The purpose of our study was to provide a novel schematized and comprehensive classification of causes and severity grading system for lumbosacral stenosis.

The MRI system proposed consisted of a severity grading scale for central and lateral (recess and foramen) stenosis, together with a schematized indication of the main causes of the disease (disc, arthritis, epidural lipomatosis, and their combinations). The system was applied to a cohort of patients from a single Institution in the last 2-years. Two radiologists evaluated all the MRIs to determine intra- and inter-observer reliability according to Cohen Kappa (Kc, for non-ordered categorical variables) and weighted Kappa (Kw, for ordered variables). Two orthopaedic surgeons clinically evaluated all patients and provided a schematic grading system with a central and lateral stenosis clinical score (CS-CS and LS-CS). Associations between ordinals were tested with chi-square test and measured with the Goodman and Kruskal’s gamma index (Gi, with 95% confidence interval [95% CI]). Lastly, the most used previous MRI systems were applied, and their performances were compared to the new system proposed.

One hundred and twelve patients were included (55 females—mean age 63.3 ± 10.7 years).

An almost perfect intra-observer agreement for the assessment of central stenosis, foramen stenosis, and lateral recess stenosis was found (Kw = 0.929, 0.928, and 0.924, respectively). The inter-observer agreement was almost perfect for central stenosis and foramen stenosis and substantial for lateral recess stenosis (Kw = 0.863, 0.834, and 0.633, respectively).

Whatever the aetiologies involved in central and lateral stenosis, the intra-observer agreement was perfect (all Kc = 1), whereas the inter-observer agreements were almost perfect for arthritis (Kc = 0.838) and lipomatosis (Kc = 0.955) and substantial for disc (Kc = 0.691) regarding central stenosis. The inter-observer agreement for the causes of lateral stenosis was lower and variable, ranging from perfect (lipomatosis) to fair (disc, Kc = 0.224).

The grading system revealed a strong association with CS-CS for both readers, with GI = 0.671 (95% CI 0.535–0.807) and 0.603 (95% CI = 0.457–0.749), respectively.

The association with MRI grading and LS-CS was moderate for foraminal stenosis and for the concomitant presence of foraminal and lateral recess stenosis, with Gi = 0.337 (95% CI 0.121–0.554) and Gi = 0.299 (95% CI 0.098–0.500), respectively. A weak association was found between lateral recess grading alone and LS-CS with Gi = 0.102 (95% CI 0.193–0.397).

The new grading systems showed higher Gi for associations with clinical symptoms, compared with previous ones, both for CS-CS and LS-CS.

A standardized visual grading system for lumbar spinal stenosis that takes into account all of the major contributing factors—including disc, arthritis, and lipomatosis, for the central canal, lateral recess, and neural foramina could be a useful and practical tool for defining the stenosis, lowering inter-observer variability, and directing the various treatment options.