Descriptive analysis to assess seasonal patterns of COVID-19 and influenza in low-income and middle-income countries in Asia, the Middle East and Latin America

Introduction

Historically, respiratory diseases, such as influenza, respiratory syncytial virus, pneumococcus and meningococcus, have displayed seasonal patterns.1–4 A disease’s seasonality, defined as a repetitive systematic yearly pattern of peaks and troughs, may differ across regions due to variations in climate and/or other factors.5 6 Vaccination campaigns and the levels of vaccine uptake also vary across regions due to inequities in the supply of and population access to vaccines, especially in low-income and middle-income countries.7 This variation may impact observed seasonal patterns. Understanding the seasonality of infectious diseases can help predict when an outbreak will occur, particularly if the infectious disease’s seasonality is driven by environmental parameters.8 Furthermore, seasonal patterns of disease are important for public health planning, such as disease surveillance and vaccination campaigns.5 A lack of a discernable seasonal pattern may support year-round or multipoint vaccination campaigns.

The COVID-19 pandemic, caused by SARS-CoV-2, continues to pose a threat to public health, with over 760 million cases and nearly 7 million deaths worldwide as of 24 July 2023.9 10 Thus far, data concerning the seasonality of COVID-19 worldwide are limited. A recent study showed that in the USA and Europe, COVID-19 infections, hospitalisations and mortality generally followed a seasonal pattern with spikes from November to April.11 However, comprehensive insights into the seasonal pattern of COVID-19 have not been reported, and limited evidence has been reported for countries outside of the USA and Europe.

Surveillance data have indicated that during the COVID-19 pandemic, the seasonal pattern of influenza was disrupted.12 Early in the pandemic, influenza transmission was significantly reduced, possibly due to non-pharmacological interventions (NPIs), such as social distancing, lockdowns and closures.13 The adoption of zero-COVID-19 policies may have further impacted the seasonality of influenza and/or other respiratory diseases. For example, in certain regions of China, such as Shanghai, following the adoption of such policies, healthcare utilisation was significantly reduced compared with that in previous years prior to the COVID-19 pandemic.14 Changes in the proportion of cases that are asymptomatic over time may also impact observed patterns of seasonality.15 16 Subsequently, influenza circulation has increased, but its pattern remains unpredictable and inconsistent, with activity late in the season in the Northern Hemisphere and early in the season in the Southern Hemisphere.13 The current seasonal pattern of influenza is unclear, and its seasonality in the post-COVID-19 pandemic era remains unpredictable, particularly in low-income and middle-income countries.13 17 Therefore, there is a need to continue evaluations of influenza seasonality patterns during the post-COVID-19 pandemic era.

The aim of this study was to examine the seasonality of COVID-19 and influenza in 12 low-income and middle-income countries. Populations living in low-income and middle-income countries typically experience a higher disease burden than those living in high-income countries due to several factors, such as poor living conditions, malnutrition and insufficient control measures, rendering vaccination critical in these populations.18 19 COVID-19 is no exception, and its disease burden is higher in developing countries due to increased infections among older adults and inadequate healthcare.20 The results of this study have the potential to improve our understanding of the continuing public health importance of influenza and COVID-19 in these countries. This study may also inform scientific exchange related to vaccination strategies and policies, including the timing of vaccination campaigns. The results will also inform future research investigating the relationships between seasonality and other key drivers of disease epidemiology.

Methods

Country selection

This study investigated the seasonality of COVID-19 and influenza in 12 low-income and middle-income countries, including countries in Southeast Asia (ie, Singapore, Thailand, Malaysia and the Philippines), Latin America (ie, Argentina, Brazil and Mexico), Africa (ie, South Africa and Morocco) and the Middle East (ie, Bahrain, Qatar and Saudi Arabia) (online supplemental figure 1). These countries were selected based on a combination of factors, including the World Bank classification, which is based on the gross national income (GNI) per capita, population size, influence of the public health system in the region and comprehensive surveillance systems for different respiratory diseases. A feasibility assessment of data availability was also conducted before finalising the country selection.

Supplemental material

Data sources

The data analysed in this study were sourced from Our World in Data (OWiD) and FluMart. All influenza cases and COVID-19 cases in these databases were included in this study, except for influenza data obtained prior to 2009, which were excluded to avoid potential bias due to the introduction of a revised influenza case definition in response to the avian influenza pandemic.21

COVID-19 data were obtained from OWiD, which is an online non-profit scientific publication that shares globally significant data for public use.22 OWiD’s COVID-19 Dashboard provides the profiles of 207 countries, including COVID-19 case numbers, deaths, vaccinations, testing and government responses.23 OWiD’s primary data source is WHO’s COVID-19 Dashboard,9 which provides information, such as case numbers, deaths and vaccination data, obtained directly from member states and official ministries of health websites. OWiD’s COVID-19 Dashboard is updated daily, and historical data are available from January 2020 to the present. OWiD includes granular clinical data, such as weekly hospitalisations, intensive care unit admissions and test positivity rates, in each included country. Recent studies have used data derived from OWiD to explore the impact of lockdown measures24 and construct models to forecast COVID-19 cases and deaths.25–27

