Results

Patients’ demographic characteristics

During the RV and SARS-CoV-2 study periods, 508 and 596 patients were included, respectively. Patients were more frequently male during both periods: 57% (289/508) and 59% (354/596), respectively. The median age was 73 [63–85] years and 60 [46–76] years during the RV and SARS-CoV-2 periods, respectively (p < 0.001). Detailed patients’ characteristics are depicted in Table 1.
Patients with a negative PCR presented similar characteristics between both periods, except for age – patients were older during the SARS-CoV-2 period (73 [63–85] and 63 [43–72] years, p < 0.001) – and cardiac failure, which was more prevalent during the SARS-CoV-2 period (13% vs 5.8%, p < 0.002; Table 1). Due to these few, but significant, differences between the two negative-PCR populations from the two periods, we conducted separate analyses of the factors associated with a positive PCR for any RV (in the RV period) and SARS-CoV-2 (in the SARS-CoV-2 period) in comparison with their own PCR-negative populations.

Virological findings

During the first study period, 216/508 (43%) patients had a positive mPCR, including 13 dual viral coinfections, leading to the following viral distribution: 68 (31%) rhinovirus, 60 (28%) influenza, 35 (16%) RSV, 30 (14%) human metapneumovirus, 28 (13%) coronaviruses, four (2%) parainfluenza, three (0.1%) adenovirus and one (0.5%) bocavirus. During the second study period, 268/596 (45%) patients had a positive PCR for SARS-CoV-2, and 70 other respiratory viral infections were also identified, including 18 co-infections with SARS-CoV-2. As patients’ populations presented some differences across the two periods, patients with a non-SARS-CoV-2 virus were analysed as SARS-CoV-2-negative patients, while those with a SARS-CoV-2 co-infection were considered as SARS-CoV-2-positive.

Patients’ characteristics associated with the detection of a non-SARS-CoV-2 respiratory virus

The median duration of symptoms was 2 [1–4] days before ED admission. The following characteristics were associated with a positive mPCR in the univariate analysis (Table 1): younger age, presence of fever, cough and expectorations, and lower white blood cell and platelet counts. These variables were also retrieved in the multivariate analysis, except for lower white blood cell count (Table 2).

Clinical and biological characteristics associated with the detection of SARS-CoV-2

The median duration of symptoms was 3 [2–7] days, and the following characteristics were associated with a positive SARS-CoV-2 PCR in the univariate analysis (Table 1): male gender, younger age, fever, chills, myalgia, cough, bilateral cracklings, diabetes, chronic lung disease, history of stroke, higher respiratory rate, CRP, lower NT-proBNP, lower leukocytes and lymphocytes counts and lower platelet count. Regarding comorbidities, chronic lung disease, diabetes and a history of stroke were associated with a positive SARS-CoV-2 PCR. In the multivariate analysis, the following features remained associated with SARS-CoV-2 detection (Table 2): younger age, male gender, fever, chills, myalgia, higher respiratory rate and absence of chronic lung disease. A positive PCR for SARS-CoV-2 was also associated with ICU admission (OR, 8.4; 95% CI, 4.6–16.8; p < 0.001) and with mortality during hospitalization (OR, 3.0; 95% CI, 1.8–5.2; p < 0.001).

Comparison of human respiratory viruses and characteristics associated with SARS-CoV-2

Under the hypothesis of SARS-CoV-2 co-circulation along with other RVs, variables associated with the virologic diagnosis (SARS-CoV-2 vs other viruses) among positive PCR were as follows (Table 2): male gender (aOR, 1.64; 95% CI, 1.10–1.44; p = 0.015), absence of chronic lung disease (0.40, 0.25–0.64, <0.001), age (0.80 per 10 years, 0.72–0.89, <0.001), presence of fever (3.27, 2.17–4.94, <0.001) and absence of expectorations (0.13, 0.07–0.26, <0.001). A clinical score based on this multivariate analysis was computed using integer coefficients based on OR logarithms (Supplementary Table S2), ranging from -6 to +5. Under the baseline scenario conditions, corresponding to the overall period of the study with a prevalence of 43% of positive PCR and a distribution of 56% and 44% of SARS-CoV-2 and non-SARS-CoV-2 viruses, respectively, among positive PCR, our score has an AUC of 0.81 (95% CI 0.77–0.85; Supplementary Figure S1). Furthermore, when using a score greater than 1 to predict SARS-CoV-2 among all viral infections, we observed a sensitivity and a specificity of 83% and 65%, respectively. To note, with a score greater than or equal to 1, the sensitivity and specificity were at 91% and 52%, respectively.
To test the robustness of our observations in case of variable prevalence of SARS-CoV-2 among all RVs, two other scenarios were added, assuming the same global positivity rate of 43% with a dominant or a limited SARS-CoV-2 distribution circulation among positive PCR at 75% and 25%, respectively. The following factors were associated with COVID-19 diagnosis in all scenarios: younger age, male gender, fever, myalgia, dyspnea, absence of expectoration and absence of chronic lung disease (Cf. Supplementary Table S1). The positive predictive value (PPV) and the negative predictive value (NPV) with a score greater than one were at 75 and 76% with the viral distribution observed in our study, respectively. Under the dominant and the limited SARS-CoV-2 distribution scenarios (i.e. 75% and 25% probability of SARS-CoV-2 among positive patients), the PPVs were at 92% and 57%, respectively, and the NPVs were at 45 and 88%, respectively.