Opracowanie i weryfikacja systemu komputerowego do identyfikacji weteranów z epilepsją

BACKGROUND: Accurate epilepsy identification in large health care systems has the potential to improve health care delivery and resource allocation. This article summarizes the creation and validation of a 3-tiered algorithm to identify veterans with epilepsy (VWE) receiving care from the Veterans Health Administration (VHA) using administrative data. METHODS: A 3-tier algorithm was developed to identify patients with epilepsy utilizing diagnosis codes and prescription data. Tier 1 integrates seizure-specific diagnostic codes and antiseizure medication data. Tier 2 includes patients with inpatient visits. Tier 3 identifies untreated or less obvious cases by including patients with multiple outpatient visits. VHA administrative databases linked to the VHA Corporate Data Warehouse were used to identify VWE. Tier 1 validation was based on 625 patients and Tiers 2 and 3 validation was based on 300 total patients. Validation was conducted by expert epilepsy clinicians (epileptologists and a nurse care coordinator) comparing algorithm classifications against the International League Against Epilepsy definition of epilepsy to ascertain positive predictive values (PPVs). Annual trends for the number of VWE cases identified by the algorithm within the VHA are also presented. RESULTS: Tier 1 demonstrated a PPV of 85.1% (95% CI, 82.1%-87.8%). Tiers 2 and 3 offered broader identification and had lower PPVs: Tier 2 PPV was 61.9% (95% CI, 53.4%-70.4%) and Tier 3 PPV was 59.8% (95% CI, 52.5%-67.1%). CONCLUSIONS: By efficiently segmenting veterans based on reliable administrative data, this 3-tiered algorithm supports enhanced surveillance, targeted health care provision, and optimal resource utilization. Though it is tailored to the VHA, this algorithmic approach holds promise for broader application in health care systems facing similar epidemiologic and administrative challenges.