Pyoderma gangrenosum (PG) is an inflammatory condition characterized by chronic cutaneous ulcerations. There are three proposed PG diagnostic frameworks (Su, PARACELSUS, Delphi); however, they lack consensus, and their performance has not yet been validated in a well-defined cohort of patients with PG. In this cross-sectional retrospective cohort study, we sought to evaluate and compare the concordance of these diagnostic frameworks within a single-institution cohort of patients with PG. There were 47 patients from an initial 76 identified by International Classification of Diseases-9 and/or International Classification of Diseases-10 codes, where two PG experts agreed in their diagnosis of PG on the basis of clinical descriptions, photographs, and pathology. This group was the PG cohort by which we evaluated the performance and concordance of the diagnostic frameworks. The PARACELSUS score identified the highest proportion of patients with PG (89% [42 of 47 patients]), followed by Delphi and Su criteria, each at 74% (35 of 47 patients). Assessment of multirater agreement found that the three criteria agreed in their diagnoses for 72% of the patients (95% confidence interval = 60-85%); chance-adjusted agreement was determined to be 0.44 (95% confidence = 0.16-0.68, Fleiss' kappa). Future research should seek to refine these diagnostic frameworks and identify targeted methods of testing to reduce rates of PG misdiagnosis and patient misclassification in clinical trials.
Copyright © 2020 The Authors. Published by Elsevier Inc. All rights reserved.