Study design: A double-blind prospective study was used to measure interobserver and intraobserver variability when interpreting lumbar spine magnetic resonance imaging studies of disc abnormalities.
Objectives: To evaluate reader consistency when interpreting disc extension beyond the interspace, and assess the effect of two distinct nomenclatures on reader consistency.
Summary of background data: Interobserver and intraobserver variability in interpretation of lumbar disc abnormalities is an important consideration in analyzing the technical efficacy of an imaging modality. However, this has not been well measured (particularly for standardized nomenclature).
Methods: Magnetic resonance imaging studies of the lumbar spine performed prospectively in 98 asymptomatic volunteers, and an additional 27 selected studies from symptomatic patients, were read blindly by two experienced neuroradiologists, using two separate nomenclatures. Only the discs were evaluated (625 interspaces). Nomenclature I was normal, bulge, herniation. Nomenclature II was normal, bulge, protrusion, extrusion. Intraobserver and interobserver variation were measured with Kappa statistic analysis.
Results: Interobserver agreement was 80% for both nomenclatures with a Kappa statistic of 0.58. Intraobserver agreement was 86% for each reader, with a Kappa statistic of 0.71 and 0.69, respectively. The most common disagreement was for normal versus bulge. The next most common disagreement (5-6%) was for bulge versus herniation (or protrusion in Nomenclature II). Herniation was read in 23% of the asymptomatic subjects. Using Nomenclature II, protrusion was seen in 27% of these subjects. Extrusion was read in only two asymptomatic subjects.
Conclusions: Experienced readers using standardized nomenclature showed moderate to substantial agreement with interpreting disc extension beyond the interspace on magnetic resonance imaging.