Among a total of 778 patients with COPD, we excluded 77 single-visit patients, 45 systemic steroid users, 15 patients who had received chemotherapy agents for cancers, and 21 patients with active TB at enrollment. An additional four patients were excluded after chart review because of the uncertainty in diagnosing TB: smear-positive, culture-negative, and TB-PCR-negative. Finally, 616 patients were entered into the study (Fig 1). The mean age at the time of entry was 65.4 ± 10.9 y. The median duration of the follow-up was 1,126 days (range, 5-4,017 days). A total of 20 patients with TB were identified during the study period: 12 in the “ICS with TB scar” group, two in the “no ICS with TB scar” group, five in the “ICS without TB scar” group, and one in the “no ICS without TB scar” group. Diagnoses of TB were made by positive culture for 13 patients; by acid-fast bacilli stain with proper radiologic findings for two patients; and by clinicoradiologic diagnosis, without bacteriologic evidence, for four patients (including two patients with positive TB-PCR). Only one patient had TB pleurisy. All four patients who had a clinicoradiologic diagnosis showed clinicoradiologic improvement after anti-TB medications. The median time period from entry to the diagnosis of TB was 1,003 days (range, 101-3,531 days). Table 1 shows the baseline characteristics of each study group. The baseline lung function, judged by FEV1, was more preserved in patients who had no radiologic scar than in patients who had a radiologic scar. In addition, lung function was more preserved in patients who did not use ICS than in patients who used ICS. A total of 41.1% of the patients showed radiologic sequelae of prior TB. However, only about 65% of patients from the “ICS with TB scar” and “no ICS with TB scar” groups had a treatment history for TB. In contrast, 3.5% of patients from the “ICS without TB scar” group and 1.6% of patients from the “no ICS without TB scar” group had a treatment history for TB, even though they did not have radiologic sequelae. The Kaplan-Meier estimates showed increased TB occurrence in the “ICS with TB scar” group (P < .001) (Fig 2). Univariate Cox regression analysis revealed that the “ICS with TB scar” group, “ICS without TB scar” group, moderate- or high-dose ICS use, class 3 sequelae of TB, and a baseline FEV1 < 25% of the predicted value (GOLD [Global Initiative for Chronic Obstructive Lung Disease] stage IV) were associated with an increased risk of TB. Baseline FEV1 (% predicted) was inversely associated with the risk of TB (Table 2). Multivariate Cox regression analysis revealed that the “ICS with TB scar” group and the “ICS without TB scar” group were independent risk factors for TB occurrence (Tables 3, 4). Among the 20 patients who developed TB, nine had a treatment history for TB and one had a repeated treatment history for pulmonary TB. A history of anti-TB medications did not alter the risk of TB occurrence posed by ICS use.