Respiratory Syncytial Virus–Associated Hospitalizations Among Young Children: 2015–2016

Inpatient surveillance was conducted at 7 pediatric medical centers from November 1, 2015, to June 30, 2016, in Rochester, New York; Cincinnati, Ohio; Nashville, Tennessee; Kansas City, Missouri; Houston, Texas; Seattle, Washington; and Oakland, California. Children <5 years old were eligible for enrollment if they resided within each hospital’s surveillance area, had an illness duration <14 days, and were enrolled within 48 hours of admission for an ARI-related condition (Supplemental Table 5). Children were excluded if they had a known nonrespiratory cause for hospitalization, had fever and neutropenia from chemotherapy, were discharged from a hospital in the previous 4 days, were transferred from another hospital after an admission of >48 hours, had never been discharged from the hospital after birth, or had previously enrolled in this study <14 days before their current admission. Study staff identified eligible children 5 to 7 days per week; each surveillance day covered admissions for a 24-hour period.

Institutional review board approvals were obtained at the CDC and at each site. Informed written consent was obtained from parents and/or guardians of eligible children for participation in standardized parent and/or guardian interviews, medical chart review, and collection and testing of midturbinate nasal and throat swabs.

Parent and/or guardian interviews for all children included information regarding patient demographics (including race and/or ethnicity), history of current illness, and social history; among children <2 years old, prematurity (<37 weeks’ gestation), gestational age at birth, and receipt of palivizumab during the 2015–2016 season were also assessed by interview. Chronic comorbid conditions, clinical interventions, and outcomes (including receipt of any supplemental oxygen above baseline during clinical care, ICU admission, and mechanical ventilation) were assessed through a standardized medical chart review.

Midturbinate nasal and throat flocked swabs were obtained and combined when both available and placed in universal transport media. For intubated patients, tracheal aspirates were accepted as alternatives to throat swabs. Specimens were transported to each site laboratory and stored at 2°C to 8°C until processed (within 72 hours). Specimens underwent testing at each site by commercial or institution-specific in-house reverse transcription polymerase chain reaction assays for RSV and other respiratory pathogens, including influenza A and B, human metapneumovirus, human parainfluenza viruses (HPIVs) 1 to 3, rhinoviruses and enteroviruses, and adenovirus. Diagnostic assay methods varied by site and included Luminex NxTAG Respiratory Pathogen Panel (in Cincinnati, Kansas City, and Oakland),¹²  BioFire FilmArray Respiratory Panel (in Seattle),¹³  Applied Biosystems TaqMan Array Microfluidic Card (in Rochester),^14,15  and in-house real-time reverse transcription polymerase chain reaction assays (in Houston^16,17  and Nashville^14,18 ). All sites conducted CDC-sponsored proficiency testing to ensure the validity and consistency of results for all viral detections across sites.

RSV infection was defined by a positive result from study-initiated RSV testing. Patients with inconclusive RSV test results due to insufficient specimen quality (as determined by assay controls) were excluded from analyses. For descriptive analyses, comparisons of patient characteristics by RSV status were performed by using χ² tests, and multivariable logistic regression analysis was used among patients who were RSV-positive to assess independent associations with the following clinical outcomes after adjusting for study site and patient age: ICU admission (regardless of the receipt of supplemental oxygen) and supplemental oxygen administration among non-ICU admissions.

Descriptive analyses included all enrolled patients, but population-based RSV-associated hospitalization rate calculations were restricted to residents of the specific counties for which additional population and care-seeking practices information was available. By site, these included Monroe County, New York (Rochester); Hamilton County, Ohio (Cincinnati); Davidson County, Tennessee (Nashville); Jackson County, Missouri (Kansas City); Harris, Fort Bend, Montgomery, Brazoria, Galveston, Liberty, Waller, and Chambers counties, Texas (Houston); King County, Washington (Seattle); and Alameda County, California (Oakland). Site-specific and overall network rates were calculated from the weighted number of RSV hospitalizations divided by the population <5 years of age for these counties multiplied by 1000^1–3 ; rates were also calculated by age group, gestational age (for children <2 years old), sex, and race and/or ethnicity groups. For each of these site-specific county areas, population denominator data were from the 2016 US bridged-race estimates¹⁹ ; denominators by gestational age were determined by applying the proportion of births by gestational age from the 2013–2016 US natality data to the 2016 bridged-race estimates, restricted to children <2 years old.^20–23  The observed number of hospitalizations was weighted to account for any weeks in which sampling was <7 days per week, the number of eligible children not enrolled, and the market share (ie, the estimated proportion of ARI hospitalizations captured by each hospital in their respective specified county areas, generated by each site by comparing their hospitalization data for ARI-related conditions to county-based data). Rates (overall and by demographic subset) were calculated with 95% confidence intervals (CIs) determined by bootstrap percentiles based on 1000 bootstrap samples for each rate.²⁴  Rate ratios and 95% CIs were also obtained by bootstrap methods and used to identify significant differences in rates. When comparing 2 rates, rate ratio CIs that did not cross 1 indicated that the rates were significantly different.

