A Methodology to Estimate the Prevalence of Epilepsy in a Managed Care Organization Using Data from aComputerized Management Information System

• Activity Number: 311
• Type: Contributed
• Date/Time: Wednesday, August 14, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology*
• Author(s): Hoang Nguyen*+ and Becky Grebosky and Ann Worley
• Affiliation(s): ORC Macro and Lovelace Clinic Foundation and Lovelace Clinic Foundation
• Keywords: epilepsy ; prevalence ; logistic regression ; ROC curve ; managed care organization ; epidemiology

Abstract

This study presents a methodology for estimating the prevalence of epilepsy using lab, pharmacy, inpatient, and outpatient data in a computerized management information system (MIS) within a managed care organization. The methodology consists of: 1.) identifying the predictor variables of epilepsy from medical procedure (CPT codes), clinical diagnosis (ICD-9-CM codes), and medications; 2.) determining the optimal mix of predictor variables with logistic regression model; and 3.) using ROC curve to determine the best cut-off point, based on the sensitivity, specificity, and positive predictive value. The "gold standard" for epilepsy status is determined by medical record review. The final model will be applied to the enrollment file to estimate the annual prevalence of epilepsy in a managed care organization, with a current enrollment of over 200,000 members. The advantages and disadvantages of this approach will be discussed in relationship to other studies that use computerized MIS to estimate the prevalence of a disease or disorder. Implications of the study results to the understanding of epilepsy will be discussed.