Burden of Influenza and Respiratory Syncytial Viruses in Suspected COVID-19 Patients: A Cross-Sectional and Meta-Analysis Study

Non-SARS-CoV-2 respiratory viral infections, such as influenza virus (FluV) and human respiratory syncytial virus (RSV), have contributed considerably to the burden of infectious diseases in the non-COVID-19 era. While the rates of co-infection in SARS-CoV-2-positive group (SCPG) patients have been determined, the burden of other respiratory viruses in the SARS-CoV-2-negative group (SCNG) remains unclear. Here, we conducted a cross-sectional study (São José do Rio Preto county, Brazil), and we collected our data using a meta-analysis to evaluate the pooled prevalence of FluV and RSV among SCNG patients. Out of the 901 patients suspected of COVID-19, our molecular results showed positivity of FluV and RSV in the SCNG was 2% (15/733) and 0.27% (2/733), respectively. Co-infection with SARS-CoV-2 and FluV, or RSV, was identified in 1.7% of the patients (3/168). Following our meta-analysis, 28 studies were selected (n = 114,318 suspected COVID-19 patients), with a pooled prevalence of 4% (95% CI: 3–6) for FluV and 2% (95% CI: 1–3) for RSV among SCNG patients were observed. Interestingly, FluV positivity in the SCNG was four times higher (OR = 4, 95% CI: 3.6–5.4, p < 0.01) than in the SCPG. Similarly, RSV positivity was significantly associated with SCNG patients (OR = 2.9, 95% CI: 2–4, p < 0.01). For subgroup analysis, cold-like symptoms, including fever, cough, sore throat, headache, myalgia, diarrhea, and nausea/vomiting, were positively associated (p < 0.05) with the SCPG. In conclusion, these results show that the pooled prevalence of FluV and RSV were significantly higher in the SCNG than in the SCPG during the early phase of the COVID-19 pandemic.


Cross-Sectional Study, Specimen Collection, and Processing
From May to September 2020, clinical samples with a clinical suspicion of COVID-19 were collected by healthcare professionals and their close contacts in the municipality of São José do Rio Preto, Brazil. During this period, the definition of suspicion of the disease was based on the presence of symptoms such as fever or acute respiratory discomfort (nasal congestion, cough, sore throat, and shortness of breath), headache, loss or alteration of smell or taste, ocular symptoms, myalgia, arthralgia, fatigue, nausea, vomiting, and diarrhea. The demographic details collected included sex, age, date of onset of symptoms, and place of collection. Clinical samples consisted of oro-nasopharyngeal swabs soaked in 3 mL of 0.9% saline sent to the Laboratório de Estudos Genômicos, Unesp, São José do Rio Preto, for molecular diagnosis. Therefore, following RNA extraction with a purification system using magnetic beads [24], a quantitative reverse transcription-polymerase chain reaction (qRT-PCR) test was performed [25]. After diagnosing for SARS-CoV-2, aliquots of samples stored at −80 • C were retrospectively screened, according to the established protocol, using a multiplex qRT-PCR assay for a panel of respiratory viruses: Alphainfluenzavirus influenzae (FluAV), Betainfluenzavirus influenzae (FluBV), and RSV [26].
Descriptive statistics were used, and demographic variables (i.e., sex and age) were used for the analysis between the groups via the chi-squared test, with a p-value of <0.05.

Systematic Review and Meta-Analysis
We performed a systematic review and meta-analysis according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) protocol (Table S1) [27]. The study protocol is available in PROSPERO with the registration number CRD42022367248 and can be viewed at https://www.crd.york.ac.uk/prospero/display_record.php?ID= CRD42022367248, accessed on 9 February 2023.

Search Strategy
After defining the research protocol, we performed a systematic search of PubMed, SciELO, and Google Scholar. Peer-reviewed articles in English, Spanish, or Portuguese were screened, and a combination of descriptors was used: ("SARS-CoV-2", OR "COVID-19", OR "SARS-CoV-2 negative", OR "non-SARS-CoV-2", OR "COVID-19 negative", OR "non-COVID-19") AND ("non-respiratory viruses" OR "influenza virus", OR "influenza", OR "respiratory syncytial virus", OR "RSV"). We also identified additional studies by screening the reference lists of the selected articles and highly cited reviews on the topic of interest.
The following inclusion criteria were used: (1) original articles published in scientific journals that contained information on molecular surveys for the detection of FluAV/FluBV and RSV in non-COVID-19 individuals during the COVID-19 pandemic; (2) studies containing data related to the proportion and rate of viral infection using laboratory tests; (3) demographic data from FluV-and RSV-positive individuals for the analysis of other factors such as sex, collection period, and age; and (4) all target studies published during the COVID-19 pandemic. The following exclusion criteria were implemented in our search strategy: (1) absence or confusing specification of the outcome of interest regarding FluV/RSV positivity in laboratory tests; (2) revisions, book chapters, and molecular and seroprevalence studies not involving humans; and (3) small-scale molecular epidemiology studies with a sample size <20.