During the 2015–2016 surveillance period, 3001 (64%) of 4716 eligible children were enrolled (Fig 1), of whom 2983 (>99%) had study-specific respiratory specimens collected and tested for RSV. Among 2969 with conclusive RSV testing results, 1043 (35%) were RSV-positive. Codetected virus cases in 132 patients who were RSV-positive (13%) included 92 (9%) rhinovirus and enterovirus, 28 (3%) adenovirus, 5 (<1%) influenza, 4 (<1%) human metapneumovirus, 3 (<1%) HPIV-3, and 2 (<1%) HPIV-1.

RSV was detected throughout the surveillance period from November 1, 2015, to June 28, 2016 (Fig 2). Overall, the proportion of RSV detections among patients with ARI was highest in December (52%), January (52%), and February (47%) and lowest in June (4%). The proportion of patients who were RSV-positive varied by site (Table 1), as did the magnitude and timing of peak RSV proportion positivity, ranging from 42% to 69% and occurring from November to February (Fig 2).

Among 1043 hospitalized children who were RSV-positive, 587 (56%) were male, 903 (87%) were <2 years old, 526 (50%) were <6 months old, and 342 (33%) were <3 months old (Table 1). Race and ethnicity information, summarized in Table 1, revealed that 71% of children who were RSV-positive were reported as having non-Hispanic or non-Latino (NH) ethnicity, of whom 59% were reported as white and 25% as African American. Most children who were infected with RSV had public insurance (60%). Interviews revealed that 79% had a history of breastfeeding of any duration, 25% lived with household members who smoked, and 59% had a maternal education of high school or less (Table 2). Among all hospitalized children with ARI, children who were RSV-positive were generally younger than children who were RSV-negative (Table 1).

Chronic comorbid conditions were documented in 21% of children who were RSV-positive, including 11% with chronic lung conditions, 4% with congenital heart disease, and 4% with neurologic and/or neuromuscular conditions (Table 2). Among 903 children <2 years old who were RSV-positive, parental report revealed that 162 (18%) were born preterm (median of 34 weeks’ gestational age [WGA]) and 21 (2%) received any palivizumab during the 2015–2016 season. Those reporting preterm birth included 70 (8%) at 35 to 36 WGA, 50 (6%) at 32 to 34 WGA, 18 (2%) at 29 to 31 WGA, 19 (2%) at <29 WGA, and 5 (1%) with no available WGA information (Table 3). Among all children who were RSV-positive, 339 (33%) had a chronic comorbid condition or were <2 years old with a history of prematurity (Table 2). Children who were RSV-positive were less likely to have a comorbid condition or history of prematurity (33%) than children who were RSV-negative (46%; P < .001).

The median length of stay for subjects who were RSV-positive was 2 days (interquartile range 1–3 days); 718 (69%) received supplemental oxygen, 182 (17%) were admitted to an ICU, and 27 (3%) required mechanical ventilation; there were no deaths (Table 2). Children who were RSV-positive were more likely to require supplemental oxygen (69% vs 50%; P < .001) or ICU admission (17% vs 13%; P = .002) than children who were RSV-negative.

