Objectives: To determine whether relevant comorbidities stratify patients with and without IE and whether these may improve the diagnostic accuracy, in addition to the modified Duke criteria.

Methods and Results: 261 consecutive patients (aged 60.1 ± 16.1 years, 62.8% male) with suspected IE were prospectively included in this single-center observational trial. Modified Duke criteria and relevant comorbidities as well as clinical characteristics, were assessed. Forty-seven patients had IE, as confirmed by a clinical event committee. Patients with IE had a higher frequency of positive blood cultures (70.2% vs. 36.9%, p < 0.0001), embolic diseases (36.2% vs. 10.8%, p < 0.0001), heart murmurs (27.7% vs. 11.7%, p = 0.01), and intensive care therapy (74.5% vs. 58.4%, p = 0.04). In receiver operating characteristics, the combination of modified Duke criteria without transesophageal echocardiography led to an area under the curve of 0.783 (0.715-0.851). The predictive value was only marginally improved by the addition of heart murmur and intensive care therapy (0.794 [0.724-0.863]). In contrast, transesophageal echocardiography alone achieved an area under the curve of 0.956 (0.937-0.977) and was further improved when adding modified Duke criteria, heart murmur, and intensive care therapy (0.999 [0.998-1.000]).

Conclusion: Modified Duke criteria provide excellent diagnostic value for evaluating suspected IE, mainly driven by transesophageal echocardiography.