FluMart is WHO’s Global Influenza Programme’s data storage system, which allows the real-time sharing, utilisation and analysis of influenza surveillance data from existing databases, such as the Pan American Health Organization (PAHO), European Surveillance System (TESSy), WHO African Region (AFRO) and Eastern Mediterranean Flu (EMFLU) databases.28 FluMart combines and integrates data from various sources into a single platform that can be easily used to create outputs, such as combined surveillance graphs, heat charts, virus detection graphs and an integrated surveillance dashboard.28 Previous studies have used data derived from FluMart,29 30 including a recent study exploring influenza patterns during the COVID-19 pandemic.31

Feasibility assessment

The feasibility of conducting a time series analysis of the data retrieved from each country was assessed. The suitability of the data was assessed by evaluating whether sufficient data were available to statistically analyse seasonality patterns in each country before and after 2020, assessing the extent of data continuity/missing data over time and evaluating the reliability of the data based on outliers in case counts and other face validity concerns. The results of the feasibility assessment indicated that the available data were insufficient for a statistical analysis. Therefore, an exploratory approach was adopted in this paper as described below.

Descriptive analyses

Descriptive analyses were conducted to (a) describe any seasonal patterns of COVID-19 and influenza, (b) identify a cocirculation pattern between COVID-19 and influenza, (c) compare the post-2009 pattern of avian influenza with the pattern of COVID-19 cases and (d) assess the seasonal pattern of influenza strains A/B. The descriptive analyses were conducted by generating plots of COVID-19 and influenza cases. COVID-19 seasonality and influenza seasonality were explored by generating plots of influenza (A/B strains combined) and COVID-19 cases, respectively. The cocirculation of COVID-19 and influenza was determined by the overlap between the COVID-19 and influenza plots. The comparison of the 2009 avian influenza pandemic and the COVID-19 pandemic was performed by exploring the overlap between the COVID-19 and influenza plots with the start of each pandemic set as ‘time 0’ (January 2009 for the influenza pandemic and January 2020 for the COVID-19 pandemic). Finally, the pattern of seasonal influenza stratified into the A and B strains was assessed by exploring the overlap in plots of the two strains.

Daily counts have high variability due to fluctuations in the daily reporting of cases, especially during weekends; historically, the number of weekend cases has been low due to staff shortages, and some countries no longer report cases on weekends. Therefore, the numbers of COVID-19 and influenza cases were smoothed using the average values over a 7-day window over time.

Patient and public involvement

No patients or the public were involved in this research.

Results

Seasonal pattern of COVID-19

Across the 3 years of the COVID-19 pandemic, only two countries displayed a seasonal pattern consistently each year from 2020 to 2023; specifically, Saudi Arabia experienced waves from May to August annually, and Mexico exhibited waves from December to February annually. No consistent 3-year pattern was observed in the other examined countries (figure 1).

Figure 1
Figure 1

Weekly average COVID-19 cases by year (2020–2023). (A) Southeast Asia, (B) Latin America, (C) Middle East and (D) Africa.

Several countries exhibited seasonal patterns in two of the 3 years of the COVID-19 pandemic. Two countries exhibited a similar pattern in 2020 and 2022 with a differing pattern in 2021 as follows: Brazil experienced COVID-19 waves from June to September and from November to December in 2020 and 2022, and the Philippines experienced waves of COVID-19 from July to December in 2020 and 2022. Similarly, Singapore experienced waves of COVID-19 from October to December in 2021 and 2022 with a differing pattern in 2020. South Africa exhibited waves from December to February in 2020 and 2021 as well as 2021 and 2022. No clear seasonal pattern of COVID-19 was observed in Argentina, Thailand, Bahrain, Malaysia, Morocco and Qatar (figure 1).

Seasonal pattern of influenza

Seasonal patterns of influenza cases (strains A and B combined) were consistently observed across several years from 2009 to 2022 in this descriptive analysis. No common pattern was observed in Malaysia and the Philippines. However, across the remaining countries, waves of influenza cases occurred during each month of the year. Three countries, that is, Argentina (June–August), South Africa (May–October) and Brazil (April–August) exhibited a pattern of influenza waves occurring mainly in autumn and/or winter. Thailand exhibited a pattern of influenza waves nearly year round, with no wave patterns occurring from May to June or in December. Similarly, Qatar exhibited waves in nearly all months, except for June, July and August (figure 2).

