Association of the C-Reactive Protein–Triglyceride–Glucose Index With All-Cause and Cardiovascular Mortality in U.S. Adults: Insights From NHANES 1999–2010

Abstract

Background

CTI combines C-reactive protein with the triglyceride–glucose index and may capture concurrent inflammatory and insulin–resistance-related metabolic disturbance. Emerging cohort evidence has linked higher CTI values to all-cause and CVD mortality, but the robustness, dose–response characteristics, and predictive relevance of these associations remain uncertain. This study evaluated the associations of CTI with all-cause and CVD mortality among U.S. adults.

Methods

Participants were selected from NHANES survey cycles conducted from 1999 through 2010, yielding an analytic sample of 10,528 adults aged 18 years or older. The CTI was computed using the formula: CTI = 0.412 × ln [CRP (mg/L)] + ln {[triglycerides (mg/dL) × fasting glucose (mg/dL)]/2}. We used weighted Cox proportional hazards models to investigate the relationships of CTI with all-cause and CVD mortality. Restricted cubic spline analyses were used to assess nonlinearity, followed by exploratory two-piecewise Cox regression to examine possible changes in slope. Beyond the primary analysis, the robustness and potential heterogeneity of these associations were further assessed using sensitivity, subgroup, and interaction analyses.

Results

During a survey-weighted mean follow-up of 164.29 ± 50.47 months, the primary survey-weighted analytic sample included 10,528 participants. In the primary DAG-informed Cox models, each 1-unit increase in CTI was associated with higher risks of all-cause mortality (HR = 1.24, 95% CI: 1.18–1.31, p < 0.001) and CVD mortality (csHR = 1.35, 95% CI: 1.20–1.53, p < 0.001). Compared with Q1, participants in Q4 had higher risks of all-cause mortality (HR = 1.54, 95% CI: 1.29–1.84, p < 0.001) and CVD mortality (csHR = 1.78, 95% CI: 1.27–2.49, p < 0.001). These associations remained significant after additional clinical adjustment. Exploratory two-piecewise analysis estimated an inflection point of 8.165 for all-cause mortality (conventional likelihood-ratio p = 0.002); because the breakpoint was selected using a data-driven procedure, this comparison was interpreted cautiously. For CVD mortality, the estimated inflection point was 8.070 and the two-piecewise model did not clearly improve fit over the linear model (p = 0.067). Time-dependent AUCs for CTI were approximately 0.60–0.62 at 5, 10, and 15 years, indicating modest discrimination.

Conclusion

After adjustment for prespecified covariates, higher CTI was associated with greater hazards of all-cause and CVD mortality. Exploratory analyses suggested a possible change in slope in the association with all-cause mortality, whereas evidence for a distinct change point in cardiovascular mortality was inconclusive. These findings extend recent population-based evidence linking CTI to mortality; however, its standalone discrimination and incremental predictive improvement beyond established clinical risk factors were modest. CTI may therefore be better viewed as an adjunctive marker of inflammatory-metabolic burden than as a major addition to existing clinical prediction models.