Respiratory Syncytial Virus-Associated Hospitalizations Among Children <5 Years Old: 2016 to 2020

NVSN surveillance methods have been described elsewhere.^5,6,13 Briefly, we conducted active, prospective surveillance ≥5 days per week for children with ARI during December 01, 2016, to September 30, 2020, at 7 pediatric medical centers in Nashville, Tennessee; Rochester, New York; Cincinnati, Ohio; Seattle, Washington; Houston, Texas; Kansas City, Missouri; and Pittsburgh, Pennsylvania. We enrolled eligible children <18 years old with ARI defined as an illness presenting to the emergency department or hospitalized with 1 or more of the following: fever, cough, earache, nasal congestion, runny nose, sore throat, vomiting after coughing, wheezing, shortness of breath, rapid, or shallow breathing, apnea, apparent life-threatening event, brief resolved unexplained event, and/or myalgia. Our analyses were restricted to hospitalized children <5 years old. We excluded children with fever and neutropenia, those with a known nonrespiratory cause for hospitalization, current illness lasting >14 days, hospitalization <5 days after a previous hospitalization, or hospitalization more than 48 hours before screening and enrollment. We also excluded newborns never discharged from the hospital and transfers from other hospitals after an admission of more than 48 hours. A parent or guardian provided informed consent before the child’s enrollment. Institutional review board approval was obtained at each site and at the CDC (CDC IRB#6770).

Structured interviews with the parent or guardian were conducted to collect the following: demographics, history of current illness, antibiotic or antiviral use, history of breastfeeding, gestational age at birth, palivizumab administration during the most recent respiratory season (ie, during the months of September through March), daycare or school attendance, household size, mother’s education level, and if there was cigarette or e-cigarette use in the household. Medical chart reviews were performed and documented chronic comorbid conditions including prematurity (<37 weeks gestational age [WGA] at birth), clinical interventions, and clinical outcomes including receipt of supplemental oxygen above baseline, ICU admission, and the need for mechanical ventilation.

Midturbinate nasal ± throat swabs were collected for research testing. If a research specimen was not obtained during the visit, a salvaged clinical specimen was used for testing if available. Specimens were tested for RSV and other respiratory pathogens, including influenza, human metapneumovirus, human parainfluenza viruses 1 through 4, rhinoviruses and enteroviruses, and adenovirus by reverse transcription polymerase chain reaction (RT-PCR).⁶ Specimens collected from children enrolled in 2020 were also tested for severe acute respiratory syndrome coronavirus disease 2 (SARS-CoV-2). CDC-supported proficiency testing was conducted by all sites to ensure the validity of results. RSV-positive results from clinical RT-PCR testing as part of routine care were also included. In the rare event of discordant results, the RSV-positive result by either test superseded the negative result. Additionally, an RSV-negative result by clinical RT-PCR testing superseded an indeterminate or absent result from research testing.

Data were collected and managed using Research Electronic Data Capture hosted at each surveillance site.^14,15 Data from each site were combined at CDC and all analyses were done using SAS 9.4 (SAS Institute, Cary, NC). Characteristics by RSV status were compared by using χ-square tests. Kolmogorov-Smirnov tests compared seasonal RSV patterns by age group. Multivariable logistic regression was used to identify independent risk factors for ICU admission or supplemental oxygen administration without ICU admission among RSV-positive children after adjusting for study site and age.

For population-based RSV-associated hospitalization rates, only residents of defined catchment area counties for the 7 sites were included; Nashville (Davidson County, TN); Rochester (Monroe County, NY); Cincinnati (Hamilton County, OH); Seattle (King County, WA); Houston (Brazoria, Chambers, Fort Bend, Galveston, Harris, Liberty, Montgomery, and Waller Counties, TX); Kansas City (Jackson County, MO); and Pittsburgh (Allegheny County, PA). County-specific population denominator data were from the 2020 US bridged-race population estimates.¹⁶ The calendar year totals for 2017, 2018, 2019, and 2020 were used as the denominators for the 2016 to 2017, 2017 to 2018, 2018 to 2019, and 2019 to 2020 seasons, respectively. The proportion of births by WGA from the 2015 to 2019 US natality data were multiplied by the US bridged race estimates for children <2 years old to obtain WGA denominators.^17,–20 Population-based numerators were obtained by inflating the observed number of hospitalizations to account for weeks with <7 days of surveillance, the percentage of eligible children not enrolled, sensitivity of RSV RT-PCR testing (87.6%, CDC unpublished data using serology as the gold standard relative to RT-PCR), and the sites’ estimated proportion of catchment area ARI hospitalizations captured. Rates were calculated per 1000 children and 95% confidence intervals (CI) were determined by bootstrap percentiles based on 1000 bootstrap samples for each rate.²¹ Rate ratios (RR) and bootstrap 95% CIs were used to identify significant differences between rates. Rates were considered significantly different if the rate ratio CIs did not include the null value of 1.00.

