To the Editor.
—Several errors in the study by Ms Gallo and colleagues1 preclude this unblinded study's conclusions. The most important is the miscalculation of sensitivity and specificity and, therefore, predictive values for detection of HIV infection.As their mathematical basis for calculation, the authors cite Vecchio,2 who published methods for statistical calculation of a single diagnostic test, yet disregard his formula. According to Vecchio, sensitivity=(diseased persons with positive test/ all diseased persons tested)ϰ 100, which yields a sensitivity of 98.8% for the OraSure device using the authors' published data. Vecchio's calculation for specificity is likewise 99.6% for their tabulated data and 99.9% for their textual data. However, the authors ignored Vecchio's formula and used equivocal or indeterminate results as both true positives and true negatives to inflate their sensitivity and specificity data.
Frank AP. Usefulness of Oral Mucosal Transudate for HIV Antibody Testing. JAMA. 1997;277(20):1591-1592. doi:10.1001/jama.1997.03540440025016