Causes of death were ascertained chiefly by death certificates and supplemented by reviews of medical records and verbal autopsy using validated instruments.
Information collected from the baseline and resurvey was directly entered into a laptop-based data entry system developed with built-in functions to avoid missing items and to minimize logic errors during the interview.
Multivariable models were stratified by age in the 5-year interval, study area, and adjusted for sex, education, marital status, and parental family history of cardiometabolic multimorbidity.