Changes in Seasonal Respiratory Illnesses in the United States During the COVID-19 Pandemic

Abstract Background Respiratory tract infections are common, often seasonal, and caused by multiple pathogens. We assessed whether seasonal respiratory illness patterns changed during the COVID-19 pandemic. Methods We categorized emergency department (ED) visits reported to the National Syndromic Surveillance Program according to chief complaints and diagnosis codes, excluding visits with diagnosed SARS-CoV-2 infections. For each week during March 1, 2020 through December 26, 2020 (“pandemic period”), we compared the proportion of ED visits in each respiratory category with the proportion of visits in that category during the corresponding weeks of 2017–2019 (“pre-pandemic period”). We analyzed positivity of respiratory viral tests from two independent clinical laboratories. Results During March 2020, cough, shortness of breath, and influenza-like illness accounted for twice as many ED visits compared with the pre-pandemic period. During the last four months of 2020, all respiratory conditions, except shortness of breath, accounted for a smaller proportion of ED visits than during the pre-pandemic period. Percent positivity for influenza virus, respiratory syncytial virus, human parainfluenza virus, adenoviruses, and human metapneumovirus were lower in 2020 than 2019. Although test volume decreased, percent positivity was higher for rhinovirus/enterovirus during the final weeks of 2020 compared with 2019; with ED visits similar to the pre-pandemic period. Discussion Broad reductions in respiratory test positivity and respiratory emergency department visits (excluding COVID-19) occurred during 2020. Interventions for mitigating spread of SARS-CoV-2 likely also reduced transmission of other pathogens. Timely surveillance is needed to understand community health threats, particularly when current trends deviate from seasonal norms.

A c c e p t e d M a n u s c r i p t  Respiratory illnesses occur frequently, and most persons in the United States (US) experience at least one acute upper respiratory tract infection (URI) annually [1]. URI symptoms include sore throat, sneezing, rhinorrhea, nasal congestion, sinus pain, cough, headache, myalgia, loss of appetite, chills, and fever [2]. Whereas URIs occur from the nasal cavity through the larynx, lower respiratory tract infections (LRTIs) occur below the larynx and include pneumonia, bronchitis, and bronchiolitis. LRTIs are responsible for a sizable proportion of emergency department (ED) visits and hospitalizations [3]. During 2001-2004, the average annual age-adjusted pneumonia-associated hospitalization rate in the US was 465 per 100,000 population, with 7.4% of these patients dying in the hospital [4]. Globally, communityacquired pneumonia is estimated to account for 15% of deaths in children < 5 years of age in 2007 [5], and in the US, bronchiolitis is the most common cause of hospitalizations among infants < 1 year old [6].
More than 200 viruses can cause respiratory infections [1, 2,7]. Rhinoviruses are the most frequent cause of the common cold, responsible for 30% to 70% of all respiratory infections [2,7]. Endemic human coronaviruses are common causes of respiratory illnesses in adults and are reported to have caused 7% to 18% of common colds prior to the novel coronavirus disease 2019 (COVID-19) pandemic [8,9]. Other common causes of respiratory illness include respiratory syncytial virus (RSV) types A and B, human parainfluenza virus (HPIV) types 1-4, adenoviruses, human metapneumovirus (HMPV) and influenza viruses A-B.
Most but not all viral respiratory pathogens exhibit seasonal prevalence patterns [8][9][10]. Influenza, RSV, and human coronaviruses typically have peak incidence in the winter, rhinoviruses usually have peak incidence in spring and fall, and enteroviruses generally have peak incidence in the summer. Seasonality of respiratory viral infections is likely affected by multiple factors, including age-dependent human behavioral changes (e.g., school calendars and to a limited degree, spending time indoors during colder months) [8,9,[11][12][13], meteorologic factors (e.g., humidity and temperature) [10], and climactic influences on host resistance (e.g., sun exposure and related vitamin D levels) [9,14]. The COVID-19 pandemic globally changed typical patterns of influenza illness. While SARS-CoV-2 infections exhibited explosive growth in 2020, influenza activity in 2020 fell below historical seasonal norms [15], and ED visits were A c c e p t e d M a n u s c r i p t 6 less frequent than in prior years, with fewer patients presenting with influenza or acute bronchitis [16].
We examined symptoms and diagnoses among ED patients as well as changes in independent clinical laboratory respiratory viral test results in order to evaluate whether typical seasonal respiratory pathogen activity (excluding SARS-CoV-2) changed during the pandemic.

