Inter-observer reliability of computed tomographic classifications of diverticulitis

The wide use of computed tomography (CT) for diverticulitis has initiated new classifications, but their reliability has not been studied. The aim of the study was to assess the inter-observer agreement of radiologists on the classifications of diverticulitis. A set of 100 CT examinations of patients with an episode of diverticulitis was used to assess inter-observer agreement. Cases were derived from two prospective trials and a retrospective cohort to comprise an evenly distributed case mix of patients. The reference standard was the modified Hinchey classification based on all available information. Three blinded radiologists independently read all CT examinations. We calculated the proportion of agreement and kappa values for the Hinchey and Ambrosetti classifications. For the Dharmarajan classification only inter-observer agreement was calculated. The agreement with the reference standard was substantial for both the modified Hinchey and the Ambrosetti classifications (P = 0.68 and P = 0.76). Overall inter-observer agreement for the modified Hinchey classification was substantial (median kappa 0.72), for the Ambrosetti classification almost perfect (median kappa 0.83) and for the Dharmarajan classification substantial (median kappa 0.76). The Ambrosetti classification is more reproducible than the modified Hinchey and Dharmarajan classifications. The Ambrosetti and modified Hinchey classifications have a substantial agreement with the reference standard and therefore produce a reliable classification. The Dhamarajan is applicable only in complicated diverticulitis and is an important complementary classification to the other more general classifications of diverticulitis