Background The Duke criteria for diagnosing infective endocarditis (IE) were recently updated by the International Society for Cardiovascular Infectious Diseases (ISCVID) with the purpose of enhancing diagnostic specificity and sensitivity. This study investigated the performance of the Duke-ISCVID criteria in relation to previous criteria. Method This retrospective single centre cohort study included patients referred to the Lund IE Team (LIET) between 2017 and 2022, and compared episodes classified according to the 2000 modified Duke, European Society of Cardiology (ESC) 2015, and 2023 Duke-ISCVID criteria. The LIET's decision to treat episodes as IE served as the reference standard. Results The study cohort comprised 661 episodes of potential IE. The LIET classified 498 (75%) episodes as IE. The Duke-ISCVID criteria classified a slightly higher proportion of episodes as definite IE (56.4%) compared to the ESC 2015 criteria (54.9%), and the modified Duke (51.1%). The Duke-ISCVID criteria had higher sensitivity (72.5%) for definite IE compared to ESC 2015 (70.7%, p = .57) and modified Duke (66.1%, p = .033). The specificities were similar for the Duke-ISCVID criteria (92.6%), compared to ESC 2015 (93.9%, p = .83) and the modified Duke criteria (95.0%, p = .49). When considering both definite and possible IE positive, specificity for all criteria was low: 32.5% for the Duke-ISCVID criteria, 41.7% for ESC 2015 (p = .086), and 42.3% for modified Duke criteria (p = .11), while the sensitivity was not significantly lower. Discussion The Duke-ISCVID criteria demonstrated slightly improved sensitivity for diagnosing IE whereas specificity remained unchanged. Further refinement of the Duke-ISCVID criteria is needed to balance sensitivity and specificity.
Read full abstract