Clinical characteristics and population‐based attack rates of respiratory syncytial virus versus influenza hospitalizations among adults—An observational study

Abstract Background The clinical significance of respiratory syncytial virus (RSV) among adults remains underinvestigated. We compared the characteristics and population‐based attack rates of RSV and influenza hospitalizations. Methods During 2018–2020, we recruited hospitalized adults with respiratory infection to our prospective substudy at a tertiary care hospital in Finland and compared the characteristics of RSV and influenza patients. In our retrospective substudy, we calculated the attack rates of all RSV and influenza hospitalizations among adults in the same geographic area during 2016–2020. Results Of the 537 prospective substudy patients, 31 (6%) had RSV, and 106 (20%) had influenza. Duration of hospitalization, need for intensive care or outcome did not differ significantly between RSV and influenza patients. RSV was more often missed or its diagnosis omitted from medical record (13% vs 1% p = 0.016 and 48% vs 15%, p > 0.001). In the retrospective substudy, the mean attack rates of RSV, influenza A, and influenza B hospitalizations rose with age from 4.1 (range by season 1.9–5.9), 15.4 (12.3–23.3), and 4.7 (0.5–16.2) per 100,000 persons among 18‐ to 64‐year‐olds to 58.3 (19.3–117.6), 204.1 (31.0–345.0), and 60.4 (0.0–231.0) per 100,000 persons among 65+‐year‐olds and varied considerably between seasons. Discussion While the attack rates of influenza hospitalizations were higher compared with RSV, RSV and influenza hospitalizations were similar in severity. Missing or underreporting of RSV infections may lead to underestimating its disease burden. Both RSV and influenza caused a substantial amount of hospitalizations among the elderly, stressing the need for more effective interventions.


| INTRODUCTION
While influenza is a widely acknowledged global health threat, the burden of respiratory syncytial virus (RSV) remains less well-charted among adults. Previously recognized as an important pathogen among children, RSV is now known to cause notable morbidity and mortality among adults, especially among immunocompromised individuals, those with cardiopulmonary diseases, and 65+-year-old elderly. [1][2][3][4][5][6][7][8] Studies suggest that RSV infection develops yearly in 3-7% of healthy elderly adults and 4-10% of high-risk adults with chronic heart or lung disease, whereas seasonal influenza infects about 10% of adults, half of whom have symptomatic infection. 6,9,10 RSV has repeatedly been among the top five causative agents of viral respiratory infections in the elderly along with influenza. 7,11,12 Influenza offers a practical point of comparison to RSV as they circulate simultaneously, and the same clinical criteria and sample can be used for testing. Both RSV and influenza epidemics occur between Weeks 41 and 20 in the northern hemisphere. 13,14 In Finland, RSV epidemics used to follow a pattern of a minor peak in April followed by a major peak in December every 2 years; however, since 2008, RSV epidemics occur every winter peaking between February and May but still exhibit a biennial pattern of major and minor epidemics. 15,16 Influenza epidemics vary considerably between seasons depending on the circulating virus type and preexisting immunity of the population. Previously, insensitive or slow diagnostic methods such as viral culture and serology were used to detect RSV in clinical studies, while tests suitable for clinical use were lacking. 5,17 Recently, molecular polymerase chain reaction (PCR)-based diagnostic methods have enabled the fast and accurate diagnostics of RSV alongside influenza testing. [17][18][19] Although RSV hospitalizations have been studied among children, elderly, and high-risk adults, few recent prospective studies have investigated all 18+ hospitalized adults. [20][21][22][23] A more complete picture of the RSV disease burden and potential vaccine target groups among adults is required as RSV vaccines likely become available soon. 24,25 We compared the clinical characteristics of adults hospitalized with RSV or influenza and calculated the population-based seasonal attack rates of RSV and influenza hospitalizations among adults to estimate their respective disease burdens in different age groups.

| Study design
Our study included two substudies: a prospective hospital-based substudy and a retrospective register-based cohort substudy of 18+ adults hospitalized with laboratory-confirmed RSV or influenza. Prospective substudy data were used to compare the clinical characteristics of RSV and influenza and to assess the proportion of all severe acute respiratory infection (SARI) hospitalizations attributable to these viruses. Retrospective register data were used to compare the age and sex distributions of patients hospitalized with RSV or influenza and to calculate population-based seasonal attack rates of RSV and influenza hospitalizations.

