key: cord-1004598-jd1cdrex
authors: Arriola, Carmen S; Kim, Lindsay; Langley, Gayle; Anderson, Evan J; Openo, Kyle; Martin, Andrew M; Lynfield, Ruth; Bye, Erica; Como-Sabetti, Kathy; Reingold, Arthur; Chai, Shua; Daily, Pam; Thomas, Ann; Crawford, Courtney; Reed, C; Garg, S; Chaves, Sandra S
title: Estimated Burden of Community-Onset Respiratory Syncytial Virus–Associated Hospitalizations Among Children Aged <2 Years in the United States, 2014–15
date: 2019-12-23
journal: J Pediatric Infect Dis Soc
DOI: 10.1093/jpids/piz087
sha: c0535776e45b8e93ce43e96b0c8cee4f1d783b1d
doc_id: 1004598
cord_uid: jd1cdrex

BACKGROUND: Respiratory syncytial virus (RSV) is a major cause of hospitalizations in young children. We estimated the burden of community-onset RSV-associated hospitalizations among US children aged <2 years by extrapolating rates of RSV-confirmed hospitalizations in 4 surveillance states and using probabilistic multipliers to adjust for ascertainment biases. METHODS: From October 2014 through April 2015, clinician-ordered RSV tests identified laboratory-confirmed RSV hospitalizations among children aged <2 years at 4 influenza hospitalization surveillance network sites. Surveillance populations were used to estimate age-specific rates of RSV-associated hospitalization, after adjusting for detection probabilities. We extrapolated these rates using US census data. RESULTS: We identified 1554 RSV-associated hospitalizations in children aged <2 years. Of these, 27% were admitted to an intensive care unit, 6% needed mechanical ventilation, and 5 died. Most cases (1047/1554; 67%) had no underlying condition. Adjusted age-specific RSV hospitalization rates per 100 000 population were 1970 (95% confidence interval [CI],1787 to 2177), 897 (95% CI, 761 to 1073), 531 (95% CI, 459 to 624), and 358 (95% CI, 317 to 405) for ages 0–2, 3–5, 6–11, and 12–23 months, respectively. Extrapolating to the US population, an estimated 49 509–59 867 community-onset RSV-associated hospitalizations among children aged <2 years occurred during the 2014–2015 season. CONCLUSIONS: Our findings highlight the importance of RSV as a cause of hospitalization, especially among children aged <2 months. Our approach to estimating RSV-related hospitalizations could be used to provide a US baseline for assessing the impact of future interventions.

Respiratory syncytial virus (RSV) is a major cause of respiratory infections and hospitalizations among young children in the United States [1] [2] [3] [4] ; RSV generally circulates from October through early May [5] . Several RSV vaccine candidates and long-acting monoclonal RSV-neutralizing antibodies are under development, with candidates in phase 1-3 trials [6] . Children aged <2 years are a priority target for vaccination as they have a high burden of severe RSV-associated disease and contribute to the spread of RSV in the community [7] . With RSV vaccines expected to be available in the upcoming years, it is important to establish nationally representative baseline estimates of RSV hospitalizations. These estimates can help to inform vaccination or monoclonal antibody strategies, and if these candidates are licensed, may be used to evaluate their impact post-introduction.

Traditionally, hospital discharge records have been used to estimate national RSV-associated hospitalizations. However, this approach has lacked individual-level laboratory-confirmed RSV data and may be subject to misclassification of RSV with discharge diagnosis likely including other causes of acute respiratory infections [8, 9] . Laboratory-confirmed, prospective, population-based RSV disease burden estimates have been limited by small catchment size [1, 10, 11] . We adapted an influenza disease-burden model that uses data from a national population-based surveillance platform of laboratoryconfirmed RSV hospitalized cases. We developed probabilistic multipliers that account for testing practices and test method sensitivities to adjust for biases in RSV case ascertainment [12] . In this pilot study, we aimed to estimate age-specific RSVassociated hospitalization rates among children aged <2 years and to describe the epidemiological and clinical characteristics of US children hospitalized with RSV.

