Risk of Subsequent Respiratory Virus Detection After Primary Virus Detection in a Community Household Study—King County, Washington, 2019–2021

Abstract

Background

The epidemiology of respiratory viral infections is complex. How infection with one respiratory virus affects risk of subsequent infection with the same or another respiratory virus is not well described.

Methods

From October 2019 to June 2021, enrolled households completed active surveillance for acute respiratory illness (ARI), and participants with ARI self-collected nasal swab specimens; after April 2020, participants with ARI or laboratory-confirmed severe acute respiratory syndrome coronavirus 2 and their household members self-collected nasal swab specimens. Specimens were tested using multiplex reverse-transcription polymerase chain reaction for respiratory viruses. A Cox regression model with a time-dependent covariate examined risk of subsequent detections following a specific primary viral detection.

Results

Rhinovirus was the most frequently detected pathogen in study specimens (406 [9.5%]). Among 51 participants with multiple viral detections, rhinovirus to seasonal coronavirus (8 [14.8%]) was the most common viral detection pairing. Relative to no primary detection, there was a 1.03–2.06-fold increase in risk of subsequent virus detection in the 90 days after primary detection; risk varied by primary virus: human parainfluenza virus, rhinovirus, and respiratory syncytial virus were statistically significant.

Conclusions

Primary virus detection was associated with higher risk of subsequent virus detection within the first 90 days after primary detection.

The circulation of respiratory viruses is associated with substantial morbidity, mortality and economic burden in the United States and globally [1–5]. During the coronavirus disease 2019 (COVID-19) pandemic, the circulation of non–severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) respiratory viral infections was reduced [6], but the ending of emergency measures and discontinuation of community mitigation efforts have led to the resurgence of these infections. The mechanisms underlying the cocirculation of respiratory viruses are complex and governed by a number of interconnected factors that may not well described. One of these factors includes viral-viral interactions, wherein one respiratory viral infection influences the risk of subsequent infections [7, 8].

The concept of viral interference, in which one viral infection reduces the risk of subsequent infection, was supported by findings during the 2009 influenza A (H1N1) pandemic. Studies showed disruptions of anticipated influenza spread in communities with heightened rhinovirus circulation [9]. The antiviral innate immunity induced by rhinovirus infection is thought to potentially confer protection against subsequent influenza virus infection at a population level, though the duration of protection as well as whether nonspecific viral interference occurs remains unknown [10]. However, studies investigating other respiratory viral infections at the individual level have shown varying results. Studies in children conducted during endemic periods of respiratory virus cocirculation suggest an increased risk of subsequent viral infections after an initial infection [11] or identified no association between initial viral infection and protection from subsequent acute respiratory illness (ARI) [12]. These findings highlight the complexity of how factors such as age, immunity, virology, and behavior may influence susceptibility to respiratory virus infection. Given the seasonal burden and pandemic potential of certain respiratory viruses, understanding the factors that mitigate or augment circulation of specific viruses could help inform public health decision making.

In the current study, we characterized the epidemiology of respiratory viral infections within a community-based household study before and during the COVID-19 pandemic. We assessed whether the risk of prior respiratory virus detection affects the risk of subsequent viral detection over 2 respiratory viral seasons.

METHODS

Study Design

We retrospectively analyzed data from a longitudinal household study designed to assess respiratory virus circulation in the Seattle, Washington, metropolitan area from 25 October 2019 to 26 June 2021 [13]. Households were recruited remotely through school newsletters and social media advertisements. Household eligibility criteria included a residence where ≥3 individuals slept in the home for ≥4 days per week in or around King County, Washington; having an adult English-speaking household member; and having ≥1 child aged 3 months to 17 years. Eligible household participants included adults and children aged ≥3 months. All adult participants provided written informed consent. Child participants enrolled after proxy consent via a parent or guardian, with assent required in those aged ≥12 years. After enrollment, all participants remotely self-screened for the onset of ARI symptoms weekly. Household study participation continued through the COVID-19 pandemic. We defined the beginning of the pandemic as 23 March 2020, the first day of the Washington state stay-at-home ordinance [14]. Of note, during the 2019–2020 influenza season, a subset of this study population participated in an influenza home-based test-and-treat interventional clinical trial [15] (ClinicalTrials.gov; trial no. NCT04141930). These studies were approved by the University of Washington (UW) Institutional Review Board.

