Statistical Analysis

The participants were divided into quartiles (Q) of equal size based on their DII scores. The baseline characteristics of the participants across the DII quartiles were compared using the Cochran–Mantel–Haenszel test for categorical variables and analysis of variance or Kruskal–Wallis for continuous variables. The intake of nutrients or food groups across the DII quartiles, which was adjusted for age, sex, and total energy consumption, was compared using a general linear regression test. To determine the association between DII and all-cause and CVD mortality risk, Cox proportional hazards model was used to estimate the adjusted hazard ratios (HRs) and 95% confidence intervals (CIs). Five models were used to assess the associations. Model 1 was unadjusted. Model 2 was adjusted for age and sex. Model 3 was adjusted for age, sex, BMI, smoking and drinking status, and work strength. Model 4 was adjusted for the variables mentioned in Model 3, with energy-adjusted salt intake as the main model. In Model 5, total cholesterol, hypertension status, and diabetes history were further adjusted. Moreover, analyses were further stratified by salt intake (＜13.2 g/day [median] and ≥ 13.2 g/day), sex (men and women), age at baseline (aged ＜65 years and ≥ 65 years), and BMI (＜25 kg/m² and ≥ 25 kg/m²). The interactions between the DII score and salt intake, age, sex, and BMI for all-cause and CVD mortality were also assessed. P-value less than 0.05 was considered statistically significant. All statistical analyses were performed using the Statistical Analysis Systems software package (version 9.4; SAS Institute, Cary, NC, USA).