predictive value remained limited.