![IconGems-Print](images/IconGems-Print.png)
654 – Evaluating and Reducing Nonsampling Errors in Surveys
Approaches for Performing Age-Adjustment in Trend Analysis
Mary Ann Bush
National Center for Health Statistics
Xianfen Li
National Center for Health Statistics
The National Center for Health Statistics (NCHS) produces publications on a wide range of health indicators from its survey data systems. Many of these reports present analyses of trends over time. Age-adjustment is used to minimize differences in observed estimates that result from changes to the age structure in a population over time. In NCHS trend analyses, age-adjusted estimates are calculated using the age distribution for the year 2000 U.S. standard population. This paper will describe two linear regression trending approaches involving age-adjusted estimates. One approach is the adjustment of the survey sample weights to reflect age-adjustment. The other approach uses age as a covariate in a regression model. The two approaches are compared to provide NCHS survey data users with options for their trend analysis work.