From 1 October 2014 through 30 April 2015, we identified laboratory-confirmed RSV-associated hospitalizations among children aged <2 years from 4 Influenza Hospitalization Surveillance Network (FluSurv-NET) sites: California, Georgia, Minnesota, and Oregon [13] . The 4 participant sites encompassed 20 hospitals and had a catchment population of approximately 4% of the US population aged <2 years. Hospitalization was defined as an admission to an inpatient hospital ward regardless of whether the patient stayed overnight. Date of hospital admission was defined as the date of admission to the inpatient ward. RSV cases were identified through hospital admission, discharge, laboratory, and infection-control practitioner logs. Laboratory testing for RSV was clinician-driven, and cases were identified by positive test results based on polymerase chain reaction (PCR), rapid test (antigen detection), viral culture, or fluorescent antibody. Children were included in surveillance if they were aged <2 years, a resident in the surveillance catchment area, and had community-onset RSV, defined as diagnosis of laboratory-confirmed RSV infection within 7 days prior to or 3 days after hospital admission. We aimed to identify acute cases (diagnosed no longer than 10 days from illness onset). Only cases identified as laboratory-confirmed RSV in the first 3 days of hospitalization were included to avoid inclusion of possible nosocomial RSV infections. Medical records were reviewed for sociodemographic data (age, sex, race, ethnicity, insurance type), RSV testing results, medical history, palivizumab receipt (any time and any number of doses during the season), and clinical course, including admission to an intensive care unit (ICU), need for mechanical ventilation (including extracorporeal membrane oxygenation), and diagnosis of coinfections with other bacterial or viral pathogens.

To adjust for under-ascertainment of cases, we collected additional data about RSV testing from a sample of patients admitted to each surveillance hospital who met our definition for acute respiratory illness (ARI). For this, we screened hospital discharge logs for predefined International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) codes (079.6, RSV; 466.11, acute bronchiolitis due to RSV; 466.19, acute bronchiolitis due to other infectious agents; and 799.02, hypoxemia for children aged <6 months and 079. 6, 466.11, 466.19, 486 [pneumonia] , and 382.9 [unspecified otitis media] for children aged 6-23 months of age). The code selection was based on a population-based study that tested prospectively all children with respiratory illness for RSV using PCR. Testing was done as part of research, and results were not shared with clinicians [11] . We looked at the most common ICD-9 codes used in children aged <2 years who were hospitalized with respiratory illness and confirmed as RSV that covered >95% of the cases. We selected a stratified random sample of at least 80 eligible patients per month (December through April) and by age group (15 from <3 months, 15 from 3-5 months, 20 from 6-11 months, and 30 from 12-23 months). For each selected patient, we collected the following information from medical charts: whether the patient was tested for RSV and, among those tested, the test type and results and hospitalization outcomes (ie, ICU admission and death).

Following a previously described method [12] , we used the FluSurv-NET site-specific surveillance catchment population of the 4 sites included in this study (overall 309 000 children aged <2 years) to estimate age-specific (ie, 0-2, 3-5, 6-11, 12-23 months) rates of RSV-associated hospitalization, adjusting for detection probabilities (ie, testing practices and test method sensitivity). Detection probabilities were estimated as the inverse product of the proportion of patients tested for RSV and the sensitivities of the tests performed, adjusted by site. The sensitivity of rapid antigen tests varies from 70% to 90% [14] [15] [16] . For the purpose of this analysis, we used an estimated sensitivity of 80% for rapid antigen tests in a pediatric population [14, 15] , 100% for molecular diagnosis [15, 17, 18] , 90% for cell culture [19] , and 85% for immunofluorescence [20, 21] . The 95% confidence intervals (CIs) for hospitalization rates were calculated by bootstrapping the number of hospitalizations and the detection multiplier, accounting for variability of testing and test sensitivities by site. Assuming the observed rates were similar to the true rates, we extrapolated these rates and their 95% CIs to the US population using 2014 census data to estimate the burden of community-onset RSV-associated hospitalizations for the entire US population. All analyses were performed using R software (version 3.4.0) and Microsoft Excel 2016 (v16.0).

