The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #63714

Kirsten Knutson, MPH, Weihong Zhang, MS, Farzaneh Tabnak, MS, PhD, and Georjean Stoodt, MD, MPH. California Department of Health Services, Cancer Detection Section, 601 North 7th Street, MS-428, PO Box 942732, Sacramento, CA 94234, 916-324-7967, kknutson@dhs.ca.gov

Problem: The California Department of Health Services, Cancer Detection Section (CDS) administers breast and cervical cancer screening programs by dividing the state into “Partnership” areas, which are aggregates of California counties. For the purposes of program planning and implementation, it is important that CDS estimate the number of women eligible for program services in each Partnership. Describing the low income, uninsured/underinsured eligible population by Partnership is problematic, however, because current county-level data containing income and health insurance status is not available. Solution: Small area estimation methodology was used to produce estimates of the eligible population per Partnership. County data on factors known to be associated with eligibility status were combined in a generalized linear mixed effect model with state data describing eligibility status. A random effect variable included in the model represents the area cluster variation that cannot be explained by the regression variables. Bootstrap methodology was applied to the model to calculate the variance. Results: In 2001, 27.12% (172,738 (90% CI 134,824, 210,652)) of Hispanic women 40 years and older in Los Angeles county are eligible for program services (an example of one Partnership’s estimated eligible population). Conclusion: This paper demonstrates the application of statistical methods to develop a quantitative description of a special population targeted for public health services. Because obtaining current, flexible, and accurate data on the income and health insurance status among demographic groups is difficult nation-wide, the method used in this paper may be helpful for public health program planners and administrators in other states.

Learning Objectives:

• At the conclusion of the session, the participant (learner) in the session will be able to:
• 1. Identify data containing information on income and health insurance status and other demographic characteristics, and recognize the lack of cross-tabulated data of these characteristics.
• 2. Describe the use and limitations of small area estimation methodology to generate estimations of local populations.
• 3. Apply the statistical methods discussed and the appropriate data to produce estimations of designated local populations for the purposes of public health program planning and implementation.

Keywords: 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.