Why do we continue to use standardized mortality ratios for small area comparisons?

Public health practitioners are often faced with the necessity to compare the mortality experience of different geographical areas. Indirect standardization, producing a 'standardized mortality ratio' (SMR) is the most commonly used technique for doing this. However, as we show, indirect standardization is inappropriate for such comparisons, as SMRs for different geographical areas have different denominators. The fact that indirect standardization is usually chosen for this type of comparison is probably based on two beliefs: (1) that direct standardization yields only a rate rather than a more easily interpreted ratio or index; (2) that direct standardization cannot be carried out in many cases because the sub-group specific mortality rates in the groups to be compared are not available or, in at least some age classes, are based upon such small numbers as to be completely unreliable. In this paper we show that a simple index (the comparative mortality figure) can be calculated from the directly standardized rate in most cases. Using a comparison of the overall mortality experience of electoral wards in Sheffield between 1980 and 1987 we demonstrate also that the advantage gained by the smaller standard error of the SMR is outweighed by the bias inherent in its construction. We recommend that the SMR is used only when absolutely necessary, that is, in the rare circumstance when data are not available for the calculation of age- and sex-specific subgroup rates in the study population.