Observer variation in the assessment of root canal curvature

To evaluate the inter- and intra-observer agreement between training/trained endodontists regarding the ex vivo classification of root canal curvature into three categories and its measurement using three quantitative methods.

Periapical radiographs of seven extracted human posterior teeth with varying degrees of curvature were exposed ex vivo. Twenty training/trained endodontists were asked to classify the root canal curvature into three categories (<10°, 10-30°, >30°), to measure the curvature using three quantitative methods (Schneider, Weine, Pruett) and to draw angles of 10° or 30°, as a control experiment. The procedure was repeated after six weeks. Inter- and intra-observer agreement was evaluated by the intraclass correlation coefficient and weighted kappa.

The inter-observer agreement on the visual classification of root canal curvature was substantial (ICC = 0.65, P < 0.018), but a trend towards underestimation of the angle was evident. Participants modified their classifications both within and between the two sessions. Median angles drawn as a control experiment were not significantly different from the target values (P > 0.10), but the results of individual participants varied. When quantitative methods were used, the inter- and intra-observer agreement on the angle measurements was considerably better (ICC = 0.76-0.82, P < 0.001) than on the radius measurements (ICC = 0.16-0.19, P > 0.895).

Visual estimation of root canal curvature was not reliable. The use of computer-based quantitative methods is recommended. The measurement of radius of curvature was more subjective than angle measurement. Endodontic Associations need to provide specific guidelines on how to estimate root canal curvature in case difficulty assessment forms.