This study is a part of the ongoing Korean Obstructive Lung Disease (KOLD) Cohort, in which a total of 800 patients with COPD or asthma will have been enrolled by 2012. For this study, we retrospectively analyzed 260 patients who had been recruited from pulmonary clinics in 11 hospitals throughout Korea from June 2005 to November 2009 and were followed up for at least one year. All 260 patients fulfilled the following three criteria: post-bronchodilator ratio of FEV₁ to forced vital capacity (FEV₁/FVC) < 0.7 after administration of 400 µg of inhaled albuterol, more than 10 pack-years of smoking history, and no or minimal abnormality on chest radiography. Moreover, none of these patients had exacerbations of COPD for at least 2 months at the time of enrollment.

Acute exacerbations were defined as worsening symptoms (dyspnea, cough, or sputum) requiring treatment with systemic steroids or antibiotics, a visit to the emergency room, and/or admission to a hospital, as decided by attending physicians.

Demographic and clinical data collected from all patients included age, sex, smoking status, cigarette smoking amount, GERD treated with medications such as proton pump inhibitors, and hospitalization due to exacerbation of COPD within 1 yr before enrollment. Comorbidities were assessed using the Charlson index, which has been shown to be a good predictor of prognosis in patients with COPD (²³). Body mass index was calculated as the weight in kilograms divided by height in meters squared (kg/m²). Productive cough was defined as an affirmative answer to the question, "Did you cough up phlegm (in the morning or during the day) for at least 3 months each year during the past two years?" Dyspnea was assessed with the Modified Medical Research Council Dyspnea Scale (MMRC). The 6-min walking distance test was performed according to American Thoracic Society guidelines, except for the short distance of 20 meters. Health-related quality of life was assessed using the St. George Respiratory Questionnaire. Spirometry was performed using the V_max 22 (SensorMedics, Yorba Linda, CA, USA; PFDX instrument; MedGraphics, St. Paul, MN, USA) (²⁴).

CT indices for COPD assessment included the emphysema index, air-trapping index and airway dimensions. (¹⁷, ^25-27). Patients underwent volumetric CT scans at full inspiration and expiration using a 16-MDCT scanner (CT machines of three manufacturers: the Somatom Sensation 16, Siemens Medical Solutions, Forchheim, Germany; the GE Lightspeed Ultra, General Electric Healthcare, Milwaukee, WI; and the Philips Brilliance 16, Philips Medical Systems, Best, Netherlands) (¹⁹). Scan parameters included 0.75-mm collimation, 100 effmAs, 140 kVP, and pitch 1.0. The scale of attenuation coefficients in this CT scanner ranges from -1,024 to 3,072 HU. Patients were scanned craniocaudally, with both scans performed in the supine position. Images were reconstructed using a soft kernel (B30f; Siemens Medical Systems) from the thoracic inlet to the lung base.

The emphysema index was determined by automatic calculation from CT data of the volume fraction of the lungs below -950 Hounsfield Units at full inspiration and the mean lung density (²⁶). The CT air-trapping index was defined as the ratio of mean lung density on expiration and inspiration (¹⁹). Airway dimensions were measured near the origin of the four segmental bronchi (RB1, LB1+2, RB10, LB10) by a consensus of two radiologists using in-house software. Airway dimensions, including the wall area (WA), lumen area (LA), and wall area percent (WA %), defined as WA/(WA+LA) × 100, were measured in each segmental bronchus.

Continuous data and the frequencies and percentages of categorical data were expressed as mean ± standard deviation. A stepwise backward multiple logistic regression was used to construct models predictive of COPD exacerbation. Clinico-physiologic indices with a univariate P value < 0.2 were considered candidates in the backward stepwise multiple logistic regression. All three radiologic indices, irrespective of the result of univariate analysis, as well as age and gender, were included in the multiple logistic regression. The predictive accuracy of the models was quantified by calculating the C-statistics. The two models were compared by bootstrap analysis using R 2.12.0 (R Development Core Team, GNU General Public License, www.r-project.org). In addition, we performed a multiple logistic regression analysis with all of the variables including both clinico-physiologic indices and radiologic indices. All other statistical analyses were performed using a statistical package (SPSS version 12.1.1; SPSS, Chicago, IL, USA).

This study was approved by the institutional review board of Asan Medical Center (Approval No. 2005-0345) and of other 10 hospitals. Written informed consent was obtained from all subjected patients.

The clinical, physiologic, and radiologic characteristics of the 260 patients are shown in Table 1. According to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines, most patients had moderate-to-severe COPD. During the previous year, 24 (9.2%) patients had been hospitalized for COPD exacerbation. The mean Charlson index score in all 260 patients was 0.24. Fifty patients (19.2%) had at least one comorbidity, with diabetes mellitus being the most frequent (Table 2).

A total of 106 (40.8%) of 260 patients had experiences of acute exacerbation of COPD during 1 yr follow-up, who were designated as exacerbators. The characteristics of exacerbators and non-exacerbatiors were summarized in Table 3. Exacerbators were older and had more experiences of hospitalization due to COPD exacerbations, more dyspnea scale, worse quality of life and exercise capacity, lower FEV₁, more emphysematous change and air-trapping on CT. Eighteen patients (17%) of 106 exacerbators were hospitalized for the treatment of COPD acute exacerbations during follow-up period.

Table 4 shows the univariate analysis for factors predictive of COPD exacerbation. Among clinico-physiologic indices, we found that old age, current smoking, hospitalization for exacerbation of COPD during the previous year, Charlson index score, BMI, MMRC dyspnea scale, six minute walk distance, the St. George Respiratory Questionnaire and FEV₁ were related to COPD exacerbation. Among the CT imaging indices, emphysema and CT air-trapping index were related to exacerbation.

We utilized multiple logistic regression analysis to develop models of risk factors for COPD exacerbation (Table 5, 6). A model using clinico-physiologic indices showed that increased age, higher Charlson Index, and lower FEV₁ were significantly associated with exacerbation of COPD, with C-statistics of 0.69. A CT imaging model showed that increased age and emphysema index were associated with COPD exacerbation, with C-statistics of 0.64. Bootstrap analysis showed that the clinico-physiologic model was significantly superior to the CT imaging indices model in predicting the occurrence of exacerbations of COPD (P = 0.04, bootstrap analysis).

When a multiple logistic regression analysis was performed with all of the variables including both clinico-physiologic indices and radiologic indices, only Charlson Index was a significant risk factor of COPD exacerbation (Table 7). When the same analysis was done for hospitalizatioin due to COPD exacerbation as a dependent variable, the St. George Respiratory Questionnaires score and emphysema index were significant (data not shown).