RESUMO
BACKGROUND: Revised diagnostic criteria for infective endocarditis (IE), the 2023 Duke-ISCVID criteria, were recently presented and need validation. Here, we compare the 2000 modified Duke criteria for IE with Duke-ISCVID among patients with bacteremia and relate the diagnostic classification to IE treatment. METHODS: We reanalyzed patient cohorts with Staphylococcus aureus, Staphylococcus lugdunensis, non-ß-hemolytic streptococci, Streptococcus-like bacteria, Streptococcus dysgalactiae, Enterococcus faecalis, and HACEK (Haemophilus, Aggregatibacter, Cardiobacterium, Eikenella, Kingella) bacteremia. Episodes were classified as definite, possible, or rejected IE with the modified Duke and Duke-ISCVID criteria. Reclassification included the microbiology criteria, positron emission tomography-computed tomography, and cardiac implanted electronic devices. To calculate sensitivity, patients treated for IE were considered as having IE. RESULTS: In 4050 episodes of bacteremia, the modified Duke criteria assigned 307 episodes (7.6%) as definite IE, 1190 (29%) as possible IE, and 2553 (63%) as rejected IE. Using the Duke-ISCVID criteria, 13 episodes (0.3%) were reclassified from possible to definite IE, and 475 episodes (12%) were reclassified from rejected to possible IE. With the modified Duke criteria, 79 episodes that were treated as IE were classified as possible IE, and 11 of these episodes were reclassified to definite IE with Duke-ISCVID. Applying the decision to treat for IE as a reference standard, the sensitivity of the Duke-ISCVID criteria was 80%. None of the 475 episodes reclassified to possible IE were treated as IE. CONCLUSIONS: The Duke-ISCVID criteria reclassified a small proportion of episodes to definite IE at the expense of more episodes of possible IE. Future criteria should minimize the possible IE group while keeping or improving sensitivity.