This study was considered nonresearch for public health purposes by the Centers for Disease Control and Prevention (CDC). Participating sites obtained approval for the FluSurv-NET RSV surveillance protocol from their respective state and local institutional review boards, as needed.

A total of 1554 laboratory-confirmed RSV cases who met the inclusion criteria were identified at the 4 FluSurv-NET sites during the study period. Of these, 40% (614), 17% (262), 19% (300), and 24% (378) were aged 0-2, 3-5, 6-11, and 12-23 months, respectively (Table 1) . Race/ethnicity varied (7) 17 (4) 76 (5) Chronic lung disease 16 (4) 10 (2) 6 (3) 12 (2) 44 (3) Abnormality of upper airway 4 (1) 6 (1) 10 (5) 10 (2) 30 (2)

Other conditions c 22 (6) . Overall, 6% (93) of admissions required mechanical ventilation (range, 2%-9% by site). The median length of hospital stay was 3 days (IQR, 2, 5). There were 5 deaths; among them, 4/5 were aged ≤6 months, 3/5 were born prematurely (range, 33-35 weeks of gestation), and we did not identify palivizumab receipt in any of these cases, although outpatient records were not reviewed. All of them (5/5) were admitted to the ICU; 3/5 received mechanical ventilation, and their hospital length of stay ranged from 2 to 120 days (further detail are provided in Supplementary Table 1) .

Eighty-nine percent of cases were tested for other viruses. Among those tested, 10% (144/1388) had other viruses detected simultaneously with RSV. The most frequent viral codetections were with rhinovirus/enterovirus (48%), followed by influenza A (24%), coronavirus (15%), adenovirus (10%), and influenza B (8%; Supplementary Table 2 ). Fortyfive percent of cases were tested for bacterial pathogens by any culture type; 43 (6%) had a positive blood culture. The most frequently identified bacterial pathogens from any culture type were Haemophilus influenzae (23%) and Streptococcus pneumoniae (12%; Supplementary Table 3) . Table 2 shows the frequency of severe outcomes during RSVassociated hospitalization, stratified by age groups. Overall, children aged 0-2 months represented the largest proportion of ICU admissions (35%; 95% CI, 23% to 48%) and mechanical ventilation (11%; 95% CI, 6% to 15%). Age-associated risk for severity was only statistically significant for mechanical ventilation; children aged 0-2 months were 3.5 times more likely to require mechanical ventilation compared with the older age groups (data not shown). We also stratified outcomes by the presence or absence of underlying medical conditions (healthy, at least 1 underlying condition, premature); these data are presented in Supplementary Tables 4-6.

The percentage of children hospitalized with ARI who were tested for RSV varied across sites and by age group (Figure 1) . The percentage of children tested was highest in the 0-2 month age group (84%; range, 76%-93%) and lowest in the 12-23 month age group Table 8) .

