Background The Birmingham Vasculitis Activity Score (BVAS) is validated to assess disease activity in systemic vasculitis, but has not been widely tested in conditions mimicking vasculitis.

Objectives To explore the utility of BVAS in screening patients with suspected vasculitis.

Methods We analysed data from the ACR/EULAR Diagnostic and Classification Criteria in Vasculitis Study (DCVAS), using the submitting physician's diagnosis (certainty >75%) as the gold-standard, with predetermined criteria to identify appropriate comparators for each vasculitis. BVAS score and frequency of individual items were evaluated in: giant cell arteritis (GCA); Takayasu arteritis (TAK); polyarteritis nodosa (PAN); and the anti-neutrophil cytoplasm antibody (ANCA) associated vasculitides (AAV) comprising granulomatosis with polyangiitis (GPA), eosinophilic GPA (EGPA) and microscopic polyangiitis (MPA); and their comparator conditions.

Results 1122 patients with vasculitis and 487 comparators were analysed. The BVAS score (median (IQR)) was significantly higher (p<0.0001) in patients with each type of AAV vs comparators: GPA 19 (12–25) vs 11 (6–17); EGPA 18 (12–26) vs 11 (6–17); MPA 19 (14–24) vs 11 (5–17). Between 5–12 individual BVAS items were found significantly more frequently in each of these vasculitides than their comparators: bloody nasal discharge/nasal crusts, sinus involvement, pulmonary infiltrate, nodules/cavities in lung, proteinuria, haematuria for GPA; pulmonary infiltrate, wheeze, sensory peripheral neuropathy, mononeuritis multiplex, heart involvement (pericarditis, ischaemic pain, cardiomyopathy or congestive failure) for EGPA; fever, bloody nasal discharge/nasal crusts, pulmonary infiltrate, proteinuria, haematuria and any renal function abnormalities for MPA. These items were used to develop classification trees whose performance is shown in the table. We analysed whether including an ANCA result in the trees altered their performance (table). This led to an increase in sensitivity for MPA (from 71% to 98%), whereas only slight changes were recorded for GPA and MPA. For PAN, GCA and TAK there was no significant difference in BVAS score compared to comparators, and few or no BVAS items were discriminative.

Conclusions BVAS consists of several items highly specific for GPA, EGPA and MPA, from which we have developed classification trees which can facilitate accurate and systematic screening of patients with suspected AAV based on clinical parameters alone. Whilst the ANCA result contributed to the classification of MPA, it had little additional value in distinguishing GPA or EGPA from their comparator conditions, limiting the value of ANCA testing for the purpose of screening for AAV. By contrast, BVAS is not effective in distinguishing PAN, GCA or TAK from comparator conditions.

Acknowledgement Italian and British societies of rheumatology for supporting first author.

Disclosure of Interest A. Floris Grant/research support from: Bursary by Italian Society of Rheumatology, N. Goodfellow: None declared, J. Sznajd: None declared, K. Wawrzycka-Adamczyk: None declared, H. Querin: None declared, A. Craven: None declared, J. Rosa: None declared, P. Merkel: None declared, R. Watts: None declared, R. Luqmani Grant/research support from: GSK, Nordic, Medimmune, Chemocentryx, Roche, UCB, Consultant for: GSK, Nordic, Medimmune, Chemocentryx