Skip to main content
. 2020 May 5;15:981–993. doi: 10.2147/COPD.S238933

Table 3.

Regression of COPD-Related Outcomes on Continuous ADI

COPD-Related Outcome Unadjusted Adjusted
Predicted Difference (95% CI) p-value Predicted Difference (95% CI) p-value
Quality of Life/Respiratory Symptoms
CAT 1.46 <0.001 0.57 <0.001
(0.87, 2.04) (0.25, 0.88)
SGRQ 3.36 <0.001 0.94 0.034
(1.63, 5.08) (0.07, 1.81)
MRC 0.07 0.006 −0.01 0.722
(0.02, 0.13) (−0.06, 0.04)
6-minute walk distance (meters) −18.21 <0.001 −11.04 <0.001
(−25.59, −10.84) (−16.98, −5.10)
Ease of cough and sputum 0.63 <0.001 0.35 <0.001
(0.47, 0.78) (0.18, 0.52)
Chest CT Imaging
Airway thickness (x10−2) 0.63 <0.001 0.33 0.029
(0.37, 0.89) (0.03, 0.63)
Percent emphysema −0.85 0.080 −0.56 0.066
(−1.81, 0.10) (−1.16, 0.04)
Percent gas trapping −1.12 0.165 −0.57 0.119
(−2.69, 0.46) (−1.29, 0.15)
Rate of Exacerbations
Any exacerbations (IRR)a 1.17 0.005 1.15 0.002
(1.05,1.31) (1.05,1.25)
Severe exacerbations (IRR)a 1.44 <0.001 1.28 <0.001
(1.24,1.67) (1.15,1.42)
Psychosocial Wellbeing
Depressive symptom score 0.44 0.005 0.10 0.386
(0.13, 0.75) (−0.12, 0.31)
Anxiety symptom score 0.38 <0.001 0.04 0.543
(0.23, 0.54) (−0.09, 0.17)

Notes: In all analyses, generalized linear mixed model with robust standard error estimator was used. For continuous outcomes (quality of life/respiratory symptoms and chest CT imaging), Gaussian family distribution and identity link function were employed; and, the predicted difference represented the difference in the level of the outcome for one standard deviation increase in ADI national rank. For count outcomes (exacerbations), negative binomial family distribution and log link function were employed, along with adjustment for the number of days in the study as an offset; and, the predicted difference represented the incidence rate ratio for the one standard deviation increase in ADI national rank. Adjusted model adjusted for age, sex, race, education, income, marital status, rural status, BMI, FEV1% Predicted, smoking status, pack years, and occupational exposure. aFor regression analysis, airway thickness was rescaled by a factor of 100 in order that the effect estimates would represent the estimates at the hundredth level.