When do we need multiple infectious disease models? Agreement between projection rank and magnitude in a multi-model setting.

Mathematical models are useful for public health planning and response to infectious disease threats. However, different models can provide differing results, which can hamper decision making if not synthesized appropriately. To address this challenge, multi-model hubs convene independent modeling groups to generate ensembles, known to provide more accurate predictions of future outcomes. Yet, these hubs are resource intensive, and how many models are sufficient in a hub is not known. Here, we compare the benefit of predictions from multiple models in different contexts: (1) decision settings that depend on predictions of quantitative outcomes (e.g., hospital capacity planning), where assessments of the benefits of multi-model ensembles have largely focused; and (2) decisions settings that require the ranking of alternative epidemic scenarios (e.g., comparing outcomes under multiple possible interventions and biological uncertainties). We develop a mathematical framework to mimic a multi-model prediction setting, and use this framework to quantify how frequently predictions from different models agree. We further explore multi-model agreement using real-world, empirical data from 14 rounds of U.S. COVID-19 Scenario Modeling Hub projections. Our results suggest that the value of multiple models could be different in different decision contexts, and if only a few models are available, focusing on the rank of alternative epidemic scenarios could be more robust than focusing on quantitative outcomes. Although additional exploration of the sufficient number of models for different contexts is still needed, our results indicate that it may be possible to identify decision contexts where it is robust to rely on fewer models, a finding that can inform the use of modeling resources during future public health crises.

Redefining pandemic preparedness: Multidisciplinary insights from the CERP modelling workshop in infectious diseases, workshop report.

In July 2023, the Center of Excellence in Respiratory Pathogens organized a two-day workshop on infectious diseases modelling and the lessons learnt from the Covid-19 pandemic. This report summarizes the rich discussions that occurred during the workshop. The workshop participants discussed multisource data integration and highlighted the benefits of combining traditional surveillance with more novel data sources like mobility data, social media, and wastewater monitoring. Significant advancements were noted in the development of predictive models, with examples from various countries showcasing the use of machine learning and artificial intelligence in detecting and monitoring disease trends. The role of open collaboration between various stakeholders in modelling was stressed, advocating for the continuation of such partnerships beyond the pandemic. A major gap identified was the absence of a common international framework for data sharing, which is crucial for global pandemic preparedness. Overall, the workshop underscored the need for robust, adaptable modelling frameworks and the integration of different data sources and collaboration across sectors, as key elements in enhancing future pandemic response and preparedness.

Local-scale phylodynamics reveal differential community impact of SARS-CoV-2 in a metropolitan US county.

SARS-CoV-2 transmission is largely driven by heterogeneous dynamics at a local scale, leaving local health departments to design interventions with limited information. We analyzed SARS-CoV-2 genomes sampled between February 2020 and March 2022 jointly with epidemiological and cell phone mobility data to investigate fine scale spatiotemporal SARS-CoV-2 transmission dynamics in King County, Washington, a diverse, metropolitan US county. We applied an approximate structured coalescent approach to model transmission within and between North King County and South King County alongside the rate of outside introductions into the county. Our phylodynamic analyses reveal that following stay-at-home orders, the epidemic trajectories of North and South King County began to diverge. We find that South King County consistently had more reported and estimated cases, COVID-19 hospitalizations, and longer persistence of local viral transmission when compared to North King County, where viral importations from outside drove a larger proportion of new cases. Using mobility and demographic data, we also find that South King County experienced a more modest and less sustained reduction in mobility following stay-at-home orders than North King County, while also bearing more socioeconomic inequities that might contribute to a disproportionate burden of SARS-CoV-2 transmission. Overall, our findings suggest a role for local-scale phylodynamics in understanding the heterogeneous transmission landscape.

Ensemble2: Scenarios ensembling for communication and performance analysis.