Data Collection

At enrollment, each participant self-collected a baseline nasal swab specimen, following a step-by-step instructional sample collection guide (Quick Start Instruction Card) included in all prepositioned nasal swab specimen collection kits [16]. For children aged ≤12 years, nasal swab specimens were collected by a parent or guardian. Participants subsequently underwent active weekly symptom surveillance, during which they self-monitored for ARI, which was defined as new or worsening cough or the presence of ≥2 respiratory symptoms reported within 72 hours after symptom onset (Supplementary Table 1). During the 2019–2020 season, for each ARI episode, the symptomatic participant was prompted to self-collect a nasal swab specimen. During the 2020–2021 season, for each ARI episode, the symptomatic participant and their household participants were prompted to collect nasal swab specimens. In addition, during the 2020–2021 season, if reverse-transcription polymerase chain reaction (RT-PCR)–confirmed SARS-CoV-2 was detected in a household, additional nasal swab specimens were collected from participants every 2 days for 14 days from the initial SARS-CoV-2–positive swab specimen. During both study years, data collection procedures were repeated if multiple respiratory illnesses were experienced during the study period. All data were collected using Project REDCap [17].

Laboratory Testing

Before October 2020, all nasal swab specimens were placed in universal transport medium and mailed at ambient temperature to the UW study laboratory. Between October 2020 and June 2021, all nasal swab specimens were shipped dry. Specimen extraction and respiratory pathogen RT-PCR testing procedures were described elsewhere [13, 18]; the multiplex RT-PCR tested for 15 viral pathogens, including influenza A, influenza B, respiratory syncytial virus (RSV), enterovirus, seasonal coronavirus, human metapneumovirus, human parainfluenza virus, rhinovirus, and adenovirus, as well as 3 bacterial pathogens. A complete list of all pathogens the multiple RT-PCR tested for are shown in Supplementary Table 2. All specimens collected on or after 1 January 2020 were tested for SARS-CoV-2 via RT-PCR (Supplementary Methods).

Data Analysis

Clinical

The prevalence of respiratory viral detections in study specimens and the Seattle metropolitan area [19, 20] by calendar time was plotted (Supplementary Methods); prevalence by virus was calculated as a ratio of the frequency of a specific viral detection to the aggregation of all viral detections at each time point. We defined a single viral detection as a nasal swab specimen collected from a unique participant in which only 1 respiratory pathogen was detected over the study period and single viral codetection as a nasal swab specimen collected from a unique participant in which >1 respiratory pathogen was concurrently detected over the study period. To capture new viral acquisitions, multiple detections were defined as ≥2 positive swab specimens collected at distinct time points from the same participant over the study period, ≥7 days apart when different pathogens were detected and ≥14 days apart when pathogens were the same. For study specimens that met the definition of multiple detections, primary detection was defined as the first virus detected and secondary detection as the subsequent virus detected. When a participant had >2 positive nasal swab specimens, a detection could be both secondary (relative to preceding detections) and primary (relative to subsequent detections). We examined the frequency of primary and secondary pairs of viral detection. Additional information on the nasal swab specimens used in this analysis is available (Supplementary Methods).