Data Sources
De-identified ED records were obtained from the National Syndromic Surveillance Program  Additionally, the following chief complaint categories were the most frequent among respiratory-related ED visits in 2019, and were thus also included in this analysis: "cough," "influenza-like illness," "shortness of breath," "sore throat," and "fever or chills." We utilized a keyword weighting approach to categorize respiratory-related ED visits by defining terms and ICD-10-CM codes that resulted in records being included or excluded in each clinical categorization (Supplemental Tables 1, 2) [20,21]. Chief complaint narratives for each visit record were scored based on a weighted keyword algorithm, with positive or negative values assigned for the presence of each keyword. Chief complaints with a score of six or greater were classified into the indicated category, and visits could be classified into more than one category. Abbreviations frequently used in medical records were included in the keyword matching algorithm [20,21]. Pathogen-specific discharge diagnosis-based categories were defined based on reported ICD-10-CM codes for "influenza," "adenovirus," "human parainfluenza virus," "rhinovirus," "human metapneumovirus" and "respiratory syncytial virus." In order to focus analyses on illnesses due to pathogens other than severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), ED records that included a diagnosis of COVID-19 were excluded from chief -complaint and discharge diagnosis-based categories. These were defined as any ED visits with discharge diagnoses that include ICD-10-CM codes  for that category was greater during 2020 than the proportion for the same week during the three prior years. We directly calculated 95% confidence intervals for PRs by exponentiating interval endpoints formed from standard error approximations of the logarithm of prevalence ratios. All analyses were conducted using R software, version 4.0.3.

Classifying laboratory findings
Laboratory results for each clinical specimen were classified into pathogen-specific groups based on Logical Observation Identifiers Names and Codes (LOINC) clinical terminology and test result description strings (Supplemental Table 3

Respiratory-related emergency department visits
We  Figure 1A). By the week beginning May 10, 2020, the percent of ED visits for non-COVID-19 A c c e p t e d M a n u s c r i p t 10 respiratory-related ED visits was below levels observed during the pre-pandemic period (PR <1.00 for each), with the exception of shortness of breath. By the week of June 14, 2020 ED visits with shortness of breath had declined to PR=1.36 (95% CI=1.35-1.37), at which time they began to rise in a second peak. PRs for all respiratory-related chief complaints increased until early July 2020, and by the end of 2020, prevalence of all non-COVID-19 respiratory conditions was lower than the prevalence during the corresponding weeks of the pre-pandemic period, with the exception of shortness of breath ( Figure 1A).
During the last four complete weeks of the 2020, 0.41% of ED visits were classified as URI, and 0.15% were classified as LRTI, a three-fold reduction compared with the pre-pandemic period: 1.24% had URI and 0.51% had LRTI during the same weeks of the pre-pandemic period (PR=0.33 and 0.29, respectively).
During March of 2020, the percent of ED visits with diagnoses for adenovirus, HMPV and rhinovirus was markedly elevated, compared with the pre-pandemic period (PR>5.00) ( Figure 1B Increases in respiratory illnesses ( Figure 1A) and viral pathogen identification ( Figure 1B) among ED patients in March and early April of 2020 coincided with an initial peak in confirmed COVID-19 cases ( Figure 1C). A second peak in confirmed COVID-19 cases occurred in July, 2020, and this coincided with a second rise in PR of ED visits for respiratory conditions.

