Hartz A J
Arch Pathol Lab Med. 1984 Jan;108(1):65-7.
The utility of a laboratory test for differential diagnosis depends on the degree that test results for different disease states overlap. I developed a new overlap index that ranges from zero, for an ideal test with no overlap, to one if the median test result is the same for the two disease states. The index does not depend on defining a normal range or estimating the percentage of observations that lie outside of the normal range. For these reasons it is superior to sensitivity and specificity for comparing laboratory tests, particularly if only limited test data are available.
实验室检查用于鉴别诊断的效用取决于不同疾病状态下检查结果的重叠程度。我开发了一种新的重叠指数,范围从0(适用于无重叠的理想检查)到1(如果两种疾病状态的检查结果中位数相同)。该指数不依赖于定义正常范围或估计超出正常范围的观察值百分比。基于这些原因,在比较实验室检查时,它优于敏感性和特异性,特别是在仅有有限检查数据可用的情况下。