OBJECTIVE: To assess the Birmingham Vasculitis Activity Score for Wegener's Granulomatosis (BVAS/WG) with respect to its selection and weighting of items. METHODS: This study used the BVAS/WG data from the Wegener's Granulomatosis Etanercept Trial. The scoring frequencies of the 34 predefined items and any "other" items added by clinicians were calculated. Using linear regression with generalized estimating equations in which the physician global assessment (PGA) of disease activity was the dependent variable, we computed weights for all predefined items. We also created variables for clinical manifestations frequently added as other items, and computed weights for these as well. We searched for the model that included the items and their generated weights yielding an activity score with the highest R(2) to predict the PGA. RESULTS: We analyzed 2,044 BVAS/WG assessments from 180 patients; 734 assessments were scored during active disease. The highest R(2) with the PGA was obtained by scoring WG activity based on the following items: the 25 predefined items rated on >or=5 visits, the 2 newly created fatigue and weight loss variables, the remaining minor other and major other items, and a variable that signified whether new or worse items were present at a specific visit. The weights assigned to the items ranged from 1 to 21. Compared with the original BVAS/WG, this modified score correlated significantly more strongly with the PGA. CONCLUSION: This study suggests possibilities to enhance the item selection and weighting of the BVAS/WG. These changes may increase this instrument's ability to capture the continuum of disease activity in WG.