Inter- and intraobserver reproducibility in radiographic diagnosis and classification of femoral trochlear dysplasia

Dejour's radiographic criteria are commonly used to diagnose and assess femoral trochlear dysplasia in case of patello-femoral instability. The aim of this study was to establish the intra- and interobserver reliability of these radiographic criteria. Sixty-eight lateral knee radiographs were examined independently by 7 observers (2 juniors, 5 seniors) to assess interobserver agreement, and the 2 juniors repeated the observations to test intraobserver agreement. These 68 true lateral views were harvested from clinical records of 64 patients who underwent a trochleoplasty because of patellofemoral instability. To evaluate the agreement on analytic data (morphologic type of trochlea) we used the kappa statistical method, and to evaluate the agreement on numerical data (depth and prominence of the trochlear groove) we used interclass correlation analysis. The "crossing sign" (between the trochear groove and the anterior aspect of both condyles) was reliable since the probability of rating as normal a pathologic trochlea was only 3.1% (0 to 8.8%). In classifying trochlear morphology interobserver agreement was slight (kappa = 0.17) and intraobserver agreement was fair (kappa = 0.3). On the other hand, the measurements of the depth and prominence of the trochlear groove were more reliable since the interclass coefficients between observers were 0.62 and 0.38 respectively. The most frequent interobserver error was related to misdiagnosis of type II. To clarify Dejour's criteria we propose a diagnosis of type II only when 5 mm or more are measured between the intersections with the medial and lateral femoral condyles. We recommand the use of the prominence of the trochlear groove to evaluate the grade of bony trochlear dysplasia.