Independent and joint associations of insulin resistance and depressive symptoms with incident cardiovascular disease: a prospective cohort study in eastern China

Insulin resistance (IR) and depression are both associated with increased risk of cardiovascular disease (CVD), but their potential interplay and value for risk identification remain unclear. We included 7,218 participants from a long-term cohort in eastern China. Baseline depression was assessed by PHQ-9 questionnaire, and IR was characterized using estimated glucose disposal rate (eGDR), triglyceride-glucose index (TyG), and non-high-density lipoprotein cholesterol (nonHDLC). A composite metabolic score integrating these three IR-related measures was constructed using confirmatory factor analysis. Cox regression was applied to examine the hazard ratios (HRs) and 95% confidence intervals (95% CIs) of IR indices and depression, as well as the joint effects on CVD risk. Exploratory effect decomposition was conducted to quantify the potential contribution of the metabolic score to the association between depressive symptoms and CVD risk, accounting for follow-up time and censoring. The incremental value of integrating depressive symptoms with each IR index for 10-year CVD risk prediction was assessed using time-dependent receiver operating characteristic curves and the corresponding area under the curves (AUC), with model calibration and clinical utility further evaluated using calibration measures and decision curve analysis (DCA). A total of 702 CVD patients were identified during follow-up. Depression and IR indices were both associated with CVD risk (per-SD HR = 1.12, 95% CI: 1.06–1.18 for depression; HR = 0.78, 95% CI: 0.71–0.87 for eGDR; HR = 1.21, 95% CI: 1.12–1.30 for TyG; HR = 1.16, 95% CI: 1.08–1.25 for nonHDLC; and HR = 1.24, 95% CI: 1.15–1.34 for metabolic score). In joint analyses, individuals with depression and higher IR had further elevated risk. The metabolic score accounted for 11.50% (95% CI: 6.58–17.30%) of the association between depressive symptoms and CVD risk, while the interaction component was not statistically significant. Integrating depressive symptoms with each IR measure improved 10-year discrimination compared with the corresponding single-component models, with the combination of depressive symptoms and the metabolic score achieving the highest AUC (0.745, 95% CI: 0.725–0.765). The combined models also showed better model fit, favorable calibration, and greater net benefit across clinically relevant decision thresholds. Both IR and depressive symptoms were independently and jointly associated with increased CVD risk. The CFA-derived metabolic score accounted for part of the association between depressive symptoms and CVD risk. Integrating depressive symptoms with IR measures improved risk discrimination and clinical utility, supporting a more comprehensive approach to CVD risk identification.