Meta-analysis of magnetic resonance imaging for the differential diagnosis of spinal degeneration.

To systematically evaluate the clinical significance of magnetic resonance imaging for the identification and diagnosis of spinal degenerative changes. We searched Cochrane Library, PubMed, EMbase, CNKI, WanFang Data, Medalink, VIP and CBM databases for clinical studies on the significance of magnetic resonance imaging for the differential diagnosis of spinal degeneration; retrieval time was from database building to October 2014. Two reviewers independently screened the literature, extracted data and evaluated methodological quality according to the inclusion and exclusion criteria. Meta-DiSc 1.4 software was used for meta-analysis. The study included six documents, 10 independent results and a total of 505 individuals. Meta-analysis showed that: In the present study, the efficacy of magnetic resonance imaging in the differential diagnosis of cervical and lumbar degeneration was firstly analyzed and discussed using the Meta-Disc 1.4 software. SPE: χ(2) = 77.59, P = 0.000, I(2) = 88.4%; SEN: χ(2) = 167.25, P = 0.000, I(2) = 94.6%; DOR: Cochran-Q = 71.64, P = 0.000. Meta-analysis of random effect model showed that: SEN merge = 0.849 [95% CI (0.816,0.878)], SPE merge = 0.745 [95% CI (0.695, 0.792)], + LR = 2.735 [95% CI (1.600, -4.676)], - LR = 0.245 [95% CI (0.122, -0.493)], DOR merge = 21.158 [95% CI (5.234, -85.529)], SROC AUC = 0.8698; the results had good stability. Then the efficacy of magnetic resonance imaging in the differential diagnosis of cervical degeneration was analyzed and the results showed that: SPE: χ(2) = 6.92, P = 0.075, I(2) = 56.6%; SEN: χ(2) = 81.73, P = 0.000, I(2) = 96.3%; DOR: Cochran-Q = 12.71, P = 0.005. Meta-analysis of random effect model showed that: SEN merge = 0.799 [95% CI (0.741, 0.850)], SPE merge = 0.769 [95% CI (0.683, -0.840)], + LR = 2.506 [95% CI (1.399, -4.489)], - LR = 0.363 [95% CI (0.149, -0.882)], DOR merge = 11.949 [95% CI (2.195, -65.036)], SROC AUC = 0.8210. The stability was good. Finally, analysis of six independent studies on the efficacy of magnetic resonance imaging in the differential diagnosis of lumbar degeneration was performed: SPE: χ(2) = 70.13, P = 0.000, I(2) = 92.9%; SEN: χ(2) = 78.35, P = 0.000, I(2) = 93.6%; DOR: Cochran-Q = 58.04, P = 0.000. Meta-analysis of random effect model showed that: SEN merge = 0.732 [95% CI (0.667, -0.791)] SPE merge = 0.883 [95% CI (0.843, -0.916)], + LR = 3.072 [95% CI (1.330, -7.091)], - LR = 0.190 [95% CI (0.063, -0.572)], DOR merge = 30.252 [95% CI (3.060, -299.13)], SROC AUC = 0.8994. Sensitivity analysis was performed by excluding each study individually and the results showed no significant changes in SEN and SPE merge, indicating good stability of the meta-analysis. Existing studies confirm that MRI had good sensitivity and specificity for the differential diagnosis of cervical and lumbar degeneration; the positive ratio in cervical and lumbar degeneration group was 3 to 10 times of that in non-degeneration control group; the efficacy for differential diagnosis was good; combined with the good maneuverability in clinical diagnosis of spinal degeneration, it can be used as effective and feasible method for clinical differential diagnosis of spinal degenerative diseases.