Epidemiology and outcomes of hospitalized adults with respiratory syncytial virus: A 6‐year retrospective study

Objectives Respiratory syncytial virus (RSV) is an important cause of morbidity and mortality in adults. Existing studies are limited by the number of seasons studied and most have focused on the immunocompromised. Methods A retrospective cohort study was conducted on all adults (≥18 years) with a positive RSV molecular test admitted from 2009 to 2015 to one hospital in Chicago, IL. Epidemiologic and outcomes data were collected after IRB approval. Results Of the 489 eligible patients, 227 had RSV A and 262 had RSV B. Patients had a median age of 61 years and comorbidity (eg, chronic lung disease [40.6%], obesity [37.8%], and cardiac disease [34.3%]). On presentation, most had cough (86.5%), fever (42.4%), and shortness of breath (38.2%). Severe disease was present in 27.6% of patients. Antibiotic was used in 76.3% inpatients and 45.8% at discharged despite few patients (4.7%) having documented bacterial infections. Supplemental oxygen and mechanical ventilation were utilized in 44.6% and 12.3%, respectively, while ICU level care was required in 26.9%. Most patients were discharged home (82.7%). Most deaths (68.4%, 13/19) were attributed to pneumonia or hypoxemia likely from RSV. Most fatal cases were seen in those with recent cancer treatment and older adults. Conclusions Respiratory syncytial virus in hospitalized adults is associated with significant morbidity and mortality with 26.9% requiring ICU level care. Antibiotics are commonly prescribed to patients with documented RSV, and antibiotics are frequently continued after diagnosis. Novel antiviral therapies are needed for RSV to improve outcomes and potentially improve antibiotic stewardship in patients without a bacterial infection.

as pneumonia. 2 RSV has been estimated to cause 177 525 excess hospitalizations each year and 14 000 excess deaths, in the United States. 1,3 Risk factors for more severe disease include old age (>65 years), solid organ (especially lung) transplant, hematopoietic stem cell transplant, and those with underlying lung disease. 4 Among nursing home residents, RSV causes significant numbers of hospitalizations and deaths each winter. 5 One study found that 12.5% of elderly patients hospitalized for an influenza-like illness had RSV detected. 6 Unlike influenza, there are no vaccines or antivirals approved for the prevention or treatment of RSV in adults. Palivizumab, a humanized monoclonal antibody against the RSV F glycoprotein, is approved for the prevention of RSV in select infants and children but has not been approved in adults as it has not been widely studied. 1,7 Several live attenuated and sub-unit RSV vaccines are in advanced stages of development. 6,8 Aerosolized ribavirin is approved for the treatment of hospitalized infants and young children while aerosolized, oral, and intravenous ribavirin has been used selectively in immunocompromised adults with RSV with variable results. 9,10 RSV-containing antibody preparations (monoclonal or intravenous immunoglobulins) have also been used in conjunction with ribavirin. 11 Currently, two small molecule drugs, presatovir (GS-5806, a fusion inhibitor) and lumicitabine (ALS-008176, a polymerase inhibitor), are undergoing phase 2 studies in hospitalized and immunocompromised adults (ClinicalTrials.gov Identifiers: NCT02135614, NCT02254421, and NCT02254408). [12][13][14] In addition, several other RSV-active antivirals are in late pre-clinical and phase 1 development and hold promise for the prevention and treatment of RSV. 15 The burden of RSV infections in immunocompromised adults that are non-HSCT is not well known. 16 Most epidemiologic studies of RSV in hospitalized adults have been retrospective single-center studies covering a single or few seasons. 2,17,18 This retrospective cohort study was conducted to better define the epidemiology and outcomes of RSV in hospitalized adults over multiple seasons.

| Patient identification
After IRB approval, a retrospective cohort study on adults hospitalized with an RSV infection was conducted. All patients admitted to either of the participating hospitals at our institution (Northwestern A small number of patients were admitted more than once during these six seasons, and if they had a repeat positive RSV molecular test, only unique admissions were included. It was thought that subsequent positive results on subsequent admissions could potentially represent prolonged viral shedding rather than new infection. 19

| Data collection
The Northwestern Medicine Enterprise Data Warehouse (EDW) was utilized to identify eligible patients and to capture key data from the electronic health record. Manual chart review was then conducted independently by two of the authors to abstract supplementary data available within the free text in the medical records (ie, presenting symptoms). Clinical data collected included demographics, comorbidities, onset of symptoms, presenting symptoms, use of supplemental oxygen requirement, ventilator support, need for renal replacement, antimicrobials, and bronchodilators, as well as duration of hospitalization, discharge location, and vital status on discharge.

