Degenerative findings in lumbar spine MRI: an inter-rater reliability study involving three raters

Abstract

Abstract Background For diagnostic procedures to be clinically useful, they must be reliable. The interpretation of lumbar spine MRI scans is subject to variability and there is a lack of studies where reliability of multiple degenerative pathologies are rated simultaneously. The objective of our study was to determine the inter-rater reliability of three independent raters evaluating degenerative pathologies seen with lumbar spine MRI. Methods Fifty-nine people, 35 patients with low back pain (LBP) or LBP and leg pain and 24 people without LBP or leg pain, received an MRI of the lumbar spine. Three raters (one radiologist and two chiropractors) evaluated the MRIs for the presence and severity of eight degenerative spinal pathologies using a standardized format: Spondylolisthesis, scoliosis, annular fissure, disc degeneration, disc contour, nerve root compromise, spinal stenosis and facet joint degeneration. Findings were identified and classified at disc level according to type and severity. Raters were instructed to evaluate all study sample persons once to assess inter-rater reliability (fully crossed design). Reliability was calculated using Gwet’s Agreement Coefficients (AC1 and AC2) and Cohen’s Kappa (κ) and Conger’s extension of Cohen’s. Gwet’s probabilistic benchmarking method to the Landis and Koch scale was used. MRI-findings achieving substantial reliability was considered acceptable. Results Inter-rater reliability for all raters combined, ranged from (Gwet’s AC1 or AC2): 0.64–0.99 and according to probabilistic benchmarking to the Landis and Koch scale equivalent to moderate to almost perfect reliability. Overall reliability level for individual pathologies was almost perfect reliability for spondylolisthesis, spinal stenosis, scoliosis and annular fissure, substantial for nerve root compromise and disc degeneration, and moderate for facet joint degeneration and disc contour. Conclusion Inter-rater reliability for 3 raters, evaluating 177 disc levels, was found to be overall acceptable for 6 out of 8 degenerative MRI-findings in the lumbar spine. Ratings of facet joint degeneration and disc contour achieved moderate reliability and was considered unacceptable.