Table 3.
Associations between large airway dimensions on computed tomography, longitudinal lung function, and incidence of first hospitalization or mortality due to chronic lower respiratory disease, among Multi-Ethnic Study of Atherosclerosis Lung Study participants without prevalent clinical chronic lower respiratory disease
| Annual Change in FEV1* (ml) (n = 1,830) | Incident COPD† (n = 1,717) Cases (n = 40 [2.3%]) | CLRD Hospitalization/Mortality‡ (n = 6,029) Events (n = 84 [1.4%]) | ||||
|---|---|---|---|---|---|---|
| Model | B per SD§ (95% CI) | P Value | OR per SD§ (95% CI) | P Value | HR per SD§ (95% CI) | P Value |
| Pi10, sequential adjustment║ | ||||||
| Unadjusted | −2.46 (−4.32, 0.60) | 0.010 | 1.49 (1.11, 2.01) | 0.009 | 1.33 (1.17, 1.52) | <0.001 |
| Partially adjusted | −2.45 (−4.26, −0.64) | 0.008 | 1.71 (1.24, 2.35) | 0.001 | 1.30 (1.13, 1.50) | <0.001 |
| Fully adjusted | −2.47 (−4.28, −0.66) | 0.008 | 1.87 (1.26, 2.78) | 0.002 | 1.38 (1.20, 1.60) | <0.001 |
| Genetic risk adjusted | −2.66 (−4.54, −0.79) | 0.005 | 2.21 (1.47, 3.33) | <0.001 | 1.41 (1.22, 1.63) | <0.001 |
| FEV1 adjusted | −2.45 (−4.36, −0.55) | 0.0118 | 2.22 (1.43, 3.45) | <0.001 | 1.57 (1.00, 2.45) | 0.0496 |
| Directly measured airway dimensions, fully adjusted║ | ||||||
| Lumen diameter | −0.58 (−2.64, 1.48) | 0.582 | 0.37 (0.25, 0.56) | <0.001 | 0.75 (0.59, 0.96) | 0.023 |
| AWT | −1.95 (−3.98, 0.07) | 0.059 | 1.97 (1.25, 3.10) | 0.004 | 1.34 (1.01, 1.78) | 0.040 |
Definition of abbreviations: AWT = airway wall thickness; B = β-estimate is milliliters per year per standard deviation increment of average wall thickness for a hypothetical airway of 10-mm lumen perimeter on computed tomography; CI = confidence interval; CLRD = chronic lower respiratory disease; COPD = chronic obstructive pulmonary disease; FEV1 = forced expiratory volume in 1 second; HR = hazard ratio; OR = odds ratio; Pi10 = average wall thickness for a hypothetical airway of 10-mm lumen perimeter on computed tomography; SD = standard deviation.
FEV1 was measured by prebronchodilator spirometry in 2004–2006 and 2010–2012. Annual change in FEV1 was analyzed by random intercept mixed models. Models exclude participants with FEV1/FVC less than 0.70 at the initial spirometry examination as well as participants with prevalent CLRD, defined as self-reported asthma or emphysema and/or inhaler use at study baseline.
Incident COPD was defined as FEV1/FVC less than 0.70 by post-bronchodilator spirometry in 2010–2012, which was present in 40 participants of 1,717 with full post-bronchodilator data (2.3%), and analyzed using logistic regression models. Models exclude participants with FEV1/FVC less than 0.70 at the initial spirometry examination as well as participants with prevalent CLRD, defined as self-reported asthma or emphysema and/or inhaler use at study baseline.
Events are defined as hospitalizations/mortality assigned the primary discharge diagnosis or underlying cause of asthma (International Classification of Diseases, 9th Revision [ICD-9], code 493; International Classification of Diseases, 10th Revision [ICD-10], codes J45–J46), COPD (ICD-9 code 496, ICD-10 code J44), chronic bronchitis (ICD-9 codes 490–491, ICD-10 codes J40–J42), or emphysema (ICD-9 code 492, ICD-10 code J43). Models exclude participants with prevalent CLRD, defined as self-reported asthma or emphysema and/or inhaler use at study baseline.
Results are per SD of log-transformed airway dimensions. For Pi10, these were equivalent on the logarithmic scale to 0.057 and 0.064 in the lung function and events analyses, respectively. For directly measured airway dimensions, these were equivalent to 0.37 and 0.13 for lumen diameter and AWT, respectively.
Models were adjusted for age, sex, race/ethnicity, height, and weight (partially adjusted), as well as for smoking status, pack-years, voxel size, computed tomography scanner type, and percent emphysema (fully adjusted). Height, weight, smoking status, and pack-years were time-varying covariates in the mixed models. The genetic risk–adjusted model is also adjusted for principal components of genetic ancestry (in the place of self-reported race/ethnicity) and COPD genetic risk score. The FEV1-adjusted model was additionally adjusted for the FEV1 percent predicted at the initial spirometry examination. For analyses of annual change in FEV1, models adjusted for the genetic risk score and the initial FEV1 were limited to 1,762. For analyses of incident COPD, models adjusted for the genetic risk score and the initial FEV1 were limited to 1,652 (cases = 39). For analyses of CLRD hospitalization/mortality, models adjusted for the genetic risk score were limited to 5,631 (events = 76); models adjusted for the initial FEV1 were further limited to 3,322 (events = 41). Directly measured airway dimensions were available in only a subset of participants with events follow-up, limiting the model to 5,626 (events = 72). Models using directly measured airway dimensions included both lumen diameter and AWT; results modeling these exposures separately yielded similar results.