Throughout the COVID-19 pandemic, scenario modeling played a crucial role in shaping the decision-making process of public health policies. Unlike forecasts, scenario projections rely on specific assumptions about the future that consider different plausible states-of-the-world that may or may not be realized and that depend on policy interventions, unpredictable changes in the epidemic outlook, etc. As a consequence, long-term scenario projections require different evaluation criteria than the ones used for traditional short-term epidemic forecasts. Here, we propose a novel ensemble procedure for assessing pandemic scenario projections using the results of the Scenario Modeling Hub (SMH) for COVID-19 in the United States (US). By defining a "scenario ensemble" for each model and the ensemble of models, termed "Ensemble2", we provide a synthesis of potential epidemic outcomes, which we use to assess projections' performance, bypassing the identification of the most plausible scenario. We find that overall the Ensemble2 models are well-calibrated and provide better performance than the scenario ensemble of individual models. The ensemble procedure accounts for the full range of plausible outcomes and highlights the importance of scenario design and effective communication. The scenario ensembling approach can be extended to any scenario design strategy, with potential refinements including weighting scenarios and allowing the ensembling process to evolve over time.

The association between influenza vaccination uptake and influenza and pneumonia-associated deaths in the United States.

BACKGROUND: The influenza mortality burden has remained substantial in the United States (US) despite relatively high levels of influenza vaccine uptake. This has led to questions regarding the effectiveness of the program against this outcome, particularly in the elderly. The aim of this evaluation was to develop and explore a new approach to estimating the population-level effect of influenza vaccination uptake on pneumonia and influenza (P&I) associated deaths. METHODS: Using publicly available data we examined the association between state-level influenza vaccination and all-age P&I associated deaths in the US from the 2013-2014 influenza season to the 2018-2019 season. In the main model, we evaluated influenza vaccine uptake in all those age 6 months and older. We used a mixed-effects regression analysis with generalised least squares estimation to account for within state correlation in P&I mortality. RESULTS: From 2013-2014 through 2018-2019, the total number of all-age P&I related deaths during the influenza seasons was 480,111. The mean overall cumulative influenza vaccine uptake (age 6 months and older) across the states and years considered was 46.7%, with higher uptake (64.8%) observed in those aged ≥ 65 years. We found that overall influenza vaccine uptake (6 months and older) had a statistically significant protective association with the P&I death rate. This translated to a 0.33 (95% CI: 0.20, 0.47) per 100,000 population reduction in P&I deaths in the influenza season per 1% increase in overall influenza vaccine uptake. DISCUSSION: These results using a population-level statistical approach provide additional support for the overall effectiveness of the US influenza vaccination program. This reassurance is critical given the importance of ensuring confidence in this life saving program. Future research is needed to expand on our approach using more refined data.

The US COVID-19 and Influenza Scenario Modeling Hubs: Delivering long-term projections to guide policy.

Between December 2020 and April 2023, the COVID-19 Scenario Modeling Hub (SMH) generated operational multi-month projections of COVID-19 burden in the US to guide pandemic planning and decision-making in the context of high uncertainty. This effort was born out of an attempt to coordinate, synthesize and effectively use the unprecedented amount of predictive modeling that emerged throughout the COVID-19 pandemic. Here we describe the history of this massive collective research effort, the process of convening and maintaining an open modeling hub active over multiple years, and attempt to provide a blueprint for future efforts. We detail the process of generating 17 rounds of scenarios and projections at different stages of the COVID-19 pandemic, and disseminating results to the public health community and lay public. We also highlight how SMH was expanded to generate influenza projections during the 2022-23 season. We identify key impacts of SMH results on public health and draw lessons to improve future collaborative modeling efforts, research on scenario projections, and the interface between models and policy.

Key Challenges for Respiratory Virus Surveillance while Transitioning out of Acute Phase of COVID-19 Pandemic.

