Acute Respiratory Illnesses in Children in the SARS-CoV-2 Pandemic: Prospective Multicenter Study

Provisional data were obtained from the New Vaccine Surveillance Network, a prospective, population-based ARI surveillance platform funded by the Centers for Disease Control and Prevention (CDC).^18,19  Surveillance for ARIs in the New Vaccine Surveillance Network was reestablished in 2015, and the current 7 US pediatric medical centers are located in the following locations: Cincinnati, Ohio; Houston, Texas; Kansas City, Missouri; Nashville, Tennessee; Pittsburgh, Pennsylvania; Rochester, New York; and Seattle, Washington. Subject recruitment occurred ≥4 days per week in the ED and ≥5 days per week in inpatient settings over 4 respiratory seasons. Analyses were restricted to subjects recruited during December 2016 to April 2017, and October to April each year from 2017 to 2020. The CDC’s and each institution’s Institutional Review Boards approved the study.

Children aged <18 years were eligible for enrollment if they resided within a site’s surveillance area and visited the ED or were admitted to the hospital within 48 hours of enrollment with ≥1 of the following symptoms: fever, cough, earache, nasal congestion, runny nose, sore throat, posttussive vomiting, wheezing, shortness of breath or rapid or shallow breathing, apnea, apparent life-threatening event, brief resolved unexplained event, or myalgias; and duration of illness <14 days.¹⁸  Children were excluded if they had a known nonrespiratory cause for their illness, had chemotherapy-associated fever and neutropenia, or had been transferred from another hospital after an admission of >48 hours, were admitted <5 days after a previous hospitalization, were never discharged from the hospital after birth, or had previous enrollment in the study <14 days before current enrollment. Of note, 3 sites restricted ED enrollment primarily to children aged <5 years during most of the periods of interest (Seattle during December 2016–April 2017, November 2017–April 2018, November 2018–April 2019, and December 2019–March 2020; Pittsburgh during December 2016–April 2018, November 2018–April 2019, and December 2019–March 2020; Kansas City during December 2016–April 2017, November 2017–April 2018, November 2018–April 2019).

After obtaining informed consent, the research staff interviewed the child’s parent and/or guardian using a standardized case report form that included demographic information (including age, race and ethnicity). Midturbinate nasal swabs and/or oropharyngeal swabs were collected and combined in viral transport media by study personnel or clinical staff from enrolled children. For intubated patients, tracheal aspirates were accepted as alternatives to oropharyngeal swabs. When nasal, oropharyngeal, or tracheal aspirate specimens were not available, clinically salvaged respiratory specimens were obtained. 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 influenza. Diagnostic assay methods varied by site and included Luminex NxTAG Respiratory Pathogen Panel (Cincinnati and Kansas City), BioFire FilmArray Respiratory Panel (Seattle), Applied Biosystems TaqMan Array Microfluidic Card (Rochester), and in-house real-time reverse transcription-polymerase chain reaction assays (Houston, Pittsburgh, and Nashville).¹⁸  All sites conducted CDC-sponsored proficiency testing to ensure the validity and consistency of respiratory viral detections at each site.^18,19 

Research enrollment activities were paused or limited for several weeks in March at all sites as institutional policies suspended clinical research to conserve supplies and personal protective equipment during the COVID-19 pandemic. Enrollment occurred at a single children’s hospital at 6 sites and at 2 children’s hospitals in Rochester; capacity to care for children was not affected in any of these hospitals during the pandemic. Table 1 reveals the dates of suspension of enrollment, school closures, and stay-at-home order implementation by study site, which were provided by principal investigators, as instituted by local orders at each site. The total number of daily eligible and enrolled children in the ED and inpatient settings during the study period were provided by all sites. Community mitigation onset was defined as the earliest of either school closures or stay-at-home orders. The community mitigation period was defined as the first full calendar week after community mitigation initiation (Table 1 ) until the end of the study period (ie, April 30, 2020 for all sites).

Descriptive statistics were summarized as frequency (percentage) for categorical variables or median (interquartile range) for continuous variables. The number of eligible and enrolled children and proportions of RSV and influenza were evaluated by calendar weeks each season. We classified the final status as inpatient setting if children who were originally enrolled in the ED were then admitted.

We sought to evaluate whether eligible case numbers and proportions testing positive for RSV and influenza were lower during the COVID-19 pandemic-related community mitigation period as compared with “baseline.” Baseline was defined as the same calendar weeks during 2016–2019. We fit generalized linear mixed-effects models (linear to model the mean number of eligible cases to obtain mean differences and binomial-logit to model the odds of patients testing positive for RSV and influenza, separately, among enrolled patients). We included random effects for study site and season and fixed effects for categorical calendar week in each model, further including for each model the 2 previous weeks of corresponding outcomes (ie, lagged outcomes) as covariates (flu and RSV cases are understood to be heavily driven by cases occurring in recent weeks).

