Respiratory Viral Infections in Pragmatically Selected Adults in Intensive Care Units: A Prospective Cohort Study

Cong-Tat Cia Department of Internal Medicine, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University I-Ting Ling Department of Pathology, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University Jen-Chieh Lee Department of Internal Medicine, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University Huey-Pin Tsai Department of Pathology, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University Jen-Ren Wang Department of Pathology, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University Wen-Chien Ko (  winston3415@gmail.com ) Department of Medicine, College of Medicine, National Cheng Kung University


Abstract Background
Respiratory viruses can be detected in 18.3% to 48.9% of critically ill adults with severe respiratory tract infections (RTIs). The present study aims to assess the clinical signi cance of respiratory viruses in pragmatically selected adults in medical intensive care unit patients and to identify factors associated with viral respiratory viral tract infections (VRTIs).

Methods
We conducted a prospective study on critically ill adults with suspected RTIs without recognized respiratory pathogens. Viral cultures with monoclonal antibody identi cation, in-house real-time polymerase chain reaction (PCR) for in uenza virus, and FilmArray® respiratory panel were used to detect viral pathogens. Multivariable logistic regression was applied to identify factors associated with VRTIs.

Results
Sixty-four (40.5%) of the included 158 critically ill adults had respiratory viruses detected in their respiratory specimens. The commonly detected viruses included in uenza virus (20), followed by human rhinovirus/enterovirus (11), respiratory syncitial virus (9), human metapneumovirus (9), human parain uenza viruses (8), human adenovirus (7), and human coronaviruses (2). The FilmArray® respiratory panel detected respiratory viruses in 54 (34.6%) patients, but showed negative results for seven of 13 patients with in uenza A/H3 infection. In the multivariable logistic regression model, patient characters associated with VRTIs included those aged < 65 years, household contact with individuals with upper RTI, the presence of fever, cough with sputum production, and sore throat.

Conclusions
Respiratory viruses were not uncommonly detected in the pragmatically selected adults with critical illness. Selective multiplex PCR testing for respiratory viruses is a practical strategy, and the detection rate could be further improved in selected patient population.

Background
Respiratory viruses can cause upper and lower respiratory tract infections (RTIs) including seasonal colds, otitis media, bronchiolitis, pneumonia, and acute deterioration of chronic lung diseases, in patients with and without suppressed immunity 1,2 . Viral pathogens may cause 24.5% of cases of communityacquired pneumonia requiring hospitalization in adults, as modern multiplex molecular diagnostic assays are applied 3,4 . Moreover, up to 10% of community-acquired pneumonia is concurrently caused by viral and bacterial pathogens 3 .
Respiratory viruses had been detected in respiratory specimens in 18.3% of critically ill adults requiring invasive mechanical ventilation 5 . The detection rates are higher, ranging from 20.5-49.0%, in patients with community or hospital-acquired lower respiratory tract infections admitted to intensive care units (ICUs) 5-14 . Human rhinovirus, in uenza virus, and human parain uenza viruses (HPIVs) are the most frequently detected viruses 5-9 . In terms of disease severities requiring timely actions, critically ill patients are considered appropriate candidates for respiratory viral panel testing 15 .
As the expense of the molecular tests remains high, universal testing for all patients with suspected RTIs may not be a cost-effective strategy, since more than half events are caused by bacteria or other noninfectious etiologies 3 . Proper selection for patients to test is required to maximize the clinical bene ts in medical care, either by early initiation of antiviral agents, avoidance of unnecessary invasive studies, or shorter antibiotic exposure 15 . Patients with known bacterial pathogens may be less likely to bene t from testing for respiratory viruses, since discontinuation of antimicrobial therapy is not a reasonable option for them. The current study aims to assess the clinical signi cance of respiratory viruses in pragmatically selected adults in medical ICUs and to delineate clinical variables associated with viral RTIs (VRTIs).