Statistical Analysis

A Cox regression model was developed to calculate risk of subsequent respiratory viral detection after a primary respiratory viral detection; primary detection was modeled as a time-dependent covariate (categorized into 4 risk periods based on the days since last primary detection: <90, 90–180, or 181–365 days or no prior detection with the primary virus within the past year), with and without adjustments for age, sex, the number of household participants, and the presence of a child <5 years old in the household; hazard ratios (HRs) were calculated, and risk ratios were approximated by HRs. Calendar time was used as the time scale, and secondary detection was used as the response variable. Separate models were developed for each type of respiratory virus for primary detection (Supplementary Table 3). For secondary detections, we excluded instances where only rhinovirus and adenovirus were detected, owing to the possibility of prolonged viral shedding. Wald tests, with robust standard errors and clustering on households, were used to evaluate the time-dependent covariates. Schoenfeld residuals were calculated to evaluate each variable in the time-dependent Cox regression models, and all models were validated. Since we had no information on prior detections at the time of enrollment, it is likely that the time-dependent covariate described above would be misclassified as “no prior detection” at the start of follow-up in some participants. Therefore, we performed sensitivity analyses where we discarded the first 90 days of follow-up and reran the Cox models.

A plot of the smoothed incident rate was constructed for secondary detection before and during the COVID-19 pandemic, defined as beginning 23 March 2020 [21]. Number of events, person-days, and incident rates were calculated for each of the 4 risk periods. We defined P < .05 as statistically significant. All data analyses were performed using SAS software, version 9.4, or R software, version 4.1.1.

RESULTS

During the 2019–2020 season, in the Seattle metropolitan area, both RSV and influenza B peaked at a similar time in late December 2019, both occurring before the peak of influenza A (Figure 1A). There was persistent detection of human parainfluenza virus, rhinovirus, and seasonal coronavirus throughout the 2019–2020 season. These community circulation patterns were also reflected by the household study specimens (Figure 1B). In the household specimens, there was a decrease in respiratory virus detection corresponding with the start of the pandemic which correlated with findings from the surrounding community. The total person-time of observation per virus in study specimens is shown in Supplementary Table 4. Rhinovirus detection persisted throughout the 2-year study period in both the household study and community data.

Between November 2019 and June 2021, 1861 participants (median age, 32 years; range, 3 months to 84 years; 60.8% adults) from 470 households were enrolled (Supplementary Figure 1), with a mean household size of 4 (standard deviation, 0.78) members. The majority of participants reported receiving an influenza vaccine for the 2019–2020 season (86.2%) and 2020–2021 season (67.1%). Overall, 4630 nasal swab specimens were returned; 4273 (92.3%) specimens represented unique viral detection episodes: 1201 (28.1%) from symptomatic participants, 1189 (27.8%) from household contacts of symptomatic participants, and 1883 (44.1%) collected from participants at baseline.

Of the 4273 specimens included in the analysis, 17.6% of specimens had ≥1 respiratory virus detected (Table 1). Overall, rhinovirus was the most frequently detected virus, though seasonal coronavirus, adenovirus, RSV, influenza B, and influenza A were also frequently detected. Most positive specimens (55.2%) were collected from symptomatic participants at specimen collection. Of note, 102 (8.6%) of the rhinovirus-positive specimens were collected from the household contacts of symptomatic participants. Viral codetection was also common, constituting 10.2% of virus-positive specimens.

There were 105 specimens collected from 51 participants that met the definition for multiple respiratory viral detections, resulting in 54 viral detection pairs (Table 2 and Table 3); of note, there were 3 specimens that served as both a secondary and primary virus. The median time between primary and secondary detections was 38 days (range, 11–540 days). The mean and median lengths of time from primary to secondary detection by primary virus and by viral detection pairs are shown in Supplementary Figure 2 and Supplementary Figure 3. The frequency of multiple detections was highest in adults aged 18–49 years, followed by in children aged 5–12, <5, and 13–17 years. No instances of multiple detections were observed in adults aged ≥50 years. Among the 54 viral detection pairs, the frequency of virus types in primary and secondary detections as well as the percentage of secondary detections, given a particular primary virus detection, are shown in Figure 2.

