The sensitivity of carpal bone indices to rotational malpositioning.

We investigated the dependence of 20 radiographic carpal measurements (carpal indices) on rotational positioning errors in posteroanterior and lateral radiographs. The measurements were made from "true perspective" digitally reconstructed radiographs created from computed tomography data. Most posteroanterior indices were not affected by rotation. Carpal height, carpal height ratio, revised carpal height ratio, capitate-radius distance, and carpal ulnar translocation were particularly robust. Lateral-view indices involving the scaphoid were the most sensitive to simulated malpositioning: radioscaphoid, scapholunate, and scaphocapitate angles were reduced from 58 degrees, 48 degrees, and 56 degrees at true lateral to 30 degrees, 24 degrees, and 34 degrees, respectively, at 20 degrees external rotation. Observers were unable to estimate the degree of malpositioning accurately in either view. Our results support use of the "scaphopisocapitate" criterion for assessing correct positioning in lateral plain radiographs.