Data Analysis
For all the selected studies, various criteria such as demographic information (i.e., first author, year of publication, place of study, and baseline characteristics of study participants including mean age, sex percentage, method of diagnosis, the proportion of positive humans investigated for respiratory virus infection, and clinical signs or symptoms) were used to analyze the data. The Newcastle-Ottawa Scale (NOS) was used to evaluate the risk of bias in the included studies [28]. The NOS evaluates group comparability and study group selection and is used to verify the exposure or outcome of interest in observational studies. NOS components were summed, and final scores were classified as follows: 0-2 low quality, 3 moderate quality, and 4-5 high quality [28].

Statistical Analyses
Data were extracted using Microsoft Excel. Several tables were generated comprising dichotomous data (presence or absence of respiratory viruses infection) for the relative and cumulative calculation of the frequencies of the outcomes of interest, with 95% confidence intervals (CIs) calculated whenever possible. Meta-analysis was conducted using STATA IC/64 version 13.1 software (Stata Corporation, College Station, TX, USA) using the metaprop, metafunnel, and metaninf commands. The prevalences of FluV, RSV, and SARS-CoV-2 were determined by dividing the confirmed number of cases by the pooled denominator, with the results expressed as percentages. The variance of each frequency estimate (known as ES (effect sizes)) was calculated as pq/n, where p is the frequency, q is 1−p, and n is the total number of individuals screened [29]. The CIs for the average ES were calculated using the formula 95% CI = ES ± 1.96 × SE, where SE is the standard error (SE = √ (pq/n)). To ensure proportionate weight distribution to the studies presenting extreme frequency (near 0 or 1), we applied the Freeman-Tukey arcsine methodology [30][31][32]. Heterogeneity, that is, any kind of variability among the results of the studies, was assessed using the I 2 index and Cochran's Q test. I 2 values of 25%, 50%, and >75% indicated low, moderate, and high heterogeneities, respectively, among the studies. Cochran's Q test p-value < 0.05 was consistent with significant heterogeneity (I 2 > 75%) [32]. Due to the nature of the studies, the existence of heterogeneity was expected; therefore, we chose to use the random-effect model for the meta-analysis as proposed by DerSimonian and Laird [33]. We performed a sensitivity analysis to test the effect of the influence of each study on the overall estimate. Furthermore, we conducted a subgroup analysis to reduce the possibility of heterogeneity. Publication bias was determined through a visual inspection of Begg's funnel plot as well as with Egger's test calculations. For all tests, p-values < 0.05 were considered to be statistically significant [34,35].

Systematic Review and Meta-Analysis: Characteristics of the Included Studies
We found 1143 reference articles in the electronic database using the descriptor combination established in the "Search Strategy" section. After the removal of duplicates (401), the title and abstract of the remaining records were screened for eligibility (681). Of these, 61 records were selected for reading their full text. As a result, 27 articles that met the eligibility criteria were included in this meta-analysis . A flowchart of this selection step is shown in the Figure S1.

Clinical Features in the SCNG and SARS-CoV-Positive Group (SCPG)
Epidemiological trends regarding the clinical characteristics of the SCNG versus the SCPG showed that fever, cough, sore throat, headache, myalgia, diarrhea, and nausea/vomiting were positively associated with the SCPG. In contrast, rhinorrhea was positively associated with SCNG. Difficulty breathing and abdominal pain showed no significant association between the two groups ( Figure 1) [42,46,47,51,54,59].

Publication Bias and Sensitivity Analysis
For the analysis of publication bias, the funnel plot indicated that the existence of a selection bias was unlikely in relation to FluV, FluBV, and sample sizes among SCNG patients (Figures S5-S7). Furthermore, using Egger's test, we also found no sign of publication bias (i.e., bias coefficient > 0.05). However, publication bias was only observed in studies that estimated FluAV and RSV positivity among SCNG patients (p < 0.05). Four studies were observed to have zero viral positivity contributing to this outcome [41,51,57,60]. Regarding the sensitivity analysis, to assess each individual study on the combined viral prevalence via the removal of individual studies, no study significantly affected the combined positivity ( Figures S8-S10).