To support the ongoing management of viral respiratory diseases while transitioning out of the acute phase of the COVID-19 pandemic, many countries are moving toward an integrated model of surveillance for SARS-CoV-2, influenza virus, and other respiratory pathogens. Although many surveillance approaches catalyzed by the COVID-19 pandemic provide novel epidemiologic insight, continuing them as implemented during the pandemic is unlikely to be feasible for nonemergency surveillance, and many have already been scaled back. Furthermore, given anticipated cocirculation of SARS-CoV-2 and influenza virus, surveillance activities in place before the pandemic require review and adjustment to ensure their ongoing value for public health. In this report, we highlight key challenges for the development of integrated models of surveillance. We discuss the relative strengths and limitations of different surveillance practices and studies as well as their contribution to epidemiologic assessment, forecasting, and public health decision-making.

Predictors of Severity of Influenza-Related Hospitalizations: Results From the Global Influenza Hospital Surveillance Network (GIHSN).

BACKGROUND: The Global Influenza Hospital Surveillance Network (GIHSN) has since 2012 provided patient-level data on severe influenza-like-illnesses from >100 participating clinical sites worldwide based on a core protocol and consistent case definitions. METHODS: We used multivariable logistic regression to assess the risk of intensive care unit admission, mechanical ventilation, and in-hospital death among hospitalized patients with influenza and explored the role of patient-level covariates and country income level. RESULTS: The data set included 73 121 patients hospitalized with respiratory illness in 22 countries, including 15 660 with laboratory-confirmed influenza. After adjusting for patient-level covariates we found a 7-fold increase in the risk of influenza-related intensive care unit admission in lower middle-income countries (LMICs), compared with high-income countries (P = .01). The risk of mechanical ventilation and in-hospital death also increased by 4-fold in LMICs, though these differences were not statistically significant. We also find that influenza mortality increased significantly with older age and number of comorbid conditions. Across all severity outcomes studied and after controlling for patient characteristics, infection with influenza A/H1N1pdm09 was more severe than with A/H3N2. CONCLUSIONS: Our study provides new information on influenza severity in underresourced populations, particularly those in LMICs.

Modeling the impact of COVID-19 nonpharmaceutical interventions on respiratory syncytial virus transmission in South Africa.

Background: The South African government employed various nonpharmaceutical interventions (NPIs) to reduce the spread of SARS-CoV-2. Surveillance data from South Africa indicates reduced circulation of respiratory syncytial virus (RSV) throughout the 2020-2021 seasons. Here, we use a mechanistic transmission model to project the rebound of RSV in the two subsequent seasons. Methods: We fit an age-structured epidemiological model to hospitalization data from national RSV surveillance in South Africa, allowing for time-varying reduction in RSV transmission during periods of COVID-19 circulation. We apply the model to project the rebound of RSV in the 2022 and 2023 seasons. Results: We projected an early and intense outbreak of RSV in April 2022, with an age shift to older infants (6-23 months old) experiencing a larger portion of severe disease burden than typical. In March 2022, government alerts were issued to prepare the hospital system for this potentially intense outbreak. We then assess the 2022 predictions and project the 2023 season. Model predictions for 2023 indicate that RSV activity has not fully returned to normal, with a projected early and moderately intense wave. We estimate that NPIs reduced RSV transmission between 15% and 50% during periods of COVID-19 circulation. Conclusions: A wide range of NPIs impacted the dynamics of the RSV outbreaks throughout 2020-2023 in regard to timing, magnitude, and age structure, with important implications in a low- and middle-income countries (LMICs) setting where RSV interventions remain limited. More efforts should focus on adapting RSV models to LMIC data to project the impact of upcoming medical interventions for this disease.

Correction: Recommended reporting items for epidemic forecasting and prediction research: The EPIFORGE 2020 guidelines.

[This corrects the article DOI: 10.1371/journal.pmed.1003793.].

Scenario Design for Infectious Disease Projections: Integrating Concepts from Decision Analysis and Experimental Design.