Among 49 139 eligible children <5 years old screened in the emergency department or inpatient wards at 7 surveillance sites, 27 686 (56%) were enrolled, nearly half of whom were hospitalized (13 651, 49%), and of those, nearly all (13 524, 99%) were tested for RSV and had conclusive results (Fig 1). Enrollment rates were similar by sex (57% and 56%, respectively), slightly higher for infants (59%) compared with children 1 to 4 years old (54%), similar for Non-Hispanic (NH) white and NH Black children (both 59%), which was slightly lower than Hispanic children (61%). Among enrolled hospitalized children <5 years old, 31% tested positive for RSV (range by season [26% to 38%] and site [26% to 36%], Table 1). Infants 3 to 5 months old had the highest proportion RSV-positive (50%) among the age groups and the proportion testing positive for RSV decreased with increasing age. Peaks in RSV-associated hospitalizations occurred during December or January each season (Fig 2) and seasonal patterns were consistent across all age groups (all Kolmogorov Smirnov tests P > .05).

Among enrolled hospitalized children <5 years old, those testing positive for RSV were more likely to be younger, NH white, and have private insurance (Table 1, all P < .001). RSV-positive children were less likely to have a chronic comorbid condition compared with those that were RSV-negative (20% and 37%, respectively, P < .001). Most RSV-positive children (65%) did not have a chronic comorbid condition or history of prematurity compared with RSV-negative children (50%, P < .001). Among RSV-positive infants, 74% did not have a chronic comorbid condition or history of prematurity. Nearly two-thirds (62%) of RSV-positive children <5 years old received supplemental oxygen and 21% were admitted to the ICU compared with 43% (P < .001) and 15% (P < .001) of those that were RSV-negative. The median length of stay did not differ by RSV status; however, 39% of RSV-positive children had length of stay ≥3 days compared with 25% of RSV-negative children (P < .001). Respiratory virus codetections were more common among children with RSV compared with RSV negative children (P < .001). There were no SARS-CoV-2 codetections. There were 17 (0.2%) in-hospital deaths among RSV-negative children and no in-hospital deaths among RSV-positive children.

Among children <2 years old, prematurity was more common in RSV-negative than RSV-positive children (23 vs 20%, P = .02, Table 1). Receipt of ≥1 dose(s) of palivizumab during the current or most recent respiratory season was reported for 23% of RSV-positive children and 29% of RSV-negative children that were born <29 WGA and/or had chronic lung disease or congenital heart disease.

Across the entire study period, the overall RSV-associated hospitalization rate was 4.0 (95% CI: 3.8–4.1) per 1000 children <5 years old and ranged between 3.3 (95% CI: 3.1–3.6) and 4.6 (95% CI: 4.3–4.8) per 1000 by season (Fig 3). The RSV-associated hospitalization rate was highest among children 0 to 2 months old (23.8 [95% CI: 22.5–25.2] per 1000) and decreased sharply with increasing age group with children 24 to 59 months old having a rate of 1.0 [95% CI: 0.9–1.1] per 1000. By single month of age, the hospitalization rate was highest among children 1 month old (31.2 [95% CI: 28.6–34.1] per 1000) and decreased with increasing months of age (Table 2). RSV-associated hospitalization rates were higher for males than females (4.4 [95% CI: 4.2–4.6] and 3.5 [95% CI: 3.4–3.7] per 1000, respectively, Table 3).