Discussion
This cross-sectional study and meta-analysis showed the prevalence of FluV and RSV in patients with a clinical suspicion of COVID-19 in 2020/mid-2021 . Interestingly, the pooled prevalence of 4% for FluV and 2% for RSV in the SCNG was significantly higher than in the co-infection group (i.e., FluV/RSV positivity plus SARS-CoV-2), which had a pooled prevalence of 0% (p < 0.001). Compared with non-COVID-19 patients, coldlike symptoms were more positively associated (p < 0.05) with confirmed cases of COVID-

Publication Bias and Sensitivity Analysis
For the analysis of publication bias, the funnel plot indicated that the existence of a selection bias was unlikely in relation to FluV, FluBV, and sample sizes among SCNG patients ( Figures S5-S7). Furthermore, using Egger's test, we also found no sign of publication bias (i.e., bias coefficient > 0.05). However, publication bias was only observed in studies that estimated FluAV and RSV positivity among SCNG patients (p < 0.05). Four studies were observed to have zero viral positivity contributing to this outcome [41,51,57,60]. Regarding the sensitivity analysis, to assess each individual study on the combined viral prevalence via the removal of individual studies, no study significantly affected the combined positivity ( Figures S8-S10).

Discussion
This cross-sectional study and meta-analysis showed the prevalence of FluV and RSV in patients with a clinical suspicion of COVID-19 in 2020/mid-2021 . Interestingly, the pooled prevalence of 4% for FluV and 2% for RSV in the SCNG was significantly higher than in the co-infection group (i.e., FluV/RSV positivity plus SARS-CoV-2), which had a pooled prevalence of 0% (p < 0.001). Compared with non-COVID-19 patients, cold-like symptoms were more positively associated (p < 0.05) with confirmed cases of COVID-19, including fever, cough, sore throat, headache, myalgia, diarrhea, and nausea/vomiting. Collectively, our study provides evidence of the early stages of the COVID-19 pandemic.
The retrospective detection of FluV and RSV, as well as other pathogenic agents, is necessary because, during the initial phase of the pandemic, all efforts were focused on the detection of SARS-CoV-2. Even though the 2567 reported cases of SARS due to FluV were fewer in 2020 than those detected in previous years [63], the positivity of FluV and RSV generated in this cross-sectional study suggests a high possibility of under-reporting in hospitalized cases.
In an analysis stratified according to FluV subtypes, a higher prevalence of FluAV than FluBV was observed, with rates of 3% (95% CI: 2-5) and 1% (95% CI: 0-2), respectively. These data agree with the literature showing a higher incidence of FluAV than FluBV [64]. Here, the 25% FluBV infection burden was close to the 23% positivity observed by Caini et al. [65] among all FluV cases. Indeed, the highest proportion of FluAV positivity is related to its pandemic risk, as new strains, with varying degrees of virulence, can arise from a large host range with "species jumpers" to humans and spread on a global scale. This was not observed for FluBV because of its smaller host range [66].
A greater burden of RSV has been recognized in some demographic groups, particularly in children, seniors, immunocompromised and hospitalized patients, and those with comorbidities [11][12][13]. In contrast, in our cross-sectional study, the highest prevalence of RSV occurred in young adults aged 23, 32, and 40 years. However, it was not possible to determine whether these RSV cases were related to other comorbidities/immunocompromised. Notably, few studies have explored the burden of RSV in a young adult population. To reduce this gap, Htar et al. [67] conducted a meta-analysis including articles evaluating patients with community-acquired pneumonia, acute respiratory infection, and general practice with adults aged <50 years. As a result, the pooled estimate of the proportion of RSV cases ranged from 1% to 9%. Regarding RSV subtypes A and B, despite the predominance of one, both often co-circulate [68,69]. In this sense, our data showed similar prevalences of RSV-A and -B. Data related to differences in virulence have suggested a greater virulence of RSV-A than -B [70,71]. Interestingly, in preliminary observations, the new RSV-A genotype ON1 has been associated with more severe disease in young children [72][73][74]. Owing to the limited data provided by these studies, it was not possible to perform an in-depth analysis of RSV genotypes versus disease severity. Consequently, we encourage future studies to clarify this gap.
FluV/RSV and SARS-CoV-2 co-infection events have shown a very low proportion of positivity during just over a year of the COVID-19 pandemic. Indeed, our data showed that the chance of FluV/RSV positivity in a mono-infection group (i.e., FluV-/RSV-positive SCNG patients) was two to four times higher than that in co-infection cases in the SCPG. However, owing to the limitation of raw data, it was not possible to perform an in-depth analysis of whether there were differences in laboratory-clinical characteristics between the two groups. In this regard, controversial data have been obtained in other studies. For instance, a meta-analysis that focused on the clinical severity of COVID-19 patients with FluV/RSV co-infection showed a positive association between FluV and SARS-CoV-2 only in terms of a higher risk of intensive care unit admission and the need for mechanical ventilation [20]. However, Krumbein et al. [21] found an increased association between co-infected patients with dyspnea and fatality.
Curiously, the higher FluV and RSV positivity in the SCNG clinical samples when compared to the SCPG samples may be due to the short co-detectability window to adequately detect all co-infection cases via nucleic acid amplification assays [22]. This is corroborated by the evidence related to the increase in co-infection among SARS-CoV-2 patients from the second half of 2021, when there was a relaxation of non-pharmaceutical interventions (NPIs) (i.e., social distancing, respiratory etiquette, and hand hygiene) and increased circulation of non-SARS-CoV-2 respiratory viruses [75].
Another possible explanation for the low co-infection positivity among SARS-CoV-2 and other respiratory viruses (i.e., FluV/RSV) could be related to negative-type viral interactions. In this scenario, the cells infected by one virus would reduce the infection and replication of another virus. More specifically, the interferon response induced by a prior viral infection would be crucial in an antiviral state, hindering the establishment of infection and replication of a second virus. Interestingly, Fage et al. observed that during simultaneous infection, SARS-CoV-2 interfered with the replication of RSV-A2a in nasal human airway epithelium cells [76]. This was not observed for FluAV (H1N1)pdm09 virus. The authors also observed that the prior infection of this cell type by SARS-CoV-2 reduced the replication kinetics of both viruses. Pizzorno et al. explored additional key factors in virus-virus interactions and demonstrated that superinfection with H1N1 slightly reduced SARS-CoV-2 infection, and superinfection with SARS-CoV-2 had a substantial impact on primary RSV infection [77]. These findings reveal an essentially relevant point in the context of the interaction between other respiratory viruses and SARS-CoV-2.
Flu vaccine coverage and NPIs, which tended to be more stringent during the early part of the COVID-19 pandemic, considerably mitigated the incidence of FluV and influenced our pooled estimate of prevalence [78]. Therefore, it is evident that changes in NPIs, including implementation dates, duration, and the rigor of following up on such measures, considerably affected the respiratory virus positivity rate. Globally, the 2020/2021 seasonal outbreak/epidemic was drastically reduced (i.e., rhinovirus, RSV, and seasonal human CoV) [79]. However, the incidence of viral seasonal outbreaks and epidemics increased in the period mid-2021/2022 owing to a reduction in NPIs [8,80,81]. Unfortunately, owing to data limitations, it was not possible to verify whether there was still a reduced trend of co-infection with SARS-CoV-2 for the group of patients from the second half of 2021 and 2022.
Although this study is relevant, including the fact that, to the best of our knowledge, this is the first meta-analysis focusing on the burden of FluV/RSV in patients who tested negative for SARS-CoV-2, we emphasize that there are limitations. First, a high degree of heterogeneity was observed across most studies. Thus, despite well-defined criteria for selecting studies to compose the review, there were variables that could influence the analyzed population, including a wide range of ages, sex ratios, collection sites, vaccination status, and collection periods. All these variables tend to be factors that contribute to the diversity of the results. Second, for several studies, it was not possible to access the raw data, thus precluding the analysis of potential risk variables associated with viral positivity, such as FluV and RSV seasonality, the vaccination status of individuals, as well as other demographic data such as sex, age, and the presence of comorbidities in the population. Third, articles published in languages other than English, Spanish, or Portuguese were not included, with the possibility of publication bias. Fourth, the influenza vaccine coverage data were not available, which may imply a bias in the comparison between the SCNG and SCPG.

