While high-resolution computed tomography (HRCT) is increasingly performed, its role in diagnosing pulmonary tuberculosis (TB) among elderly patients with community-acquired pneumonia (CAP) has not been fully elucidated. This study aimed to determine HRCT features that can differentiate pulmonary TB from non-TB CAP in elderly patients. This study included consecutive elderly patients (age > 65 years) admitted to two teaching hospitals for pulmonary TB or non-TB pneumonia who met the CAP criteria of the American Thoracic Society/Infectious Diseases Society of America guidelines. After propensity score matching for clinical background between patients with pulmonary TB and those with non-TB CAP, their HRCT features were compared. This study included 151 patients with pulmonary TB and 238 patients with non-TB CAP. The presence of centrilobular nodules, air bronchograms, and cavities and the absence of ground-glass opacities and bronchial wall thickening were significantly associated with pulmonary TB. The negative predictive values of centrilobular nodules, air bronchograms, and cavities for pulmonary TB were moderate (70.6%, 67.9%, and 63.0%, respectively), whereas the positive predictive value of cavities was high (96.6%). In elderly patients, although some HRCT features could differentiate pulmonary TB from non-TB CAP, no useful findings could rule out pulmonary TB with certainty.

Background Non-conversion of sputum culture or smear within 2 months after the start of treatment is a known poor prognostic factor of pulmonary tuberculosis. In elderly patients, sputum conversion may be delayed because of the age-related decline in immune competence. This study aimed to assess how a long interval to sputum conversion predicts in-hospital mortality in elderly patients with pulmonary tuberculosis. Methods Consecutive elderly patients (age > 65 years) who were admitted to our institution for bacteriologically confirmed pulmonary tuberculosis were included. The association between sputum conversion within 30, 60, 90, or 120 days from the start of treatment and in-hospital mortality were analyzed by Cox proportional-hazards regression after adjustment for other potential variables. Results This study included 262 patients, and 74 patients (28%) died during hospitalization. Multivariate analyses showed that sputum non-conversion within 90 days (adjusted hazard ratio 0.424, 95% CI 0.252–0.712, p = 0.001) or 120 days (0.333, 0.195–0.570, p < 0.001) was independently associated with in-hospital mortality, whereas that within 60 days was not (p = 0.890). Conclusions In elderly patients with tuberculosis, 2 months may be insufficient when evaluating sputum conversion as a prognostic factor. Sputum non-conversion within 90 days or longer may predict in-hospital mortality more accurately.