Stroke: Who's counting what?

Introduction: Patients with stroke are often selected for epidemiological r eporting and research using ICD-9-CM (ICD-9) diagnostic codes. This study a ddresses the accuracy of these codes in identifying patients with stroke. M ethods: A sample of 279 patients with new stroke and 392 non-stroke (no evi dence of new stroke) patients were identified by medical record review from II Veterans Affairs Medical Centers. Administrative records containing ICD -9-CM diagnoses were matched with this sample. Coding sensitivity and speci ficity were determined using individual ICD-9 codes and two coding algorith ms. Results: Significant variation was found in the accuracy of cerebrovasc ular ICD-9-CM codes in identifying patients diagnosed with stroke. Two codi ng algorithms were identified with the following performance statistics bas ed on the sampled populations: 1) 91-percent sensitivity, 40-percent specif icity; and 2) 54-percent sensitivity, 87-percent specificity. Discussion/Co nclusions : Variability in identifying patients with stroke using ICD-9 cod es has been reported in the literature and confirmed. Two coding algorithms for maximizing sensitivity or specificity are proposed. Caution is urged w hen using ICD-9-coded administrative data to identify patients with stroke.