A decision rule for identifying patients at high risk for impaired lung diffusion capacity after COVID-19

Abstract

Aim. To elaborate a decision rule for identifying the main predictors of impaired lung diffusion capacity after COVID-19. Materials and methods. The retrospective study included 341 patients without underlying lung diseases (median age 48 years) who experienced COVID-19 with bilateral pneumonia. The median extent of parenchymal lesion in the acute phase of COVID-19 (CTmax) was 50%. Spirometry, body plethysmography, and lung diffusion capacity for carbon monoxide (DLCO) test were performed. The data were analyzed by descriptive statistics, correlation analysis, one-dimensional logistic regression analysis with an assessment of odds ratios (OR), and multivariate logistic regression analysis. Receiver operating characteristic (ROC) analysis was used to assess the quality of the binary classifier model.Results. The initial model for predicting reduced DLCO (< 80% of predicted) included the following predictors: CTmax, time interval from the COVID-19 onset, gender, age, body mass index. Backward stepwise regression was applied, and a binary classifier model that includes CTmax was obtained. The sensitivity and specificity of the model for the training sample were 80 and 67%, respectively, for the test sample – 79 and 70%, respectively. The analysis of OR showed that OR > 1 was observed at СTmax > 40%.Conclusion. The decision rule was obtained for predicting impaired lung diffusion capacity after COVID-19 with virus-associated lung damage in patients without underlying bronchopulmonary diseases. Patients with CTmax > 40% require more thorough clinical follow-up with DLCO monitoring after the acute phase of COVID-19