The age-specific adjusted community-onset RSV hospitalization rates per 100 000 population for the 2014-2015 season were 1970 (95% CI, 1787 to 2177), 897 (95% CI, 761 to 1073), 531 (95% CI, 459 to 624), and 358 (95% CI, 317 to 405) for children aged 0-2, 3-5, 6-11, and 12-23 months, respectively (Table 3) Our overall hospitalization rates for children aged <2 years (678 per 100 000 children) were consistent with those reported in 2013 by the CDC-supported New Vaccine Surveillance Network (NVSN), a prospective, active, population-based surveillance network for children hospitalized with ARI that had been implemented in 3 hospitals by 2005, each in a different state, and had a surveillance population of 132 085 children aged <2 years [1, 22] . Hall et al [1] estimated an annual average of 520 RSV-associated hospitalizations per 100 000 children aged <2 years between 2000 and 2005. Moreover, Iwane et al [22] reported 1850 and 740 RSV-associated hospitalizations per 100 000 children aged 0-5 months and 6-11 months, respectively, for 1 season. Similar to FluSurv-NET, RSV was identified by molecular assays; however, NVSN systematically tests those who meet the ARI case definition, whereas FluSurv-NET collects data on patients with an RSV-confirmed diagnosis who were tested at the discretion of clinicians, hence, the need for case ascertainment adjustments. Other studies that used different estimation methods presented similar findings [8, 9, [23] [24] [25] [26] . Nonetheless, compared with our study, Homaira et al [25] , in New South Wales, Australia, used administrative datasets that followed 870 314 children aged <5 years from 2001 to 2010 and reported higher annual incidence rates for RSV hospitalizations among healthy children aged 0-3, 4-6, 7-11, and 12-24 months to be 2560, 1670, 470, and 280 per 100 000, respectively [25] . Similarly, a prospective severe respiratory illness surveillance study in 3 hospitals in 2 South African provinces reported an incidence of RSV-associated hospitalizations of 6593, 3411, 1494, and 409 per 100 000 for children aged 0-2, 3-5, 6-11, and 12-23 months, respectively [27] . Hospitalization rates from these 2 studies were higher compared with our estimate. In Australia they did not have laboratory-confirmed cases and relied on administrative codes for acute lower respiratory tract infections and RSV, which could partly explain the higher rates. There could also be differences due to the underlying characteristics of the study populations, including a high prevalence of human immunodeficiency virus in South Africa [28] . Notably, for all of these approaches and in all settings, the highest rates were in the youngest age group, which is an important consideration for targeting interventions. Estimates of RSV-associated hospitalization rates for children aged <12 months are higher than estimates of influenzaassociated hospitalization rates for children aged <12 months using the same methodology. Rates of hospitalization per 100 000 children for RSV and influenza were 1970 vs 328 for children 0-2 months, 897 vs 117 for 3-5 months, and 531 vs 101 for 6-11 months [29] . A limitation to this comparison is that the above-mentioned influenza data are from 2003 to 2012, whereas the RSV data are from the 2014-15 season. Despite this limitation, a number of other studies have found RSV to have a higher incidence among children aged <5 years compared with influenza [4, 8, 24, 26, 29, 30] . Our findings reaffirm that among young children in the United States, RSV represents a major cause of respiratory disease-associated hospitalizations.

The RSV clinical course described in our study follows trends that are similar to those presented in other studies, but site-specific variation in severity was observed [3, [31] [32] [33] [34] [35] . Severity indicators such as respiratory distress (59% in Georgia to 85% in Oregon) and wheezing (19% in Georgia to 44% in Oregon) varied by site, which could indicate different criteria for admission or different underlying characteristics of the populations at each study site. We observed that ICU admission varied substantially among our 4 study sites (from 13% to 45%), which could be explained by different thresholds for admission and different testing practices among participating FluSurv-NET hospitals [36] . Overall, 90% of the ICU-admitted children were tested for RSV compared with 75% among those not admitted to the ICU. Testing rates among ICU-admitted children and non-ICU-admitted children were similar among the very young age group, but differences up to approximately 20% can be seen among older age groups (Supplementary Table  9 ). When we looked at variation of testing by site and ICU admission, Georgia and Oregon were 40% more likely to test children in the ICU than those not in the ICU. Other factors, including gestational age at delivery, age at hospital admission, and type of insurance, may also influence ICU admission practices [33] . Most importantly, Georgia had the highest rate of ICU admissions and children requiring mechanical ventilation (9% compared with a low of 2% in California), and 4 of the 5 identified deaths occurred in Georgia. Overall, the proportion of ICU admissions in our study was higher than described in the US study by Rodriguez-Fernandez et al (10% vs 27% reported in our study), which was based on 2 pediatric hospitals [34] , but with a substantially smaller sample size (253 children aged <2 years compared with our sample of more than 1500 children) that may explain the variation. Similar to other studies [3, 32] , mechanical ventilation was used in <10% of the cases. Furthermore, 67% of all hospitalized cases had no clear risk factor for severe disease, consistent with a previous report [35] and highlighting the importance of developing new interventions that can be broadly applied to infants.

