Echocardiography in Staphylococcus aureus Bacteremia

To the Editor—In their recent article, Kaasch et al [1] propose a set of predictive criteria that, if absent, identify patients with nosocomial Staphylococcus aureus bacteremia (SAB) who are at low risk of infective endocarditis (IE) and in whom transesophageal echocardiography (TEE) may be unnecessary. This is based on post hoc analysis of 2 prospectively gathered cohorts of SAB. We submit 4 cautions regarding their findings. First, it is not clear whether the predictive criteria were chosen prospectively and evaluated using the cohort data, or whether they were chosen because they provided the best fit to the existing data. The second approach would clearly be subject to multiple testing bias, in terms of both detection of effects and measurement of their magnitude [2]. Second, neither the predictive criteria nor the data presented address other well-recognized risk factors for IE such as preexisting valvular disease or intravenous drug use. This may limit the