Discussion
In this study, we focused on medically attended RSV infections, which was based on the presence of clinical diagnosis codes for RSV-associated diseases, including pneumonia and bronchiolitis or positive lab test results for RSV infections that are documented in patient’s EHR and required medical care or hospitalisation. Children with mild infections are often not sick enough to present to a healthcare setting. Even in a healthcare setting, most providers do not test for RSV unless the patient is being hospitalised. Therefore, these medically attended RSV infections are more likely severe RSV infections. In fact, among the 19 936 incident cases of medical attended RSV cases occurred in children aged 0–5 years during October 2022–December 2022, 14 103 (70.7%) were RSV-associated diseases, including RSV-associated bronchiolitis (54.5%). In this study, we used first-time medically attended RSV infections in order to not double count two different clinical visits for the same RSV infection, since it is difficult to discern whether two close RSV visits recorded in patient EHRs were for two different or the same RSV infection.
In 2020, the incidence rate of RSV infection was low throughout the year, which is likely due to non-pharmaceutical interventions such as lockdown, masking and social distancing that prevented RSV from spreading. Computational models that simulated non-pharmaceutical interventions and associated immunity debt predicted a large outbreak in the 2021 winter.45 46 However, our data showed that the seasonal pattern returned in 2021, but the incidence rate was lower for severe diseases, including RSV-associated bronchiolitis than expected. Positive lab test-confirmed RSV infection reached prepandemic levels for children aged 0–1 year and was higher for children aged 0–5 years. Certain COVID-19 preventative measures remained in place in 2021 that limited the spread of RSV infections. In April 2022, the CDC lifted the mask mandate but still recommended that people wear masks at public transportation settings.47 Interestingly, very young children aged 0–1 year (as of 2021), many of whom were born after 2020, also showed increased RSV infection in 2021. Waning maternal immunity due to low RSV exposure during the COVID-19 pandemic and the consequent decrease in transplacental RSV antibody transfer may have contributed to increased RSV infections in 2021 compared with 2020. However, RSV infections in 2021 did not reach the levels in 2022, largely because of the preventive measures and fewer COVID-19-infected children.
In 2022, medically attended RSV infection including both clinically diagnosed severe RSV diseases and positive lab test-confirmed infections reached a historically high rate, higher for RSV diseases than the positive lab test-confirmed RSV infection. Among children aged 0–1, the peak incidence rate of RSV-associated diseases in November 2022 was 47% higher than lab test-confirmed RSV. These data suggest that the 2022 RSV surge was disproportionately driven by more severe cases of RSV diseases, which could not be fully explained by possibility of increased testing practices, awareness or transmission through day-care or siblings alone. While non-pharmaceutical interventions in 2021 and immunity debt might have contributed to the increased rate of RSV infection, these factors alone could not fully explain the huge surge in November 2022, representing 4–5 times as many severe RSV-associated diseases as in 2021. For children aged 0–1 year (as of 2022), if the immune debt due to waning maternal immunity was the main contributor, we would expect that the level of RSV infection in 2022 to be similar to that in 2021. Instead, the peak incidence rate of severe RSV diseases in children aged 0–1 year was 2285 cases per 1 000 000 person-days in November 2022, a 161% increase compared with the peak rate of 874 cases per 1 000 000 person-days in August 2021. These data suggest that together with the effects of RSV-specific immunity debt and other factors, the large buildup of COVID-19-infected children and the potential long-term adverse effects of COVID-19 on the immune and respiratory systems7–11 may have contributed to the 2022 winter surge of severe RSV diseases. Future work is warranted to examine whether prior COVID-19 infection was associated with increased risk for medically attended RSV infections while accounting for other risk factors.
RSV is the most common virus associated with bronchiolitis. However, many other viruses cause bronchiolitis, including human rhinovirus, coronavirus, metapneumovirus and adenovirus.48 In the winter of 2015, there was a marked increase in incidence rate of unspecified bronchiolitis, which remained stable from 2015 through 2019. Starting in 2017–2018, there was an increase in the incidence rate of lab test-confirmed RSV but was not accompanied by a proportional increase in clinically diagnosed RSV diseases until in 2019–2020. The underlying reasons for these changes remain unknown but may be attributable to advances in genomic surveillance, increased detection of respiratory viruses in the laboratory, the emergence of new viral strains, increased awareness, among others. The linear time-series regression model used a linear term to model the increasing time trend; however, the observed peak incidence rate of RSV in 2022 was 143% higher than expected, suggesting that these factors may not be the sole contributing factor for the 2022 surge.
Several studies from the USA and other countries reported disrupted seasonality of RSV infection during 2020–2021.16 19–23 Our study is the first to report the long-term time-series data of medically attended first-time RSV infection among young children from 2010 to 2023. We show that our EHR-based incidence rate of medically attended RSV infection corresponds closely with the CDC-reported RSV-associated hospitalisations. In addition, our study provides information that is complementary to the CDC data. First, the EHR-based data reported both laboratory-confirmed cases and RSV-associated diseases, which allowed use to perform separate analyses for different types of RSV infection and compared with unspecified bronchiolitis. Second, our study population was drawn from a nation-wide database of EHRs collected in diverse clinical settings across 50 states in the US, representing demographically and clinically diverse RSV cases. The database also contains longitudinal patient data of more than 20 years, which allowed us to build robust time-series models. The CDC RSV-NET began data collection in children in 2018 based on voluntary reporting from state health departments. The EHR database is updated daily, which allows for real-time surveillance. Together, these complementary data resources provide a comprehensive overview of updated RSV activity and outcomes at both medical and general population levels in the USA.
Currently, the changes in the epidemiology of RSV especially the unusual surge in 2022 remain unknown. Multiple factors may have contributed to RSV circulation and its associated diseases in the community, including non-pharmaceutical interventions, change in human behaviours, viral interferences, immunity debt, waning population-level immunity and the long-term adverse effects of COVID-19 infection on the immune and respiratory systems. While non-pharmaceutical interventions have been largely lifted, other factors may still play a significant role, which makes the postpandemic monitoring of respiratory virus infections necessary and complicated.
Limitations
Our study has several limitations: first, the paediatric population from the TriNetX network are those who had medical encounters with healthcare systems that contribute data to TriNetX. Though the population is large (1.7 million children aged 0 to 5 years in our study), they do not necessarily represent the entire US population. Although our study showed a time trend that is well-aligned with that from the CDC population-based surveillance data, results from this study need to be validated in other populations. Second, though the EHR data were drawn from 34 healthcare organisations across 50 states, covering diverse geographic regions, we were unable to further break down the trend patterns by region due to TriNetX’s deidentification restrictions. Third, this is a retrospective cohort study of monthly incidence of medically attended RSV infections based on analysing patient EHRs and there may be overdiagnosis/misdiagnosis/underdiagnosis of RSV infection and associated diseases in patient EHRs.