Computed tomographic scanning of cervical spine fractures: does it influence treatment?

Objective: To determine whether the superior sensitivity of computed tomography (CT) results in changes in treatment plans for cervical spine fractures that have been diagnosed on plain films alone.

Design: Retrospective review of radiographic studies for cervical spine trauma.

Setting/participants: An orthopaedic spine surgeon (SS), an orthopaedic traumatologist (OT), an orthopaedic spine fellow (SF), and an orthopaedic chief resident (CR) were independently presented thirty-nine cases of cervical spine trauma imaged with adequate plain radiographs and with CT.

Main Outcome Measures: Agreement was measured by calculation of kappa coefficients.

Results: The detection rate of total fractures on plain radiographs alone ranged from 47 percent to 71 percent, and the diagnosis changed an average 53 percent of cases. Change in treatment plans ranged from 10 percent (SS) to 46 percent (CR) of cases. Of these changes, undertreatment occurred as follows: SS =3 percent, OT =8 percent, SF =36 percent, and CR = 46 percent. The mean kappa coefficient for intraobserver agreement of treatment plans was 0.69. The experienced observers demonstrated "excellent" agreement with an average kappa coefficient of 0.85, whereas the mean coefficient for inexperienced observers was 0.54 or "moderate" agreement. Complete diagnostic agreement occurred between the experienced observers after review of both the plain films and CT scans. The interobserver agreement of treatment plans for the experienced observers increased from 0.79 to 0.88.

Conclusions: CT scanning afforded additional information for all observers. Experienced observers can reliably determine treatment plans for cervical spine trauma diagnosed on plain films alone, whereas inexperienced observers are less reliable. For the experienced observers, interobserver agreement on treatment plans increased after the addition of CT.