Table 6

Multiple logistic regression to control for potential confounders

P

OR

AUC

Sensitivity (%)

Specificity (%)

Main PA diameter ≥29 mm

0.0196

4.8

0.93

77.4

89.6

Left PA diameter ≥24 mm

0.2160

2.6

0.92

77.4

87.5

Right PA diameter ≥25 mm

0.4461

1.9

0.91

73.6

93.8

RDPA diameter ≥19 mm

0.0059

7.0

0.93

83.0

85.4

True RDPA diameter ≥16 mm

0.0487

4.1

0.92

83.0

87.5

True LDPA diameter ≥21 mm

0.0075

15.5

0.93

79.2

91.7

RV free wall ≥6 mm

0.0303

30.5

0.95

81.0

91.9

RV lumen ≥30 mm

0.0915

5.8

0.95

92.9

73.0

LV free wall ≥15 mm

0.1607

5.2

0.95

85.7

83.8

LV lumen ≥57 mm

0.3945

3.0

0.94

76.2

88.9

Hilar diameter ≥124 mm

0.2968

2.2

0.92

81.1

75.0

RV wall/LV wall ratio ≥0.32

0.0141

8.8

0.96

78.6

83.8

RV lumen/LV lumen ratio ≥1.28

0.0196

28.8

0.95

85.7

86.1

L apical artery/bronchus ratio ≥1.75

0.2851

3.5

0.92

75.5

87.5

Hilar/thoracic ratio ≥0.52

0.0757

3.7

0.92

75.5

87.5

Main PA/AA ratio ≥0.84

0.0208

6.0

0.93

73.6

91.7

Main PA/DA ratio ≥1.29

0.0269

5.7

0.93

77.4

89.6

IV septum bowing into LV (yes or no)

0.1053

11.6

0.95

81.0

89.2

Main PA/AA ratio >1

0.0085

9.1

0.93

86.8

79.2

Regression of the outcome variable (PH vs. no-PH) to the predictors:

ULN for each hypothesized predictor of PH, age, sex, ascending aorta diameter (AA), BSA, thoracic diameter (TD), and pulmonary wedge pressure (PWP) >15 mmHg.

Note: one logistic regression model was created for each hypothesized predictor of PAH. For example, the analysis in the first row above included the predictors: Main PA diameter ≥29, age, sex, AA, BSA, TD, and PWP >15. See table 1 for definitions of radiographic metrics.

Chan et al. BMC Medical Imaging 2011 11:7   doi:10.1186/1471-2342-11-7