POTENTIAL PITFALLS IN INTERPRETING MAPS OF STABILIZED RATES

There have been many advances in the mapping of disease mortality and morbidity rates in recent years. An increasingly common approach is to calculate ''stabilized'' rates for individual geographic areas by inc orporating information from the entire set of data. One such method, a dopted by the National Cancer Institute for its mortality maps, is des cribed and applied to national end-stage renal disease data. Some poss ible difficulties in interpretation are noted, resulting from: (1) ins ufficient shrinkage, (2) highly skewed rate distributions, (3) differe nt denominator distributions, and (4) variations in population density .