Diagnostic accuracy of lumbar CT and MRI in the evaluation of chronic low back pain without red flag symptoms

Background

Low back pain (LBP) is one of the most frequent reasons for medical consultation. Most of the patients will have nonspecific LBP, which usually are self-limited episodes. It is unclear which of the diagnostic imaging pathways is most effective and costeffective and how the imaging impacts on patient treatment. Imaging techniques are usually indicated if symptoms remain after 6 weeks. Magnetic resonance imaging (MRI) is the diagnostic imaging examination of choice in lumbar spine evaluation of low back pain; however, availability of MRI is limited.

Objectives

To evaluate the diagnostic accuracy of computed tomography (CT) with MRI (as standard of reference) in the evaluation of chronic low back pain (LBP) without red flags symptoms. To compare the results obtained by two radiologists with different grades of experience.

Materials and methods

Patients with chronic low back pain without red flags symptoms were retrospectively reviewed by two observers with different level of experience. Patients included had undergone a lumbar or abdominal CT and an MRI within a year. Once the radiological information was collected, it was then statistically reviewed. The aim of the statistical analysis is to identify the equivalence between both diagnostic techniques. To this end, sensitivity, specificity and validity index were calculated. In addition, intra and inter-observer reliability were measured by Cohen’s kappa values and also using the McNemar test.

Results

340 lumbar levels were evaluated from 68 adult patients with chronic low back pain or sciatica. 63.2% of them were women, with an average age of 60.3 years (SD 14.7). CT shows high values of sensitivity and specificity (>80%) in most of the items evaluated, but sensitivity was low for the evaluation of density of the disc (40%) and for the detection of disc herniation (55%). Moreover, agreement between MRI and CT in most of these items was substantial or almost perfect (Cohen’s kappa-coefficient > 0’8), excluding Modic changes (kappa = 0.497), degenerative changes (kappa0.688), signal of the disc (kappa = 0.327) and disc herniation (kappa = 0.639). Finally, agreement between both observers is mostly high (kappa > 0.8). Foraminal stenosis, canal stenosis and the grade of the canal stenosis were overdiagnosed by the inexperienced observer in the evaluation of CT images.

Conclusions and significance

CT is as sensitive as lumbar MRI in the evaluation of most of the items analysed, excluding Modic changes, degenerative changes, signal of the disc and disc herniation. In addition, these results are obtained regardless the experience of the radiologist. The rising use of diagnostic medical imaging and the improvement of image quality brings the opportunity of making a second look of abdominal CT in search of causes of LBP. Thereby, inappropriate medical imaging could be avoided (2). In addition, it would allow to reduce MRI waiting list and prioritize other patients with more severe pathology than LBP.