The characteristics of patients who were RSV-positive were further assessed for associations with ICU admission or receipt of supplemental oxygen for non-ICU admissions. Initial univariate analyses were used to identify significant study site differences for both outcomes (P < .001). Subsequent multivariable analyses were adjusted for age and study site (Supplemental Tables 6 and 7). Children who were RSV-positive who were outside the 12- to 23-month age group were more likely to be admitted to the ICU compared with those 12 to 23 months old (0–2 months: adjusted odds ratio [aOR] = 2.4 [95% CI: 1.4–4.1]; 3 to 5 months: aOR = 2.2 [95% CI: 1.2–4.1]; 6 to 11 months: aOR = 2.1 [95% CI: 1.1–3.8]; 24 to 59 months: aOR = 2.6 [95% CI: 1.4–4.9]); these differences remained statistically significant when including having any comorbid condition in the model. Those with a comorbid condition or history of prematurity were more likely to be admitted to the ICU than those who did not (aOR = 1.5 [95% CI: 1.1–2.2]). Among patients who were RSV-positive not admitted to the ICU, congenital heart disease was more common among those administered supplemental oxygen compared with those with no supplemental oxygen receipt (aOR = 2.5 [95% CI: 1.1–5.7]). In addition, NH African American children were less likely to require supplemental oxygen compared with NH white (aOR = 0.5 [95% CI: 0.4–0.8]) and Hispanic or Latino children (aOR = 0.6 [95% CI: 0.4–0.9]). Relative to 0- to 2-month-old children, 6- to 11-month-old (aOR = 1.7 [95% CI: 1.1–2.7]) and 24- to 59-month-old children (aOR = 1.7 [95% CI: 1.1–2.8]) were more likely to require supplemental oxygen; however, the difference for 24- to 59-month-old children did not remain statistically significant when including having any comorbid condition in the model. Remaining comparisons revealed no associations with supplemental oxygen receipt or ICU admission for day care and/or school attendance, other children living in the same household, household members who smoked, or history of breastfeeding.

More than three-quarters (78.4%) of children who were RSV-positive resided in the counties used to calculate hospitalization rate estimates. The rate of RSV-associated hospitalizations across all sites was 2.9 (95% CI: 2.8–3.1) per 1000 children <5 years old and 6.3 (95% CI: 5.9–6.7) per 1000 children <24 months old (Supplemental Table 8, Table 4). Younger infants had the highest rates of hospitalization, with 14.7 (95% CI: 13.6–15.9) hospitalized per 1000 children <6 months old and 18.9 (95% CI: 17.0–20.9) per 1000 children <2 months old. Individual site rates of RSV hospitalizations ranged from 1.8 to 6.5 per 1000 children <5 years old and 8.2 to 32.4 per 1000 children <6 months old (Supplemental Table 8).

Estimated rates of RSV-associated hospitalization among children <24 months old by chronological month of age were highest in 1-month-old infants (25.1 [95% CI: 21.1–29.3] per 1000). The next highest rates were in <1- and 2-month-old infants and decreased by subsequent months through 10 months of age (Table 4).

Among children <24 months old, the RSV-associated hospitalization rate for those with a reported preterm birth (9.6 [95% CI 8.1–11.1] per 1000 children) was higher than for term-born children (5.8 [95% CI: 5.4–6.3] per 1000). Further analyses stratified by gestational age yielded rates of 15.4 (95% CI: 7.3–24.3) per 1000 for those <29 WGA, 13.8 (95% CI: 8.1–20.8) per 1000 for 29 to 31 WGA, 11.3 (95% CI: 8.1–14.6) per 1000 for 32 to 34 WGA, and 7.4 (95% CI: 5.8–9.2) per 1000 for 35 to 36 WGA. Rates for gestational age groups <35 weeks were statistically significantly higher than those for the ≥37 WGA group (Supplemental Table 9).

Among children <5 years old, the RSV-associated hospitalization rate was higher among male (3.3 [95% CI: 3.0–3.6] per 1000) than female (2.6 [95% CI: 2.3–2.9] per 1000) children. Similarly, higher rates were observed among Hispanic or Latino children (3.3 [95% CI: 3.0–3.7] per 1000) and NH African American children (3.4 [95% CI: 2.9–3.8] per 1000) compared with NH white children (2.5 [95% CI: 2.3–2.8] per 1000). The higher rate estimates for Hispanic or Latino and NH African American children relative to NH white children were also observed in the 6- to 11- and 24- to 59-month age groups (Supplemental Table 9).

RSV was associated with more than one-third of ARI hospitalizations among US children <5 years old during the 2015–2016 season of multicenter surveillance. An estimated 2.9 per 1000 children <5 years old were hospitalized with RSV, with the highest rates observed among the youngest infants (25.1 per 1000 infants 1 month old). Extrapolating to the US population, we estimate 58 000 (95% CI: 56 000–62 000) RSV-associated annual hospitalizations among children <5 years old, including 50 400 (87%; 95% CI: 47 000–53 000) in those <2 years old. Among children <2 years old, higher hospitalization rates were associated with preterm birth at <35 weeks’ gestation. However, most hospitalized children who were RSV-positive had no history of prematurity or underlying medical condition, suggesting that otherwise healthy young children, in addition to children at high risk, could benefit from interventions against RSV.

