Can routine laboratory tests discriminate SARS‐CoV‐2‐infected pneumonia from other causes of community‐acquired pneumonia?

Abstract Background The clinical presentation of SARS‐CoV‐2‐infected pneumonia (COVID‐19) resembles that of other etiologies of community‐acquired pneumonia (CAP). We aimed to identify clinical laboratory features to distinguish COVID‐19 from CAP. Methods We compared the hematological and biochemical features of 84 patients with COVID‐19 at hospital admission and 221 patients with CAP. Parameters independently predictive of COVID‐19 were calculated by multivariate logistic regression. The receiver operating characteristic (ROC) curves were generated and the area under the ROC curve (AUC) was measured to evaluate the discriminative ability. Results Most hematological and biochemical indexes of patients with COVID‐19 were significantly different from patients with CAP. Nine laboratory parameters were identified to be predictive of a diagnosis of COVID‐19. The AUCs demonstrated good discriminatory ability for red cell distribution width (RDW) with an AUC of 0.87 and hemoglobin with an AUC of 0.81. Red blood cell, albumin, eosinophil, hematocrit, alkaline phosphatase, and mean platelet volume had fair discriminatory ability. Combinations of any two parameters performed better than did the RDW alone. Conclusions Routine laboratory examinations may be helpful for the diagnosis of COVID‐19. Application of laboratory tests may help to optimize the use of isolation rooms for patients when they present with unexplained febrile respiratory illnesses.

F I G U R E 1 General characteristics of hematological parameters among healthy control (HC), CAP patients, and COVID-19 patients. *P < .05, **P < .01, n.s, not significant T A B L E 1 Laboratory values at the time of admission to hospital for COVID-2019 patients and CAP patients

Variable
Healthy control (n = 120) CAP (n = 221) COVID-2019 (n = 84) P The SARS-CoV-2 is considered a relative of the deadly Middle East respiratory syndrome (MERS) and severe acute respiratory syndrome (SARS) coronaviruses. The COVID-19 patients are characterized by pneumonia symptoms, such as fever, radiographic evidence of pneumonia, respiratory symptoms, and possibly transmitted from animal to human. [3][4][5] The public health authorities proposed COVID-19 definitions that combined clinical features (e.g., fever, cough, and anhelation) with epidemiological factors (e.g., travel to a seafood and wet animal wholesale market in Wuhan or direct contact with another patient with COVID-19) to improve diagnostic accuracy. 6,7 Unfortunately, these epidemiological features are not specific and have poor positive predictive value during the outbreak.
The COVID-19 appears to cause symptoms similar to other etiologies of community-acquired pneumonia (CAP) based on clinical data from 41 COVID-19 patients, 3 and can spread from humans to humans. 3,8 Distinguishing COVID-19 from other etiologies of CAP is one of the major challenges of the COVID-19 outbreak. Despite recommendations that examining hematological and biochemical parameters as part of the diagnostic workup for COVID-19, 3,9 it is urgent to evaluate the ability of these features to accurately discriminate cases of COVID-19 from cases of CAP. Thus, we conducted the current study aiming to evaluate the ability of routine laboratory tests for distinguishing COVID-19 from other etiologies of CAP and help health workers to effectively, quickly, and calmly deal with COVID-19.

Patients and data collection
To determine the ability of routine laboratory tests to differentiate COVID-19 from CAP due to other causes, we compared the hematological and biochemical data of COVID-19 and CAP patients. The 84 COVID-19 patients presented to our hospital from December 26, 2019 to January 30, 2020. The patients were laboratory confirmed SARS-CoV-2 infection by real-time RT-PCR. The CAP group consisted of 221 patients who visited our hospital from January 2018 to December 2018. All of the medical records were reviewed retrospectively. The inclusion criteria of CAP group consisted of (a) patients had ≥2 symptoms and signs of CAP and had evidence of pneumonia revealed by the emergency department physician or internal medicine consultant, (b) patients with a complete record of hematological and biochemical indicators, and (c) hospitalized patients. Exclusion criteria were as follows: (a) patients deficient in clinical hematological and biochemical data and (b) outpatient. Patients were diagnosed according to the World Health Organization interim guidance for COVID-19 and divided into mild and severe groups. Healthy controls included 120 healthy people who made the physical check-up in our hospital from December 13, 2019 to December 17, 2019. The clinical data collection from patients was approved by the Ethics Committee of Zhongnan Hospital of Wuhan University. Written informed consent was