This study highlights the opportunity to adapt an existing platform for influenza hospitalization surveillance in order to monitor RSV in the United States. In 2014-2015, the RSV and influenza seasons largely overlapped (the RSV season nationally ranged from 11 October 2014 to 9 May 2015, peaking on 7 February 2015) [5] . However, this is not the case every year and would need to be taken into account to avoid substantial underestimation of RSV-associated burden. We applied a previously described method to adjust for inherent factors of the surveillance system such as under-ascertainment of cases due to variability in testing practices and diagnostic test sensitivity [12, 15, 16] . Although RSV testing overall was high among children aged <2 years with respiratory symptoms, we observed a substantial decrease in testing among children aged ≥6 months, necessitating the adjustment. The study included several limitations. First, due to limited funding, the study was limited to 4 of the 13 states that currently participating in FluSurv-NET [13] . Nonetheless, the 4 sites together provided a diverse population, and their catchment population reflected 4% of the US population aged <2 years. However, results may not be generalizable to the entire US population. Including additional hospitals in the future would allow for more robust assessments, lessening the impact of variability in clinical practices among healthcare settings. Second, surveillance was limited to 1 season. Several years of surveillance are needed for robust burden estimates because there can be variations in RSV circulation and severity from year to year [11] . Third, we only obtained data on RSV testing practices from December 2014 to April 2015 and did not include October 2014 and November 2014. While this may have underestimated our rate calculation, the observed variability for the underdetection multipliers by month is small, particularly among children aged 0-2 months (Supplementary Table 8 ). Fourth, we used 90% sensitivity for viral culture based on a previous reference [19] . Weinberg et al [37] found low sensitivity for RSV diagnosis by culture (36%) when compared with real-time polymerase chain reaction (RT-PCR) [37] . However, for the analytic part of this study, culture represented only approximately 6% of those tested. Finally, we based the correction of case ascertainment on ICD-9 codes that were preselected based on the most common codes identified through surveillance among RSV cases from other studies [1] . As such, we may have inadvertently missed RSV cases who did not meet the preselected criteria, which could have led to a slight underestimation of hospitalization rates. On the other hand, since several of the preselected ICD-9 codes are not specific for RSV, we may have included hospitalizations associated with other respiratory viruses that cocirculated during this time [38] . Despite these limitations, our estimates of RSV-associated hospitalization rates were similar to those previously reported in systematic surveillance following the ARI case definition (sudden symptom onset with fever and/or respiratory symptoms and other related conditions including apnea and rule-out sepsis) [22, 39] , validating the capability of FluSurv-NET to provide estimates of RSVassociated hospitalization rates.

Our findings emphasize the importance of RSV as a cause of hospitalization in young children, especially those aged 0-2 months, regardless of their gestational age at birth. This laboratory-based approach that adjusts for testing practices and diagnostic testing sensitivity may produce robust estimates of RSV disease burden to inform prevention strategies and, when used over the years (as it is the case for influenza), could allow for the assessment of the impact of public health interventions. Future studies using this approach could consider a correction factor for differential testing based on severity, as we reported above to be the case for ICU-admitted children in some sites, and should consider expanding surveillance to best capture RSV seasons that do not overlap with circulation of influenza viruses.

Supplementary materials are available at Journal of the Pediatric Infectious Diseases Society online.

