Adjusted mortality rates: a tool for creating more meaningful league tables for stillbirth and infant mortality rates.

A number of problems associated with league tables of performance indicators have been discussed in the literature. This paper attempts to address these problems for stillbirth and infant mortality rates in order to produce meaningful and useful information for the government, general public and health professionals. Composite stillbirth and infant mortality rates, low birth-weight and very low birth-weight rates were determined for the 100 English Health Authorities for 1996-1997. Townsend deprivation scores for these districts were also obtained. The mortality rates were adjusted by multiple regression for very low birth-weight and Townsend score separately and together. Confidence intervals were calculated for the dual-adjusted rates. Almost 60% of the variability in mortality rates were explained by Townsend score and very low birth-weight rates together. Adjusted league tables showed how the individual and combined predictors affect the individual mortality rates for each Health Authority. There was considerable overlap in the confidence intervals for the adjusted rates although there were a few Health Authorities whose mortality rates were clearly below most others. We conclude that fairer and more useful information is provided by geographically based league tables which give both crude rates and rates adjusted for single and multiple predictor variables. The inclusion of confidence intervals aids interpretation of annual random variations and knowledge of differences in the effects of the individual predictors enables better resource targeting.