Rhinovirus and seasonal coronavirus were the most frequently detected viruses for primary detection; seasonal coronavirus was the most frequently detected virus for secondary detections. There were 8 instances with rhinovirus as the primary and seasonal coronavirus as the secondary virus; 4 instances each of rhinovirus as the primary virus with influenza A and RSV as the secondary viruses. There were 5 instances in which seasonal coronavirus was the primary virus and secondary virus. Additional viral pairs with influenza A, influenza B, human metapneumovirus, and human parainfluenza viruses as both primary and secondary infections were identified in no more than 2 instances each during the study period.

The risk of secondary detection varied depending on the type and timing of the most recent primary virus detection (Table 4). There was a significantly higher risk of a secondary detection within the first 90 days from the primary viral detection, after adjustment for age, sex, number of household participants, and presence of a child <5 years-old in the household, for primary detection of rhinovirus (HR, 1.38 [95% confidence interval, 1.07–1.77]), human parainfluenza virus (1.66 [1.05–2.61]), and RSV (2.06 [1.06–3.99]). The adjusted models also showed lower secondary detection risk with increasing age, although sex and the number of participants in a household were not significantly related to the risk of secondary viral detections. Unadjusted Cox models gave similar results by primary virus type (Supplementary Table 5).

Unadjusted and adjusted models in which the primary viral detection was seasonal coronavirus, enterovirus, influenza B, and influenza A showed a higher risk of secondary detection within the first 90 days after primary virus detection, but the differences were not statistically significant. Owing to the COVID-19 pandemic and associated respiratory prevention practices, the number of respiratory viruses detected likely decreased with calendar time (Figure 3 and Supplementary Figure 4) and this, combined with limited follow-up, may be associated with fewer secondary detections in the periods 90–180 and 181–365 days after the primary virus detection. Overall, there was no consistent pattern of increased or decreased risk of infections after 90 days past the primary virus detection. We performed sensitivity analyses in which we discarded the first 90 days of follow-up to avoid misclassification of primary detection status. Results from these sensitivity analyses were consistent with the main analysis (Supplementary Table 6).

DISCUSSION

In this household study before and during the COVID-19 pandemic, the risk of subsequent respiratory viral detection was higher up to 90 days after a primary respiratory viral detection, irrespective of the respiratory viral pathogens, compared with the risk in participants who had not had viral detection previously in follow-up. There was a statistically significant lower risk of subsequent viral detection with increasing age, adjusted for sex and household size, which is consistent with the descriptive clinical characteristics of multiple viral detections presented here. During the COVID-19 pandemic, there was a decrease in seasonal respiratory virus detection in both the household study and Seattle metropolitan community specimens.

Prior epidemiological and community surveillance studies have noted that the peak incidence of one respiratory virus may be delayed following the peak incidence of a different respiratory virus [7, 9, 10, 21], which supports the concept of respiratory viral interference at the population level. However, at an individual level, pediatric cohort studies have identified no association between primary viral infection and a subsequent period of protection from secondary viral infection [12]. One study described an increased risk of secondary viral infection following a primary viral infection, varying by primary viral etiology [11], indicating possible viral potentiation. In that study, the authors hypothesized that the developing immune system, leading to greater ARI susceptibility, may put children at increased risk of secondary infection. These disparate findings highlight the complicated dynamics underlying multiple viral infections and underscore the importance of examining viral interference and potentiation at both a population and an individual level.

The model presented here examines the risk of any subsequent respiratory viral detection, excluding rhinovirus and adenovirus, at various time points following an initial viral detection.

