Obtaining life table data from cohort analyses: A critique of current methods1