Abstracts

This study examines whether statewide administrative billing data can be used for surveillance of seizure and epilepsy cases in South Carolina (SC). A critical issue is whether seizure and epilepsy diagnoses are accurately coded in such a database. Inpatient and emergency department (ED) ICD-9-CM discharge codes indicating possible seizure or epilepsy are being reviewed for accuracy. We utilized a dataset of all 2001 inpatient and ED discharges from non-federal hospitals in SC. This database is housed at SC Office of Research and Statistics (ORS). A sample of approximately 35% of 345.x (epilepsy) diagnoses, 5% of 780.3 (convulsions), 1% of 780.2 (syncope and collapse), and 5% of 293.0 (acute delirium) was selected - approximately 3000 charts. If an individual had multiple visits, the highest-level diagnosis chronologically listed first was chosen. 2543 charts (85%) were abstracted by SC Department of Health [amp] Environmental Control. Data was de-identified at ORS and sent to the Medical University of South Carolina (MUSC). At MUSC epilepsy specialists reviewed the data for appropriateness of diagnostic coding according to International Classification of Epileptic Seizures from the International League Against Epilepsy. The data presented is based on information from 547 (22%) of the abstractions reviewed thus far. In 43% of the cases, the reviewer agreed with the listed seizure- or epilepsy-related code, in 47% the reviewer felt the code was incorrect, and in 10% of the charts there was insufficient information to determine correctness. Of the 47% incorrect, however, most (86%) had a 780.39 code (seizure not otherwise specified) and a history of seizures, but not enough information to characterize the type of epilepsy. It is felt that these could have been more properly coded as 345.9 (epilepsy, unspecified). Whether a seizure was new onset is critical in determining incidence. 17% were new onset, 73% had a history of seizures, and in 10% of the charts there was insufficient information to make a determination. Preliminary analysis appears to indicate an under diagnosis of epilepsy. While administrative data has limited information, it has the advantages of representing the entire state and ease of use. While the present data does not include outpatient visits, there are additional outpatient datasets ORS may access that represent approximately one-third of the state population. It is planned to include this data to estimate incident and prevalent cases of seizure and epilepsy in SC. This study will help determine an accurate estimate of the number of cases of seizure and epilepsy, which will inform public policy on the appropriate allocation of resources. (Supported by Cooperative agreement with CDC, # U36/CCU319276)