Confirmed or unconfirmed cases of 2019 novel coronavirus pneumonia in Italian patients: a retrospective analysis of clinical features

Background Since December 2019, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged as a novel etiologic agent of viral pneumonia. We aimed to compare clinical features of 165 Italian patients with laboratory confirmed or unconfirmed 2019-nCoV pneumonia. Methods On March 31, 2020, hospitalized patients who presented with fever and/or respiratory symptoms, exposures, and presence of lung imaging features consistent with 2019-nCoV pneumonia were included. Before admission to a hospital ward, patients underwent RT-PCR based SARS-CoV-2 RNA detection in their nasopharyngeal swab samples. Results Of 165 patients studied, 119 had positive RT-PCR results and 46 were RT-PCR negative for 2 days or longer (i.e., when the last swab sample was obtained). The median age was 70 years (IQR, 58–78), and 123 (74.6%) of 165 patients had at least one comorbidity. The majority of patients (101/165, 61.2%) had a mild pneumonia, and the remaining patients (64/165, 38.8%) a severe/critical pneumonia. We did not find any substantial difference in symptoms, incubation periods, and radiographic/CT abnormalities as well as in many of the biological abnormalities recorded. However, at multivariable analysis, higher concentrations of hemoglobin (OR, 1.34; 95% CI, 1.11–1.65; P = 0.003) and lower counts of leukocytes (OR, 0.81; 95% CI, 0.72–0.90; P < 0.001) were statistically associated with confirmed COVID-19 diagnosis. While mortality rates were similar, patients with confirmed diagnosis were more likely to receive antivirals (95% vs 19.6%, P < 0.001) and to develop ARDS (63% vs 37%, P = 0.003) than those with unconfirmed COVID-19 diagnosis. Conclusions Our findings suggest that unconfirmed 2019-nCoV pneumonia cases may be actually COVID-19 cases and that clinicians should be cautious when managing patients with presentations compatible with COVID-19.


Background
The 2019 novel coronavirus (2019-nCoV), named severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), became notorious since December 2019 as a new etiologic agent of viral pneumonia [1]. In early illness stages, patients with coronavirus disease 2019 (COVID- 19) present with symptoms of acute respiratory infection, which can progress to acute respiratory distress syndrome (ARDS) and other serious complications [2]. Because of substantial pneumonia-related morbidity and mortality [3], testing for SARS-CoV-2 infection of patients who meet the suspected-case definition for COVID-19 [4] is central for their management. Accordingly, provision of supportive care (e.g., oxygenation, ventilation, and fluid therapy) and/or administration of antiviral agents may be decisive [5].
Real-time reverse-transcriptase-polymerase-chain-reaction (RT-PCR) based SARS-CoV-2 RNA detection in respiratory samples (e.g., nasopharyngeal swabs) is the reference diagnostic method to confirm COVID-19 [6]. However, one or more negative results do not exclude the likelihood of SARS-CoV-2 infection [4]. Currently published studies suggest lung imaging, biomarkers, and other non-microbiological tests as ancillary diagnostic methods [6], encouraging further investigation to understand the value of radiological or laboratory findings to diagnose COVID-19. We comparatively explored the clinical features of 165 patients with laboratory confirmed or unconfirmed 2019-nCoV pneumonia admitted to COVID-19 wards of the Fondazione Policlinico A. Gemelli IRCCS, which is a tertiary care university hospital in Rome, Italy. Thus, we investigated the prospect that cases with a negative RT-PCR test result are actually cases of 2019-nCoV pneumonia or, in other words, are not to distinguish from those without confirmation test performed.

