INTEROBSERVER RELIABILITY OF DETECTING LUMBAR INTERVERTEBRAL DISC HIGH-INTENSITY ZONE ON MAGNETIC-RESONANCE-IMAGING AND ASSOCIATION OF HIGH-INTENSITY ZONE WITH PAIN AND ANULAR DISRUPTION

Study Design. Retrospective analysis of a spine imaging center's recor ds of patients with chronic low back pain referred by tertiary care fa cilities. Objectives. 1) To assess the interobserver reliability of de tecting lumbar intervertebral disc high-intensity zone on T-2-weighted magnetic resonance imaging, and 2) to assess the relation between hig h-intensity zone and discography or post-computed tomography in sympto matic patients with low back pain. Summary of Background Data. Two of the three previous studies on this subject found an association betwee n high-intensity zone and the presence of Grade 4 anular disruption wi th discographic reproduction of patients' exact low back pain. Methods . Records of patients with low back pain who had undergone lumbar spin e discography injection and post-computed tomography from June 1995 to August 1996 were reviewed. Two independent observers were asked to id entify the presence of an high-intensity zone from the T12-L1 disc to L5-S1 on T-2-weighted magnetic resonance images. With this data, inter observer reliability was assessed with the kappa statistic. Concordant high-intensity zone results were then compared with the Dallas Discog ram rating for anular disruption and to patients' subjective pain resp onse to discography injection. With this data, the sensitivity, specif icity, and predictive values of high-intensity zone for detecting disc disruption and pain response were calculated. Results. The interobser ver reliability for detecting a high-intensity zone in a given disc wa s fair to good (kappa = 0.57; 95% confidence interval = 0.44, 0.70). T he sensitivity of high-intensity zone for detecting Grade 4 anular dis ruption and exact pain was poor (31%) but its specificity was relative ly high (90%). The positive predictive value of a high-intensity zone was low (40%) for a severely disrupted and exactly painful disc. Concl usions. The interobserver reliability of detecting a high-intensity zo ne and the positive predictive value of the presence of a high-intensi ty zone for detecting a severely disrupted and exactly painful disc we re much lower than previous studies have shown. The relatively low pos itive predictive value may be attributable to differences in sample ch aracteristics or procedural variations, or suggest that a high-intensi ty zone is not indicative of exactly painful internal intervertebral d isc disruption.