Computing exact excess death rates from a published mortality study.

We wish to estimate the associated excess death rate (EDR) or mortality ratio (MR) from a published study of persons with a given medical condition. This requires computation of the expected mortality in the study population. If age- and sex-specific person years of data are available, this task is straightforward. Most often, however, we have only descriptive statistics--percentage male, average and standard deviation of age--at the beginning of follow up. We show here how this limited information can be used to compute an exact EDR or other quantities of interest.