Severity of Illness in Bronchiolitis Amid Unusual Seasonal Pattern During the COVID-19 Pandemic Free

This cross-sectional study included US children’s hospitals participating in the Pediatric Health Information System database (PHIS; Children’s Hospital Association, Lenexa, KS). The database includes billing and administrative codes for patient encounters at children’s hospitals across the United States. For this study, dates, diagnoses, procedures, and disposition were captured from the database. Diagnoses and procedures were identified using International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), current procedural terminology (CPT), and PHIS-specific clinical transaction classification (CTC) codes.¹⁶  Billing codes cannot specifically identify high-flow nasal cannula; therefore, it was not included in the data.^16,17  Laboratory results including viral testing are not available in this database. Invasive mechanical ventilation was defined using the PHIS flag for mechanical ventilation. Noninvasive ventilation was defined as any of the following continuous positive airway pressure (PHIS CTC 521162), bilevel positive airway pressure (PHIS CTC 521164), noninvasive ventilation (PHIS CTC 521165), and noninvasive positive pressure ventilation (CPT 5A0957).^16,18 

We included children <24 months old presenting to a participating hospital between October 2016 and September 2021 with either (1) a primary diagnosis of bronchiolitis (J21) or (2) a primary diagnosis of acute respiratory failure (J96), apnea (R06.81), or cardiac arrest (I46) and a secondary diagnosis of bronchiolitis. We measured case counts of emergency department (ED). To assess the severity of illness, the outcomes included rates of hospitalization, ICU admission, noninvasive ventilation, and invasive mechanical ventilation.

We used descriptive statistics to characterize the demographic features of the patients, using frequencies with proportions and medians with interquartile ranges for categorical and continuous variables, respectively. To compare demographics between the time periods of October 2020 to September 2020 and October 2016 to September 2021, encounters across years, we used χ² and Kruskal-Wallis rank sum tests.

To evaluate trends over time in the of outcomes, the primary analysis consisted of simple logistic regression models with the outcome as the dependent variable and year as the independent variable. To allow for winter epidemics to be captured contiguously, years were defined from October of 1 year to September of the following year. The models compared patients presenting in October 2020 to September 2021 to those presenting in October 2016 to September 2020. Secondary models were adjusted for age in months, to account for higher median age among patients presenting in the most recent study year. Clustered sandwich SE estimates were used, which allow for intrahospital correlation among observations.

To examine seasonal trends, we estimated Fourier Poisson regression models separately for October 2016 to September 2020 and October 2020 to September 2021.¹⁹  Fourier models allow for the exploration of seasonality, or predictable fluctuations in observations over time, by incorporating trigonomic functions. We used case count as the dependent variable, sine and cosine terms as the independent variables, and month as the unit of analysis. Before the analysis, we elected to include a single combination of sine and cosine terms, because there is a singular annual surge in bronchiolitis. We jointly tested the null hypothesis that the sine and cosine terms from October 2020 to September 2021 were equal to those from October 2016 to September 2020.

α was set at 0.001 given the large sample size and all tests were 2-tailed. Analyses were conducted by using Stata SE 16.1 (StataCorp, College Station, TX).

The study included 46 children’s hospitals in the United States. In the 5-year period, 389 411 ED visits for bronchiolitis and 167 018 hospitalizations were included. Median age of patients presenting in October 2020 to September 2021 was higher than those presenting from October 2016 to September 2020 (P < .001, Table 1). In the year from October 2020 to September 2021, the odds of hospitalization, ICU admission, invasive mechanical ventilation, and noninvasive ventilation did not differ in unadjusted analysis or analysis adjusted for age (all P > .05, Table 2).

Seasonality of the overall number of ED visits for bronchiolitis differed significantly in the year from October 2020 to September 2021 compared to previous years (P < .001). Seasonal peaks in bronchiolitis occurred in January 2017, December 2017, December 2018, and December 2019. After the absence of a peak in the winter season of 2020 to 2021, an atypical rise in bronchiolitis cases occurred in March to September 2021 with the highest number of cases in August 2021 (Fig 1).

This study of 46 US children’s hospitals provides evidence for unusual seasonality in bronchiolitis in 2021. Following the absence of a winter epidemic in 2020 to 2021, cases began to rise in March 2021. Yet, severity of illness in infants with bronchiolitis from October 2020 to September 2021 was consistent with previous annual epidemics, as evidenced by no differences in hospitalization, ICU admission, invasive mechanical ventilation, or noninvasive ventilation.

The patterns of bronchiolitis cases in 2021 identified are consistent with previous literature on the unusual seasonal patterns in RSV described in several countries.^10–14  We do not have viral testing data to compare. The absence of an expected seasonal epidemic in winter 2020 to 2021 also validates previous literature. These trends may be driven by reduction in efforts to prevent COVID-19 infections or reduced viral immunity among children in the absence of a winter surge. Insufficient data are available at this time to know if August 2021 will be the peak of current trend.

The median age of infants presenting in the most recent season is higher than in previous epidemics. A potential explanation for this observation is infants and young children have not experienced routine childhood viral illnesses because of efforts to mitigate the spread of COVID-19. The impact of this reduced viral immunity on the magnitude and severity of the subsequent viral epidemics is not known. This study provides no evidence for increased severity this year thus far. However, it is essential to note that cases may rise after the end of the study period, and severity may vary at that time. This trend could potentially represent the beginning of a prolonged, more severe epidemic that continues through the winter of 2021 to 2022. This warrants close attention given the warnings from experts regarding the potential of a prolonged, more severe epidemic in the setting of lower herd immunity.

Strengths of this study include the large, national database with recent information on patients at 46 children’s hospitals. Limitations of this study include the administrative nature of the data, lack of viral testing results to assess trends in viral pathogens causing bronchiolitis, and generalizability beyond children’s hospitals.

The implications of the recent unusual seasonality of bronchiolitis remain unknown. This uncertainty limits the ability of institutions to plan for surges in hospitalizations and critical illness in children, which is essential component of healthcare system readiness during the COVID-19 pandemic. Future research to describe emerging trends in viral illnesses in children and compare severity of illness to previous epidemics is critical to support health care system preparedness.