We also aimed to evaluate whether eligible ARI cases and proportions testing positive for RSV and influenza during the COVID-19 pandemic-related community mitigation period were the same as would have been predicted on the basis of data from previous years. Regression models for eligible cases (linear) and proportions of patients testing positive for RSV and influenza (logistic) were based on data from all sites before mitigation. Each model included 2 weeks of lagged outcomes, a two-way interaction between study site and season alone with each lower-order term, and a restricted cubic spline on calendar week with 6 knots positioned at equally spaced quantiles. The fitted regression models were then used to generate predicted means or proportions for the community mitigation period, which were graphically represented in conjunction with premitigation predictions using a locally weighted scatterplot smoother. We used a nominal significance level of 0.05 (two-tailed) for all analyses. All statistical analyses were performed by using Stata software version 15.1 (Stata Corp, College Station, TX), and R version 4.0.2.

Over 4 respiratory seasons from 2016 to 2020, 45 759 eligible children with ARIs were identified: 27 114 (59%) from the ED and 18 645 (41%) from the inpatient setting. Overall, 25 415 (56%) were enrolled and had respiratory specimens collected and tested for RSV and/or influenza. The cohort median age was 19 months (interquartile range: 7–50), 56% were boys, and 33% were non-Hispanic white. Overall, 25% of the samples were RSV-positive and 14% tested positive for influenza. A total of 12 366 subjects were enrolled from ED and then discharged from the hospital; 18% and 19% tested positive for RSV and influenza, respectively. A total of 13 049 were enrolled, and their final status was the inpatient setting; 32% and 8% tested positive for RSV and influenza, respectively.

The number of eligible and enrolled subjects and the proportions of RSV and influenza by calendar week stratified by season are shown in Fig 1. In 2020, each study site had an estimated average of 10.6 (95% confidence interval [CI]: 5.92–15.2; P < .001) fewer eligible ARI cases per calendar week per site during the community mitigation period as compared with the corresponding periods in previous years.

The cumulative proportions of weekly RSV and influenza detections by study season are shown in greater detail in Fig 2. Compared with previous seasons, RSV and influenza detections in 2020 stopped after week 15. In 2020, no RSV or influenza detections were observed in weeks 15–18 (April 5 to April 30) in either the ED or inpatient settings (Fig 3). Among enrolled patients, the odds of testing positive for RSV was estimated to be 63.9% (95% CI: 23.5%– 82.9%; P = .008) lower in the community mitigation period of 2020 as compared with the corresponding time periods in previous years; for influenza, the odds were 45.8% (95% CI: 7.64%–68.2%; P = .024) lower.

Regression models were developed for all 7 sites by study year for total eligible (Fig 4A), proportion with positive test results for RSV (Fig 4B), and proportion with positive test results for influenza (Fig 4C). On the basis of fitted lines by season, for all sites except Seattle, the number of eligible ARI cases and proportion of patients who tested positive for RSV and influenza among those enrolled in the 2020 community mitigation period was lower than what would have been predicted on the basis of data from site- and time-adjusted trajectories that were present premitigation (Fig 4 A–C).

In this prospective multicenter population-based ARI surveillance study, we found a substantial and sustained decline in medically attended ARI visits and the proportion of enrolled children with RSV or influenza infection in 7 distinct geographic cities in the United States. These declines occurred promptly after community mitigation measures were implemented at each site in response to the COVID-19 pandemic.

Published studies have documented that these measures effectively decreased the reported burden of COVID-19 cases, with a robust temporal association between the timing of these strategies and the decrease in the trend in the epidemic growth rate of reported COVID-19 cases.^4,5,20  The majority of community mitigation strategies in each of the 7 study sites were implemented around mid-March 2020; after these dates, we noted significantly fewer estimated eligible cases and lower odds of children with RSV- and influenza-positive test results compared with the same periods in previous seasons. A reduction in eligible case numbers during community mitigation periods can be explained in part by several other factors, including changes in health care-seeking behaviors and limited access to medical care.^10–13  However, it is less plausible that such factors explain the reduction in the proportions of positive detections of RSV and influenza among enrolled subjects or in the frequency of cases of severe ARIs that are hospitalized. Our study results are consistent with the hypothesis that community mitigation can slow the spread of severe RSV and influenza infections, similar to the experiences with SARS-CoV-2.

