4 Methods and Design
|The data in this analysis are for events between 1976
(the earliest possible date with this data set) and 1993. (Note 3) For each of these
years, we can identify an individuals age, whether (s)he entered the data base,
died, or was lost from the data base for some other reason and censored. We can calculate
the number alive in each year-long interval at their birthday, and the number dying before
their next birthday. With these life table functions, we calculate age-specific mortality
The logic of the evaluation of the age specific mortality rates is that consistent mortality rates across regions for the same races and sexes is evidence that the age-at-death data are consistent. Regions referred to here are the regions where SSNs were received. We assume that the regions represent more homogeneous populations than if the country as a whole were to be considered, especially for the older population. At the time when most of this population received their SSNs, the U.S. population was much less mobile, and regional ethnic compositions and economic conditions were well differentiated. One problem of note with using the region where the individuals obtained their card is that the populations of the west and southwest were small at this time, and this is reflected in the stability of mortality rates at the oldest ages.
Age-specific mortality rates are estimated using the following equation:
Mx = Dx / ((Kx + Kx+1)/2)
where Kx = the number of persons alive at the beginning of interval x and Dx = the number of deaths in age interval x.
|Evaluation of U.S. Mortality Patterns at Old
Using the Medicare Enrollment Data Base.
Allan M. Parnell and Cynthia R. Owens
© 1999 - 2000 Max-Planck-Gesellschaft ISSN 1435-9871