Chronic granulomatous disease presenting as severe sepsis due to Burkholderia gladioli.

words, the sensitivity of the Duke criteria actually appears to be better than that of the Beth Israel criteria modified with the addition of echocardiographic findings. Furthermore, modification of the Beth Israel criteria may result in the use of various unvalidated and possibly incomparable diagnostic classifications. Another point raised by Martos-Perez et a1. is that of the specificity of the Duke criteria. As previously stated [2], I agree that the Duke criteria should not be substituted for the Beth Israel criteria until their specificity has been evaluated. We initiated such an evaluation in a retrospective study of 100 patients with suspected IE, and we found that the specificity of the Duke criteria was 99% (95% CI, 97%-100%) [3]. It would be interesting to evaluate specificity of both the Duke