Epidemiology, clinical features, and resource utilization associated with respiratory syncytial virus in the community and hospital.

Abstract Background The epidemiology, clinical features, and resource utilization of respiratory syncytial virus (RSV) cases in the community and the hospital are not fully characterized. Methods We identified individuals of all ages with laboratory‐confirmed RSV from two sources, a community cohort undergoing surveillance for acute respiratory infections (ARIs) and hospitalized patients from the same geographic area of New York City between 2013 and 15. The epidemiology, clinical features, and resource utilization (antibiotic/steroid/ribavirin usage, chest X‐rays, respiratory‐support (continuous positive airway pressure [CPAP], mechanical ventilation or extracorporeal membrane oxygenation [ECMO]), and indicators of disease severity (respiratory‐support, and/or ICU admission or death)) were compared among age groups using univariate and bivariate analyses. Results In the community cohort (1777 people with 1805 ARIs), 66(3.7%) tested RSV‐positive (3.8% of <1‐year‐olds; 3.8% of adults ≥65); 40.9% were medically attended, and 23.1% reported antibiotic usage. Among 40,461 tests performed on hospital patients, 2.7% were RSV‐positive within ± 2 days of admission (37.3% <1 year old; 17.4% ≥65 years old). Among RSV‐positive hospitalized adults ≥65%, 92.7%, 89.6% and 78.1% received a chest X‐ray, antibiotics and/or steroids respectively, compared with 48.9%, 45.7%, and 48.7% of children <1. Severe illness occurred in 27.0% RSV‐positive hospitalized <1‐year‐olds and 19.8% ≥65‐year‐olds. Conclusions Respiratory syncytial virus had a demonstrated impact in the community and hospital. Only 40% of RSV community cases were medically attended. In the hospitalized‐cohort, <1‐ and ≥ 65‐year‐olds accounted for the majority of patients and had similar rates of severe illness. In addition, resource utilization was high in older adults, making both young children and older adults important potential RSV vaccine targets.


| INTRODUC TI ON
Respiratory syncytial virus (RSV) has been well-characterized in children with an annual estimated 57 527 hospitalizations and 2.1 million outpatient visits for children <5 years in the US. 1 However, recent studies have identified increased burden of RSV among adults ≥65 years old, [2][3][4] especially those in contact with young children. 2,5 One model estimates that RSV accounts for approximately 10 000 deaths in ≥65-year-olds annually in the US 6 While studies of RSV epidemiology have been conducted among medically attended cases, 1,2,7,8 a fuller understanding of RSV impact would be gained by examining infections and resource utilization in both the community and hospital setting in tandem specifically examining higher risk populations including adults ≥65 and young children. The objectives of this study were to use data from a community surveillance study of respiratory viral infections and from hospitalizations from the same geographic area to: (a) characterize RSV in a community cohort (including non-medically attended cases) and hospitalized patients, (b) describe the clinical features and resource utilization of RSV, and (c) examine seasonal trends.  9 Households in this study were primarily multigenerational, Latino and publicly insured. Briefly, households were queried twice-weekly via text-messaging for ARI symptoms.

| Study design, study subjects, and study sites
Mid-turbinate nasal swabs were obtained at a home-visit by research staff, the majority within 24-48 hours of symptom onset, for those who meet symptomatic criteria-defined as ≥2 of the following: fever/feverishness, cough, sore throat, runny nose/nasal congestion or body aches. Children <1 year old were also swabbed if they had rhinorrhea/congestion only.

| Viral diagnostic testing
Mid-turbinate nasal swabs from the community and nasopharyngeal swabs from hospitalized subjects were analyzed by multiplex RT-PCR using the same FDA-approved FilmArray Respiratory-panel 1.7 (BioFire Diagnostics, Inc). The community samples were brought to one of the co-investigator's laboratory (PSL) in universal transport media within 4 hours of collection for testing. The hospital samples were tested in the CUMC Clinical Microbiology Laboratory.

| Clinical features and resource utilization in community and hospital subjects with RSV
For the community cohort, in addition to the ARI symptoms outlined above, these additional symptoms were captured: chills, fatigue, headache, wheezing, dyspnea, hoarseness, earache, conjunctivitis, rash, vomiting, diarrhea, and loss of appetite. On follow-up calls conducted starting 10 days after illness report and continuing intermittently until symptom resolution, self-reported visits (primary care, urgent-care, Emergency Department and/or hospitalizations), antibiotic usage, and missed school/work days were collected to capture a summary of the entire illness. Index cases were defined as the first symptomatic household member.
The electronic medical records (EMR) of hospitalized patients positive for RSV were queried for select medications (antibiotics, steroids, albuterol, racemic epinephrine, and ribavirin) received from 2 days prior to 7 days after the positive RVP, chest X-rays within ± 2 days of positive RVP, respiratory, blood and urine culture results within ± 2 days of positive RVP, type of respiratory-support older adults, making both young children and older adults important potential RSV vaccine targets.