RSV continues to be a major cause of hospitalized ARI among young children worldwide.^{1–3,8,25,26}  Although previous US estimates of RSV disease relied on administrative databases,^4–8  active prospective population-based surveillance generates more precise burden estimates.^1,2  Our rates are comparable to earlier US estimates from 3 of the same surveillance sites that employed similar methodology (Rochester, Nashville, and Cincinnati), which included an overall rate of 3.0 hospitalizations per 1000 children <5 years old in 2000–2004¹  and the highest rates by month of age also observed in the youngest infants in 2000–2005 (as high as 25.9 per 1000 children 1 month old).²  Current study enrollment was similarly based on admission with ARI-related conditions and collection of nasal and throat swabs. However, the current study features 4 additional surveillance sites, making the current network more geographically representative of the United States. This and other differences (eg, current molecular diagnostics and one-year surveillance period) limit further direct comparisons to the previous studies. However, our data reveal the continued substantial burden of pediatric RSV infections in the United States.

A history of prematurity was another factor associated with higher rates of RSV-associated hospitalization, in addition to younger age. Among children <24 months old, hospitalization rates were higher among children born at <35 WGA compared with term-born children. However, because of the relatively low case numbers from one season of data, no conclusions could be made regarding differences among rates for gestational age groups <35 WGA. Point estimates for 29 to 31 WGA and 32 to 34 WGA were higher than previously reported, but conclusive comparisons were not possible given the relatively small case numbers in these groups.²  In addition, although the proportion of children <2 years old who were RSV-positive with reported palivizumab administration (2%) was comparable to that in 2000–2005 (3%),²  these data were obtained by unverified parental history, limiting further analyses.

In considering other characteristics previously associated with severe RSV disease, such as male sex, race, underlying medical conditions, and environmental factors (including exposure to smoking and contact with other children),^27–35  we found higher RSV-associated hospitalization rates among male children, NH African American children, and Hispanic or Latino children that were statistically significant but small in magnitude. Although higher rates of RSV hospitalization in ≥12-month-old children have been previously observed among NH African American versus NH white children at 3 surveillance sites (Rochester, Nashville, and Cincinnati) over multiple years, analyses of Hispanic or Latino children were not possible in those studies because of low numbers.³⁰  The addition of 4 new surveillance sites raised the proportion of Hispanic or Latino children closer to that of the US population¹⁹  and allowed for additional rate estimates by Hispanic or Latino ethnicity.

Limitations of our study include that surveillance started on November 1, 2015, after RSV season had begun in certain regions (in late September to October) containing a number of our sites (Nashville, Cincinnati, Houston, and Rochester),³⁶  which would contribute to the underestimation of hospitalization rates, and covered one winter season, which precluded assessment of seasonal variation. In addition, surveillance was performed in major urban pediatric medical centers that may not fully represent other areas of the United States. Despite these limitations, our estimates rely on a sensitive surveillance platform optimized to capture pediatric RSV disease, which includes the use of a broad case definition that ensures the capture of RSV infections that present without fever, seen particularly among younger infants.³⁷  In addition, prospective testing using current molecular diagnostic techniques and specimen collection methods ensured sensitive detection not dependent on clinical testing that can vary by practitioner or institution and, in cases of bronchiolitis, is not recommended routinely.³⁸ 

Our findings represent renewed surveillance efforts at geographically diverse study locations to characterize RSV burden in young US children. With new immunoprophylaxis, antiviral, and vaccine products in development,^10,11,39  continued surveillance is important to establish robust estimates of RSV burden so that the impact of future products can be assessed. In particular, stratification of infant rates by month of age should help inform planning for prevention measures such as vaccines and immune intervention strategies (eg, maternal vaccines and specifically targeted monoclonal antibodies) that target infants <6 months old.^9,11  An improved understanding of the epidemiology of RSV disease among young children should help prioritize these interventions in the context of overall public health strategies. Because most children hospitalized with RSV have no underlying medical conditions or history of prematurity, interventions against RSV used to broadly target all young infants are likely to result in the greatest impact.

We thank the children and parents who participated in this study. We also thank all the members of the New Vaccine Surveillance Network, including Monika Johnson at the University of Pittsburgh Medical Center Children’s Hospital of Pittsburgh and Jessica L. Anderson and Dean D. Erdman at the CDC.

aOR

adjusted odds ratio

ARI

acute respiratory illness

CDC

Centers for Disease Control and Prevention

CI

confidence interval

HPIV

human parainfluenza virus

NH

non-Hispanic or non-Latino

RSV

respiratory syncytial virus

WGA

weeks’ gestational age