Comparative incidence and burden of respiratory viruses associated with hospitalization in adults in New York City

Abstract Background Although the burden of influenza is well characterized, the burden of community‐onset non‐influenza respiratory viruses has not been systematically assessed. Understanding the severity and seasonality of non‐influenza viruses, including human coronaviruses, will provide a better understanding of the overall disease burden from respiratory viruses that could better inform resource utilization for hospitals and highlight the value of preventative strategies, including vaccines. Methods From October 2017 to September 2019, a retrospective study was performed in a pre‐defined catchment area to estimate the population‐based incidence of community‐onset respiratory viruses associated with hospitalization. Included patients were ≥18 years old, resided in New York City, were hospitalized for ≥24 hours, and had a respiratory virus detected within 3 calendar‐days of admission. Disease burden was measured by hospital length of stay (LOS), intensive care unit (ICU) admissions, and in‐hospital mortality and compared among those with laboratory‐confirmed influenza versus those with laboratory‐confirmed non‐influenza viruses (human coronaviruses, parainfluenza viruses, respiratory syncytial virus, human metapneumovirus, and adenovirus). Results During the study period, 4232 eligible patients were identified of whom 50.9% were ≥65 years of age. For each virus, the population‐based incidence was highest for those ≥80 years of age. When compared to those with influenza viruses detected, those with non‐influenza respiratory viruses detected (combined) had higher population‐based incidence, significantly more ICU admissions, and higher in‐house mortality. Conclusions The burden of non‐influenza respiratory viruses for hospitalized adults is substantial. Prevention and treatment strategies are needed for non‐influenza respiratory viruses, particularly for older adults.


| BACKG ROU N D
Respiratory viruses cause significant morbidity and mortality in adults, especially among frail older adults and those with chronic comorbid conditions. 1,2 While the burden of influenza viruses is best studied, non-influenza viruses such as respiratory syncytial virus (RSV), human metapneumovirus (hMPV), human coronaviruses (CoV), adenoviruses (AV), and parainfluenza viruses (PIV) are responsible for a substantial burden of illness in adults. [3][4][5][6] However, the population-based incidence and burden of community-onset non-influenza respiratory viruses associated with hospitalization in adults have not been systematically assessed. Previous approaches have utilized weekly laboratory surveillance and syndromic surveillance associated with discharge data in statistical models to estimate rates of hospitalization and mortality associated with respiratory viruses. 4,7,8 Yet, these methods may lack precision as specific viruses associated with acute respiratory infections (ARIs) are not routinely laboratory-confirmed nor consistently reported in discharge records. Thus, the aims of this study were to estimate the population-based incidence of different laboratory-confirmed respiratory viruses in hospitalized adults and to describe the disease burden associated with different viruses measured by hospital length of stay (LOS), intensive care unit (ICU) admissions, and in-hospital mortality. These parameters were compared in patients with influenza versus patients with non-influenza respiratory viruses in efforts to provide an enhanced understanding of the overall disease burden from respiratory viruses and better inform resource utilization for hospitals and highlight the value of preventative strategies, including vaccines.

| Study design, sites, and subjects
From October 2017 to September 2019, a retrospective study was performed to identify hospitalized adults who had respiratory viruses detected from nasopharyngeal swab specimens using a multiplex polymerase chain reaction (mPCR) assay (described below). Study sites are located in Northern Manhattan, are part of the same academically affiliated hospital system, and include three hospitals: a ~750-bed tertiary care hospital caring for adults, a ~200-bed community hospital caring for adults, and a ~250-bed children's hospital.
To estimate the population-based incidence of community-onset respiratory viruses associated with hospitalization, eligible patients were 18 years of age and older, resided in New York City, as defined by the zip code of their residence, were hospitalized for at least 24 hours, and had a respiratory virus detected within 3 calendar-days of admission. Patients were identified from the electronic health record (EHR). The Columbia University Irving Medical Center institutional review board approved this study with a waiver of informed consent.  To improve the incidence estimate's reliability, the catchment area was defined as the eight zip codes in which the hospitals had ≥ 60% market share (10 032, 10 033, 10 034, 10 040, 10 452, 10 453, 10 463 and 10 471).

| Data analysis
To assess the clinical burden of specific viruses, the following outcomes associated with each virus were determined: the median hospital LOS, the proportion of patients who had an ICU admission, the ICU LOS, and in-hospital all-cause mortality. Time spent in the emergency department prior to being admitted to an inpatient unit was included in LOS calculations. If more than one virus was de- Patients with co-detections of influenza and non-influenza viruses were excluded from these comparative analyses.

| Study population
During the study period, 4232 adults were hospitalized with a laboratory-confirmed respiratory virus and met the community-onset case definition. Testing for a respiratory virus occurred on hospital day 1 for 90.7% of patients, on hospital day 2 for 5.0% of patients, and on hospital day 3 for 2.0% of patients, and 2.3% of patients had testing performed in ambulatory settings affiliated with the hospitals.
Selected demographic characteristics and the distribution of the zip codes of patients are shown in Table 1. Demographic characteristics were similar in year 1 and year 2; 50.9% of patients were ≥65 years of age, and 23.5% of patients were ≥80 years of age.
Overall, 2447 (57.8%) of patients resided in zip codes that represented ≥ 60% of the hospital's market share.

