Recorded presentation

Ken Keppel, PhD and Jeffrey Pearcy, MS. Division of Health Promotion, National Center for Health Statistics/CDC, 3311 Toledo Road, Hyattsville, MD 20782, 301 458-4310, KKeppel@cdc.gov

The second goal of Healthy People 2010 is to eliminate disparities among subgroups of the population and this initiative provides a framework for monitoring trends in disparity over time. Disparities are measured primarily in terms of relative differences from the “best” group rate in each domain. Statistical tests of the difference between group rates will be performed when estimates of standard errors are available. In order to facilitate comparisons across Healthy People objectives, disparities are consistently measured in terms of adverse events. These approaches focus on measuring differences between groups.

However, two related questions remain. How can we conclude that a disparity between groups is not “important”, or that a difference between groups has been “eliminated” over time? Assuming that differences from the “best” group rate will not be exactly 0, criteria for the absence of an important difference could be based on: A target value, an absolute difference, a relative difference, a statistical test of the difference between group rates, or a statistical test of the equivalence between group rates. These criteria are discussed as a basis for determining whether a disparity is unimportant and examples are provided of conclusions that can be drawn from statistical tests of difference and equivalence. The power of a test of differences may be limited by the size of the sample for each specific group. Similarly, when both difference and equivalence tests are not significant, we cannot conclude whether the disparity is important.

Learning Objectives:

• List five criteria that could be used to determine that a disparity is not important.
• Understand the basic calculation for a statistical test of equivalence.
• Understand the implications of the four possible outcomes of statistical tests of difference and equivalence.

Keywords: Health Disparities, Statistics

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.

Recorded presentation