Reliability of the radiographic variables in the International Spinal Cord Injury Spinal Column Injury Basic Data Set compared with the AO classification.

Abstract

Intra- and interrater reliability study for radiological variables of the International Spinal Cord Injury (SCI) Spinal Column Injury Basic Data Set. To test reliability of the radiological variables in the International SCI Spinal Column Injury Basic Data Set and compare it with the Arbeitsgemeinschaft für Osteosynthesefragen (AO) classification. The database of Eastern Denmark Regional SCI Referral Center, Copenhagen, Denmark. Ratings of the International SCI Spinal Column Injury Basic Data Set radiological variables and AO classification were obtained by two international observers for all the surgically treated spine trauma patients between 1st October 2010 and 31st December 2012 at the Spine Unit, Rigshospitalet, Denmark. Statistical analyses for intra- and interrater crude agreement and Cohen's unweighted kappa (κ) coefficients were performed. For 283 spine injuries, the intra- and interrater reliability for the individual radiological variables of the International SCI Spinal Column Injury Basic Data Set was at least substantial (κ=0.67-0.97 for interrater, κ=0.79-0.89 for the intrarater agreement). For the AO classification, intrarater reliability was moderate-to-substantial (κ=0.57-0.75), whereas interrater reliability was substantial (κ=0.67-0.69). The crude intra- and interrater agreement for a combined radiographic SCI Spinal Column Injury Basic Data Set variable showed no significant difference compared with the AO classification (P=0.067-0.895). The reliability of International SCI Spinal Column Injury Basic Data Set radiological variables is comparable to the AO classification system. We encourage its use for spinal column injury description, thus facilitating data collection and comparison between centres and countries.