Table 2:

Diagnostic performance of imaging attributes

SensitivitySpecificityDiagnostic Odds RatioaP Valueb
Estimate95% Confidence IntervalEstimate95% Confidence IntervalEstimate95% Confidence Interval
Neutral MRIc70% (14/20)(48%–85%)100% (17/17)(82%–100%)78.1(4.0–1505.9)<.0001
Neutral and flexion MRI71% (15/21)(50%–86%)100% (17/17)(82%–100%)83.5(4.3–1605.1)<.0001
LOA on neutral imagesc65% (13/20)(43%–82%)94% (16/17)(73%–99%)29.7(3.2–273.4).0004
Anterior dural shift with flexiond76% (16/21)(55%–89%)94% (16/17)(73%–99%)51.2(5.4–488.7)<.0001
Asymmetric cord flattening48% (10/21)(28%–68%)65% (11/17)(41%–83%)1.7(0.4–6.2).521
Cord T2 signal33% (7/21)(17%–55%)94% (16/17)(73%–99%)8.0(0.9–73.2).053
Abnormal curvaturec50% (10/20)(30%–70%)65% (11/17)(41%–83%)1.8(0.5–6.9).510
C8 most affected cervical myotome at EMG with moderate-to-severe changee50% (10/20)(30%–70%)88% (14/16)(64%–97%)7.0(1.3–39.1).0317
  • a The odds ratio for neutral MRI and neutral and flexion MRI was computed by adding 0.5 to each cell (ie, the empiric odds ratio) to account for the zero cell that resulted from 100% specificity.

  • b P values were derived from a 2-tailed Fisher exact test.

  • c One patient did not have neutral images for assessment.

  • d Recorded as present, defined as any degree of dural shift, or absent.

  • e One patient did not have an EMG on record.