Across many fields, scenario modeling has become an important tool for exploring long-term projections and how they might depend on potential interventions and critical uncertainties, with relevance to both decision makers and scientists. In the past decade, and especially during the COVID-19 pandemic, the field of epidemiology has seen substantial growth in the use of scenario projections. Multiple scenarios are often projected at the same time, allowing important comparisons that can guide the choice of intervention, the prioritization of research topics, or public communication. The design of the scenarios is central to their ability to inform important questions. In this paper, we draw on the fields of decision analysis and statistical design of experiments to propose a framework for scenario design in epidemiology, with relevance also to other fields. We identify six different fundamental purposes for scenario designs (decision making, sensitivity analysis, value of information, situational awareness, horizon scanning, and forecasting) and discuss how those purposes guide the structure of scenarios. We discuss other aspects of the content and process of scenario design, broadly for all settings and specifically for multi-model ensemble projections. As an illustrative case study, we examine the first 17 rounds of scenarios from the U.S. COVID-19 Scenario Modeling Hub, then reflect on future advancements that could improve the design of scenarios in epidemiological settings.

Antigenic drift and subtype interference shape A(H3N2) epidemic dynamics in the United States.

Influenza viruses continually evolve new antigenic variants, through mutations in epitopes of their major surface proteins, hemagglutinin (HA) and neuraminidase (NA). Antigenic drift potentiates the reinfection of previously infected individuals, but the contribution of this process to variability in annual epidemics is not well understood. Here we link influenza A(H3N2) virus evolution to regional epidemic dynamics in the United States during 1997-2019. We integrate phenotypic measures of HA antigenic drift and sequence-based measures of HA and NA fitness to infer antigenic and genetic distances between viruses circulating in successive seasons. We estimate the magnitude, severity, timing, transmission rate, age-specific patterns, and subtype dominance of each regional outbreak and find that genetic distance based on broad sets of epitope sites is the strongest evolutionary predictor of A(H3N2) virus epidemiology. Increased HA and NA epitope distance between seasons correlates with larger, more intense epidemics, higher transmission, greater A(H3N2) subtype dominance, and a greater proportion of cases in adults relative to children, consistent with increased population susceptibility. Based on random forest models, A(H1N1) incidence impacts A(H3N2) epidemics to a greater extent than viral evolution, suggesting that subtype interference is a major driver of influenza A virus infection dynamics, presumably via heterosubtypic cross-immunity.

Estimating the time-varying reproduction number for COVID-19 in South Africa during the first four waves using multiple measures of incidence for public and private sectors across four waves.

OBJECTIVES: The aim of this study was to quantify transmission trends in South Africa during the first four waves of the COVID-19 pandemic using estimates of the time-varying reproduction number (R) and to compare the robustness of R estimates based on three different data sources, and using data from public and private sector service providers. METHODS: R was estimated from March 2020 through April 2022, nationally and by province, based on time series of rt-PCR-confirmed cases, hospitalisations, and hospital-associated deaths, using a method that models daily incidence as a weighted sum of past incidence, as implemented in the R package EpiEstim. R was also estimated separately using public and private sector data. RESULTS: Nationally, the maximum case-based R following the introduction of lockdown measures was 1.55 (CI: 1.43-1.66), 1.56 (CI: 1.47-1.64), 1.46 (CI: 1.38-1.53) and 3.33 (CI: 2.84-3.97) during the first (Wuhan-Hu), second (Beta), third (Delta), and fourth (Omicron) waves, respectively. Estimates based on the three data sources (cases, hospitalisations, deaths) were generally similar during the first three waves, but higher during the fourth wave for case-based estimates. Public and private sector R estimates were generally similar except during the initial lockdowns and in case-based estimates during the fourth wave. CONCLUSION: Agreement between R estimates using different data sources during the first three waves suggests that data from any of these sources could be used in the early stages of a future pandemic. The high R estimates for Omicron relative to earlier waves are interesting given a high level of exposure pre-Omicron. The agreement between public and private sector R estimates highlights that clients of the public and private sectors did not experience two separate epidemics, except perhaps to a limited extent during the strictest lockdowns in the first wave.

Behavioral factors and SARS-CoV-2 transmission heterogeneity within a household cohort in Costa Rica.

