New diagnostic efficiency statistics: Comparative sensitivity and predictive/prevalence ratio

A major issue in evaluating the validity and utility of a criterion for a target diagnostic syndrome requires that its efficiency be compared to those of other diagnoses. A criterion may prove quite sensitive for diagnosis A, but this level may turn out to be a lesser magnitude than for diagnosis B and C. A new statistic, that of 'comparative sensitivity', introduces a formula for comparing the relative diagnostic sensitivity of a criterion when several diagnostic categories are being evaluated simultaneously. Another issue of concern relates to the positive predictive power of diagnostic criterion; the significance and utility of this power is a function largely of the prevalence or base-rate of the diagnostic syndrome for which it is intended to serve as an index. Accordingly, a second new statistic, that termed the 'predictive/prevalence ratio' is presented; it provides a standard gauge for evaluating the effective utility of the PPP statistic. The logic, methodology, and several examples are presented for these new statistics.