| Study setting and population
Jorvi Hospital is a tertiary care hospital belonging to HUS Helsinki University Hospital (HUS) and serving a catchment area consisting of the populations of the municipalities of Espoo, Kirkkonummi, and Kauniainen. All residents of these three municipalities requiring hospi- Since autumn 2019, the use of RIADT was discontinued due to low sensitivity observed during previous years, and only direct RT-PCR testing for both influenza (A and B) and RSV was conducted.
In March 2020, the COVID-19 pandemic spread to the study area after which all hospitalized SARI patients were tested for SARS-CoV-2 while diagnostic testing for influenza or RSV was not always done. However, all patients included in the prospective substudy since March 2020 were tested for influenza and RSV whenever a respiratory sample was available (66/94, 70%).

| Data analysis
Using prospective substudy data, we compared the clinical characteristics, symptoms, and outcomes of RSV and influenza patients.
Categorical variables were compared using Fisher's exact test and continuous variables using Mann-Whitney's U test. Data analysis was performed using SPSS version 26.0 (IBM SPSS statistics ® ).
With the use of retrospective register data, we calculated seasonal attack rates (or cumulative incidences) of hospitalizations by dividing the number of hospitalized patients positive for RSV or influenza in a certain age group during an epidemic season by the base population in that age group on December 31 during that season.
Hypertension, pulmonary, and cardiovascular diseases were the most common comorbidities in both RSV and influenza patients. Data suggested that kidney diseases were more common among RSV than influenza patients (8, 26% vs. 9, 8%, p = 0.025), and RSV patients were slightly more likely to have cancer (11, 35% vs. 20, 19%, p = 0.085). McCabe score was 2 or 3 indicating the presence of serious comorbidities in 42% of RSV and 25% of influenza patients Almost a third of both RSV and influenza patients had been hospitalized in the previous 12 months (10, 32% vs. 29, 27%). Smoking was as common in both groups with six (19%) of RSV and 25 (24%) of influenza patients being current smokers. Uptake of seasonal influenza vaccine (14+ days before hospitalization) did not differ significantly between RSV and influenza patients (Table 1).
Sudden onset of symptoms was less common among RSV than influenza patients (23% vs. 61%, p < 0.001, Table 1). Only 74% of the RSV patients had fever ≥38 C compared with 91% of influenza patients (p = 0.031). Cough was the most common symptom among both RSV and influenza patients followed by deterioration of general condition and dyspnea.

