[Skip to Content]
[Skip to Content Landing]
June 9, 1997

Analysis of ROC Curves Applied to the Diagnosis of Lyme Disease-Reply

Author Affiliations

Valhalla, NY

Arch Intern Med. 1997;157(11):1267. doi:10.1001/archinte.1997.00440320176020

This article is only available in the PDF format. Download the PDF to view the article, as well as its associated figures and tables.


Solomon has misinterpreted the column head "Any Positive Band" in our Table 2 as meaning a cut point in the test indicating a positive result when any 1 band is reactive. Receiver operating characteristic curves are calculated across all possible cut points; hence, the term any refers to all possible cut points (0, 1, 2, or 3... positive bands) among any of the measurable immunoblot bands, as opposed to all potential cut points of 3 specific bands (0, 1,2, or 3 positive bands). The ROC curve area is independent of the chosen cut point and, hence, does not change depending on the number of reactive bands considered to indicate a positive result. We agree that the test improves when the number of bands in the test is less restrictive. The appropriate cut point is that which results in the greatest sensitivity and specificity.

Although several statistical programs will easily calculate

First Page Preview View Large
First page PDF preview
First page PDF preview