DR Miller, CHQOER, VAMC, Bedford, MA; LM Pogach, VAMC, East Orange, NJ; A Hendricks, CHQOER, VAMC, Bedford, MA; W West, CHQOER, VAMC, Bedford, MA; G Hawley, HAIG, VAMC, Milwaukee, WI; D Gilden, Jen Associates, Boston, MA; M Safford, VAMC, East Orange, NJ

Workshop Objectives: Valid estimates of disease prevalence in patient populations are crucial to health services researchers and to health care managers in planning services and programs and in setting policies. Medical care system administrative databases represent a potential wealth of information for estimating disease rates and identifying cases, but there a many challenges to overcome in making effective and appropriate use of these resources. These problems include incomplete enumeration of the patient population, lack of complete and accurate diagnostic information, data entry errors, data inaccuracies related to financial incentives, and enormous data processing and computing requirements. Many VA investigators have grappled with these problems and a methodology is emerging which can be used to make valid and effective use of computerized medical and claims records to estimate disease prevalence in the VA patient population. We have applied this methodology in processing and linking VA and HCFA databases to estimate prevalence of diabetes and associated complications among VA patients. In this workshop, we will inform participants on concepts, resources, process, problems, and solutions in using VA and non-VA administrative data to estimate disease prevalence in a patient population, and we will present our work in estimation of diabetes prevalence among VA patients as a model. Participants will be better informed and prepared to use administrative databases for epidemiologic research and/or to interpret findings from such research.

Workshop Activities: In this workshop we will review epidemiologic principles, inventory VA and non-VA databases which may provide information on disease prevalence of VA patients, and discuss problems and solutions in processing and merging databases to obtain valid data for disease rate estimation. We will present our work to date in estimating national diabetes prevalence rates using VA computerized visit data, pharmacy records, laboratory records, large patient surveys, and linked Medicare (HCFA) data. In addition to our diabetes work, presentations will include examples from studies of other disease.

Target Audience: This workshop will be of interest to epidemiologists and health services researchers who wish to study the prevalence, distribution, and course of diseases in a patient population, and to health care administrators and clinical managers who wish to estimate prevalence for purposes of planning and resource allocation. It will have particular relevance for those with an interest in diabetes.

Audience Familiarity: Participants should have familiarity with the VA health care system.