This model of any subsequent viral detection was used to maximize the analysis sample size, as the COVID-19 pandemic may have reduced circulation of other viruses and may have decreased the number of viral detections available to study. Our findings show an increased risk of subsequent viral detection in the first 90 days after detection of a primary respiratory virus. We also found a statistically significant decrease in risk of subsequent viral detection with increasing age, adjusted for sex and household size. Nonetheless, sample size limitations restricted our ability to examine specific viral pairings, some of which have been hypothesized to result in viral interference [7, 9, 10, 21–23]. Thus, additional studies across multiple respiratory seasons, and in nonpandemic seasons, are needed to better understand the epidemiology of multiple viral infections and to further elucidate whether specific viral pairings or interactions between specific respiratory viruses confer a period of protection or a period of increased risk after primary infection.

Multiyear observational studies have historically formed the basis for our current understanding of the seasonal patterns of respiratory virus cocirculation in the general population [24–27]. Following the development of RT-PCR, longitudinal studies of households have been shown as a valuable way to study respiratory virus seasonality and ARI frequency [27–29]. In our household cohort, the 2-year seasonal detection of respiratory viruses appear to be representative of respiratory viral infections in the Seattle metropolitan area. Our study also demonstrates the ability of remote, longitudinal household surveillance during the ongoing COVID-19 pandemic; we were able to detect SARS-CoV-2 and demonstrate changes in viral cocirculation during the first year of the pandemic, including persistent rhinovirus detection, which is consistent with surveillance from the Seattle metropolitan community, as well as other studies [30–32].

Our study is subject to several limitations. First, it was primarily conducted during the COVID-19 pandemic, which may have reduced viral cocirculation, limiting sample size. Second, the pandemic likely affected participant behavior, including COVID-19 vaccination, use of nonpharmaceutical interventions, and other factors that may affect exposure and infection and were not accounted for in this analysis. Thus, the generalizability of our findings may be limited to the pandemic period and location of study. Third, we were unable to distinguish between distinct virus species, lineages, or subtypes owing to virologic assay limitations, which prevented us from examining primary viral detection and secondary detection risk at a more specific level. Fourth, the analyses presented here are limited to examining common seasonal respiratory viruses; there may be instances of bacterial-viral codetection, nonseasonal respiratory viral codetection, or seasonal respiratory viruses below the limit of detection unaccounted for here.

Fifth, while we used robust case definitions for primary and secondary viral detections, there may have been misclassification, including persistent viral shedding misclassified as unique detection instances. Sixth, participant symptom status was only known for some study specimens because symptom data were not obtained at baseline sample collection or not always provided by participants or household contacts during active surveillance, which prevented us from accurately adjusting for symptom status in the model. Seventh, while our models adjust for important covariates, there may be unmeasured confounders associated with the risk of primary and subsequent detections, therefore, the associations do not demonstrate a causal association. Finally, because no information was available about viral detection status before study enrollment, it is possible that misclassification of the “no prior infection” variable occurred, which may bias the HR estimates presented here toward the null. However, sensitivity analyses showed similar results when the first 90 days of follow-up were discarded.

These results suggest that within the first 90 days after detection of a respiratory virus, there is a higher risk of subsequent respiratory virus detection. The statistical significance of the risk of subsequent virus detection varied by primary virus etiology; rhinovirus, human parainfluenza virus, and RSV as the primary virus significantly increased the risk of secondary respiratory virus detection. Overall, these findings highlight the important role of household studies in understanding respiratory virus cocirculation and epidemiology. Further epidemiological studies are needed to understand the complex dynamics of respiratory virus cocirculation.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We would like to thank all the household volunteers for their time, effort, and dedication as well as members of the Seattle Flu Study investigators team.

Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the US Centers for Disease Control and Prevention. Gates Ventures was not involved in the design of the study and does not have any ownership over the management and conduct of the study, the data, or the rights to publish.

Financial support. This work was supported by Gates Ventures (the Seattle Flu Study is funded by Gates Ventures), and a US Centers of Disease Control and Prevention contract. This work utilized REDCap at the Institute of Translational Health Sciences (ITHS) which is supported by the National Center for Advancing Translational Sciences of the National Institutes of Health under award UL1 TR002319.

References

Author notes