INTRODUCTION: Variability in household secondary attack rates and transmission risks factors of SARS-CoV-2 remain poorly understood. METHODS: We conducted a household transmission study of SARS-CoV-2 in Costa Rica, with SARS-CoV-2 index cases selected from a larger prospective cohort study and their household contacts were enrolled. A total of 719 household contacts of 304 household index cases were enrolled from November 21, 2020, through July 31, 2021. Blood specimens were collected from contacts within 30-60 days of index case diagnosis; and serum was tested for presence of spike and nucleocapsid SARS-CoV-2 IgG antibodies. Evidence of SARS-CoV-2 prior infections among household contacts was defined based on the presence of both spike and nucleocapsid antibodies. We fitted a chain binomial model to the serologic data, to account for exogenous community infection risk and potential multi-generational transmissions within the household. RESULTS: Overall seroprevalence was 53% (95% confidence interval (CI) 48-58%) among household contacts. The estimated household secondary attack rate is 34% (95% CI 5-75%). Mask wearing by the index case is associated with the household transmission risk reduction by 67% (adjusted odds ratio = 0.33 with 95% CI: 0.09-0.75) and not sharing bedroom with the index case is associated with the risk reduction of household transmission by 78% (adjusted odds ratio = 0.22 with 95% CI 0.10-0.41). The estimated distribution of household secondary attack rates is highly heterogeneous across index cases, with 30% of index cases being the source for 80% of secondary cases. CONCLUSIONS: Modeling analysis suggests that behavioral factors are important drivers of the observed SARS-CoV-2 transmission heterogeneity within the household.

When living in the same house with known SARS-CoV-2 cases, household members may change their behavior and adopt preventive measures to reduce the spread of SARS-CoV-2. To understand how behavioral factors affect SARS-CoV-2 spreading in household settings, we focused on household members of individuals with laboratory-confirmed SARS-CoV-2 infections and followed the way SARS-CoV-2 spread within the household, by looking at who had antibodies against the virus, which means they were infected. We also asked participants detailed questions about their behavior and applied mathematical modeling to evaluate its impact on SARS-CoV-2 transmission. We found that mask-wearing by the SARS-CoV-2 cases, and avoiding sharing a bedroom with the infected individuals, reduces SARS-CoV-2 transmission. However, caring for SARS-CoV-2 cases, and prolonged interaction with infected individuals facilitate SARS-CoV-2 spreading. Our study helps inform what behaviors can help reduce SARS-CoV-2 transmission within a household.

Inferring the differences in incubation-period and generation-interval distributions of the Delta and Omicron variants of SARS-CoV-2.

Estimating the differences in the incubation-period, serial-interval, and generation-interval distributions of SARS-CoV-2 variants is critical to understanding their transmission. However, the impact of epidemic dynamics is often neglected in estimating the timing of infection-for example, when an epidemic is growing exponentially, a cohort of infected individuals who developed symptoms at the same time are more likely to have been infected recently. Here, we reanalyze incubation-period and serial-interval data describing transmissions of the Delta and Omicron variants from the Netherlands at the end of December 2021. Previous analysis of the same dataset reported shorter mean observed incubation period (3.2 d vs. 4.4 d) and serial interval (3.5 d vs. 4.1 d) for the Omicron variant, but the number of infections caused by the Delta variant decreased during this period as the number of Omicron infections increased. When we account for growth-rate differences of two variants during the study period, we estimate similar mean incubation periods (3.8 to 4.5 d) for both variants but a shorter mean generation interval for the Omicron variant (3.0 d; 95% CI: 2.7 to 3.2 d) than for the Delta variant (3.8 d; 95% CI: 3.7 to 4.0 d). The differences in estimated generation intervals may be driven by the "network effect"-higher effective transmissibility of the Omicron variant can cause faster susceptible depletion among contact networks, which in turn prevents late transmission (therefore shortening realized generation intervals). Using up-to-date generation-interval distributions is critical to accurately estimating the reproduction advantage of the Omicron variant.

