Assessing the risk of infective endocarditis in Staphylococcus aureus bacteremia: a comparison of 4 scoring systems

Diagnosing infective endocarditis (IE) in patients with Staphylococcus aureus bacteremia (SAB) is crucial for appropriate management. Several scores have been developed to assess IE risk, guide the use of echocardiography and help to limit unnecessary echocardiography.

We retrospectively included SAB cases hospitalized between January 2017 and December 2022. IE was diagnosed using the modified Duke criteria. Four predictive scores of IE (PREDICT at day 1 and day 5, VIRSTA, POSITIVE and LAUSTAPHEN) were calculated and compared.

Among 887 SAB patients, 75 (8.4%) had IE and 725 (81.3%) underwent echocardiography. VIRSTA and LAUSTAPHEN both achieved 100% sensitivity and negative predictive value (NPV). PREDICT D5 and POSITIVE had sensitivities of 92% and 84%, and NPVs of 98.3% and 98.1%, respectively. AUC ROC curve analysis showed that VIRSTA (90.4) and POSITIVE (89.4) had the highest discriminatory ability. The application of these scoring systems would have profoundly altered echocardiography utilization patterns, leading to up to a 389.1% increase in TEE use (VIRSTA) and a 95.4% reduction in TTE use (PREDICT D1) compared with observed clinical practice.

In our cohort, the VIRSTA score reliably identified patients with S. aureus bacteremia (SAB) at low risk of infective endocarditis (IE). The LAUSTAPHEN score demonstrated comparable performance, warranting further validation. However, even with highly discriminative models, limited specificity may lead to overuse of transesophageal echocardiography (TEE), suggesting that its use should be guided by case-by-case clinical judgment rather than by scoring systems alone.