Re: “Comparison of 2 Approaches for Determining the Natural History Risk of Brain Arteriovenous Malformation Rupture”

We have with great interest read the article, ‘‘Comparison of 2 Approaches for Determining the Natural History Risk of Brain Arteriovenous Malformation Rupture,’’ by Kim et al. (1), published in the June 15, 2010, issue of the American Journal of Epidemiology. We agree with the authors that it is conceptually attractive to use age at the first arteriovenous malformation hemorrhage to assess the natural course of disease, and we published a paper using this concept in 1997 (2). However, as with all evaluations of retrospective data, the study by Kim et al. must be observed for the risk of selection bias. The use of life table statistics in this case is flawed with the lack of information from all patients that died from the initial hemorrhage, before reaching the hospital, and also in cases where the arteriovenous malformation is not diagnosed. We therefore preferred an analytical approach, using the age distribution at first hemorrhage rather than life table statistics to elucidate the annual risk for hemorrhage from these lesions (2). This has the advantage of being insensitive to the relation between arteriovenous malformations that have or have not bled. It can also be used to calculate the age dependence, if such exists. However, assessment of the annual risk is accurate only if the assumption of arteriovenous malformation’s being a congenital condition is correct. If so, there is an unambiguous relation between the age distribution of the age at first hemorrhage and the risk for hemorrhage. If not, the calculated risk is lower than the true annual risk for hemorrhage. Furthermore, it implies that an insignificant number of patients develop an arteriovenous malformation after the peak of the age distribution, that is, later than 25 years of age in our material. The most important finding in our study was the strong relation between age and risk for arteriovenous malformation rupture, the risk increasing with increasing patient age. It would be very interesting to combine the patient materials and further analyze the important question of the natural course of brain arteriovenous malformations.