963. External Validation of the 2023 Duke - International Society for Cardiovascular Infectious Diseases (ISCVID) Diagnostic Criteria for Infective Endocarditis (IE)

Abstract Background The 2023 Duke-ISCVID criteria for IE were recently introduced to improve classification of IE for research and clinical purposes (Table 1). We compared the diagnostic accuracy of the new criteria with the 2000 modified Duke criteria, the 2015 European Society of Cardiology (ESC) criteria and a clinically adjudicated reference standard by a panel of international experts on IE.Table 1:Overview of changes in criteria from 2000 modified Duke to 2023 Duke-ISCVIDAdapted from: Fowler et al, Clinical Infectious Diseases, 2023 Methods The study population consisted of consecutively enrolled patients suspected to have IE and discussed by the IE team of the Amsterdam University Medical Center (AMC) between October 2016 and March 2021. A panel of international experts on IE assigned a final diagnosis, which served as the reference standard. We compared the definite classification of the 2023 Duke-ISCVID criteria against this reference standard. We also evaluated accuracy of the new criteria, excluding data obtained from cardiac surgery and pathology (“Clinical Criteria”). Lastly, we quantified the added value of the proposed changes to the criteria, and compared the 2023 Duke-ISCVID to previous criteria using the McNemar test. We performed sensitivity analyses in patients who underwent cardiac surgery and patients who underwent TEE. Results 595 patients with suspected IE were included, of whom 399 (67%) were adjudicated as having IE. 111 patients (19%) had prosthetic valve IE, 48 (8%) had cardiac implantable electronic device IE. The 2023 Duke-ISCVID criteria were more sensitive than the modified Duke or ESC criteria (84.2 vs 74.9 vs 80%,respectively, p < 0.001 for ISCVID vs ESC criteria) without significant loss of specificity (Table 2). After excluding cardiac surgery/pathology results, the 2023 Duke-ISCVID Clinical Criteria also had significantly better sensitivity than the modified Duke and ESC criteria, again without losing specificity. The changes in the 'major microbiological' and 'imaging' criteria had the most impact (Figure 1). Findings were robust in sensitivity analyses (Table 3). When classifying ‘definite’ and ‘possible’ IE as a positive test, the 2023 Duke-ISCVID criteria had 99% sensitivity compared to the reference standard, but specificity decreased to 21%.Table 2:Diagnostic accuracy of modified Duke, ESC and Duke-ISCVID criteria Sensitivity and specificity are against the reference standard (adjudication panel). The Clinical Cri