Study design and subjects
We enrolled adult patients with pulmonary TB from the cohort study of pulmonary tuberculosis (COSMOTB) between November 2016 and September 2018 to compare the clinical characteristics of active and subclinical TB. Briefly, COSMOTB is a prospective observational cohort study to assess the prevalence of discordant results of phenotypic and molecular drug susceptibility tests [9]. COSMOTB was conducted at three university-affiliated tertiary hospitals in South Korea that participated in the public-private mix project for TB control in South Korea. TB specialist nurses under this project educated TB patients and monitored them for medication adherence and adverse drug reactions. The inclusion criteria are as follows: (1) age ≥19 years, (2) a diagnosis or suspicion of pulmonary TB, and (3) receiving anti-TB treatment for less than one month. The exclusion criteria are as follows: (1) age ≤18 years, (2) extrapulmonary TB without pulmonary involvement, (3) patients who were finally diagnosed as inactive TB or pulmonary diseases other than TB, and (4) voluntary withdrawal from study participation. Inactive TB was diagnosed when a follow-up chest radiography showed no pulmonary lesions changes or if previous chest images revealed unchanged lesions without microbiological evidence of Mycobacterium tuberculosis infection [10].
Definition of subclinical and active diseases
Patients were categorized as having active TB or subclinical TB. Active TB was defined as the presence of clinical TB-related symptoms with radiographic abnormalities or microbiologic evidence of M. tuberculosis. Subclinical TB was defined as the presence of radiographic or microbiologic test results consistent with TB without clinical symptoms. We implemented a two-stage symptom assessment using a predefined checklist, which listed TB-related symptoms, such as cough, sputum, fever, general weakness, dyspnoea, chest pain, body weight loss, and haemoptysis. First, TB patients met a TB specialist nurse at the hospital, who interviewed and identified patients’ TB-related symptoms. Subsequently, patients met with a physician at the clinic, who reconfirmed their symptoms and their duration. As patients were identified as asymptomatic after two-stage assessment, they were categorized as subclinical TB disease.
Data collection
Participants were evaluated at each hospital on study entry. Demographic, clinical, and laboratory data were prospectively collected from enrolled patients using a case report form upon study entry. Microbiological tests were performed after the first clinical assessment by a physician. Acid-fast bacilli (AFB) smears using light and fluorescent microscopy and nucleic acid amplification test (NAAT) were conducted at each hospital. Mycobacterium culture testing using both solid (3% Ogawa media) and liquid (BACTEC MGIT 960 system, BD, NJ, USA) cultures were performed at the reference laboratory. Culture-based phenotypic drug susceptibility tests were performed using the absolute concentration method on Löwenstein-Jensen medium.
Statistical analyses
Continuous variables were presented as means and standard deviations or medians and interquartile ranges, whereas discrete variables were presented as frequencies or percentages. The baseline characteristics of patients with active or subclinical TB were compared; univariate analysis was performed using Chi-square test for categorical variables and Mann-Whitney U test for continuous variables. Subsequently, we selected age, sex, and other clinical variables with p-values <0.20 [11] based on the univariate analysis and further performed multivariate binary logistic regression to evaluate the possible association between variables and subclinical TB. For regression, unknown data were regarded as missing values. A p-value <0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 17.0 (Statistical Product and Service Solutions, Chicago, IL, USA).
Sample size
We selected eight variables a priori for inclusion into our model, such as age at diagnosis, sex, foreigners, body mass index, chronic respiratory disease, AFB smear, culture, and NAAT results. Eighty events of subclinical diseases are required to ensure a minimum of 10 events per variable, which are needed to minimize bias in logistic regression models [12]. Assuming that proportions of subclinical disease are 18-21% [3], 381 – 445 patients with pulmonary TB were required for sample size.
Treatment outcomes
Participants were evaluated at 2 and 4 weeks, 2, 4, 6, 9, 12, and 24 months after initiating anti-TB treatment to document their treatment outcome. Those with successful outcomes were also followed for at least 1 year to identify recurrence. Treatment outcomes were defined according to the Korean TB guidelines adopted from the WHO’s definition [13]. Treatment success was the sum of cured patients and those that completed treatment within 1 year of anti-TB treatment. Favourable outcome was defined as patients who had achieved treatment success without recurrence within the 1-year post-treatment follow-up period. We evaluated the association between subclinical disease and treatment outcome in the drug-susceptible cohort comprised of patients with positive culture results susceptible to both isoniazid and rifampin and clinically diagnosed TB patients without microbiological evidence using binary logistic regression and adjusting for age and sex.