Figure 1

Classification tree analysis of 427 patients with antineutrophil cytoplasmic antibody-associated vasculitis in model 1. The dominant clinical feature (over 80%) is used to name each class. The observed number of individuals allocated to each class is shown in each column. The algorithm was initiated on the basis of ANCA positivity and was subsequently allocated according to organ involvement. Among 47 patients with PR3-ANCA, 34 were assigned to Cluster 1. Among 15 patients with ANCA negativity, 8 were assigned to Cluster 2. Among 49 patients with MPO-ANCA and ENT symptoms, 48 were assigned to Cluster 3 or 7. Among 43 patients with MPO-ANCA and skin symptoms, 34 were assigned to Cluster 4. The rest were assigned to predicted clusters by cluster analysis. The overall concordance rate was 93%. ENT ear, nose and throat, ILD interstitial lung disease, MPO-ANCA myeloperoxidase-antineutrophil cytoplasmic antibody, PR3-ANCA proteinase-3-antineutrophil cytoplasmic antibody.