Reliability of classification systems for subaxial cervical injuries

STUDY DESIGN  Interobserver and intraobserver reliabilityObjective: To measure and compare the interobserver and intraobserver reliability of the cervical spine injury severity score (CSISS), the subaxial injury classification (SLIC) and severity scale, and the Allen-Ferguson system in patients with subaxial cervical spine injuries presenting to the emergency department. METHODS  Five examiners independently reviewed c-spine x-rays (CT/MRI) of 50 consecutive patients with subaxial cervical-spine injuries. They classified each case using CSISS, SLIC, and the Allen-Ferguson system. Examiners also documented if they believed the case required surgical management. At least 6 weeks later, the above steps were repeated for ten randomly chosen cases. RESULTS  The interobserver and intraobserver reliability for the total CSISS and total SLIC score are excellent. There is poor interobserver reliability and excellent intraobserver reliability when a total kappa score is calculated using all 21 groups for the Allen-Ferguson system. With respect to surgical management decisions, the interobserver agreement is moderate and the intraobserver agreement is excellent. CONCLUSIONS  There is no universally accepted classification scheme for subaxial cervical-spine injuries. A useful classification system must have excellent reliability to consistently and accurately describe injury patterns between different observers and allow for comparison across systems or cohorts. Both the CSISS and the SLIC and severity scale are promising classification systems with excellent interobserver and intraobserver reliability. Future studies will need to determine if their quantitative scores correlate with management and clinical outcomes.