Study design and population
We conducted a prospective study from May 2017 to December 2018 at a 42-bed medical ICU in a tertiary hospital with more than 1300 beds in southern Taiwan. The cases aged at least 20 years with suspected VRTIs were considered for inclusion. Similar to the real-world practice, those with positive results of microbiological tests indicative of speci c causative pathogens, such as positive rapid antigen tests, blood cultures, or signi cant bacteria on Gram stain for endotracheal aspirates, were excluded. Written informed consents were obtained from the patients or surrogate decision-makers. The study was approved by the institutional review board of National Cheng Kung University Hospital (IRB No.: B-ER-105-350). All experiments were performed in accordance with relevant guidelines and regulations Clinical data collection Electronic medical records of included patients were reviewed for demographic variables, clinical manifestations, contact history, chronic illness, laboratory results within 48 hours upon presentation, virological studies, causes of respiratory distress, mechanical ventilation, acute respiratory distress syndrome (ARDS), concurrent non-viral respiratory pathogens, severity scores grading by Acute Physiology and Chronic Health Evaluation II (APACHE II) on the rst ICU day and Sequential Organ Failure Assessment (SOFA) score on the day of rst respiratory specimen collection, length of ICU or hospital stay, duration of mechanical ventilation, ICU and 28-day mortality. In addition, one of the authors (CT Cia or JC Lee) obtained symptom details from the patients or their relatives upon inclusion.
The number of critically ill adults with virologically con rmed in uenza during the study period was retrieved from severe in uenza noti cation records, to quantify the clinical burden of in uenza, as some cases might be excluded from this study due to positive rapid tests for in uenza.
De nitions ARDS and its severities were de ned according to the Berlin de nition 16 . Vasopressor use was de ned as norepinephrine, vasopressin, epinephrine, dopamine, or phenylephrine administration for 120 minutes or longer. Septic shock was de ned as vasopressor use and hyperlactatemia without hypovolemia, based on the Sepsis-3 consensus 17 .
Virological studies Nasopharyngeal swab (NPS), throat swab (TS), or bronchoalveolar lavage (BAL) from ICU patients were examined by viral cultures, real-time polymerase chain reaction (PCR) for in uenza virus, and BioFire® FilmArray® respiratory panel (BioFire Diagnostics, Salt Lake City, Utah, USA). The swabs were placed into in-house or commercially available viral transport medium (Copan Diagnostics, Murrieta, California, USA). Flocked swabs, instead of conventional cotton swabs, were used for the collection of nasopharyngeal or throat samples since April 2018. The decision to perform a bronchoscopy was made by the treating physicians. Other microbiological studies were ordered according to clinical needs.
Respiratory specimens in transport medium were inoculated to cell lines, including human lung carcinoma (A549), human embryonic rhabdomyosarcoma (RD), and Madin-Darby canine kidney (MDCK) cells. Tubes with cytopathic effect were con rmed by D 3 ® Ultra 8™ DFA (direct uorescent antibody) Respiratory Virus Screening and ID Kit (Diagnostic Hybrids, Athens, Ohio, USA), which was able to detect in uenza A and B viruses, respiratory syncytial virus (RSV), human adenovirus (HAdv), HPIV 1-3, and human metapneumovirus (HMPV). Those without cytopathic effects were blindly stained by the same kit after inoculation for 10 days.
Nucleic acid ampli cation of in uenza viruses was performed on LightCycler® 480 real-time PCR machine (Roche Diagnostics, Rotkreuz, Switzerland) after RNA extraction with LabTurbo 48 Compact auto-extraction System (Taigen Bioscience Corp., Taipei, Taiwan). The sequences of in uenza primers and probes were provided by the World Health Organization and the Taiwan Centers of Disease Control (Additional File 1, Table S1).

Statistical analysis
All statistical analyses were performed using R 4.0.2 (R Foundation for Statistical Computing, http://www.R-project.org/, Vienna, Austria). Continuous variables were presented as means ± standard deviations or medians (1st − 3rd quartile). Student's t test or Wilcoxon rank-sum test was applied for the between-group comparisons for normally or non-normally distributed variables, respectively. Category variables were compared by Pearson's χ2 test or Fisher's exact test when appropriate. The Wilson score was applied for con dence interval (CI) estimates of binomial proportions. A p value of less than 0.05 was considered to be statistically signi cant.
The multivariable logistic regression was performed to identify the factors associated with the detection of respiratory viruses. The parameters not readily available at the time of virological testing were not included into the regression models. The cases with missing data were excluded from the corresponding analyses. Of continuous variables such as age, C-reactive protein, and procalcitonin, we tried modeling with clinically relevant cut-off values to enhance their clinical applicability. Factors with a p value of less than 0.15 in simple regressions were included into the multivariable model. The effect sizes were presented by odds ratios (OR) with 95% CIs.

Results
One hundred and fty-eight critically ill adults with 167 respiratory specimens (NPS 154, TS 2, and BAL 11) were included during the study period. All patients were tested by the FilmArray® panel, while one and seven patients did not underwent nucleic acid ampli cation for in uenza viruses and viral cell cultures, respectively. The rst specimen collection was obtained at a median of one day (0-1 day) after ICU admission.
Respiratory viruses were detected in 64 (40.5%) patients, i.e., 64 patients with VRTIs. The comparisons of clinical data between patients with or without detection of respiratory virus are demonstrated in Table 1. Patients with VRTIs were younger (median age: 62.5 vs. 69.0 years, p = 0.003), but there were no differences in terms of chronic illness, disease severity, and clinical outcomes, including mortality rate and length of ICU or hospital stay between the two groups. The most commonly detected virus was in uenza virus (20 patients: A/H1 6, A/H3 13, B 1), followed by RV/EV (11), RSV (9), HMPV (9), HPIVs (8), HAdV (7), and HCoVs (2) (Fig. 1). To be noted, 57 adults admitted to medical ICUs with con rmed in uenza were not included in the study, usually due to positive results in rapid antigen tests. The FilmArray® respiratory panel detected respiratory viruses in 54 patients and M. pneumoniae in one patient. Nine patients with negative results of the FilmArray® respiratory panel, but had VRTIs due to respiratory viruses detected by other tests, including in uenza A/H3 (7), in uenza A/H1 (1) and HAdV (1) association of the number of the above variables with probability of VRTI was shown in Fig. 2 and Additional File 1, Table S2.

