Paper #III-5

SPINAL STENOSIS CT PARAMETERS: AN ANALYSIS OF INTEROBSERVER VARIABILITY

Andrew Furey, MD

St. John's, NL

Diagnosis of lumbar spinal stenosis is a common clinical problem, as it represents both an anatomic and clinical diagnosis. Although the gold standard is considered to be myelography, many new tests have developed which are less invasive and offer potentially more preoperative information. The purpose of this study was to assess the degree of reliability in using one of these modalities, namely CT scanning. Twenty five CT scans reported as being stenotic were randomly chosen from the past five years in two tertiary care centers. The CT's were reviewed by two orthopedic surgeons, one senior orthopedic resident, two radiologists and one neurosurgeon who measured the AP and IP diameters of marked images representing L3-4, L4-5, and L5-S1. They were also asked to classify the spinal canal as normal, mild, moderate or severe stenosis. The results were first tabulated and analyzed using a weighted kappa test for the spinal canal as a whole. The weighted kappa achieved was 0.51 with a 95% confidence interval of (0.43,0.60). A second weighted kappa was performed assessing the agreement between orthopaedic surgeons (0.58) and between radiologists (0.58). Weighted kappas were performed considering the Verbiest classification at all three levels L3-4, L4-5, and L5-S1 were 0.25, 0.22, and 0.26 respectively. A series of four ANOVA analyses were performed to determine the degree of agreement in the AP measurements when considering how the scans were originally classified as normal, mild, moderate or severe. It was concluded that the degree of agreement between observers in assessing the degree of stenosis for the spine overall was moderate while the agreement between measurements and assessing individual level was less reliable.