Multiple models for outbreak decision support in the face of uncertainty.

Policymakers must make management decisions despite incomplete knowledge and conflicting model projections. Little guidance exists for the rapid, representative, and unbiased collection of policy-relevant scientific input from independent modeling teams. Integrating approaches from decision analysis, expert judgment, and model aggregation, we convened multiple modeling teams to evaluate COVID-19 reopening strategies for a mid-sized United States county early in the pandemic. Projections from seventeen distinct models were inconsistent in magnitude but highly consistent in ranking interventions. The 6-mo-ahead aggregate projections were well in line with observed outbreaks in mid-sized US counties. The aggregate results showed that up to half the population could be infected with full workplace reopening, while workplace restrictions reduced median cumulative infections by 82%. Rankings of interventions were consistent across public health objectives, but there was a strong trade-off between public health outcomes and duration of workplace closures, and no win-win intermediate reopening strategies were identified. Between-model variation was high; the aggregate results thus provide valuable risk quantification for decision making. This approach can be applied to the evaluation of management interventions in any setting where models are used to inform decision making. This case study demonstrated the utility of our approach and was one of several multimodel efforts that laid the groundwork for the COVID-19 Scenario Modeling Hub, which has provided multiple rounds of real-time scenario projections for situational awareness and decision making to the Centers for Disease Control and Prevention since December 2020.

Host heterogeneity and epistasis explain punctuated evolution of SARS-CoV-2.

Identifying drivers of viral diversity is key to understanding the evolutionary as well as epidemiological dynamics of the COVID-19 pandemic. Using rich viral genomic data sets, we show that periods of steadily rising diversity have been punctuated by sudden, enormous increases followed by similarly abrupt collapses of diversity. We introduce a mechanistic model of saltational evolution with epistasis and demonstrate that these features parsimoniously account for the observed temporal dynamics of inter-genomic diversity. Our results provide support for recent proposals that saltational evolution may be a signature feature of SARS-CoV-2, allowing the pathogen to more readily evolve highly transmissible variants. These findings lend theoretical support to a heightened awareness of biological contexts where increased diversification may occur. They also underline the power of pathogen genomics and other surveillance streams in clarifying the phylodynamics of emerging and endemic infections. In public health terms, our results further underline the importance of equitable distribution of up-to-date vaccines.

Direct and indirect mortality impacts of the COVID-19 pandemic in the United States, March 1, 2020 to January 1, 2022.

Excess mortality studies provide crucial information regarding the health burden of pandemics and other large-scale events. Here, we use time series approaches to separate the direct contribution of SARS-CoV-2 infection on mortality from the indirect consequences of the pandemic in the United States. We estimate excess deaths occurring above a seasonal baseline from March 1, 2020 to January 1, 2022, stratified by week, state, age, and underlying mortality condition (including COVID-19 and respiratory diseases; Alzheimer's disease; cancer; cerebrovascular diseases; diabetes; heart diseases; and external causes, which include suicides, opioid overdoses, and accidents). Over the study period, we estimate an excess of 1,065,200 (95% Confidence Interval (CI) 909,800-1,218,000) all-cause deaths, of which 80% are reflected in official COVID-19 statistics. State-specific excess death estimates are highly correlated with SARS-CoV-2 serology, lending support to our approach. Mortality from 7 of the 8 studied conditions rose during the pandemic, with the exception of cancer. To separate the direct mortality consequences of SARS-CoV-2 infection from the indirect effects of the pandemic, we fit generalized additive models (GAM) to age- state- and cause-specific weekly excess mortality, using covariates representing direct (COVID-19 intensity) and indirect pandemic effects (hospital intensive care unit (ICU) occupancy and measures of interventions stringency). We find that 84% (95% CI 65-94%) of all-cause excess mortality can be statistically attributed to the direct impact of SARS-CoV-2 infection. We also estimate a large direct contribution of SARS-CoV-2 infection (≥67%) on mortality from diabetes, Alzheimer's, heart diseases, and in all-cause mortality among individuals over 65 years. In contrast, indirect effects predominate in mortality from external causes and all-cause mortality among individuals under 44 years, with periods of stricter interventions associated with greater rises in mortality. Overall, on a national scale, the largest consequences of the COVID-19 pandemic are attributable to the direct impact of SARS-CoV-2 infections; yet, the secondary impacts dominate among younger age groups and in mortality from external causes. Further research on the drivers of indirect mortality is warranted as more detailed mortality data from this pandemic becomes available.