We found that the proportion of influenza-positive cases in the ED or inpatient setting in the 2020 community mitigation period is lower than what would have been predicted at most site locations, except Seattle. However, there were no cases of influenza detected in Seattle after week 15, consistent with the other sites. Similar nonmedical interventions were found effective in previous influenza pandemics, including the 1918–1919 influenza pandemic, when medical interventions were unavailable, and in more recent severe epidemics, when they were used together with medical interventions.^3,21–23  A surveillance study that monitored influenza-like illnesses and laboratory-confirmed influenza in New York revealed that influenza-like illnesses decreased with an average decrease of 96.9% to 98.9% in daily laboratory-confirmed influenza case rate per 100 000 cases in different regions in New York during the COVID-19 pandemic.²⁴  Similarly, Sakamoto et al⁶  evaluated the effects of social isolation and adherence to personal hygiene on influenza weekly activity in all age groups in Japan. Compared with the same periods in preceding years in their study, influenza activity was significantly lower in January to March 2020, coinciding with mitigation efforts. Young et al²⁵  evaluated the 2019–2020 influenza season compared with previous years and noted steeper declines in new influenza cases after the seasonal influenza peak in China and Italy, which coincided with a substantial increase in COVID-19 cases in these countries. Although the mainstay for preventing influenza remains annual vaccination, our findings suggest that, in future severe influenza pandemics, similar community mitigation measures might reduce the incidence of severe influenza infection in children.

In this study, we compared the proportion of RSV-infected children detected before and after social isolation measures in 7 cities in the United States. RSV is a major cause of ARI hospitalizations in young children during the fall and winter months and is associated with significant morbidity.^16,17 We found a >70% decrease in RSV-positive ARIs in the community mitigation period in 2020 compared with the same period in previous surveillance seasons during which similar enrollment protocols were used. Baker et al²⁶  used laboratory surveillance data from 2020 and estimated that that RSV transmission declined by at least 20% in the United States. Further active surveillance is needed during community mitigation periods to understand if RSV circulation in the United States can be affected by these measures.

Several reports indicated that parents were reluctant to take their children to health care facilities during this pandemic because of fear of SARS-CoV-2 exposure,^12,13,27  which could, in part, contribute to the decline in the number of eligible subjects with ARIs. However, we assume that the main effects would be on mild ARIs that could be managed outside of the inpatient settings. Although the number of eligible hospitalized ARI subjects decreased in our study during the period of COVID-19 mitigation, health care-seeking behavior was less likely to be responsible if we assume presentation of severely ill children would not have been demonstrably altered during this period. In addition, local health care providers and researchers altered their practices and priorities during the early part of the COVID-19 pandemic, which posed limitations in our enrollment.¹⁹  Nonetheless, all sites recorded the number of eligible subjects, even when enrollment was not possible because of institution-specific mandates in research curtailment. These data revealed a significant decrease in the number of eligible subjects in 2020, as well as in the proportions positive for RSV and influenza. Additionally, a sensitivity analysis only including the 3 sites with no pauses in enrollments revealed significant decreases for eligible and RSV cases and similar point estimates for influenza cases, albeit not significant, which may be explained by insufficient power, given the reduction in sample size. It might be expected, if confounding by limitations of enrollment to only the most ill children was influential, that the proportions of enrolled children with RSV and influenza detection might have increased, not decreased, especially because RSV and influenza are important causes of severe ARIs in children.^{16–18 ,28–31} 

Both RSV and influenza activity have natural fluctuations from season to season such that the decrease in RSV and influenza detections in our study might be partially explained by seasonal variation or even differences in virulence of circulating viral strains.^14,15,17  The use of 4 seasons of data from the same study sites by using similar enrollment criteria across wide geographic settings makes secular seasonal and virulence trends less likely to explain our findings of abrupt curtailment of RSV and influenza circulation among children in the study sites. We likewise accounted for between-site and between-year variations in prospective surveillance data. Also, we focused on comparing outcomes from the community mitigation period to comparable periods from previous years, and, by including lag variables from 2 weeks leading up to the community mitigation period to reduce the confounding effects of a hiatus in enrollment as the mitigation period began. The major strengths of our study included the use of sensitive, validated molecular virological testing, analysis of a large sample size from 7 geographically diverse cities across the United States, and the use of consistent enrollment procedures over 4 years of active, prospective, ARI surveillance.

In summary, after COVID-19 community mitigation procedures were implemented, we noted a consistently sharp decline in children with ARIs seeking medical care in EDs or being admitted to hospitals at our study institutions across 7 US cities early in the COVID-19 pandemic. Furthermore, there were no RSV or influenza cases reported ∼2 to 4 weeks after these measures were implemented in weeks 11 to 13, which contrasted with predictions based on data from previous seasons when mitigation measures were not in effect. This study validates the importance of ongoing active ARI surveillance in children to determine trends in viral positivity before, during, and after interventions, including implementation of community mitigation strategies during public health emergencies and pandemics. Future studies are needed to confirm our findings, especially in the upcoming 2020-2021 RSV and influenza respiratory season because our data suggest that community mitigation measures may have contributed to decreased RSV and influenza disease activity in young children.

We thank the children and parents who participated in this study.

ARI

acute respiratory illness

CDC

Centers for Disease Control and Prevention

CI

confidence interval

COVID-19

coronavirus disease 2019

ED

emergency department

RSV

respiratory syncytial virus

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2