Impact of the COVID-19 Pandemic on the Epidemiological Situation of Pulmonary Tuberculosis–Using Natural Language Processing

Abstract

BACKGROUND: We aimed to analyze the impact of the COVID-19 pandemic on pulmonary tuberculosis (TB) using artificial intelligence. To do so, we compared the real-life situation during the pandemic with the pre-2020 situation. METHODS: This non-interventional, retrospective, observational study applied natural language processing to the electronic health records of the Castilla-La Mancha region of Spain. The analysis was conducted from January 2015 to December 2020. RESULTS: A total of 2592 patients were diagnosed with pulmonary tuberculosis; 64.6% were males, and the mean age was 53.5 years (95%CI 53.0-54.0). In 2020, pulmonary tuberculosis diagnoses dropped by 28% compared to 2019. In total, 62 (14.2%) patients were diagnosed with COVID-19 and pulmonary tuberculosis coinfection in 2020, with a mean age of 52.3 years (95%CI 48.3-56.2). The main symptoms in these patients were dyspnea (27.4%) and cough (35.5%), although their comorbidities were no greater than patients with isolated TB. The female sex was more frequently affected, representing 53.4% of this patient subgroup. CONCLUSIONS: During the first year of the COVID-19 pandemic, a decrease was observed in the incidence of pulmonary tuberculosis. Women presented a significantly higher risk for pulmonary tuberculosis and COVID-19 coinfection, although the symptoms were not more severe than patients diagnosed with pulmonary tuberculosis alone.

Ähnliche Arbeiten