Figure 2
Figure 2

Weekly average influenza cases by year (2009–2023). (A) Southeast Asia, (B) Latin America, (C) Middle East and (D) Africa.

All remaining countries exhibited a pattern of influenza waves occurring primarily in the fall and winter beginning as early as September and ending as late as March. For example, Saudi Arabia exhibited waves from October to December, Singapore exhibited waves from November to February, Bahrain exhibited waves from November to March, Mexico exhibited waves from December to March and Morocco exhibited waves from November to March (figure 2).

Covid-19 and influenza cocirculation and pandemic patterns

A COVID-19 and influenza cocirculation pattern was only observed in Bahrain, which experienced a wave of both COVID-19 and influenza from May to September 2022, and Saudi Arabia, which experienced a wave of both COVID-19 from May to August 2022 and a wave of influenza from May to July 2022 (figure 3). No COVID-19 and influenza cocirculation pattern was observed in any other country.

Figure 3
Figure 3

Covid-19-influenza cocirculation, weekly average cases (2020–2023). (A) Southeast Asia, (B) Latin America, (C) Middle East and (D) Africa.

Similar seasonal patterns of pandemic for COVID-19 and 2009 avian influenza were observed in only two countries. Singapore exhibited an overlap in year 1 of the pandemics in May, and South Africa exhibited overlaps in the waves of each pandemic in year 1 from May to September, year 2 from June to October and year 3 from May to July (figure 4).

Figure 4
Figure 4

COVID-19 and influenza pandemic patterns, weekly average cases. 2009 is year 1 of the influenza pandemic, and 2020 is year 1 of the COVID-19 pandemic. (A) Southeast Asia, (B) Latin America, (C) Middle East and (D) Africa.

Seasonal patterns of influenza strains A and B

The overlap in influenza waves caused by influenza strains A and B was assessed from 2009 to 2022 (figure 5). South Africa exhibited the greatest degree of overlap with similar patterns between influenza strains A and B in 12 of the years examined (all years, except for 2019–2020), followed by Thailand with similar patterns in 8 years and Malaysia with similar patterns in 7 years. In contrast, the Philippines exhibited the lowest degree of overlap, with a similar pattern between the strains in only 1 year (2010–2011). Mexico, Qatar and Saudi Arabia each exhibited an overlap between the strains in three of the years examined. The remaining countries exhibited similar patterns of influenza strains A and B in 4 years (Argentina) or 5 years (Singapore, Bahrain and Morocco). No overlap between influenza A and B was observed in Brazil.

Figure 5
Figure 5

Weekly average influenza cases by strain (2009–2023). (A) Southeast Asia, (B) Latin America, (C) Middle East and (D) Africa.

Discussion

Available knowledge regarding the seasonality of COVID-19 is limited. In this study, a descriptive analysis was undertaken to describe any seasonality trends and/or overlap between influenza and COVID-19 in 12 low-income and middle-income countries. Overall, no clear COVID-19 seasonality pattern was observed in all countries, except for Saudi Arabia and Mexico, which exhibited consistent seasonal peaks in all 3 years from 2020 to 2023. Saudi Arabia exhibited waves of COVID-19 from May to August; although Hajj season occurs within this time frame, we did not observe increases in cases occurring specifically during Hajj likely due to the early response by the Saudi Arabian government, including the implementation of travel bans and suspension of religious activities.32 33

Previous studies have explored COVID-19 seasonality in select countries, including low-income and middle-income countries. Some studies identified that meteorological factors and NPIs affected the seasonality and spatiality of COVID-1934 35 and influenza.35 Global analyses of the impact of environmental factors on COVID-19 seasonality suggest that the occurrence of COVID-19 outbreaks could be predicted based on a seasonal pattern of temperature variations.36 The seasonality of COVID-19 has been explored in temperate countries, demonstrating a reduction in COVID-19 mortality during the summer of 2020 and 2021 with increased mortality in the autumn of 2020.37 This pattern is similar to the historical pattern of the influenza virus, which had the greatest impact during the winter in temperate countries.38 In Russia, temperature seasonality (in the humid continental region) and the diurnal temperature range (in the sub-Arctic region) were found to have the greatest impact on COVID-19 transmission.39

Seasonal trends of COVID-19 have been observed in high-income countries, such as the USA and European countries, where COVID-19-related hospitalisations and deaths peaked from November to April, representing the typical seasonality of respiratory viruses.11 Many factors could account for the differences in seasonality between high-income countries and low-income and middle-income countries, including access to vaccines, the implementation and impact of NPIs, hygiene and sanitation, and resource availability.40

In contrast to COVID-19, in this study, most countries exhibited seasonal patterns of influenza, with influenza waves in the spring or summer in South Africa, Argentina and Brazil and influenza waves in the fall or winter in the remaining countries, except for Malaysia and the Philippines, which did not exhibit influenza seasonality. It is well documented that influenza displays seasonal patterns, although the precise mechanisms underlying its seasonality remain unclear.1 4 41 Compared with influenza, which has fewer variants, the seasonality of COVID-19 could be impacted by the evolution of SARS-CoV-2. The emergence of new variants and subvariants has resulted in waves of COVID-19, thereby impacting the accurate detection of seasonal patterns.11 The continuous emergence and circulation of new variants globally may further complicate the observation of seasonal patterns until the disease becomes endemic, with fewer dominating variants in circulation. Altogether, these findings highlight the importance of maintaining year-round vaccination rather than implementing seasonal campaigns.

