Cohort restriction based on prior enrollment: Examining potential biases in estimating cancer and mortality risk