Figure 3b: GCPs for diagnostic performance of (a) ultrasonography (US) and (b) CT for atypical acute appendicitis (39). Solid line = positive test, dashed line = negative test. For a given pretest probability, posttest probability of a positive or negative test is derived by drawing a perpendicular line up to the solid line or dashed line, respectively, and across to the y-axis. Disease present = 1.0 (100%), disease absent = 0.0 (0%). (a) Performance of US, with summary statistics from six studies (CEBM level 4 evidence [25,36,38]). If the emergency room physician is operating at a point of maximum clinical uncertainty (0.50 pretest probability), posttest probability of a positive test is >0.9 and confirms diagnosis. Posttest probability of a negative test is 0.10; this might be adequate to allow the patient to be observed. However, if the clinical data set favors acute appendicitis or intraabdominal sepsis (appendicitis likely, 0.75 pretest probability), posttest probability of a negative test is 
0.25—too high to dismiss. (b) Performance of CT, with summary statistics from four studies (level 4 evidence). Again, if the test is positive, appendicitis is ruled in. If the physician is operating at maximum clinical uncertainty, posttest probability of a negative test is <0.10, and if pretest probability is 0.75, posttest probability is <0.20. This is not perfect, but it is a better test performance than that with US. With very high pretest probability, further investigation may be indicated (eg, laparoscopy) even if the CT scan is negative. Images from both tests appear easier to interpret correctly when positive. Clinical information is needed for optimal interpretation of negative US and CT scans.