Hispanic children <6 and 6 to 11 months old had higher RSV-associated hospitalization rates (23.7 [95% CI: 21.8–25.7] and 9.2 [95% CI: 8.0–10.5] per 1000, respectively) compared with NH children (16.2 [95% CI: 15.3–17.0] and 6.8 [95% CI: 6.2–7.4] per 1000, respectively). NH white children had lower RSV-associated hospitalization rates compared with the other race/ethnicity groups except compared with NH Black children <6 months old, NH other race (ie, race other than white or Black or identified as multiracial) children 6 to 11 months old, and Hispanic children 24 to 59 months old (Table 3, Supplemental Fig 4). The RSV-associated hospitalization rate was higher for Hispanic children <6 months old compared with NH Black children but did not significantly differ between the 2 groups among children ≥6 months old.

Among children <2 years old, the RSV-associated hospitalization rate was 8.5 (95% CI: 8.1–8.8) per 1000. Within this age group, children born at <37 WGA had nearly double the RSV-associated hospitalization rate of children at term (14.6 [95% CI: 13.5–15.8] versus 7.5 [95% CI: 7.2–7.9] per 1000, Supplemental Table 4). Among preterm children, RSV-associated hospitalization rates were highest in the most premature children (<29 WGA) at 3 times the rate of children born at term (RR = 3.32 [95% CI: 2.62–4.06]).

Risk factors for ICU admission among RSV-positive children included age <6 months (0–2 months, adjusted odds ratio [aOR] = 1.97 [95% CI: 1.54–2.52], 3–5 months, aOR = 1.56 [95% CI: 1.18–2.06] compared with children 24–59 months), prematurity (aOR = 1.32 [95% CI: 1.08–1.60]), and having ≥1 underlying comorbid condition (aOR = 1.35 [95% CI: 1.10–1.66], Supplemental Table 5). Risk factors for supplemental oxygen among RSV-positive children not admitted to the ICU included chronic lung disease (aOR = 1.71 [95% CI: 1.31–2.24]), congenital heart disease (aOR = 2.11 [95% CI: 1.40–3.20]), and a history of prematurity (aOR = 1.31 [95% CI: 1.08–1.60], Supplemental Table 6). There were differences by site in the use of supplemental oxygen and ICU admission.

In this 4-year study of more than 13 000 hospitalized children <5 years old across 7 US pediatric medical centers, we found nearly 1 of every 3 ARI hospitalizations were associated with RSV. Most RSV-positive children (3 of every 5) in our study had no history of a chronic comorbid condition or prematurity, a finding consistent with previous studies.^5,11,12 Notably, three-quarters of RSV-positive infants <1 year old were otherwise healthy with no chronic comorbid conditions or history of prematurity. These findings highlight that among the youngest children, most severe RSV disease (ie, leading to hospitalization) occurs among children that were previously healthy.

The highest mean annual RSV-associated hospitalization rate of 31.2 per 1000 occurred among 1 month old infants and hospitalization rates decreased with increasing age. The average annual RSV-associated hospitalization rate in infants 0 to 5 months old was 18.6 per 1000, more than twice the rate of infants 6 to 11 months old (7.6 per 1000). These findings were consistent with previous studies using similar methods and underline the importance of young age as a risk factor for severe RSV disease.^5,11,12 

In addition to higher RSV-associated hospitalization rates associated with younger age, we found that among children <2 years old, prematurity was associated with higher RSV-associated hospitalization rates. Children <32 WGA had the highest RSV hospitalization rates and rates decreased with increasing gestational age at birth. Although rates were higher among premature infants, premature children comprised a small percentage (20%) of children <2 years old hospitalized for RSV, with most RSV-associated hospitalizations occurring in healthy, term infants.

We also analyzed RSV-associated hospitalization rates by race and ethnicity. Overall, rates were higher for Black, other race, and Hispanic children compared with NH white children. Hispanic children <6 months old had a higher rate compared NH Black and NH white children. Rates between NH Black and Hispanic children 6 to 59 months old did not significantly differ and were higher for both groups compared with NH white children. Rha et al noted a higher RSV-associated hospitalization rate for Hispanic children and NH Black children during 1 surveillance season and Iwane et al noted higher rates among Black children compared with white children based on 7 surveillance seasons.^5,22 Our hospitalization rates, and those from Iwane et al and Rha et al, were not adjusted for potential confounding factors such as prematurity, underlying comorbid conditions, or socioeconomic factors because of insufficient population denominator data for those characteristics.