Discussion
ED and laboratory data suggest that seasonal patterns of respiratory illness activity were profoundly disrupted throughout 2020 compared with earlier years. The incidence of SARS-CoV-2 has been well-documented elsewhere [19], and this study examined the impact of the pandemic on other respiratory pathogens. Beginning in mid-to late-March 2020, the proportion of respiratory-related ED visits for illnesses that lacked a COVID-19 diagnosis declined sharply, and generally remained below A c c e p t e d M a n u s c r i p t 12 seasonal norms throughout 2020. During the final four weeks of 2020, URIs and LRTIs without COVID-19 diagnoses accounted for a third as many ED visits as they did during the corresponding weeks of the pre-pandemic period. Likewise, ED visits and laboratory testing positivity indicated that influenza virus, RSV, HMPV, and HPIV circulated at levels far below seasonal norms. This is consistent with reports of decreased influenza activity during the COVID-19 pandemic [15].  [23,24]. In the UK, rhinovirus positivity declined following a national lockdown in late March of 2020, but rose sharply following re-opening of state primary and secondary schools in early September 2020, suggesting that the social distancing measures that were implemented did not prevent rhinovirus transmission in schools [25]. In addition, rhinovirus might be less susceptible to source-control methods, such as face coverings. One study found that  [28,29]. These reports are consistent with our observation of decreased influenza detection through 2020, but the impacts of viral interference and COVID-19 mitigation measures on the findings presented in this report are uncertain.
Our findings should be interpreted in the context of several limitations. Prevalence ratios of respiratory-related ED visits facilitate comparison between time periods. However, this has the potential to introduce bias from differences in comparison groups. We attempted to exclude patients with SARS-CoV-2 infection from ED visit analyses in order to highlight changes in respiratory activity from pathogens prevalent in the pre-pandemic period. We excluded visits with diagnostic codes specific to COVID-19, which were introduced on April 1, 2020 [30]; however, it is likely many ED patients with COVID-19 did not receive these diagnosis codes, particularly early in the pandemic when SARS-CoV-2 testing capacity was reduced. PRs in ED visits for respiratory conditions exhibit peaks in March-April of 2020 and June-July of 2020, which is similar to the timing of peaks in reported COVID-19 cases ( Figure   1A, C) which would be expected if some patients with SARS-CoV-2 infections did not receive COVID-19 diagnostic codes, particularly early in the pandemic. This could contribute to elevated prevalence ratios for ED visits with shortness of breath in 2020, and our findings likely underestimate actual decreases in ED visits for non-COVID-19 respiratory visits.
Ascertainment bias might influence these results, particularly regarding viral testing in EDs (Figure 1 B). Diagnoses of specific respiratory pathogens among ED patients is incomplete, since viral tests are not performed for every patient with respiratory illness, and negative results are not documented in diagnosis codes. Testing protocols vary between facilities and likely change over time. We note a substantial peak in identification of vital respiratory pathogens in EDs in March of 2020 ( Figure 1B), and while the number of viral tests performed in EDs is not available, the number of tests performed in independent clinical laboratories displays a substantial increase in testing, but not positivity, during this time period (Figure 3), which suggests that increased awareness of COVID-19, simultaneous with limited availability of tests for SARS-CoV2 led to increased testing for multiple respiratory pathogens in this time period. Increased identification of viral pathogens among ED patients during March of 2020 is therefore likely to be a consequence of increased testing rather than increased viral prevalence. Collectively, these findings document substantial changes in circulation of respiratory pathogens other than SARS-CoV-2 and in related health-seeking behavior. The long-term implications of these changes are uncertain, but it is expected that interventions designed to mitigate the spread of COVID-19 will continue to impact multiple respiratory pathogens. Understanding current viral circulation aids clinicians in understanding the most probable pathogens affecting patients, and continued surveillance is essential for guiding mitigation strategies.
A c c e p t e d M a n u s c r i p t 15

Disclaimer
The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention.   proportion of visits during 2020 for that category was greater during 2020 than the proportion for the same week during the three prior years.
b Indicates the date the patient the emergency initiated an ED visit, by week, with the week start date for 2020 shown.