Multivariable Cox regression model was adjusted for sex, age at baseline (year, continuous), smoking status (never, former, current, and missing), education (no formal education or primary education, secondary education, trade or technical education, university education or above, and missing), physical activity (none or almost none, low, intermediate, high, and missing), alcohol intake (nondrinker, drinker, and missing), and baseline health conditions (cerebrovascular disease, coronary heart disease, cancer, hypertension, and diabetes). HR indicates hazard ratio.