Computed Tomography Observer Agreement in Staging Malignant Lymphoma

To determine pretreatment computed tomography observer agreement in patients with newly diagnosed lymphoma. Forty-nine computed tomography scans were reviewed by 3 experienced radiologists, with each scan assessed twice by 1 observer. Predefined nodal and extranodal regions were assessed, and Ann Arbor stages were assigned. K-statistics were defined as poor (κ < 0.2), fair (κ > 0.2 to κ ≤ 0.4), moderate (κ > 0.4 to κ ≤ 0.6), substantial (κ > 0.6 to κ ≤ 0.8), and almost perfect (κ > 0.8 to κ ≤ 1). Nodal interobserver agreement varied from 0.09 for infraclavicular involvement to 0.95 for para-iliac involvement; intraobserver agreement was substantial to almost perfect, except for infraclavicular nodes. Extranodal interobserver agreement varied from 0.56 to 0.88; intraobserver agreement was substantial to almost perfect. Ann Arbor stage interobserver agreement varied from 0.57 to 0.69; intraobserver agreement was substantial. Computed tomography observer agreement in staging malignant lymphoma appears to be suboptimal.