Cholesterol and Egg Intakes with Cardiometabolic and All-Cause Mortality among Chinese and Low-Income Black and White Americans

We examined the associations of dietary cholesterol and egg intakes with cardiometabolic and all-cause mortality among Chinese and low-income Black and White Americans. Included were 47,789 Blacks, 20,360 Whites, and 134,280 Chinese aged 40–79 years at enrollment. Multivariable Cox models with restricted cubic splines were applied to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for mortality outcomes using intakes of 150 mg cholesterol/day and 1 egg/week as the references. Cholesterol intake showed a nonlinear association with increased all-cause mortality and a linear association with increased cardiometabolic mortality among Black Americans: HRs (95% CIs) associated with 300 and 600 mg/day vs. 150 mg/day were 1.07 (1.03–1.11) and 1.13 (1.05–1.21) for all-cause mortality (P-linearity = 0.04, P-nonlinearity = 0.002, and P-overall < 0.001) and 1.10 (1.03–1.16) and 1.21 (1.08–1.36) for cardiometabolic mortality (P-linearity = 0.007, P-nonlinearity = 0.07, and P-overall = 0.005). Null associations with all-cause or cardiometabolic mortality were noted for White Americans (P-linearity ≥ 0.13, P-nonlinearity ≥ 0.06, and P-overall ≥ 0.05 for both). Nonlinear inverse associations were observed among Chinese: HR (95% CI) for 300 vs. 150 mg/day was 0.94 (0.92–0.97) for all-cause mortality and 0.91 (0.87–0.95) for cardiometabolic mortality, but the inverse associations disappeared with cholesterol intake > 500 mg/day (P-linearity ≥ 0.12; P-nonlinearity ≤ 0.001; P-overall < 0.001 for both). Similarly, we observed a positive association of egg intake with all-cause mortality in Black Americans, but a null association in White Americans and a nonlinear inverse association in Chinese. In conclusion, the associations of cholesterol and egg intakes with cardiometabolic and all-cause mortality may differ across ethnicities who have different dietary patterns and cardiometabolic risk profiles. However, residual confounding remains possible.