Measuring cancer inequality with the cumulative rate

Background: There is widespread interest in comparing populations with respect to cancer incidence and mortality. How does the incidence of cancer one’s own region compare with the incidence in the rest of the state or county? How does the mortality of cancer before an intervention compare with the mortality some years after the intervention? Such questions arise in making decisions about resource allocation. \n \nAim: Usually such comparisons are based on age-standardized rates or cumulative risks. Our aims are (i) to state our reservations about both age standardized rates and cumulative risks, and (ii) to present an alternative approach based on cumulative rates. \n \nMethods: In 1976, N. Day introduced the method of cumulative rates into the cancer literature. Cumulative incidence (or mortality) rates can be calculated easily using basic demographic data and incidence (or mortality) data stratified by age groups. The method leads to results that are similar to those that one would obtain by using age-standardized rates or cumulative risks. However, no special assumptions are required to use the method of cumulative rates or to interpret the results. \n \nResults: We will illustrate how the method works in practice by measuring inequalities in mortality from colorectal cancer (ICD C18-C20) in Australia. A hard copy of the calculations will be available for delegates at the Congress. \n \nConclusions: The assumptions in age-standardization involve introducing weights from a standard population. However, applying arbitrary weightings is unjustifiable when a better measure is available that obviates that need in the first place. In addition, the key assumption in cumulative risk measurement - namely, that the only cause of death is cancer - can be misleading to the wider public. The cumulative rate does not share these disadvantages of the age-standardized rate and cumulative risk. It can be used to measure inequalities in incidence or mortality between different regions, or between the same region at different times, or between men and women. It leads to confidence intervals and tests of statistical hypotheses. Furthermore, the method is very easy to apply. Based on our experience, we commend the method of cumulative rates for use in impact evaluation studies in cancer care, and as a basis for allocating resources. \n