The epidemiology of brain arteriovenous malformations

OBJECTIVE: Common estimates of the prevalence rate for pial arteriovenous malformations (AVMs) of the brain vary widely, and their accuracy is questionable. Our objective was to critically review the original sources from which these rates were derived and to establish best estimates for both the incidence and prevalence of the disease. METHODS: We reviewed all of the relevant original literature: autopsy series, the Cooperative Study of Intracranial Aneurysms and Subarachnoid Hemorrhage and related analyses, and other population-based studies. We also modeled the confidence intervals of estimates for a process of low prevalence such as AVMs. RESULTS: Many of the prevalence estimates (500-600/100,000 population) were based on autopsy data, a source that is inherently biased. Other estimates (140/100,000 population) originated from an inappropriate analysis of data from the Cooperative Study. The most reliable information comes from a population-based study of Olmsted County, MN, but prevalence data specific to AVMs was not found in that study. CONCLUSION: The estimates for AVM prevalence that are published in the medical literature are unfounded. Because of the rarity of the disease and the existence of asymptomatic patients, establishing a true prevalence rate is not feasible. Owing to variation in the detection rate of asymptomatic AVMs, the most reliable estimate for the occurrence of the disease is the detection rate for symptomatic lesions: 0.94 per 100,000 person-years (95% confidence interval, 0.57-1.30/100,000 person-years). This figure is derived from a single population-based study, but it is supported by a reanalysis of other data sources. The prevalence of detected, active (at risk) AVM disease is unknown, but it can be inferred from incidence data to be lower than 10.3 per 100,000 population.