A method of comparison for standardized rates of low-incidence events

Abstract

The authors present an approximate but simple approach to comparing age-gender standardized rates of low-incidence events (mortality or morbidity) rates across several geographic areas. The presented method will be useful particularly when a person-level database, one that includes unique person identifiers, is not available. We specify conditions under which comparisons can be made without calculating empirical standard errors and worrying unduly about recurrent events. To compare indirectly standardized rates, only information on the size, the crude rate, and the standardized rate of the areas are needed. For comparing directly standardized rates, which requires stratum-specific rates, the method also requires stratum-specific total numbers of events and individuals but does not require person-level information. The proposed approach builds on previous work comparing rates using person-level data.