The results of this study show that, to date, there have not been clear seasonal patterns of COVID-19 in 10 of the 12 countries examined. Similarly, no clear COVID-19-influenza cocirculation or pandemic patterns were observed. The lack of a clear COVID-19-influenza cocirculation pattern may be unsurprising given the lockdown, social distancing and other NPI measures adopted early in the pandemic. From 2020 to 2021, in certain regions worldwide, the incidence of influenza was very low compared with that in previous years possibly due to NPIs.31 42 Studies have shown that high mask usage was associated with a low influenza incidence and that influenza re-emerged following the relaxation of NPIs.31 43 Future data could help discern whether this lack of a cocirculation pattern will persist in the future.

Certain limitations should be considered when interpreting the results of this study. Importantly, an exploratory descriptive analysis was conducted in this study using visual inspection of data. Statistical analyses, such as a time series analysis, could have provided more robust observations and comparisons across countries. However, the feasibility analysis indicated that the available data were insufficient for a statistical analysis. Nevertheless, this exploratory analysis could support and pave the wave for future studies using statistical analyses if such data become available. COVID-19 has only been circulating for 3 years, so the length of time for which data are available may be insufficient to observe a clear seasonal pattern. It may be that after more time has elapsed, COVID-19 will develop into a seasonal pattern that can be observed with a longer time horizon. With the easing of COVID-19-related restrictions, testing and reporting have also decreased over time, which might lead to an underestimation of cases in later years. The analysis was further limited by the availability and reliability of surveillance data from low-income and middle-income countries. The lack of adequate diagnostic facilities and clinical laboratories able to properly identify pathogens poses challenges to surveillance efforts.44 COVID-19 testing and reporting continue to be insufficient, and there is a lack of adequate monitoring of COVID-19-related hospitalisations and deaths.45 These data may be further impacted by the increasing availability of at-home COVID-19 tests, likely resulting in the under-ascertainment of COVID-19 cases.46 This study is descriptive in nature, so it is expected that the observed patterns in the number of COVID-19 and influenza cases may be subject to substantial heterogeneity across countries. Specifically, stringency/mobility measures and vaccination programmes varied across countries during the COVID-19 pandemic period. Furthermore, country geography (eg, latitude and climate variations even within the same country) may affect the observed patterns over time and render any assessment of seasonal patterns challenging. Finally, given the changes in world dynamics introduced by COVID-19, whether the future pattern will resemble the pre-COVID-19 pattern remains unclear. Thus, no robust conclusions or accurate predictions may be drawn from this investigation.

This investigation was intended to be exploratory to identify and describe the seasonal patterns of COVID-19 and influenza in low- and middle-income countries.

Despite the exploratory nature of this study, several key merits of this study warrant consideration. First, the seasonality of diseases, particularly novel diseases, such as COVID-19, must be systematically described, preferably based on data with high temporal resolution.8 This study is among the first to explore the seasonality and cocirculation of COVID-19 and influenza in low-income and middle-income countries. Although this analysis is preliminary, this study paves the way for future studies. To predict future disease outbreaks, predictive models with accurate parameter inputs need to be constructed. Such models rely on available data. Therefore, the quality and reliability of available data must be improved, which could be achieved with the establishment of surveillance/sentinel networks in select countries in emerging market regions. This approach would require the development of partnerships across academia, governmental bodies and industry. Furthermore, data collection and analysis methodology should be improved to define and classify disease peaks more accurately. Data with high temporal resolution, perhaps at the weekly level, could be useful to better explore the seasonality of disease and the underlying factors.8

Vaccination remains critical for the prevention of severe disease and the spread of SARS-CoV-2.7 ,47 Therefore, the seasonality of COVID-19 may be impacted by vaccination campaigns and vaccine uptake of both primary and booster doses. Decision-makers formulating vaccination strategies could benefit from an understanding of disease seasonality.5 The results of this study indicate a lack of an evident seasonal pattern and support the implementation of year-round vaccination campaigns, especially in low-income and middle-income countries suffering from a higher disease burden.

This post was originally published on https://bmjopen.bmj.com