Performance of the recent Duke criteria versions in patients with suspected infective endocarditis and sufficient cardiac imaging: The more things change, the more they remain the same
Elisavet Stavropoulou, Maria Padrutt, Berit Siedentop, Pierre Monney, Michelle Frank, Georgios Tzimas, Nicolas Fourré, Virgile Zimmermann, Nicoleta Ianculescu, Laurence Senn, Lars Niclauss, Matthias Kirsch, Mathias Van Hemelrijck, Omer Dzemali, Benoit Guery, Barbara Hasse, Matthaios Papadimitriou-OlivgerisAbstract
Background
Even after the major revisions incorporated into the International Society for Cardiovascular Infectious Diseases (ISCVID) and 2023 European Society of Cardiology (ESC) Duke criteria, approximately one third of cases remain categorized as “possible” IE. We evaluated the diagnostic performance of the Duke criteria in a cohort of patients who underwent guideline-concordant multimodality cardiac imaging.
Methods
This retrospective study at two Swiss university hospitals (2014-2025), included adult patients with suspected IE who underwent guideline-concordant cardiac imaging. The reference standard was the final diagnosis established by the Endocarditis Team. Each episode was classified according to 2015 Duke-ESC, 2023 Duke-ISCVID, and 2023 Duke-ESC clinical criteria as definite, possible, and rejected IE. Diagnostic performance was assessed using two approaches: (1) definite IE versus possible/rejected IE, and (2) definite/possible IE versus rejected IE.
Results
Among 3,448 of suspected IE, 1,302 (38%) were adjudicated as IE. The proportion of episodes classified as possible IE remained high: 2015 Duke-ESC: 29%; 2023 Duke-ISCVID: 33%; and 2023 Duke-ESC: 29%. Sensitivity (definite IE vs possible/rejected IE) increased from 63% (95% CI 60-66%) with the 2015 Duke-ESC criteria to 74% (71-76%) with the 2023 Duke-ISCVID criteria and 71% (68-73%) with the 2023 Duke-ESC criteria. Specificity (definite/possible IE versus rejected IE) declined from 72% (70-74%) to 59% (57-61%) and 66% (64-68%), respectively.
Conclusions
In this large cohort of patients who underwent guideline-concordant imaging, both 2023 Duke criteria increased sensitivity but failed to reduce the persistent diagnostic gray-zone of “possible” IE, highlighting an ongoing limitation with direct clinical and research implications.