Discussion
The current study assessed the prevalence of VRTIs in pragmatically selected adults with critical illness, when other causes were unable to fully explain the patients' respiratory distress and clinical presentation. A positivity rate of 40.5% in the present study is higher than that of non-selected patients receiving mechanical ventilation in ICUs 5 . The median time from ICU admission to specimen collection for multiplex test is one day, indicating that the tests were ordered according to clinical information obtained right after ICU admission, rather than reserving the tests for patients with negative results of bacterial cultures.
To our knowledge, the host variables predictive of VRTIs among critically ill patients had not been adequately investigated before, as previous studies included mainly demographic information, comorbidities, or laboratory results 7-9 . Age < 65 years old, household contact with an individual having upper RTI and the presence of fever, productive cough, or sore throat were linked to VRTIs. In the patients presenting three of the above ve characters, the probability of VRTIs is 60%. Considering the limited accuracy of clinical information from severely ill patients on respiratory support and their families, these associations of clinical symptoms and VRTIs may be more signi cant than that in the present study, which suggests the indispensability of thorough history taking even in critical care settings. The age factor could be attributed to a higher rate of non-infectious causes of respiratory distress in the elderly 18,19 . It is not surprising that traditional laboratory tests failed to detect respiratory viruses in clinical samples, since bacterial etiological pathogens could be noted in only 13.2% of the included patients.
In spite of the exclusion of 57 cases of severe in uenza, in uenza virus remained to be the most commonly viral pathogen. The weighting of in uenza virus is obviously underestimated in the viral etiologies of RTIs. Thus, our prevalence data supports empiric administration of anti-in uenza agents to selected critically ill patients with suspected VRTIs. However, the FilmArray® respiratory panel failed to detect in uenza A/H3 in 8 (61.5%) out of the 13 patients diagnosed by the in-house PCR using updated primers. As in uenza viruses evolve rapidly, physicians should be cautious that commercially available molecular assays may provide unreliable results, if not updated frequently 20 .
As in uenza virus, other respiratory viruses can cause signi cant morbidity and mortality in critically ill patients 21,22 , and in this study these viruses accounted for 68.8% of all VRTIs. Six of seven patients died in the ICUs were infected by non-in uenza viruses. However, our case number is insu cient to compare clinical characteristics among different respiratory viruses and further studies are warranted.
Previous studies showed that clinical specimens from upper and lower respiratory tract specimens may yield discordant testing results, as noted in 2 (25%) of our 8 patients with both NPS and BAL samples tested for respiratory viral pathogens, in 11-36% of critically ill or immunosuppressed hosts 5,6,23,24 . Accordingly, etiological surveys using both upper and lower respiratory tract specimens ought to be considered in patients with highly suspected VRTIs.
Clinical application of multiplex respiratory viral assays had been expected to facilitate discontinuation or de-escalation of antibiotics 25 . Interestingly, several studies with the detection rates of < 25% for respiratory viruses invariably show no signi cant decline in antibiotic use [21][22][23]. When the prevalence of VRTIs exceeded 25%, the utilization of these assays was associated with favorable results on antimicrobial stewardship program (Additional File 1, Table S3) [26][27][28][29][30][31][32][33] . As none of these studies focus on ICU patients, our detection rate of 40.5% for viral pathogens in the selected patients suggested the need of further investigations into the impact of multiplex viral testing on reduction of unnecessary antibiotic use in selected critically ill patients.
Our study inherited several limitations. First, this is a single-center observation study and the study result cannot be generally applied to other clinical settings or healthcare facilities, since the distribution of respiratory viruses varies seasonally and geographically. Second, only 7% of our included patients had lower respiratory tract specimen tested, and thus the rate of VRTIs are likely to be underestimated 5,6 .
Third, the case of concurrent viral and non-viral RTIs were largely excluded since according to the study protocol, the initial clinical presentation would preclude respiratory virus tests.

Conclusions
VRTIs were not uncommon in pragmatically selected adults with critical illness. Age < 65 years old, household contact with a person having upper RTI, and the presence of clinical symptoms of fever, productive cough, or sore throat, were associated VRTIs. To maximize clinical bene t, intensivists can consider selective application of multiplex molecular assays for respiratory viral pathogens, based on the above predictive variables. The sensitivity and speci city of the predicative model and clinical impact of VRTIs need to be further studied on larger prospective cohorts.