Methods
This retrospective, single-center observational study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Fondazione Policlinico A. Gemelli IRCCS (reference number 17057/20), and written informed consent was obtained from each enrolled patient. All patients who were hospitalized for suspected 2019-nCoV pneumonia [7] on March 31, 2020 were considered for recruitment. Inclusion criteria were fever and/or respiratory symptoms, exposures, and presence of lung imaging features consistent with 2019-nCoV pneumonia [8]. At the emergency room (before admission to a hospital ward), all patients had undergone nasal and oropharyngeal swabs for detection of one or more SARS-CoV-2 specific nucleic acid targets [9], using the Korean Ministry of Food and Drug Safety approved Allplex 2019-nCoV assay (Arrow Diagnostics S.r.l., Genova, Italy). Samples resulted negative had been repeated after 48-72 h. To rule out the presence of infections due to common viral (adenovirus, coronaviruses 229E, HKU1, NL63, OC43, influenza viruses, rhinovirus/enterovirus, etc.) or bacterial (Bordetella pertussis, Chlamydophila pneumoniae, Legionella pneumophila, Mycoplasma pneumoniae) pathogens, patients' samples had also been tested with the GenMark's ePlex Respiratory Pathogen Panel assay. Eleven patients yielding positive results for the etiological agents above mentioned were not included.
We retrieved demographic, clinical, laboratory, imaging, treatment, and outcome data from the patients' medical chart records. We classified cases as mild (see above specification), severe (i.e., dyspnea, respiratory rate ≥ 30 breaths/min, blood oxygen saturation ≤ 93%, and partial pressure of arterial oxygen [PaO 2 ]/fraction of inspired oxygen [FiO 2 ] ≤300 mmHg), or critical (i.e., respiratory failure, septic shock, and/or multiple organ failure) pneumonia. We recorded chest X-ray or computed tomography (CT) features using the Fleischner Society terminology [10], we defined ARDS based on timing, lung imaging, origin of edema, and oxygenation as specified in the Berlin definition [11], and we defined liver injury as transaminase elevation two-to three-times the upper limit of normal (e.g., 45 U/L for alanine aminotransferase). According to the Italian Society of Infectious Diseases guidelines for COVID-19 treatment [12], we administered antiviral agents (e.g., lopinavir/ritonavir) to all patients with severe/critical disease or to patients with mild disease who had comorbidities including age > 70. Categorical variables were expressed as number with percentage and were compared using the χ 2 test, and continuous variables were expressed as median with interquartile range (IQR) and were compared using the Mann-Whitney U-test. A two-sided P value of < 0.05 was considered statistically significant. We performed univariate analysis using the aforementioned tests, and all significant variables (among demographics and baseline characteristics) were included in a multivariable logistic regression model to identify those variables that were statistically associated with confirmed COVID-19 diagnosis. Odds ratio (OR) values with 95% confidence intervals (CIs) were calculated. All analyses were performed with Stata software version 11.1 (Stata-Corp, College Station, TX, USA).

Discussion
We tested the hypothesis that negative patients did not differ from SARS-CoV-2 RNA positive patients by comparing features of 165 cases with clinically diagnosed 2019-nCoV pneumonia in our hospital. We did not find any substantial difference in symptoms, incubation periods, and radiographic/CT abnormalities as well as in many of the biological abnormalities recorded. However, blood/serum test results showed that patients with laboratory-confirmed diagnosis of 2019-nCoV pneumonia were more likely to have higher levels of hemoglobin and lower levels of leukocytes. Additionally, the proportion of ARDS in the group of COVID-19 confirmed patients was significantly higher than in the group of COVID-19 unconfirmed patients.
Large or small descriptive studies published in 2020 mainly focused on patients with laboratory-confirmed 2019-nCoV pneumonia [2,13,14]. Nonetheless, among 72,314 cases (as of February 11, 2020) from the COVID-19 outbreak in China, 16,186 (22%) and 10,567 (15%) of them could not receive laboratory confirmation for COVID-19 and, then, were classified as suspected cases or clinically diagnosed cases, respectively [8]. Although testing for SARS-CoV-2 in our laboratory was not restricted [15], 46 (27.8%) of 165 patients with 2019-nCoV pneumonia did not have a laboratory-confirmed diagnosis in our study. Despite a well-documented active virus replication in the upper respiratory tract [16], swab samples may have a limited sensitivity to identify cases. Sampling or testing related factors may be responsible for false-negative RT-PCR results, necessitating additional sample collection from the lower respiratory tract specimens including sputum [4].
It is plausible that our SARS-CoV-2 negative patients were outside the window of peak shedding in the upper respiratory tract samples or did not have symptoms highly suggestive for upper respiratory tract infection [16]. Accordingly, eight of 119 patients with positive RT-PCR results became positive only with swabs taken on subsequent days after the first (negative) sampled swab, and their median time from symptom onset did not differ from that of 111 remaining patients (7 days vs 7 days, P = 0.76). Furthermore, the sensitivity of the Allplex 2019-nCoV assay might have limited by its requirement that three genes were all detectable for a positive result. In our study, all 46 patients had a radiological evidence of pneumonia not attributed to any typical respiratory viral infection agents, including the human coronaviruses HKU1, OC43, NL63, 229E, the influenza virus A and B, and others (data not shown). It is worthy to note that the finding of typical ground glass opacities in chest CTs of clinically diagnosed patients [17] prompted the Chinese authorities, at one point in early February 2020, to count these patients as confirmed cases [18]. This allowed identifying and quarantining patients as early as possible.
In conclusion, using the clinical diagnosis as the reference standard, the RT-PCR testing allowed to correctly identify two thirds of our patients as COVID-19 while one third was not correctly  Laboratory-based confirmation of 2019-nCoV pneumonia was done by SARS-CoV-2 RNA detection using a well-established RT-PCR assay [9] b According to multivariable logistic regression model analysis (see text for details) c Includes anemia, endocrine disorders, inflammatory bowel disease, and obesity d CT findings in all 112 patients were assessed according to imaging features described elsewhere [10] e Not calculated because the relative variable was not entered into multivariable logistic regression model (see text for details) f Patients were admitted between 3/6/2020 and 3/31/2020, with follow-up through 4/23/2020 identified. However, as undocumented SARS-CoV-2 infections may be a relevant source of transmission among hospitalized patients [19], we believe that clinicians should exceed on the side of caution when managing patients with presentations compatible with COVID-19.