Which patients with possible endocarditis should be treated as if the diagnosis were definite? An international modified Delphi study

OBJECTIVES: To identify which combinations or numbers of diagnostic criteria in possible infective endocarditis (IE) are associated with the decision to treat as IE, that is, to give a full course of endocarditis antibiotic treatment. METHODS: In a two-round modified Delphi study, international IE experts rated 47 standardised vignettes covering possible IE, of which 39 formed the basis of the main analysis. On a nine-point Likert scale, they indicated whether they would treat the patient as if he or she had IE. Consensus in favour required ≥70% of ratings at 7-9 and ≤15% at 1-3; consensus against was defined symmetrically. Associations with criteria were explored by modified Poisson regression. RESULTS: Of 137 experts, 103 completed round 1 (84 [83%] from Europe; 71 [69%] infectious disease specialists) and 98 of 102 round 2. Consensus was reached for 24 of 39 vignettes (62%): in favour of treating as IE in 22 (56%) and against in 2 (5%). Consensus in favour rose with accumulating criteria: 33% with minor criteria only, 45% with one major and one minor, and 74% with one major and two minor. Fifteen vignettes (38%) reached no consensus. In an exploratory model, consensus in favour was most strongly associated with typical (adjusted risk ratio [aRR] 12.40, 95% CI 2.11-72.93) and consistent (aRR 8.27, 1.17-58.54) major microbiological criteria and with major imaging criteria (aRR 7.37, 1.62-33.51). The criteria associated with consensus differed by valve type: imaging in native-valve and microbiological in bioprosthetic-valve vignettes. CONCLUSIONS: Expert consensus on treating possible IE as IE was reached in 62% of scenarios, almost exclusively in favour of treatment, and was most strongly associated with major criteria. These findings describe current expert views under uncertainty and do not establish the clinical benefit of treating possible IE as definite.