K E Y W O R D S
community surveillance, medically attended, respiratory syncytial virus received including continuous positive airway pressure (CPAP), mechanical ventilation, extracorporeal membrane oxygenation (ECMO) and/ or ICU admission. Use of bilevel positive airway pressure (BiPAP) and oxygen supplementation was excluded as accurate data for their use could not be readily obtained in structured secondary sources. Blood and respiratory cultures were only considered positive if the identified species was a known pathogen (eg, Pseudomonas aeruginosa) and not a common contaminant (eg, Staphylococcus epidermidis).

| Factors associated with increased severity of illness
In the community cohort, medically attended illness and missed school/work days were considered markers of more serious illness.
Potential risk factors and co-variates collected at enrollment included demographic characteristics (sex, race, ethnicity, language, age, and insurance), and chronic respiratory conditions. Among hospitalized patients, severe illness was defined as respiratory-support with CPAP, ventilation or ECMO, and/or ICU admission or death during the RSV hospitalization. 10,11 Patients' International Classification of Diseases 9 and 10 (ICD) codes were queried to assess comorbid conditions and primary reason for hospitalization. Conditions were classified into chronic comorbid categories (CCC) tabulated by the study-team based on consensus (Table   S1) and included: cardiovascular, respiratory, endocrine, gastrointestinal, genetic/congenital neurologic including neuromuscular, renal, hematologic, transplant/immunosuppression, HIV, malignancy, metabolic, failure to thrive, obesity, prematurity (<37 weeks gestation in <2-year-olds), cystic fibrosis, and Down's syndrome. 4,5,10-13 If a specific condition was coded for <15 patients or did not fit one of the CCCs, it was disregarded.
The primary-ICD diagnosis for each patient was categorized as respiratory or non-respiratory by study-investigators. There were 196 patients for whom there was no primary-ICD code designated.
For these patients, their additional ICD codes were examined to categorize the diagnosis as respiratory or non-respiratory. For ambiguous diagnoses (eg, fever and dehydration), a chart-review was conducted to assess the patient's primary symptoms for proper classification.

| Data analysis
In the community cohort, bivariate analyses assessed the associations between demographic characteristics, viral co-detections, and chronic conditions with more serious illness associated with RSV detection. The small number of illnesses detected precluded multivariate analysis.
Among hospitalized patients, bivariate analyses assessed the associations of demographic characteristics (eg, age, sex, race, ethnicity, age, primary language, and insurance), clinical factors (eg, primary respiratory diagnosis), CCCs (number and types) and presence of viral co-detections with severe illness associated with RSV detection. Multivariable logistic regression analyses with forward-selection were conducted to examine the associations between any variables positive in the bivariate analyses plus age (a priori inclusion), and severe illness. Chi-square test was conducted to examine the difference among clinical features and resource utilization (ie, chest X-rays, antibiotic, and steroid usage) among age groups.
To assess seasonal trends, RSV activity overall and monthly was compared in the hospital and community groups. Epidemic curves and bivariate analysis were used to compare age groups (children <18 years vs adults ≥18 years) to month of illness onset or hospital admission. Analyses were conducted using SPSS v.23.

| Study population
The community MoSAIC cohort consisted of 371 households

| Associated clinical features and resource utilization of RSV cases
In the community cohort, 40.9% of participants had medically at-  Table 1).

F I G U R E 1 Method for inclusion in hospital and community study population based on respiratory viral panel (RVP) results
Among hospitalized patients, children <1 were more likely to be treated with CPAP (20.7%) vs patients ≥65 years (9.9%) (P = .001).
Among those ≥65 years old, 92.7% had a chest X-ray, and 89.6% and 78.1% received antibiotics and/or steroids, respectively. Among children <1 year, 48.9% had a chest X-ray, and 45.7% and 48.7% received antibiotics and/or steroids, respectively (Table 3). Adults 18-64 had the highest percentage of ribavirin administration (8.2%) compared with 3.1% in adults ≥65 and 0% in children <1. Patterns were similar when the sample was restricted to those from the catchment-area.

| Respiratory syncytial virus epidemiology
Both hospital and community RSV-detections began in the early fall, peaked during the end of November/early December and tapered off by spring, without lag time between RSV in the community and hospital ( Figure 2). Hospital detections of RSV occurred year-round among children; however, none was detected among hospitalized adults nor in the community cohort during the summer months.
The subsequent resurgence of detections in the community corresponded to the rise in hospital detections in the early fall. This pattern was similar when restricting to those from the catchment-area.
There was a significant difference between timing of RSV detected in adults (≥18 years) and children (<18 years) at both the hospital and community levels (P = .001) (Figure 2). While RSV was detected consistently beginning in September in children, RSV in adults was detected one month later on average. There are limitations in this study. The community cohort was primarily Latino and lived in a low-income, urban neighborhood; thus, the community findings may not be generalizable to a broader population.

| D ISCUSS I ON
Also, the clinical features and resource utilization of the community cohort were captured solely by interview and could not be verified by medical-record documentation. While the hospital and community cohort are in the same geographic area, no one from the community cohort was hospitalized during the study period, potentially due to the relatively smaller sample size. In addition, findings from the hospital group likely reflect referral center bias; however, patterns were similar when assessing just those from the catchment-area. Also, due to EMR limitations, we were unable to abstract accurate data on all types of respiratory-support, calculate percent positive by age or analyze the results of chest X-rays in the hospitalized group. In addition, by narrowing our selectivity to RSV-positive RVP within ± 2 days of admission and collecting data on resource utilization with ± 2 days of positive RVP, we may have underestimated the true burden of RSV among hospitalized patients. Finally, our small sample size for community cases limited our ability to analyze adults ≥65 years old.
In conclusion, this study suggests that hospitalization with RSV disproportionately affects children <1 year old as well as adults ≥65 as judged by comparing the age distribution of cases with that in a community cohort. In addition, among hospitalized patients, severe illness associated with RSV mostly affected adults ≥65 and children <1 year old; however, resource utilization in the hospital was highest for older adults. Thus making, both older adults and young children important targets for RSV vaccines. Furthermore, trends suggest that in New York City, RSV cases in both the inpatient and outpatient setting are beginning earlier than previously anticipated.

CO N FLI C T O F I NTE R E S T S
The authors have no conflict of interest to declare. L. Saiman and P.

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section.