Research Article

A Nomogram to Predict Benign/Malignant Mediastinal Lymph Nodes Based on EBUS Sonographic Features

Table 5

Univariate and multivariate logistic regression analyses of the training set in the whole cohort.

CharacteristicsUnivariateMultivariate
OR (95% CI) valueOR (95% CI) value

Long-axis2.87 (2.21–3.72)<0.0012.33 (1.63–3.35)<0.001
Short-axis
 ≤1Reference
 >14.73 (3.21–6.96)<0.0012.38 (1.3–4.35)0.005
Long-axis/short-axis ratio
 <1.5Reference
 ≥1.50.79 (0.53–1.17)0.239
Number of passes per node1.12 (0.94–1.32)0.198
Echo intensity
 HypoechoicReference
 Hyperechoic1.05 (0.38–2.90)0.919
 Isoechoic0.41 (0.09–1.87)0.250
Echogenicity
 HomogeneousReference
 Heterogeneous5.51 (3.83–7.93)<0.0015.65 (3.6–8.85)<0.001
Margin
 IndistinctReference
 Distinct1.29 (0.84–2.00)0.249
Blood flow
 LackingReference
 Rich1.89 (1.27–2.81)0.0021.57 (0.93–2.66)
Fusion
 AbsentReference
 Present2.02 (1.29–3.14)0.0020.36 (0.19–0.68)0.002
Central hilar structure
 PresentReference
 Absent9.44 (6.46–13.78)<0.00115.59 (9.06–26.83)<0.001