| Definitions
Each season was defined as May 1 through April 30 of the next year.
Severe illness was defined as a patient who was given non-invasive positive-pressure ventilation (NIPPV), intubated, or admitted to the ICU. The presence of abnormal imaging was defined as the presence of pulmonary infiltrates described by a radiologist on a chest X-ray or computed tomography (CT) scan of the chest. Nosocomial RSV infection defined as cases where symptoms began while a patient was in a hospital or other healthcare facility for ≥72 hours prior to a positive test for RSV.

| Statistical analysis
Simple descriptive statistics (frequency and percent for categorical variables; mean and standard deviation or median and range for continuous variables) summarized sample demographic characteristics and clinical measures. Analyses employed Pearson's chi-squared test to assess the association between severe disease status and symptoms, and between chronic disease and death, discharge status, ICU admission, and intubation status.
Analyses further utilized Fisher's exact method when expected cell counts were less than five for a given cross-tabulation. All analyses assumed a 5% level of significance. We did not adjust for multiple hypothesis testing since the study is exploratory in nature and controlling for type II error rate is of more importance than type I error rate. Analyses were performed using R: A language and environment for statistical computing (R Foundation for Statistical Computing, Vienna, Austria; version 3.3.2).

| Clinical presentation
The most common symptom on presentation was cough (86.

| Management and outcomes
Supplemental oxygen was utilized in 44.6% (218) patients with an additional 11.9% (58) requiring non-invasive ventilation, such as CPAP, and 12.3% (60) requiring mechanical ventilation at some point during their hospitalization (see Table 2 Table S2).  Table S1 for details). Oncology and hematopoietic stem cell transplant recipients had a higher mortality rate than the general population (see Table S2). presentation were less likely to develop severe disease (see Table 3).

| Risk factors for outcomes
No pre-existing condition was significantly over-represented in patients requiring ICU admission, although the presence of a co-infection (26.5% vs 10.6%, P < 0.001) was more common among ICU admissions.
No factor was statistically associated with intubation, although there was a trend toward higher rates of intubation in patients with cancer treatment, diabetes, and a documented co-infection (see Table 4).

| D ISCUSS I ON
This study represents one of the largest studies of RSV in hospitalized adults with patients sampled over multiple seasons. About a quarter of patients required ICU level care, and death was more common among those with recent cancer treatment and age >65 years. The finding that nasal congestion, rhinorrhea, cough, and myalgias were associated with a lack of severe disease may help identify patients who may require less supportive measures and can be discharged more quickly. Likewise, the association between older age, prior cardiac disease, and co-infection and need for higher level of care on discharge likewise may help risk stratify patients on  In comparison with other studies, this study suggests similar data regarding symptoms at presentation, median length of stay, and rates of co-infection, antimicrobial therapy, and death. The patients in our study were somewhat younger than other studies that included hospitalized adults as well. 17,18 However, our study suggests that rates of ICU admission and mechanical ventilation were higher overall than what is previously described. 17 The relatively high proportion of immunocompromised patients included in this study may, in part, explain this high rate of ICU admission and ventilator use.
The time from symptom onset to presentation was also similar to what has been documented in other studies. 18 Additionally, the finding that ICU and intubation rates were lower among patients with recent chemotherapy or HSCT compared to healthier patients is interesting. While death rates were higher in the immunocompromised patients, other risk factors such as underlying lung disease may play a larger role in respiratory decompensation. It is also possible that the immunocompromising conditions result in less inflammation which could be protective against the need for ICU care or intubation. Intubation is associated with higher mortality in immunocompromised patients, and the lower rate may be due to avoidance of invasive procedures in these populations. 24 The lower death rate among healthy individuals is otherwise unremarkable given the inherent physiologic reserve these patients may have to rebound from illness.
An interesting finding is that patients who died had a longer period of time between symptom onset and diagnostic testing.
The reason for this is not entirely obvious with the data collected in this study. The likely factor that was difficult to obtain in the retrospective analysis was the clinical decision making of the providers. Many of the fatal cases occurred in patients who were immunosuppressed or recently received chemotherapy which may make symptoms less severe. Further, nosocomial acquisition of infection is also a possibility. This may be likely because the length of stay prior to ICU transfer was relatively long in many of the