The standardized mortality ratio revisited: improvements, innovations, and limitations.

Evaluation of patient outcome for end-stage renal disease patients is an important component of the comparison of treatment protocols. End-stage renal disease mortality comparisons are facilitated by availability of the relevant data in administrative databases. The standardized mortality ratio has been widely used by the renal community for such evaluation. Accurate methods for evaluating adjusted mortality with the standardized mortality ratio are shown, even when there are few deaths expected. Both confidence interval and probability value calculations are shown. Classification errors for the identification of elevated mortality will occur even with accurate calculations. Criteria for choosing adjustment factors are reviewed. Only those factors that result in unavoidable variation in mortality should be adjusted for. Analysts using the US Renal Data System death rate tables for standardized mortality ratio computations must calculate mortality and patient follow-up with the same methods used by the US Renal Data System. In particular, patient follow-up must include postwithdrawal and posttransfer data but not data for the first 90 days of end-stage renal disease. Methods for comparison to a regional norm are presented.