Diagnostic performance of three versions of the Duke criteria (2015 Duke-ESC, 2023 Duke-ISCVID, 2023 Duke-ESC) for infective endocarditis (IE) was compared in 3,448 adult episodes of suspected IE at two Swiss university hospitals (2014–2025) where guideline-concordant multimodality cardiac imaging had been performed.
Both 2023 criteria improved sensitivity for definite IE (2015 Duke-ESC: 63%; 2023 Duke-ISCVID: 74%; 2023 Duke-ESC: 71%) but at the cost of lower specificity (72% → 59% and 66%, respectively); approximately one-third of cases remained in the diagnostic gray-zone of 'possible' IE across all three criteria versions.
Retrospective design at two Swiss tertiary centers limits generalizability; the reference standard (Endocarditis Team adjudication) is itself subject to expert bias; the cohort is enriched for patients who already received guideline-concordant imaging, so performance may differ in lower-resource or lower-acuity settings.
Even with optimal multimodality cardiac imaging, the 2023 Duke criteria updates do not resolve the persistent 'possible' IE category (~29–33% of cases), so clinicians should not rely on criteria classification alone and must continue to exercise individualized clinical judgment for diagnostically uncertain cases.
Explore related topics