| Viruses detection
The proportion of patients who had specific viruses detected is shown in Table 1 The cumulative proportion of all coronaviruses was similar in both years, but seasonal differences occurred in the relative proportions of CoV229E, CoVHKU1, and CoVOC43 as illustrated in Figure 1. In year 1, CoVHKU1 was most common and had a sharp winter peak.
In year 2, CoV229E peaked in winter, but continued to be detected throughout the spring.
Overall, co-detection of two or more respiratory viruses oc-

| Population-based Incidence
The population-based incidence for each virus overall and by age strata for year 1 and year 2 is shown in Figure 2A

| Clinical burden associated with different viruses
The median LOS, the proportion of subjects admitted to an ICU, the ICU LOS, and in-hospital mortality for each virus are shown in Those with RV/EV (n = 1,667 patients) and those with influenza/ non-influenza co-detection (n = 67 patients) were excluded from the analysis comparing the burden of influenza viruses (784 patients) with that of the non-influenza viruses (1714 patients). Compared to patients with influenza, a significantly higher proportion of patients with the non-influenza viruses CoV, RSV, PIV, hMPV, and AV were admitted to the ICU (11.0% vs. 16.7%, respectively, P = .002).
Compared to patients with influenza, a significantly higher proportion of patients with the non-influenza respiratory viruses died (3.2% and 5.2%, respectively, P = .025).

| D ISCUSS I ON
This study provides further insights for the population-based incidence of laboratory-confirmed respiratory viruses associated with hospitalization in adults. Our findings may provide more precise estimates of the incidence of non-influenza respiratory viruses than previous studies, which have often relied on diagnostic codes or syndromic surveillance, rather than systematic laboratory testing with a multiplex-PCR assay, and thus may have underestimated the case burden. 11 We also evaluated the incidence of respiratory viruses in the subset of patients ≥ 80 years of age and found that the incidence in this oldest age strata was consistently more than double that of patients aged 65-79 years. These findings suggest that the impact of non-influenza respiratory viruses is likely  Furthermore, lack of diagnostic testing could lead to lack of appropriate infection prevention and control and potential nosocomial transmission.
We found that influenza viruses predictably exhibited a high seasonal incidence with different types each year. The burden of non-influenza respiratory viruses was potentially higher than that of influenza viruses. Non-influenza respiratory viruses collectively had a higher incidence than that of influenza viruses collectively, although it is likely that the influenza vaccine reduced hospitalizations and mortality, particularly when there was a good match between the vaccine and circulating influenza strains. 12 We found that 2.8 times more patients were hospitalized with non-influenza viruses (excluding RV/EV) than with influenza viruses. Overall, the proportion of patients admitted to an ICU was significantly higher for non-influenza than for influenza viruses. Mortality was also significantly higher for non-influenza respiratory viruses. Notably, our analysis of the collective burden non-influenza viruses could be perceived as an underestimate as we excluded RV/EV, which was the most commonly detected virus, as similarly described by others. 13 We chose to exclude RV/EV from the analysis of the collective burden because RV/EV were the most commonly co-detected viruses and the mPCR assay used cannot distinguish rhinovirus from enterovirus nor unique subtypes. Furthermore, while RV/EV are associated with hospitalizations for ARIs 14 and exacerbations of underlying cardiac and pulmonary comorbidities, [13][14][15] prolonged viral shedding is also well described, 16 which could decrease the causal relationship of detecting these viruses with burden.
Comparison of our findings with other reports is challenging due to different methodologies including case finding and study population. For example, past studies exclusively examined older adults ≥ 65 years old and high-risk adults with congestive heart failure and chronic obstructive pulmonary disorder. 6 Furthermore, much of the morbidity and mortality associated with non-influenza respiratory viruses has been described in adult long-term care facilities experiencing outbreaks associated with high attack rates and high death rates. 17,18 Nonetheless, our incidence estimates of influenza hospitalizations were comparable to the Centers for Disease Control and Prevention (CDC) national estimates for both years. 19,20 The current study's rate of ICU admissions from RSV (16.1%) was similar to that described in previous prospective studies of RSV by Widmer and colleagues (10%) 3 and Falsey and colleagues (15%). 6 The current study's ICU admission rate of 11.5% associated with influenza was also similar to that previously described (6%, 3 12% 6 ) as was the 16.5% rate associated with HMPV (13% 3 ). In conclusion, the burden of non-influenza respiratory viruses is substantial, particularly for older adults who lack durable immunity for respiratory viruses and frequently have comorbid conditions that increase their risk of severe disease. Thus, prevention and treatment strategies for non-influenza respiratory viruses, including effective vaccines, are needed. Future research should further assess the clinical impact of specific human coronaviruses and the potential impact of specific coronavirus types on the epidemiology and impact of SARS-CoV-2. formal analysis (lead); investigation (lead); methodology (equal); software (lead); writing-original draft (lead); writing-review and editing (equal). Connor R. Goldman: Conceptualization (supporting); data curation (supporting); formal analysis (supporting); investigation (supporting); methodology (supporting); project administration (supporting); validation (supporting); writing-original draft (supporting); writing-review and editing (supporting).

Pe e r Rev iew
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/irv.12842.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.