The influence of the criterion of abnormal DLco value on the prediction of impaired lung diffusion capacity after SARS-CoV-2 infection
Aim. To predict impaired lung diffusion capacity after SARS-CoV-2 infection depending on the criteria of pathological deviation of DLco value (carbon monoxide transfer factor). Methods. The retrospective study included 341 patients (median age was 48 years, 76.8% of the participants were men) after...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Siberian State Medical University (Tomsk)
2025-04-01
|
| Series: | Бюллетень сибирской медицины |
| Subjects: | |
| Online Access: | https://bulletin.ssmu.ru/jour/article/view/5965 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Aim. To predict impaired lung diffusion capacity after SARS-CoV-2 infection depending on the criteria of pathological deviation of DLco value (carbon monoxide transfer factor). Methods. The retrospective study included 341 patients (median age was 48 years, 76.8% of the participants were men) after SARS-CoV-2-associated lung injury. The median volume of lung injury during the acute phase of COVID-19 was 50%. All patients underwent a diffusion test. Descriptive statistics, logistic regression analysis were applied, taking into account the previously obtained model for prognosis of abnormal DLco (<80% of the predicted value (%pred.)) [11]. In the present study on the same sample of patients, the prognosis of abnormal DLco was studied depending on the criterion 1: DLco < 80%pred. or criterion 2: DLco < predicted – 1.645SD (SD — standard deviation. ROC analysis was used to assess the quality of the binary classifier models. Results. The coefficients of the logistic regression equations were obtained on the training sample with regard to the chosen criterion of pathological deviation of DLco. The ROC analysis procedure showed that, when applying criterion 1, area under curve (AUC) was 0.776, p < 0.001 (0.707–0.824 95% confidence interval (CI)), sensitivity and specificity of the training model were 81% and 66%, respectively. When applying criterion 2, AUC was 0.759, p < 0.001 (0.701–0.817 95% CI), sensitivity and specificity of the training model were 83.4% and 59%, respectively. Conclusions. The criterion for determining the lower limit of normal DLco (LLNDLco) does not significantly affect the quality of the model for impaired lung diffusion capacity prognosis after SARS-CoV-2-associated lung injury. It is advisable to give preference to a method that is easier to apply in practice. |
|---|---|
| ISSN: | 1682-0363 1819-3684 |