Skip to main content

Table 4 Densitometric progression adjusted for lung volume using random coefficient model

From: Detection of emphysema progression in alpha 1-antitrypsin deficiency using CT densitometry; Methodological advances

Variable

Annual change, mean ± SE (95% CI)

t

p value for annual rate

p value for volume

Whole lung analysis

    

WL Perc15

-2.13 ± 0.44 (-3.01, -1.24)

-4.82

< 0.0001

< 0.0001

WL VI-950

0.90 ± 0.19 (0.52, 1.29)

4.72

< 0.0001

< 0.0001

WL VI-910

1.16 ± 0.25 (0.66, 1.65)

4.67

< 0.0001

< 0.0001

Single slice analysis

    

UZPerc15

-2.02 ± 0.53 (-3.08, -0.97)

-3.82

0.0003

< 0.0001

UZ VI-950

0.60 ± 0.21 (0.18, 1.02)

2.84

0.0057

< 0.0001

UZ VI-910

0.94 ± 0.33 (0.29, 1.59)

2.88

0.0051

< 0.0001

LZ Perc15

-1.93 ± 0.58 (-3.07, -0.78)

-3.34

0.0013

< 0.0001

LZ VI-950

1.02 ± 0.32 (0.39, 1.65)

3.22

0.0019

< 0.0001

LZ VI-910

1.34 ± 0.38 (0.58, 2.1)

3.52

0.0007

< 0.0001

Targeted sampling

    

UT Perc15

-2.54 ± 0.62 (-3.77, -1.32)

4.13

< .0001

< 0.0001

UT VI-950

0.67 ± 0.22 (0.24, 1.10)

3.11

0.0027

< 0.0001

UT VI-910

1.28 ± 0.44 (0.42, 2.15)

2.95

0.0042

< 0.0001

MT Perc15

-2.94 ± 0.49 (-3.91, -1.97)

-6.04

< 0.0001

< 0.0001

MT VI-950

1.20 ± 0.24 (0.73, 1.67)

5.08

< 0.0001

< 0.0001

MT VI-910

1.78 ± 0.35 (1.08, 2.47)

5.12

< 0.0001

< 0.0001

LT Perc15

2.85 ± 0.61 (-4.07, -1.64)

-4.68

< 0.0001

< 0.0001

LT VI-950

1.28 ± 0.26 (0.76, 1.80)

4.96

< 0.0001

< 0.0001

LT VI-910

1.57 ± 0.29 (0.99, 2.15)

5.37

< 0.0001

< 0.0001

  1. WL, whole lung; UZ, upper zone single slice; LZ, lower zone single slice; ATl, apical third; MT, middle third; LT, lower third; SE, standard error; Perc15, 15th percentile point (measured in Hounsfield Units); VI-950, voxel index at a threshold of -950HU (measured in %); VI-910, voxel index at a threshold of -910HU (measured in %). The random coefficient model consists of outcome measurement as the dependent variable, time (year) as the fixed effect, volume as a time-dependent covariate, intercept and time (year) as random effects. The slope (coefficient for time variable) from the model is the estimated annual change adjusted for volume.