Page 236 - Libro vascular I

App-B.qxd 29~8~04 14:58 Page 227
Appendix B
Sensitivity and specificity
227
When a new test is developed, a threshold at which the results are considered to indicate the presence or absence of the disease has to be selected. To do this the results of the test are compared to another method of detecting the disease, often known as ‘the gold standard’. Unfortunately, there is usually an overlap between the results obtained in the pres- ence of the disease with the results obtained in the absence of disease. This means that some patients will be falsely diagnosed as having the disease whereas others will be falsely diagnosed as not having the dis- ease. Selecting the best threshold for the test depends partly on the prevalence of the disease within the group tested and also on the consequence of a false classification (i.e., missing the disease or incorrectly diagnosing disease in a normal patient). The per- formance of a diagnostic test can be measured by its sensitivity and specificity, which are defined as follows:
Number of patients with the disease correctly diagnosed Number of patients with the disease
Number of patients without the disease correctly diagnosed Number of patients without the disease
The value of the sensitivity and specificity will vary as the threshold of the test is altered. For example, if the threshold is changed to improve sensitivity, the specificity will fall. Altering the threshold and
Sensitivity
Specificity