Statistical analysis
Statistical analyses were conducted using IBM SPSS version 22.0 software. Statistical analysis for the results was performed using the Student's t-test for only two groups or using one-way analysis of variance for more than two groups. The receiver operating characteristic (ROC) curves were generated, and the area under the ROC curve (AUC) was measured to evaluate the discriminative ability. 10 Higher AUC was considered to show better discriminatory ability as follows: excellent, AUC of ≥0.90; good, 0.80 ≤ AUC < 0.90; and fair, 0.70 ≤ AUC < 0.80. A P-value of <0.05 was considered statistically significant.

RESULTS
We  Table 1 and Figure 1. Both COVID-19 and CAP patients had lower mean lymphocyte and platelet counts than healthy control. COVID-19 patients had significantly lower mean values for white blood cell The proportion of patients with abnormal laboratory features is presented in Table 2. Both COVID-19 (81.0%) and CAP (56.6%) patients had lymphopenia. A significantly higher proportion of COVID-19 patients presented reduced WBC and EO, normal basophile, increased AST and CRP, and normal PCT, whereas a significantly higher proportion of CAP patients had a decrease of RBC, HGB, HCT, ALB, and increase of neutrophil, monocyte, and RDW.
Multivariate analysis demonstrated the laboratory indexes independently discriminating COVID-19 from CAP. The OR of the factors to predict COVID-19 versus CAP is shown in Table 3. The ROC curves and AUC (Figure3) demonstrated that RDW, MCHC, and HGB had good discriminatory ability. RBC, ALB, EO, HCT, ALP, and MPV had fair discriminatory ability. Furthermore, the cutoff value for RDW could be 13.35 that provided a reasonable sensitivity (79.8%) or specificity (84.6%). When ROC curves were calculated for combi-nations of these three parameters, improvement in AUC was presented, with the maximum AUC being 0.89.

DISCUSSION
This study indicates that several hematological and biochemical abnormalities occur more frequently in CAP patients than in COVID-19 patients. The statistically significant difference in mean values was noted for most laboratory features tested. However, to be useful in diagnosis, the overlap in the distribution of the results for patients with COVID-19 and those with CAP must be small. SARS-CoV-2 spread quickly in China since the first official announcement in December 2019 by the Wuhan Municipal Health Commission. 11,12 Possibly, the Spring Festival travel rush that millions of people were on the way home and SARS-CoV-2 would spread fast especially along with people coming out from Wuhan. Because the confirmation diagnosis of COVID-19 now mainly relies on PCR assays for the detection of SARS-CoV-2, 13 initial medical examination of potential COVID-19 depends primarily on clinical, radiographic, and epidemiological features. 9 Physicians and public health workers keep struggling with the difficult task of evaluating patients for COVID-19 presenting with unknown febrile respiratory illnesses. As a result, public health authorities went on revising COVID-19 definitions to improve the accuracy of diagnosis.
Our findings demonstrated that nine laboratory features independently predictive of discernibility between COVID-19 and CAP. The identification of the RDW, HGB, and MCHC as the best discriminatory ability. The discriminatory ability likely results from elevated HGB and MCHC while low RDW seen in COVID-19. Our study also highlighted laboratory parameters that are common in both COVID-19 and CAP, and therefore not useful in differentiating the two diseases. Lymphocytopenia is characteristic and of similar magnitude for both COVID-19 and CAP. However, lymphocytopenia in COVID-19 was also accompanied by depletion of EO and normal basophile, whereas it was accompanied by reduced RBC, elevated neutrophil count, and monocyte count in CAP.
Because the CAP cohort did not have laboratory indexes over time, trends in laboratory values were unable to perform after hospital admission. Therefore, more significant differences in the laboratory indexes might occur later in the illness between COVID-19 and CAP patients. Our findings indicate that simple laboratory tests may help to distinguish COVID-19 from CAP. Application of these tests together with epidemiological data may be helpful to avoid misdiagnosis of COVID-19 as CAP and shorten the time of isolation of patients with respiratory symptoms.

ACKNOWLEDGMENTS
This work was supported by the National Key Research and Development Program of China (2018YFE0204500). We acknowledge all health care workers involved in the diagnosis and treatment of patients in Wuhan.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.