Defining the seasonality of respiratory syncytial virus around the world: National and subnational surveillance data from 12 countries

Background Respiratory syncytial virus (RSV) infections are one of the leading causes of lower respiratory tract infections and have a major burden on society. For prevention and control to be deployed effectively, an improved understanding of the seasonality of RSV is necessary. Objectives The main objective of this study was to contribute to a better understanding of RSV seasonality by examining the GERi multi‐country surveillance dataset. Methods RSV seasons were included in the analysis if they contained ≥100 cases. Seasonality was determined using the “average annual percentage” method. Analyses were performed at a subnational level for the United States and Brazil. Results We included 601 425 RSV cases from 12 countries. Most temperate countries experienced RSV epidemics in the winter, with a median duration of 10–21 weeks. Not all epidemics fit this pattern in a consistent manner, with some occurring later or in an irregular manner. More variation in timing was observed in (sub)tropical countries, and we found substantial differences in seasonality at a subnational level. No association was found between the timing of the epidemic and the dominant RSV subtype. Conclusions Our findings suggest that geographical location or climatic characteristics cannot be used as a definitive predictor for the timing of RSV epidemics and highlight the need for (sub)national data collection and analysis.


Funding information Sanofi/AstraZeneca
Background: Respiratory syncytial virus (RSV) infections are one of the leading causes of lower respiratory tract infections and have a major burden on society.
For prevention and control to be deployed effectively, an improved understanding of the seasonality of RSV is necessary.
Objectives: The main objective of this study was to contribute to a better understanding of RSV seasonality by examining the GERi multi-country surveillance dataset.
Methods: RSV seasons were included in the analysis if they contained ≥100 cases. Seasonality was determined using the "average annual percentage" method. Analyses were performed at a subnational level for the United States and Brazil.
Results: We included 601 425 RSV cases from 12 countries. Most temperate countries experienced RSV epidemics in the winter, with a median duration of 10-21 weeks. Not all epidemics fit this pattern in a consistent manner, with some occurring later or in an irregular manner. More variation in timing was observed in (sub)tropical countries, and we found substantial differences in seasonality at a subnational level. No association was found between the timing of the epidemic and the dominant RSV subtype.
Conclusions: Our findings suggest that geographical location or climatic characteristics cannot be used as a definitive predictor for the timing of RSV epidemics and highlight the need for (sub)national data collection and analysis. high-risk adults. 3,4 With no existing vaccine, prevention methods include hand washing, avoiding contacts with those infected with RSV, and passive immunization through the administration of the monoclonal antibody (mAb)-palivizumab. 5 The latter, requiring monthly administration during the RSV season, is recommended in the first year of life in high-risk infants. 5 For prevention methods to be deployed effectively and to improve the planning of health services, an improved understanding of the seasonality of RSV is required.
The lack of global surveillance data results in few studies exploring the regional or global RSV seasonality using a multicountry dataset and uniform method. Studies that do transcend the national level are limited to reviews or meta-analyses based on previously published national seasonality estimates, [6][7][8] and few examples exist that are based on surveillance data. [9][10][11][12][13][14][15] Importantly, the range of statistical methods used to define seasonality in the literature hampers a robust comparison across studies.
Previous studies have found that the seasonality of RSV in temperate climates is stable over time, with most countries experiencing a distinct peak in RSV cases per season. Temperate countries in the Northern Hemisphere generally experience the start of the RSV season between September and January and those in the Southern Hemisphere between March and June, closely aligned with colder temperatures. 7 The seasonality of RSV in (sub)tropical climates appears less consistent-some studies indicate that the peak of RSV epidemics is closely aligned with the rainy season, 6,10 and others have found increased RSV activity with higher temperatures. [16][17][18] Additionally, studies have explored the impact of RSV subtype dominance on seasonality, but the association between an RSV A or B dominant season and the timing of epidemics remains unclear. [19][20][21] Several (other) climatic associations have been sought to explain variation within climate zones, but a clear predictor of seasonality has yet to be established.
The Global Epidemiology of RSV in Hospitalized and Community care (GERi) study collects detailed surveillance data to better describe the epidemiology of RSV and support prevention and control measures. Here, we aim to describe the start, end, peak, and duration of RSV epidemics in 12 countries around the world. We also assess the impact of RSV subtype dominance on the timing of RSV seasonality.

| Global RSV virological surveillance data
The GERi network, consisting of 16 countries, and the surveillance systems from which the virological data have emerged have been described previously. 22 Briefly, data were stratified by level of care (hospitalized and community) and both case definitions (ILI at community level and SARI at hospitalized level), and diagnostic methods were largely similar across participating countries. For the current analysis, seasons containing <100 cases at national level were excluded. Large countries for which we had subnational data (Brazil and the United States) were considered as separate areas, and seasonality was assessed at a subnational level.

