Accuracy and interobserver agreement for determinations of rabbit posterolateral spinal fusion.

Abstract

The accuracy and interobserver agreement of fine detail radiography and computed tomography (CT) determination of spinal fusion were evaluated in an established animal spine fusion model. To determine the accuracy and interobserver agreement of radiographic determinations of spinal fusion in rabbit posterolateral spine fusion. The rabbit posterolateral intertransverse process spine fusion model is an established animal model for evaluating bone graft alternatives for spinal fusion. However, little is known regarding the accuracy and interobserver agreement of radiographic determinations of spondylodesis in this model. Forty-two New Zealand White rabbits underwent posterolateral spinal fusion. The animals were killed at 9 weeks and the lumbar spine harvested. Manual manipulation, fine detail radiography, and CT images were used to assess spinal fusion. Using manual palpation testing as the standard by which to assess fusion, there was high sensitivity and negative predictive value for both radiographic methods. Positive predictive value, however, was poor (26% fine detail radiography, 61% CT scan). CT correlated better with manual palpation testing when compared with fine detail radiographs. There was substantial interobserver agreement of successful fusion using CT scan imaging (kappa = 0.63) and moderate interobserver agreement radiographs (kappa = 0.52). Both radiographic techniques used in the study recorded high sensitivity and negative predictive value. However, positive predictive value was poor, especially with fine detail radiographs. Nevertheless, CT with reformatted images did appear to be superior to fine detail radiographs in accurately identifying nonunions in this animal model.