Recorded presentation

James L. Wilson, PhD, Christopher J. Mansfield, PhD, Denise Kirk, MS, Luke Schwankl, BA, and Michael L. Gwaltney, BA. Center for Health Services Research and Development, East Carolina University, Building, Greenville, NC 27858, 252-816-2785, WilsonJa@mail.ecu.edu

Background. CDC’s Behavioral Risk Factor Surveillance System uses a telephone interview survey to track health risk behaviors among U.S. adult populations. Since 1994 it has been employed in all states and provides valuable information for public health surveillance and planning at the state level. Because sample sizes at the county level are usually small in each annual survey, however, few states provide estimates of risk factors at the local level. Synthetic estimates are inaccurate and stratified sampling is prohibitively expensive, particularly for counties with small populations. We have developed a simple, reliable way of aggregating annual data into multi-year samples to provide regional and county-level samples that reveal statistically significant variations between counties and regions for many variables of interest.

Method. We aggregate responses to repeated questions from the BRFSS over the four most recent years for each county. Using SUDAAN, to account for variations in probability sampling, weightings in the original stratification are preserved. Counties are examined individually and aggregated into regional clusters of interest. As a check of validity, our estimates are then aggregated up to the state level for comparison to state reported data.

Results. Using Body Mass Index in Pitt County, North Carolina, as an example, 42.9% of African Americans are obese compared to 17.4% of Whites (p<.05); values for the state are 33.9% and 20.0%, respectively.

Conclusions: Aggregation of BRFSS data provides an accurate picture of county to county variation. It is analogous to using a longer exposure to take snapshots in reduced light.

Learning Objectives:

• Understand BRFSS sampling procedures
• Understand the limitations of using state-level BRFSS data for local planning
• Provide local planners with a way of deriving county- and regional-level data from the BRFSS system

Keywords: Community Health Planning, Risk Factors

Related Web page: www.chsrd.med.ecu.edu

Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: East Carolina University Center for Health Services Research and Development
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