| Average annual percentage
To estimate the start and end of the epidemic, we calculated the "average annual percentage" (AAP) of cases for each week. 13,23 The AAP was calculated as follows: where i denotes the week, n denotes the number of cases, and x is the total number of weeks in a given season. Weeks were then sorted in descending order, and the first weeks to add up to at least 75% of the AAP were identified as epidemic weeks. To identify the start and end of an epidemic, the epidemic weeks were sorted by time. The start and end of the epidemic were then identified as the first and last week of the longest consecutive epidemic weeks with a 2-week gap allowed. The number of weeks from the start to end of the epidemic was defined as the duration.
If multiple weeks had the same AAP i -and not all of these were required to make up 75% of the AAP-only those adjacent to the epidemic were included as epidemic weeks. The peak week was defined as the week with the majority of cases-if there were multiple weeks with the same number of cases, a 3-week average was calculated with these weeks as the central week. 23 The central week with the highest 3-week average was then defined as the peak week.

| RSV subtype classification
The occurrence of RSV A or B dominant seasons was explored using both a ≥60% or ≥70% threshold to define a subtype as being dominant. Seasons were excluded if <100 cases were subtyped. We investigated whether there were differences in the start, end, and duration of RSV A or B dominant epidemics using a regression analysis with robust standard errors, which took into account the potential clustering of individual country results.

| RESULTS
After exclusion, the analysis included 12 countries and 210 seasons between 2000 and 2020 (Cameroon, Romania, Vietnam, and Russia were excluded), in both the Northern and Southern Hemispheres (Table S1). This included 131 seasons on a subnational level. The dataset consisted of 501 425 cases of RSV with a median of 1585 cases per country, region, and season. The median, end, start, peak, and duration of RSV epidemics are summarized by hemisphere, climate zone, and country in Table S1.

| Temperate climate zones
In countries in the Northern Hemisphere experiencing a temperate climate, the median start of epidemics consistently occurred in December or January and ended in February or March ( Figure 1A,B and Table S1). This resulted in a median epidemic duration of 10 to The median duration of RSV activity ranged between 11 and 17 weeks, with the longest RSV seasons occurring in HHS4 (Atlanta) and the shortest in HHS8 (Denver).

| (Sub)tropical climate zones
Less stability in seasonality was found across countries experiencing a (sub)tropical climate ( Figure 3A latter study found RSV epidemics in the East to be later than in the West (i.e., a longitudinal association). 13 The positive association between latitude and RSV seasonality is largely shown in our data, but several exceptions were noted. In the United States, RSV activity appeared to start in the most Southern region (HHS4) with a later start in higher latitudes (e.g., HHSs 1,8,or 10 10 One study found that this might be due to higher levels of precipitation and humidity during the months when RSV circulates. 9 Several studies have explored the association between RSV activity and weather conditions, especially in the (sub)tropics where RSV activity appears less consistent. 9,13,27 Such associations not only explain variation in the timing of RSV epidemics between countries but also season-to-season variation on a (sub)national level. These studies have suggested that RSV epidemics occur in conjunction with the humid and/or rainy season, which for countries located near the equator explains the tendency for year-round or long-lasting RSV epidemics. 7,28 The latter provides a potential explanation for the year-round circulation of RSV that has been observed in Singapore.
Similarly, a study conducted in the Netherlands showed how some differences in the timing of RSV epidemics from season to season could be explained by relative humidity as well as temperature variations. 27 Though the association between low temperatures and RSV activity in temperate countries has been well documented, evidence regarding this association in the (sub)tropics has been conflicting. 28 We were able to assess the timing of RSV epidemics on a subnational level for two countries: the United States and Brazil. On a national level, the United States, a temperate country in the Northern Hemisphere, experienced RSV epidemics typical for its location-with the majority of RSV activity occurring in the winter. However, on a subnational level, there was considerable seasonal variation, with the RSV season starting in the South East and then moving to the North West. The start of the RSV season differed by approximately 2 months. This pattern has been described by other studies 16,29 ; however, a clear explanation for the earlier start in Florida is unknown. 16 Part of the explanation may lie in its climate. The way in which we have defined climate zones implies that Florida experiences a temperate climate based on its latitude; however, the South of Florida, the most populous area (e.g., Miami), is known to experience a tropical climate. 30 On a national level, Brazil appears to fit the association with latitude on a national level. However, this was not the case on a subnational level, as seasons appeared to first start in the Midwest and the latest onset was experienced in the South To the best of our knowledge, limited research has been done on the impact of the RSV subtype on the seasonality of RSV. One study found no such relation 26 whereas one study in Beijing found an association for longer and earlier epidemics during RSV A dominant seasons. 19 We did not find this association in our current and previous study, 20 13 Our study highlights how the diversity in approaches used to estimate seasonality results in varying, and sometimes contradictory, outcomes, which underlines the need to harmonize seasonality estimation methods.
This study is one of the few studies that has used a multicountry dataset for the estimation of (sub)national RSV seasonality.
The strength of the study is that it uses a diverse dataset spanning 12 different countries around the world. One method could therefore be applied to analyze RSV seasonality allowing for comparison between countries. A limitation of an analysis that compares seasonality across countries might be the discrepancies in surveillance systems as previously described, 22 with factors such as case definitions and the surveillance setting potentially having implications on the estimation of RSV seasonality. 32 Another limitation of our study is the lack of representation of several regions of the world as no data from countries in North Africa or the Middle East were included.