Measuring reduction in disparity among several rates

The Healthy People 2010 initiative challenges us to measure progress toward the reduction of disparities among subgroups in the population. Pair-wise comparisons provide specific information about absolute or relative differences between rates. Changes in disparity between two groups can be measured by comparing the disparity between rates at one point in time with the disparity between comparable rates at a second point in time. However, when there are more than two groups, it can be difficult to draw a general conclusion about changes in disparity over time based on a series of pair-wise comparisons. Summary measures can be used to quantify the differences among several rates, both at a point in time and over time. Summary measures can also be used to compare the degree of disparity across indicators, areas, and populations; and under some constraints across population domains. The mean deviation is a measure of the average of absolute differences between group rates; the index of disparity expresses the mean deviation relative to (as a percent of) the total population rate; and the average of ratios is a measure of the average of relative differences. Measures of disparity among several groups are described as extensions of the techniques used to measure differences between two groups. The significance of choosing a reference point for measuring disparity, the estimation of reliability for summary measures, and the need to interpret summary measures in terms of underlying rates are discussed.