Coronavirus disease 2019 (COVID-19) is characterized by a distinctive blood leucocyte pattern and B-lines on lung ultrasound (LUS) as marker of alveolar-interstitial syndrome. We aimed to evaluate the accuracy of blood leucocyte count alone or in combination with LUS for COVID-19 diagnosis. We retrospectively enrolled consecutive patients diagnosed with community acquired pneumonia (CAP) at hospital admission to derive and validate cutoff values for blood cell count that could be predictive of COVID-19 before confirmation by the nucleic acid amplification test (NAAT). Cutoff values, generated and confirmed in inception (41/115, positive/negative patients) and validation (100/180, positive/negative patients) cohorts, were ≤17 and ≤10 cells/mm 3 for basophils and eosinophils, respectively. Basophils and/or eosinophils below cutoff were associated with sensitivity of 98% (95%CI, 94–100) and negative likelihood ratio of 0.04 (95%CI, 0.01–0.11). In a subgroup of 265 subjects, the sensitivity of B-line on LUS was 15% lower ( p < 0.001) than that of basophils and/or eosinophils below cutoff. The combination of B-lines with basophils and eosinophils below cutoff was associated with a moderate increase of the positive likelihood ratio: 5.0 (95%CI, 3.2–7.7). In conclusion, basophil and eosinophil counts above the generated cutoff virtually rule out COVID-19 in patients with CAP. Our findings can help optimize patient triage pending the NAAT results.
【저자키워드】 COVID-19, community acquired pneumonia, Lung ultrasound, eosinophils, basophils, 【초록키워드】 Coronavirus disease 2019, Pneumonia, lung, eosinophil, sensitivity, Accuracy, NAAT, COVID-19 diagnosis, Community, Hospital admission, nucleic acid amplification, moderate, patients, Blood, marker, basophil, Combination, blood cell count, leucocyte, Predictive, 95%CI, patient triage, syndrome, help, cohorts, Cutoff, likelihood ratio, cutoff value, positive, likelihood, enrolled, evaluate, diagnosed, characterized, subjects, consecutive patient, COVID-19 in patient, 【제목키워드】 Ultrasound, count, role,