Rapidly shifting immunologic landscape and severity of SARS-CoV-2 in the Omicron era in South Africa.

South Africa was among the first countries to detect the SARS-CoV-2 Omicron variant. However, the size of its Omicron BA.1 and BA.2 subvariants (BA.1/2) wave remains poorly understood. We analyzed sequential serum samples collected through a prospective cohort study before, during, and after the Omicron BA.1/2 wave to infer infection rates and monitor changes in the immune histories of participants over time. We found that the Omicron BA.1/2 wave infected more than half of the cohort population, with reinfections and vaccine breakthroughs accounting for > 60% of all infections in both rural and urban sites. After the Omicron BA.1/2 wave, we found few (< 6%) remained naïve to SARS-CoV-2 and the population immunologic landscape is fragmented with diverse infection/immunization histories. Prior infection with the ancestral strain, Beta, and Delta variants provided 13%, 34%, and 51% protection against Omicron BA.1/2 infection, respectively. Hybrid immunity and repeated prior infections reduced the risks of Omicron BA.1/2 infection by 60% and 85% respectively. Our study sheds light on a rapidly shifting landscape of population immunity in the Omicron era and provides context for anticipating the long-term circulation of SARS-CoV-2 in populations no longer naïve to the virus.

Global and national influenza-associated hospitalisation rates: Estimates for 40 countries and administrative regions.

Background: WHO estimates that seasonal influenza epidemics result in three to five million cases of severe illness (hospitalisations) every year. We aimed to improve the understanding of influenza-associated hospitalisation estimates at a national and global level. Methods: We performed a systematic literature review of English- and Chinese-language studies published between 1995 and 2020 estimating influenza-associated hospitalisation. We included a total of 127 studies (seven in Chinese) in the meta-analysis and analyzed their data using a logit-logistic regression model to understand the influence of five study factors and produce national and global estimates by age groups. The five study factors assessed were: 1) the method used to calculate the influenza-associated hospitalisation estimates (rate- or time series regression-based), 2) the outcome measure (divided into three envelopes: narrow, medium, or wide), 3) whether every case was laboratory-confirmed or not, 4) whether the estimates were national or sub-national, 5) whether the rates were based on a single year or multiple years. Results: The overall pooled influenza-associated hospitalisation rate was 40.5 (95% confidence interval (CI) = 24.3-67.4) per 100 000 persons, with rates varying substantially by age: 224.0 (95% CI = 118.8-420.0) in children aged 0-4 years and 96.8 (95% CI = 57.0-164.3) in the elderly aged >65 years. The overall pooled hospitalisation rates varied by calculation method; for all ages, the rates were significantly higher when they were based on rate-based methods or calculated on a single season and significantly lower when cases were laboratory-confirmed. The national hospitalisation rates (all ages) varied considerably, ranging from 11.7 (95% CI = 3.8-36.3) per 100 000 in New Zealand to 122.1 (95% CI = 41.5-358.4) per 100 000 in India (all age estimates). Conclusions: Using the pooled global influenza-associated hospitalisation rate, we estimate that seasonal influenza epidemics result in 3.2 million cases of severe illness (hospitalisations) per annum. More extensive analyses are required to assess the influence of other factors on the estimates (e.g. vaccination and dominant virus (sub)types) and efforts to harmonize the methods should be encouraged. Our study highlights the high rates of influenza-associated hospitalisations in children aged 0-4 years and the elderly aged 65+ years.