| Results of the retrospective register-based substudy
Age and sex distributions of the patients hospitalized with RSV or influenza A or B did not differ significantly over the whole study period ( Table 2) Table S1.
There was a considerable season-to-season variation in both RSV and especially influenza hospitalizations (Figures 2 and 3).  seems to affect younger patients than A(H3N2), as also observed in our prospective substudy. 35 Thus, the possible age difference between RSV and influenza patients may depend on the circulating influenza A subtype.
Consistently with previous research, the majority (30/31, 97%) of RSV patients had at least one preexisting comorbidity. 22,23,31,33,36,37 Serious comorbidities (McCabe 2 or 3) were somewhat more common (42% vs. 25%, p = 0.114) among RSV than influenza patients. As in previous studies, chronic kidney diseases were more common among RSV than influenza patients. 20,21 As in other studies, cancer and immunosuppression tended to be more common among RSV than influenza patients; however, these differences did not reach statistical significance, and the lack of patients with acute hematologic malignancies or recent transplant may have affected our results. 20,21,33 Another recent study reported that hematologic malignancies and solid cancer are risk factors of RSV pneumonia. 38 The clinical pictures of RSV and influenza were similar except that fever was less common among RSV patients as in previous studies. 5,9,31,34 Sudden onset of symptoms was less common, and hospitalization occurred on average 1 day later in RSV than influenza SARI. This later presentation and hospitalization of RSV patients have been documented in several studies. 11,21,22,31,33 Consistently with other studies, the duration of hospitalization and need for ICU admission did not differ significantly between RSV and influenza patients. 6,[20][21][22]33 Notably, in our prospective substudy, 13% of RSV cases were missed based on clinical diagnostics alone, and the virus-specific diagnosis was omitted from medical records by treating physicians in 48% of cases compared with 1% missed diagnoses and 15% of omitted In our prospective substudy, none of either RSV or influenza patients died during hospitalization, whereas in other prospective studies, in-hospital mortality has ranged from 0% to 8% for RSV and from 1% to 7% for influenza. 6,20,22,36 Furthermore, during follow-up, none of either RSV or influenza patients died within 30 days, and only 2 (6%) of RSV and 1 (1%) of influenza patients died within 3 months from admission. In contrast, in a retrospective cohort study from Hong Kong, 30-day and 60-day all-cause mortality of RSV and influenza patients were 9.1% and 8.0% and 11.9% and 8.8%, respectively. 31 Also, two recent studies from China have found higher 30-day mortality (10.9-13.7%) among RSV than influenza patients (5.0-6.2%). 21,40 Our low mortality rates are likely related to the underrepresentation of elderly patients as institutionalized patients, and those treated at secondary care hospital were excluded from our prospective substudy. Additionally, severely ill patients may have been unable to give their informed consent; however, we did try to overcome this limitation by accepting consent from their next of kin.
In our register-based cohort study, among adults, attack rates of both RSV and influenza hospitalizations rose with age consistently with previous studies. 8,39,41,42 Altough most RSV hospitalizations occurred among the elderly, in our study as well as previous studies, work-age adults with comorbidities were also at risk of being hospitalized with RSV. 6 Our prospective substudy setup offered several advantages. SARI patients were systematically screened and tested for both influenza and RSV by RT-PCR, which is more sensitive than previously used viral detection methods. We included all 18+ adults, whereas many previous studies have focused on the elderly. This substudy also had limitations, most importantly its small sample size. All eligible patients did not consent to participate, and elderly patients were underrepresented. As our SARI definition did not include runny nose or wheezing commonly observed in RSV infection, RSV cases may have been missed. 5,7,11 Strengths of our register-based cohort substudy include utilizing data from four different epidemic seasons, which enabled the observation of the seasonal variation of these viruses, and comprehensive coverage of our background population. We believe our attack rates are accurate estimates of community-acquired laboratory-confirmed influenza hospitalizations given that the study site is the only regional reference center for the inhabitants needing hospital care and has the policy to test all hospitalized SARI cases with RT-PCR, leading to the clinical detection of 99% of influenza cases in our prospective substudy.
Limitations of our retrospective cohort substudy include testing at clinician's discretion, likely leading to some missed diagnoses. Our attack rates of acute RSV-associated hospitalizations are probably underestimates because based on our prospective substudy, 13% of RSV diagnoses were missed when only diagnostic clinical samples were considered and because patients with severe immunosuppression were treated at another hospital. Additionally, patients whose positive RSV or influenza test was taken before hospitalization at primary care or at a previous visit to JEC not leading to hospitalization are missing from attack rate calculations. However, RT-PCR testing for influenza and RSV is uncommon at primary care. Our focus was on hospitalizations with acute laboratory-confirmed RSV or influenza; thus, we did not include later hospitalizations due to, for example, secondary bacterial infections, exacerbations of preexisting conditions, or cardiorespiratory complications unless the patients were tested for and remained PCR-positive for RSV or influenza, which becomes more unlikely the more time passes since the onset of infection. 47 Thus, our attack rates should be interpreted as representative of the lower bounds of the true attack rates of RSV and influenza hospitalizations.
In conclusion, our study adds to the growing evidence indicating that besides influenza, RSV is an important and underestimated cause of hospitalizations for acute respiratory infection among the elderly.
RSV vaccines and more effective influenza vaccines are needed to decrease this significant burden of severe respiratory infections in the elderly population and to alleviate the seasonally varying and thus unpredictable strain put on hospitals. A better understanding of RSVassociated morbidity among adults is necessary to guide the use of preventive measures, diagnostics, and possible treatments of RSV. supervision.

PATIENT CONSENT STATEMENT
Informed consent to participate was obtained from all participants of the prospective substudy. Informed consent included consent to publish study data of the prospective substudy.

SOURCES
No material from other sources was reproduced.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/irv.12914.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request respecting GDPR and with permission from HUS and THL.