Reliability of surgeon-reported MRI findings to a national spine register.

Abstract

Spine registries contribute to valuable knowledge and research; however, the data quality has been questioned. MRI findings are crucial for diagnostics and grading of degenerative spinal disorders. We aimed to explore the reliability of surgeon-reported MRI findings in a national spine registry (NORspine). We assessed the reliability of MRI findings from three spine centres. Two spine surgeons re-examined previously surgeon-reported MRI findings for a sample of NORspine patients. We assessed the inter-rater reliability and the reliability between the NORspine registry and each study rater by Cohen's Kappa (ƙ). Two spine surgeons reassessed preoperative MRI of the lumbar spine for 276 previously enrolled NORspine patients; 92 at each treating centre equally distributed by three categories of spinal procedures: removal of disc herniation (LDH), decompression of lumbar spinal stenosis (LSS), and lumbar fusion. The inter-rater reliability varied from fair (0.21) to substantial (0.75) (most reliable for detecting LDH and LSS). The reliability between the NORspine registry and each rater varied from slight (0.13) to substantial (0.75). The highest reliability was found for LDH and LSS (ƙ = 0.72-0.75), while degenerative disc (DDD), foraminal stenosis (FS) and Modic changes had lower reliability (ƙ = 0.27-0.49). The reliability for surgeon-reported MRI diagnostics to the NORspine registry varied and was substantial for LDH and LSS, moderate for DDD and FS, and slight or fair for Modic changes.