In this and prior NVSN studies, we estimated lower RSV-associated hospitalization rates compared with other studies using administrative claims data and/or modeling methods.²³ Among the reasons for the differences is that active surveillance may have a higher specificity for excluding children that may be misclassified as RSV-associated by other methods. Since this study conducted active prospective surveillance at 7 large pediatric medical centers, it did not rely on less precise medical claims or discharge diagnosis data, a heterogeneous mix of hospital types, or on assumptions or imputations that other study designs use.

It should be noted that our study’s RSV-associated hospitalization rates were slightly higher (15% to 20%) relative to the previous NVSN studies.^5,11,12 This is likely because of several factors. NVSN expanded from 3 sites during 2000 to 2009 to 7 sites beginning in 2015. Year-round surveillance for ARI commenced in December 2016. Because of increased routine clinical RT-PCR testing, we included children that had a positive clinical RSV RT-PCR result but were not positive or were not tested by the research laboratory, increasing our RSV-positive totals by 3%. We also adjusted for RSV RT-PCR sensitivity as an additional inflation factor when calculating the population-based rates. By using a universal estimate of test sensitivity, we did not account for potential differences in respiratory panel sensitivity by site. The RT-PCR sensitivity adjustment was added to be more consistent with methods used to estimate influenza-associated hospitalization rates in the United States.^24,25 

As in prior studies, we found that RSV-positive children were more likely to receive supplemental oxygen and to be admitted to the ICU.^5,11,12,26 Among RSV-positive children, infants <6 months old had a higher odds of ICU admission compared with those ≥6 months old. Being born premature and most comorbid conditions were also associated with increased odds of ICU admission. Among RSV-positive children not admitted to the ICU, chronic lung disease, congenital heart disease, and prematurity were associated with a higher odds of supplemental oxygen administration. There were differences by site in the use of supplemental oxygen and ICU admission, likely reflecting differences in clinical practice and thresholds for ICU admission in different health systems. Although comorbid conditions and prematurity increased the odds for these more intensive patient care outcomes, most RSV-positive children receiving these interventions did not have any of these risk factors.

Surveillance did not start until December 1, 2016. This likely resulted in underestimating the hospitalization rate for that season by 24% to 36% based on subsequent seasons that included October and November. Likewise, the arrival of the coronavirus disease 2019 pandemic in the United States during early 2020 likely shortened the duration of 2019 to 2020 season because of masking, school closures, and other nonpharmaceutical interventions.²⁷ The 7 sites where surveillance was conducted may not be representative of the US population of children <5 years old. For example, the surveillance sites may be skewed toward representing more urban populations given the locations of the pediatric medical centers. However, the sites are geographically dispersed across the continental United States and the study population was racially and ethnically diverse. Multiple years of surveillance allowed for robust estimates but also resulted in large sample sizes, making even modest differences statistically significant. To offset this, we attempted to highlight meaningful differences rather than reporting all statistically significant results. Although we described palivizumab use among recommended risk groups, the results were solely based on parental recall and should be interpreted with caution. Finally, there may have been systematic differences among children enrolled and those not enrolled. However, comparisons of basic characteristics collected during screening did not reveal substantial differences between the groups.

Nearly one-third of ARI hospitalizations among children <5 years old were associated with RSV. Younger infants and children born prematurely experienced the highest rates of RSV-associated hospitalization. However, most children hospitalized with RSV were otherwise healthy, suggesting that universal administration of safe and effective RSV prevention products may significantly reduce morbidity among children in the United States.

We thank all members of the New Vaccine Surveillance Network as well as the children and parents that participated in the study.

aOR

adjusted odds ratio

ARI

acute respiratory infection

CDC

Centers for Disease Control and Prevention

CI

confidence interval

NVSN

New Vaccine Surveillance Network

RSV

respiratory syncytial virus

RT-PCR

reverse transcription polymerase chain reaction

WGA

weeks gestational age