Predictors of active pulmonary tuberculosis among hospitalized patients with atypical symptom and sign and underlying diseases having impact on the outcome of the COVID-19

Background This study aimed to focus on the diagnostic use of high-resolution computed tomography (HRCT) to identify active pulmonary tuberculosis (aPTB) with atypical symptom and sign among the hospitalized patients with the underlying diseases having the impact on the outcome of the Coronavirus disease 2019 (COVID-19). Methods Within the study period (2018.01.01-2021.12.31), for patients with underlying diseases having the impact on the outcome of the COVID-19, chest –x-ray (CXR) / HRCT scans along with their patients’ charts were reviewed. These patients (n = 4,380) were classified into the [aPTB] group I (G1, n = 277) and pulmonary disease without aPTB (G2, n = 4103). Lung morphology, and lobar (segmental) distribution using CXR/HRCT, the underlying diseases and clinical symptom/sign were analyzed. To identify independent variables associated with G1, multivariate analysis was performed. Independent variables were used to generate prediction scores, which were used to develop models for predicting G1. Results For the HRCT model, multivariate analysis revealed cavitation, clusters nodules/mass (CNM) of the right/left upper lobe or ground-glass opacity were useful predictors for the G1. The negative predictive value of the HRCT model, and the CNM model for the GI were 99.3%, and 97.5%, respectively. However, the CNM model has the highest positive predictive value of 95.4%. Conclusions The CNM model may play an auxiliary role for the identification of G1 with atypical symptom and sign among the patients with underlying diseases having the impact on the outcome of the COVID-19.