CT, MRI prone to false-positive, false-negative interpretations
Mallee W. J Bone Joint Surg. 2011. doi:10.2106/JBJS.I.01523.
CT and MRI had similar diagnostic characteristics, and both were better at excluding scaphoid fractures than confirming them, according to researchers.
Wouter Mallee, MSc, of Academic Medical Center of Amsterdam in The Netherlands, and colleagues followed 34 consecutive patients with a suspected scaphoid fracture. The patients had scaphoid tenderness and normal radiographic findings after a fall on the outstretched hand. They underwent CT and MRI within 10 days of injury.
The reference standard for a true scaphoid fracture is 6-week follow-up radiographs taken in four views, the authors wrote. A panel of surgeons and radiologists agreed on a diagnosis for each imaging type. They viewed each image in a randomly ordered, blinded fashion, independent of the other imaging types.
According to the reference standard, there were six true fractures. CT identified five fractures, with one false-positive, two false-negative and four true-positive results. MRI identified seven fractures, with three false-positive, two false-negative and four true-positive results.
For the CT, sensitivity was 67%, specificity was 96% and accuracy was 91%. For MRI, sensitivity was 67%, specificity was 89% and accuracy was 85%. These differences were not significant, according to the McNemar test for paired binary data. According to the Bayes formula, the positive predictive value was 0.76 for CT and 0.54 for MRI. The negative predictive value was 0.94 for CT and 0.93 for MRI.
“CT and MRI had comparable diagnostic characteristics,” the authors wrote. “Both were better at excluding scaphoid fractures than they were at confirming them, and both were subject to false-positive and false-negative interpretations. The best reference standard is debatable, but it is now unclear whether or not bone edema on MRI and small unicortical lines on CT represent a true fracture.”