Use of ICD-9 coding for estimating the occurrence of cerebrovascular malformations

BACKGROUND AND PURPOSE: Accurate epidemiologic data concerning cerebrovascular malformations are scarce. Our goals were to determine the distribution of lesions in the International Classification of Diseases, Ninth Revision, (ICD-9) code for cerebrovascular malformations and to evaluate the use of state discharge registries for estimating their detection rate. METHODS: We reviewed records of all patients discharged from our center between January 1, 1992, and June 30, 1999, whose diagnoses included the ICD-9 code for cerebrovascular anomaly (code 747.81) to determine the accuracy of the coding. Hospital admission rates for cerebrovascular anomaly were calculated by using the 1995-1999 state discharge databases of California and New York. RESULTS: Of 804 patients with this code, 706 (88%) had a lesion consistent with the diagnosis. Five lesions accounted for 99% of the diagnoses; the two most common were AVM (66%) and cavernous malformation (13%). The ratio of AVMs to all cerebrovascular anomalies was similar to that in a prior population-based study. The sensitivity of identifying a patient with cerebrovascular malformation by using ICD-9 coding was 94%; the false-positive rate was 1.7 cases per 100,000 person-years. For California and New York, rates of first hospital admission for cerebrovascular malformation were 1.5 and 1.8 cases per 100,000 person-years, respectively. CONCLUSION: Rates of admission for cerebrovascular malformations calculated from state discharge databases are consistent with disease detection rates in the range of 1 case per 100,000 person-years. However, the false-positive rate for coding is in the same range as the disease detection rate. Thus, current state discharge registries cannot serve as sources of detailed epidemiologic data.