Conclusions
In summary, our study revealed a low FluV and RSV prevalence rate of 2% and 0.27%, respectively, among SCNG patients. Regarding the meta-analysis, the pooled prevalence of 4% for FluV and 2% for RSV in the SCNG was relatively low. However, the prevalence was significantly higher than that of the tested SCPG. The results of this study and continued molecular surveillance across different regions in the coming months will help to inform and deepen our understanding of the effects of viral interactions among the non-SARS-CoV-2 viral infections versus SARS-CoV-2 positivity concerning the epidemiological profile of both viruses.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/v15030665/s1, Supplementary File S1. Figure S1-PRISMA 2009 checklist showing guidelines for systematic reviews and meta-analysis. Figure S2-Forest plot containing positivity according to FluV and RSV subtypes. Figure S3-FluV in relation to the year of collection of biological samples. Figure S4-RSV positivity in relation to the year of collection of biological samples. Figure S5-Funnel plot for FluV positivity among SCNG. Figure S6-Funnel plot for FluAV positivity among SCNG. Figure S7-Funnel plot for FluBV positivity among SCNG. Figure  S8-Sensitivity analysis for the effect of removing one study at a time on the overall estimation of FluV positivity in SCNG. Figure S9-Sensitivity analysis for the effect of removing one study at a time on the overall estimation of FluAV positivity in SCNG. Figure S10-Sensitivity analysis for the effect of removing one study at a time on the overall estimation of FluBV positivity in SCNG. Table  S1-PRISMA 2009 checklist. Table S2-The characteristics of the studies included in the systematic review and meta-analysis. Table S3-Methodologic quality of cohort and case-control studies included in the meta-analysis.  Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee, following Brazilian regulations ethical standards (protocol number: 33202820.7.1001.5348, 08/2020).