Arthrography: 5 clinical studies (251 joints) and 5 autopsy studies (208 joints: 179 anterior and 29 sideways or rotational displacement) were used to evaluate diagnostic outcome; 1 study (30 joints) was used to assess observer performance.
CT: 7 clinical studies (377 joints) and 2 autopsy studies (30 joints) were used to evaluate diagnostic outcome.
MRI: 7 clinical studies (293 joints) and 3 autopsy studies (55 joints: 15 anterior and 40 sideways or rotational displacement) were used to evaluate diagnostic outcome; 1 study (unspecified number of joints) was used to assess observer performance.
Arthrography.
The mean values from clinical studies were sensitivity 0.94, specificity 0.94, PPV 0.98 and NPV 0.73. The mean values from autopsy studies were sensitivity 0.9, specificity 0.8, PPV 0.88 and NPV 0.82. In terms of observer performance (7 observers), intra and inter-observer agreement gave kappa indices from 0.30 to 0.95 and from 0.54 to 0.68, respectively.
CT.
The mean values from clinical studies were sensitivity 0.97, specificity 0.76, PPV 0.97 and NPV 0.86. The mean values from autopsy studies were sensitivity 0.66, specificity 0.68, PPV 0.66 and NPV 0.74.
MRI.
The mean values from clinical studies were sensitivity 0.92, specificity 0.83, PPV 0.98 and NPV 0.4. The mean values from autopsy studies were sensitivity 0.81, specificity 0.87, PPV 0.82 and NPV 0.88. Intra- and inter-observer performance agreement gave kappa indices of 0.93 and 0.88, respectively.
After weighting.
For anterior position, the mean positive likelihood ratio was 4.5 for arthrography, 2.3 for MRI and 2.1 for CT; the mean negative likelihood ratios were 0.13, 0.22 and 0.5, respectively. For sideways and rotational displacements, the mean positive likelihood ratio was 3.8 for arthrography and 6.2 for MRI; the mean negative likelihood ratios were 0.43 and 0.22, respectively.