Comparison of mortality outcomes derived from survival curves calculated by the mean covariates and corrected group prognosis methods.

As creators of mortality abstracts, we assume the mortality and survival curves from source articles have been created by the appropriate methodology. Two methods can be used to generate adjusted curves--the mean of covariates and the group prognosis methods. The use of the former, although simple to implement, may be mathematically problematic. A comparative mortality analysis is created from a study population using curves adjusted by both methods. For this cohort, utilizing data from direct measurement of enlarged survival curves produced no difference in mortality outcomes. Small differences in percent cumulative survival derived from each method do not translate into important differences in mortality ratios and excess deaths.