Respiratory syncytial virus-associated hospitalizations among children less than 24 months of age

Influenza and the rates of hospitalization for respiratory disease among infants and young children

Respiratory syncytial virusassociated hospitalizations among infants and young children in the United States

Community-acquired pneumonia requiring hospitalization among U.S. children

Respiratory syncytial virus seasonality -United States

Progress toward a respiratory syncytial virus vaccine

On the relative role of different age groups during epidemics associated with respiratory syncytial virus

Estimates of hospitalization attributable to influenza and RSV in the US during 1997-2009, by age and risk status

Hospitalizations associated with influenza and respiratory syncytial virus in the United States

Evaluation of recent New Vaccine Surveillance Network data regarding respiratory syncytial virus hospitalization rates in US preterm infants

The burden of respiratory syncytial virus infection in young children

Estimating influenza disease burden from population-based surveillance data in the United States

The US Influenza Hospitalization Surveillance Network

Diagnostic accuracy of rapid antigen detection tests for respiratory syncytial virus infection: systematic review and meta-analysis

Respiratory syncytial virus: infection, detection, and new options for prevention and treatment

Rapid antigen-based testing for respiratory syncytial virus: moving diagnostics from bench to bedside?

Molecular diagnosis of respiratory viruses

Multiplex PCR for typing and subtyping influenza and respiratory syncytial viruses

Current best practices for respiratory virus testing

Rapid viral diagnosis of acute respiratory infections: comparison of enzyme-linked immunosorbent assay and the immunofluorescence technique for detection of viral antigens in nasopharyngeal secretions

Detection, pathogenesis, and therapy of respiratory syncytial virus infections

New Vaccine Surveillance Network. Population-based surveillance for hospitalizations associated with respiratory syncytial virus, influenza virus, and parainfluenza viruses among young children

Global burden of acute lower respiratory infections due to respiratory syncytial virus in young children: a systematic review and meta-analysis

Estimating the hospitalization burden associated with influenza and respiratory syncytial virus

High burden of RSV hospitalization in very young children: a data linkage study

Hospitalizations associated with influenza and respiratory syncytial virus among patients attending a network of private hospitals in South Africa

The role of HIV in influenza-and respiratory syncytial virus-associated hospitalizations in South African children

RSV Global Epidemiology Network. Global, regional, and national disease burden estimates of acute lower respiratory infections due to respiratory syncytial virus in young children in 2015: a systematic review and modelling study

Influenza Hospitalization Surveillance Network. The burden of influenza hospitalizations in infants from

Disparities between black and white children in hospitalizations associated with acute respiratory illness and laboratoryconfirmed influenza and respiratory syncytial virus in 3 US counties-2002-2009

Respiratory syncytial virusassociated mortality in hospitalized infants and young children

The epidemiology and clinical characteristics of young children hospitalized with respiratory syncytial virus infections in Guatemala

Respiratory syncytial virus hospitalization outcomes and costs of full-term and preterm infants

Respiratory syncytial virus genotypes, host immune profiles, and disease severity in young children hospitalized with bronchiolitis

Factors associated with disease severity in children with bronchiolitis

Survey of influenza and other respiratory viruses diagnostic testing in US hospitals

Superiority of reverse-transcription polymerase chain reaction to conventional viral culture in the diagnosis of acute respiratory tract infections in children

Centers for Disease Control and Prevention. The National Respiratory and Enteric Virus Surveillance System (NREVSS). 2019. Available at

New Vaccine Surveillance Network Study Group. Epidemiology of respiratory infections in young children: insights from the new vaccine surveillance network

We thank all of those involved in respiratory syncytial virus hospitalization surveillance at each of the Influenza Hospitalization Surveillance Network sites, including Stepy Thomas, Suzanne Segler, Emily Fawcett, and Shelton Bartley from the Georgia Emerging Infections Program and Charisse Nitura from the Centers for Disease Control and Prevention (CDC).Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC.Potential conflicts of interest. E. J. A. has received research funding unrelated to this article from Novavax, Pfizer, GSK, Merck, Sanofi-Pasteur, Regeneron, Micron, and Medimmune and has also consulted for AbbVie and Pfizer unrelated to this article. All remaining authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed