Global Viral Lower Respiratory Disease Episodes, Hospitalisations, and Clinical Outcomes by Aetiology, 2010–2021

Abstract

Introduction

Viral lower respiratory tract disease (LRTD) is a major cause of global morbidity and mortality, but pathogen-specific estimates remain limited. We aimed to quantify the global burden of viral LRTD episodes, hospitalisations, and severe clinical outcomes by aetiology from 2010 to 2021.

Methods

We used the Global Burden of Disease (GBD) 2021 modelling framework to generate estimates of viral LRTD incidence, hospitalisations, and clinical outcomes from 2010 to 2021 for 204 countries and territories, 21 regions, and seven super-regions. Aetiology-specific estimates were generated for five viral categories: influenza, respiratory syncytial virus (RSV), human metapneumovirus (hMPV), SARS-CoV-2, and “other” viral pathogens. Data inputs included surveillance systems, clinical informatics, published literature, surveys, and vital registration. Viral LRTD incidence was estimated using DisMod-MR 2.1, a GBD Bayesian meta-regression tool for disease incidence. We then applied location-specific admission scalars, adjusted for healthcare access, to derive hospitalisations. Aetiology-specific clinical outcome proportions—including intensive care unit (ICU) need, invasive mechanical ventilation (IMV) need, and in-hospital mortality—were estimated using meta-regression—Bayesian, regularised, trimming models. These were then applied to the hospitalisation estimates. COVID-19 incidence and hospitalisations were estimated using established GBD COVID-19 methods. All estimates are reported with 95% uncertainty intervals (UIs).

Results

Globally, viral LRTD episodes increased from 82.3 million (95% UI 76.4–88.2) in 2010 to 94.9 million (88.8–101.2) in 2019, while episode rates remained relatively stable at approximately 1200 per 100,000 population. Over the same period, hospitalisation rates increased by 40%, from 103.7 (96.1–111.4) to 145.4 (134.6–156.4) per 100,000 population. In 2020 and 2021, the emergence of SARS-CoV-2 resulted in sharp increases in burden, with episodes exceeding 1.1 billion in 2020 and 1.5 billion in 2021. During this period, 2020–2021, episode and hospitalisation rates associated with influenza, RSV, and hMPV declined by nearly 50%. Between 2010 and 2019, global rates of ICU need increased from 16.1 (14.3–18.0) to 23.7 (21.0–26.6) per 100,000 population, IMV need from 7.5 (6.7–8.4) to 11.2 (10.0–12.6), and in-hospital mortality from 9.4 (8.4–10.5) to 13.9 (12.3–15.6). In 2020–2021, SARS-CoV-2 dominated severe outcomes, with in-hospital mortality reaching 87.7 (83.3–92.2) per 100,000 population in 2021. Across the study period, older adults consistently experienced the highest rates of severe outcomes, and substantial regional variation was observed.

Conclusions

Before 2020, global incidence of viral LRTD was stable; however, hospitalisations and severe outcomes increased substantially, suggesting rising clinical severity and healthcare demand. The COVID-19 pandemic profoundly altered the global viral LRTD landscape, driving unprecedented increases in hospitalisations, intensive care use, and mortality, while suppressing other respiratory viruses in the years 2020 and 2021. These findings demonstrate a substantial burden of severe viral LRTD globally, with marked age and regional variations, underscoring the importance of sustained aetiology-specific surveillance, adequate healthcare capacity, and equitable access to preventive and therapeutic interventions.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40121-026-01406-4.

Key Summary Points

Why Carry out this study?

Viral lower respiratory tract diseases (LRTDs) are a major cause of global morbidity and mortality, yet comprehensive, pathogen‑specific estimates of incidence, hospitalisation, and clinical severity over time have been limited.

Using the Global Burden of Disease (GBD) 2021 modelling framework, this study provided the first internally consistent, globally comprehensive assessment of viral LRTD episodes, hospitalisations, and severe clinical outcomes of hospitalised cases by viral aetiology across 204 countries and territories from 2010 to 2021.

What was learned from the study?

Between 2010 and 2019, viral LRTD episode rates remained relatively stable, while hospitalisations and severe outcomes, including intensive care unit need, invasive mechanical ventilation need, and in‑hospital mortality, increased substantially. These results indicate rising clinical severity and health‑care demand before the COVID‑19 pandemic.

During 2020–2021, SARS‑CoV‑2 drove unprecedented increases in viral LRTD burden and severe outcomes, while episodes and hospitalisations associated with influenza, respiratory syncytial virus, and human metapneumovirus declined by approximately 50%. These findings establish a baseline for understanding post-pandemic re-emergence pattern and shifts in population susceptibility.

Viral LRTDs impose a substantial and evolving global burden with pronounced age and regional disparities, highlighting the need for sustained aetiology‑specific surveillance, strengthened health‑care capacity, and equitable access to preventive and therapeutic interventions.

Key limitations of this analysis include heterogenous input data due to case definition and reporting, reliance on hospitalisation scalars derived from three high-income settings, and constraints on diagnostic specificity which may be disproportionately present in low-resource environments; estimates should therefore be interpreted within this modelling framework context.

Introduction

Viral lower respiratory tract disease (LRTD) represents a major global health burden, affecting millions of individuals annually and contributing substantially to morbidity and mortality [1]. LRTDs target the respiratory system below the larynx, encompassing the trachea, bronchi, bronchioles, and alveolar structures where critical gas exchange occurs [2]. The clinical outcomes of LRTDs range from mild respiratory symptoms to life-threatening complications, with patients typically presenting with persistent cough, dyspnoea, sputum production, wheezing, and tachypnoea, often accompanied by systemic symptoms such as fever and fatigue [2]. Disease severity exists along a spectrum from outpatient-manageable cases to severe presentations characterised by hypoxemia requiring hospitalisation with supplemental oxygen, and in the most critical cases, intensive care support including mechanical ventilation.

Viral LRTD disproportionately affects vulnerable populations, including individuals with underlying chronic cardiovascular or pulmonary diseases, immunocompromised patients, and those with limited access to preventive measures such as vaccines, face masks, clean cooking fuels, and adequate sanitation [3]. Multiple viral pathogens contribute to this burden, including seasonal influenza viruses, respiratory syncytial virus (RSV), human metapneumovirus (hMPV), and since late 2019, SARS-CoV-2, which fundamentally altered the global viral LRTD landscape through the COVID-19 pandemic [1].

Previous epidemiological studies have documented the substantial global impact of LRTD through comprehensive surveillance efforts. The Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) demonstrated that all-cause LRTD resulted in 2.5 million deaths and 97.2 million disability-adjusted life-years globally in 2019, establishing it as the fourth leading cause of both mortality and morbidity worldwide [4]. However, significant knowledge gaps persist in our understanding of viral LRTD burden, particularly regarding the specific contributions of individual viral pathogens to overall disease burden. The COVID-19 pandemic has fundamentally disrupted historical disease patterns and healthcare systems, yet comprehensive assessments of pandemic-era viral LRTD burden remain limited [1]. Current estimates inadequately capture pathogen-specific contributions to LRTD, limiting our ability to prioritise targeted interventions, allocate resources effectively, and prepare for future respiratory pathogen emergencies. The indirect effects of the pandemic during 2020–2021 (including disrupted vaccination programmes, altered healthcare-seeking behaviours, deferred care, and overwhelmed health systems) have likely influenced LRTD epidemiology in ways that are not yet fully characterised [5].

To address these critical evidence gaps, we conducted a comprehensive analysis to estimate the global burden of viral LRTD from 2010 to 2021, spanning the pre-pandemic and pandemic periods. We estimated the incidence of viral LRTD episodes and hospitalisations attributable to five key aetiologies: influenza, RSV, hMPV, SARS-CoV-2, and a residual “other viral” category encompassing remaining viral pathogens. We further characterised the clinical severity of hospitalised cases by estimating intensive care need (ICU), invasive mechanical ventilation (IMV) need, and in-hospital mortality. All estimates were generated with 95% uncertainty intervals (UIs) and stratified by age and sex for global totals, 204 countries and territories, 21 GBD regions, and seven GBD super-regions. This study aims to address evidence gaps by providing the first internally consistent, globally comprehensive assessment tracking five distinct viral aetiologies and three severe clinical endpoints—ICU need, IMV need, and in-hospital deaths—simultaneously over a 12-year period using a unified framework.

Methods

Overview

This study utilised the GBD 2021 modelling framework to estimate global viral LRTD incidence, hospitalisations, and clinical outcomes [6]. We defined LRTD as a clinician-diagnosed episode of pneumonia or bronchiolitis, consistent with the GBD definition for the “lower respiratory infection” cause category. The case definitions and additional details can be found in Supplementary Materials Sect. 2.1.1 and Table S2. Aetiology-specific estimates for influenza, RSV, and SARS-CoV-2 were informed by existing GBD 2021 aetiological case definitions (Supplementary Materials Sect. 2.1.2 and Table S3). This analysis complies with the GATHER statement, as detailed in the Supplementary Materials Table S1.

Data Sources

Input data included surveillance data, clinical informatics data, scientific literature, surveys, and vital registration, consistent with GBD 2021 LRTD and COVID-19 estimation methods (Supplementary Materials Table S5). Additional hMPV and clinical severity data were identified through systematic and targeted literature searches (Supplementary Materials Sect. 2.3).

Analytical Strategy

Viral LRTD Incidence (Excluding COVID-19)

Total viral LRTD incidence was estimated using a Bayesian meta-regression epidemiological disease modelling tool (DisMod-MR 2.1) integrating data on prevalence, incidence, remission, and mortality [6]. DisMod-MR 2.1 was selected because direct epidemiological data are frequently unavailable, sparse, or inconsistent across the 204 tracked countries. It is a specialised Bayesian tool explicitly developed to address these real-world data gaps by synthesizing heterogeneous data types while enforcing strict internal mathematical consistency between incidence, prevalence, remission, and mortality; and it is well suited to generate globally comparable estimates. We ran separate models by sex, age, and time, incorporating the Socio-demographic Index (SDI) and Healthcare Access and Quality (HAQ) Index as covariates [7, 8]. We assumed 7.79-day average symptom duration (range 6.2–9.64 days) based on GBD 2021 systematic review findings, applied consistently across viral aetiologies for prevalence–incidence conversions [1].

For aetiology-specific estimates, we used proportion-only models for hMPV, influenza, RSV, and “other” viruses within DisMod-MR 2.1 [6]. The aetiological proportions were rescaled to ensure they summed to 100%, and these normalised proportions were then multiplied by the viral LRTD incidence estimates to generate aetiology‑specific outputs. Pandemic-related reductions in non-SARS-CoV-2 viral LRTD were quantified by applying influenza-specific disruption scalars—developed in prior work through a multistep process that interpolated reported influenza cases, adjusted for under-reporting, generated pandemic-free counterfactual estimates, and produced annual disruption scalars—to each aetiology, under the assumption that non-SARS-CoV-2 respiratory viruses experienced broadly similar transmission reductions during the pandemic [1].

Hospitalisation Estimation (Excluding COVID-19)

Aetiology-specific hospitalisation incidence used location- and year-specific scalars representing hospitalisation proportions, adjusted for healthcare access using the location-specific HAQ Index as a predictive covariate. Scalars were modelled using age- and sex-specific slope-only linear models, with HAQ Index estimates predicting ratios of inpatient admissions to total LRTD encounters. Ratios were derived from hospitalisation data from the USA (Merative MarketScan), Poland (Poland National Health Fund Patient Claims), and Taiwan (Taiwan National Health Insurance Claims), with counts deduplicated for readmissions and including primary and non-primary diagnoses of viral LRTD. This approach dynamically scales hospitalisation proportions according to regional healthcare capacity such that locations with limited healthcare access and weaker infrastructure are modelled with down-scaled hospitalization ratios.

COVID-19 Estimation

We used a separate methodology for modelling COVID-19 incidence, developed as part of the GBD 2021 framework specifically for SARS-CoV-2. Our COVID-19 estimation process is described comprehensively in prior publications [6, 7]. In brief, daily infections were estimated using the meta-regression—Bayesian, regularised, trimming tool (MR-BRT) [9]. Estimates were adjusted for infection-to-confirmation and confirmation-to-hospitalisation delays by aligning reported cases and hospitalisations with the inferred timing of infections. We then calculated infection–detection, infection–hospitalisation, and infection–fatality ratios using the GBD composite infection framework; when deriving infection–fatality ratios-based infections, we applied the standard 24-day infection-to-death interval used in the GBD methodology. Total infections were multiplied by asymptomatic infection proportions to derive symptomatic cases. As described previously, key assumptions in our COVID-19 modelling included 80% seropositivity among individuals with at least two vaccine doses; 30–70% cross-variant immunity between escape variants (Beta, Gamma, Delta) and others; and 10–13 days from infection to diagnosis.

Clinical Outcomes

Clinical outcomes assessed among hospitalised viral LRTD cases were ICU need, IMV need, and in-hospital deaths (Supplementary Materials Table S4). For each of these outcomes, we estimated the proportion of cases meeting the clinical endpoint using MR-BRT. The models included age and covariates of HAQ Index and SDI as predictors. We implemented quadratic fixed-effects models with three spline terms, with interior knots placed according to data density or clinical expertise. Separate models were run for each aetiology–outcome combination (12 total for hMPV, influenza, RSV, and “other”). For SARS-CoV-2, ICU need and in-hospital death proportions were obtained from the IHME COVID-19 Forecasting team. These in-hospital death proportions were derived by comparing estimated infections to observed hospitalisations and deaths, reflecting the fraction of infections that progress to critical illness or result in death among hospitalised cases. IMV need was estimated using a separate MR-BRT model.

We then applied the estimated outcome proportions to aetiology-specific hospitalisation estimates to generate annual counts and rates by age, sex, location, and year. Because comparable country-level data on intensive care and IMV use are limited, these proportions were derived from global empirical data and applied across all locations using age-specific ratios. This approach ensures consistent estimation where direct utilisation data are sparse. The resulting estimates reflect the clinical need for intensive care and IMV, rather than observed utilisation. Definitions of each clinical outcome are provided in Supplementary Materials Table S4. All rates are expressed per 100,000 population.

95% UIs

Uncertainty was quantified using 95% intervals from 1000 draws per posterior parameter distribution, with the 25th and 975th values as bounds. Intervals reflect data availability, sample sizes, and cross-source consistency.

Ethics Declaration

This analysis used fully deidentified data and did not involve direct interaction with human participants; it was determined to be exempt from formal ethical review and informed consent requirements in accordance with applicable institutional and regulatory guidelines. The authors confirm they had full institutional permission to access and analyse all underlying databases utilized in this study.

Results

Viral LRTD Episodes and Hospitalisations

Globally, total estimated viral LRTD episodes increased by 15% from 82.3 million (95% UI 76.4–88.2) in 2010 to 94.9 million (88.8–101.2) in 2019, while rate of episodes increased by 4% from 1185.0 (1100.0–1269.7) per 100,000 population in 2010, to 1226.9 (1147.3–1307.3) by 2019. In the same period, however, hospitalisation rates from LRTD increased by 40%, from 103.7 (96.1–111.4) to 145.4 (134.6–156.4) per 100,000 population (Table 1). Furthermore, the proportion of LRTD episodes resulting in hospitalisation increased gradually over time, with the highest proportions observed in high-income locations and the lowest in sub-Saharan Africa (Supplementary Materials Figure S2).

Table 1.

Global count, rate per 100,000 population, and annual rate of change in viral LRTD episodes and hospitalisations for all ages, 2010–2021

	Episodes	Hospitalisations	Episodes	Hospitalisations	Episodes	Hospitalisations
Year	Count (95% UI)		Rate per 100,000 (95% UI)		Annual rate of change (%)a
2010	82,297,991 (76,395,316–88,182,149)	7,205,191 (6,671,558–7,739,109)	1185.0 (1100.0–1269.7)	103.7 (96.1–111.4)	–	–
2011	83,559,949 (77,590,143–89,532,048)	7,549,267 (6,992,812–8,107,208)	1188.3 (1103.4–1273.2)	107.4 (99.4–115.3)	0.3%	3.6%
2012	84,688,194 (78,681,212–90,652,466)	7,911,171 (7,328,778–8,495,925)	1189.4 (1105.0–1273.1)	111.1 (102.9–119.3)	0.1%	3.4%
2013	85,593,800 (79,590,727–91,467,880)	8,280,288 (7,671,275–8,886,853)	1187.1 (1103.9–1268.6)	114.8 (106.4–123.3)	− 0.2%	3.3%
2014	86,431,275 (80,418,332–92,207,669)	8,682,328 (8,045,918–9,312,650)	1184.0 (1101.6–1263.1)	118.9 (110.2–127.6)	-0.3%	3.6%
2015	87,844,180 (81,777,804–93,642,493)	9,100,956 (8,437,427–9,757,740)	1188.6 (1106.5–1267.1)	123.1 (114.2–132.0)	0.4%	3.5%
2016	89,613,298 (83,519,984–95,541,471)	9,593,028 (8,893,104–10,286,683)	1198.0 (1116.5–1277.2)	128.2 (118.9–137.5)	0.8%	4.1%
2017	91,606,020 (85,496,470–97,612,275)	10,166,668 (9,426,038–10,902,979)	1210.2 (1129.5–1289.6)	134.3 (124.5–144.0)	1.0%	4.8%
2018	93,615,906 (87,448,133–99,726,100)	10,749,543 (9,960,542–11,536,123)	1222.8 (1142.2–1302.6)	140.4 (130.1–150.7)	1.0%	4.5%
2019	94,947,901 (88,788,116–101,172,600)	11,255,496 (10,414,730–12,104,922)	1226.9 (1147.3–1307.3)	145.4 (134.6–156.4)	0.3%	3.6%
2020	1,109,365,942 (1,053,735,669–1,163,499,052)	21,422,620 (19,775,747–23,068,509)	14,193.2 (13,481.5–14,885.8)	274.1 (253.0–295.1)	1056.8%	88.5%
2021	1,527,769,967 (1,448,995,897–1,604,696,144)	33,045,789 (30,787,694–35,394,131)	19,375.4 (18,376.3–20,350.9)	419.1 (390.5–448.9)	36.5%	52.9%

^aAnnual rate of change is calculated based on the rate per 100,000 population for episodes and hospitalisations, respectively; for example, the annual rate of change from 2010 to 2011 for episodes is 1188.3/1185.0–1 = 0.3%

Across the 21 global regions modelled in the GBD, between 2010 and 2019, South Asia had the highest average annual LRTD episode rate (2152.6 [95% UI 1993.2–2308.4] per 100,000 population), while Australasia had the lowest (341.7 [318.9–364.0] per 100,000 population). For hospitalisations in the same period, high-income North America had the highest average annual rate (217.3 [202.8–232.3] per 100,000 population) and central sub-Saharan Africa the lowest (36.2 [33.1–39.4] per 100,000 population). Additionally, between 2020 and 2021, Eastern Europe experienced disproportionately high rates of hospitalisations (1426 [1250–1602] per 100,000 population), while Australasia showed smaller surges (25 [20–31] per 100,000 population). These patterns underscore the regional heterogeneity in viral LRTD burden that persisted across the study period. All regional rates and counts for viral LRTD episodes and hospitalisations can be found in Supplementary Materials Table S7.

In 2020, with the emergence of SARS-CoV-2, episodes of viral LRTD rose sharply to 1.10 billion (95% UI 1.05–1.16) and reached 1.5 billion (1.44–1.60) in 2021. From 2019 to 2020, the average episode rates rose by 1057% and hospitalisation rates by 89%.

Aetiology of Viral LRTD Episodes

From 2010 to 2019, annual global episode rates associated with influenza, RSV, hMPV, and other viral aetiologies were largely stable, with modest increases observed for some pathogens (Supplementary Materials Table S6). Influenza showed the most consistent upward trend over this period, whereas RSV and hMPV remained relatively stable, with annual percentage changes ranging from – 2.0% to 3.3% depending on aetiology and year. Across the pre-pandemic period, distinct age-specific patterns were observed by aetiology (Fig. 1A). Influenza incidence was highest among adults aged 60 years and older. RSV episode rates were highest among children younger than 5 years, with a marked decline beyond the neonatal period. hMPV exhibited a bimodal age distribution, with higher incidence at the extremes of age and higher episode rates than RSV among adults aged 60 years and older. Before 2020, males generally had higher episode rates than females across all aetiologies (Supplementary Materials Figure S1).

Fig. 1.

Incidence of aetiology-specific viral LRTD episodes (A) and hospitalisations (B) rate per 100,000 population, by age and year (2010, 2019, and 2021). Panel A displays the global incidence rates of lower respiratory tract disease (LRTD) episodes stratified by five distinct viral categories: influenza, RSV, hMPV, SARS-CoV-2, and a residual other viruses category. Panel B displays the global rates of inpatient hospital admissions for each respective viral aetiology. Results are stratified by 26 standard age groups, ranging from early neonatal to adults aged 95 plus. Early neonatal is defined as ages 0–6 days, and late neonatal is defined as ages 7–27 days. Black vertical lines positioned on each bar denote the 95% uncertainty intervals (UIs)

Following the emergence of SARS-CoV-2 in 2020, substantial declines were observed in non-COVID viral LRTD episode rates (Supplementary Materials Table S6). Influenza rates decreased from 469.7 (95% UI 442.1–500.1) to 227.8 (190.1–266.8), while RSV episode rates declined from 161.0 (147.6–175.8) per 100,000 population in 2019 to 79.0 (65.6–92.0) per 100,000 population in 2020. These reductions persisted into 2021. During this period, SARS-CoV-2 accounted for the majority of viral LRTD episodes across all ages and both sexes (Fig. 1A). Pandemic-related changes varied by age group and aetiology (Fig. 1A). Declines in RSV episode rates between 2019 and 2021 were most pronounced among children younger than 5 years (~ 64%) and adults aged 70 years and older (> 65%) (Fig. 1A). Influenza showed similar age-specific reductions, particularly among adults aged 60 years and older (> 60%), alongside a relative broadening of incidence into younger age groups in 2021 compared with 2019 (Fig. 1A). In 2021, males continued to have higher episode rates than females for all aetiologies except SARS-CoV-2, for which females had higher rates across all age groups younger than 85 years (Supplementary Materials Figure S1).

Aetiology of Viral LRTD Hospitalisations

From 2010 to 2019, global hospitalisation rates for viral LRTDs increased steadily across all aetiologies, with annual increases ranging from 3.3% to 4.8% (Table 2). The magnitude of these increases varied by pathogen, with influenza-associated LRTD hospitalisation rates increasing by 4.5% to 6.0% per year, RSV by 1.5% to 3.7% per year, hMPV by 2.0% to 3.8% per year, and other viral aetiologies by 2.6% to 4.0% per year (Table 2). Before 2020, age-specific patterns of viral LRTD hospitalisation rates were broadly consistent with those observed for episode incidence (Fig. 1B). Hospitalisation rates were highest among children younger than 5 years and adults aged 65 years and older across most aetiologies and were consistently higher in males than in females (Supplementary Materials Figure S1).

Table 2.

Global viral LRTD hospitalisation rate per 100,000 population, count, and annual rate of change by aetiology for all ages, 2010–2021

Year	Influenza	RSV	hMPV	Other viral	SARS-CoV-2
Count (95% UI)
2010	2,948,974 (2,713,823–3,194,476)	746,072 (691,904–799,859)	936,490 (871,755–1,001,094)	2,573,655 (2,395,749–2,751,199)	–
2011	3,121,872 (2,872,394–3,380,451)	778,559 (722,524–836,757)	968,567 (900,951–1,036,297)	2,680,270 (2,493,159–2,867,698)	–
2012	3,307,143 (3,041,145–3,583,736)	810,581 (751,922–870,939)	1,002,610 (931,431–1,071,877)	2,790,837 (2,592,704–2,983,647)	–
2013	3,502,454 (3,219,000–3,798,696)	839,040 (777,815–901,435)	1,038,025 (962,543–1,109,566)	2,900,768 (2,689,833–3,100,691)	–
2014	3,715,798 (3,413,735–4,033,670)	870,241 (807,259–935,946)	1,080,232 (1,000,257–1,156,141)	3,016,057 (2,792,765–3,228,000)	–
2015	3,942,663 (3,626,417–4,281,129)	906,403 (842,001–974,886)	1,118,372 (1,034,348–1,199,999)	3,133,518 (2,898,095–3,362,226)	–
2016	4,215,784 (3,877,317–4,578,748)	941,850 (873,655–1,010,857)	1,166,399 (1,077,644–1,252,828)	3,268,995 (3,020,248–3,511,225)	–
2017	4,524,543 (4,159,422–4,917,597)	980,866 (908,855–1,054,543)	1,223,711 (1,129,583–1,315,439)	3,437,549 (3,173,134–3,695,226)	-
2018	4,841,584 (4,446,460–5,268,498)	1,017,629 (941,477–1,092,811)	1,277,198 (1,178,499–1,374,507)	3,613,132 (3,333,916–3,888,415)	-
2019	5,139,236 (4,705,434–5,608,830)	1,042,958 (964,137–1,122,396)	1,321,272 (1,217,834–1,421,056)	3,752,030 (3,458,297–4,035,387)	-
2020	2,623,255 (2,172,665–3,040,507)	529,566 (437,063–609,601)	586,626 (463,252–725,950)	1,892,642 (1,494,600–2,342,146)	15,790,531 (15,129,261–16,418,809)
2021	1,125,648 (748,385–1,562,053)	315,304 (226,659–420,524)	327,826 (217,955–454,922)	1,125,648 (748,385–1,562,053)	29,666,652 (28,516,854–30,741,798)
Rate per 100,000 (95% UI)
2010	42.5 (39.1–46.0)	10.7 (10.0–11.5)	13.5 (12.6–14.4)	37.1 (34.5–39.6)	–
2011	44.4 (40.8–48.1)	11.1 (10.3–11.9)	13.8 (12.8–14.7)	38.1 (35.5–40.8)	–
2012	46.4 (42.7–50.3)	11.4 (10.6–12.2)	14.1 (13.1–15.1)	39.2 (36.4–41.9)	–
2013	48.6 (44.6–52.7)	11.6 (10.8–12.5)	14.4 (13.3–15.4)	40.2 (37.3–43.0)	–
2014	50.9 (46.8–55.3)	11.9 (11.1–12.8)	14.8 (13.7–15.8)	41.3 (38.3–44.2)	–
2015	53.3 (49.1–57.9)	12.3 (11.4–13.2)	15.1 (14.0–16.2)	42.4 (39.2–45.5)	–
2016	56.4 (51.8–61.2)	12.6 (11.7–13.5)	15.6 (14.4–16.7)	43.7 (40.4–46.9)	–
2017	59.8 (54.9–65.0)	13.0 (12.0–13.9)	16.2 (14.9–17.4)	45.4 (41.9–48.8)	–
2018	63.2 (58.1–68.8)	13.3 (12.3–14.3)	16.7 (15.4–18.0)	47.2 (43.5–50.8)	–
2019	66.4 (60.8–72.5)	13.5 (12.5–14.5)	17.1 (15.7–18.4)	48.5 (44.7–52.1)	–
2020	33.6 (27.8–38.9)	6.8 (5.6–7.8)	7.5 (5.9–9.3)	24.2 (19.1–30.0)	202.0 (193.6–210.1)
2021	20.4 (13.9–27.7)	4.0 (2.9–5.3)	4.2 (2.8–5.8)	14.3 (9.5–19.8)	376.2 (361.7–389.9)
Annual rate of change, %a
2010	–	–	–	–	–
2011	4.5	3.7	2.2	2.7	–
2012	4.5	2.7	2.2	2.9	–
2013	4.7	1.8	2.1	2.6	–
2014	4.7	2.6	2.8	2.7	–
2015	4.7	3.4	2.0	2.7	–
2016	5.8	2.4	3.3	3.1	–
2017	6.0	3.2	3.8	3.9	–
2018	5.7	2.3	3.1	4.0	–
2019	5.1	1.5	2.4	2.8	–
2020	− 49.4	− 49.6	− 56.1	− 50.1	–
2021	− 39.3	− 41.2	− 44.0	− 40.9	86.2

^aAnnual rate of change is calculated based on the rate per 100,000 population; for example, the annual rate of change from 2010 to 2011 for influenza hospitalisations is 44.4/42.5 − 1 = 4.5%

Following the emergence of SARS-CoV-2 in 2020, hospitalisation rates associated with non-COVID viral LRTDs declined markedly (Table 2). hMPV-related LRTD hospitalisation rates decreased from 17.1 (95% UI 15.7–18.4) per 100,000 population in 2019 to 7.5 (5.9–9.3) in 2020, RSV-related rates from 13.5 (12.5–14.5) to 6.8 (5.6–7.8), and influenza-related rates from 66.4 (60.8–72.5) to 33.6 (27.8–38.9). These reductions continued into 2021 (Table 2). Pandemic-related changes in hospitalisation rates varied by age and aetiology (Fig. 1B). The largest reductions in RSV-, influenza-, hMPV-, and other virus-associated LRTD hospitalisations were observed among children younger than 5 years and adults aged 65 years and older. In contrast, SARS-CoV-2-associated hospitalisation rates peaked among older adults, differing from the age distribution of SARS-CoV-2 episodes, which peaked among individuals aged 15–24 years (Fig. 1A). In 2021, hospitalisation rates across all aetiologies were higher in males than in females (Supplementary Materials Figure S1).

Global Clinical Outcomes of Viral LRTD

Between 2010 and 2019, the global burden of viral LRTD increased steadily across all three clinical severity outcomes—ICU need, IMV need, and in-hospital death. Over this period, ICU need rose by 47% from 16.1 (95% UI 14.3–18.0) to 23.7 (21.0–26.6) per 100,000 population. IMV need increased by 49% from 7.5 (6.7–8.4) to 11.2 (10.0–12.6) per 100,000 population, and in-hospital mortality increased by 48% from 9.4 (8.4–10.5) to 13.9 (12.3–15.6) per 100,000 population. In absolute terms, this equated to an estimated increase of 1.83 million (1.63–2.06) cases with ICU need, 870,000 (732,000–925,000) with IMV need, and 1.07 million (0.952–1.21) in-hospital deaths in 2019 (Table 3).

Table 3.

Global viral LRTD clinical outcome rates per 100,000 population, counts, and annual rate of change for all ages, 2010–2021

	Intensive care unit need			Invasive mechanical ventilation need			In-hospital deaths
Year	Count	Rate	Annual rate of change, %a	Count	Rate	Annual rate of change, %	Count	Rate	Annual rate of change, %
2010	1,120,338 (994,315–1,252,019)	16.1 (14.3–18)	–	520,255 (462,108–581,399)	7.49 (6.65–8.37)	–	654,452 (581,022–731,244)	9.42 (8.37–10.5)	–
2011	1,179,099 (1,046,684–1,317,981)	16.8 (14.9–18.7)	4.3	548,777 (487,516–613,390)	7.8 (6.93–8.72)	4.0	688,939 (611,749–769,949)	9.8 (8.7–10.9)	4.1
2012	1,241,634 (1,102,997–1,388,265)	17.4 (15.5–19.5)	3.6	579,191 (514,899–647,574)	8.13 (7.23–9.09)	4.1	725,638 (644,782–811,203)	10.2 (9.06–11.4)	4.2
2013	1,306,541 (1,161,550–1,461,759)	18.1 (16.1–20.3)	4.0	610,886 (543,544–683,425)	8.47 (7.54–9.48)	3.9	763,768 (679,190–854,390)	10.6 (9.42–11.9)	4.2
2014	1,376,379 (1,223,583–1,540,837)	18.9 (16.8–21.1)	4.4	645,019 (573,950–722,011)	8.84 (7.86–9.89)	3.8	804,843 (715,676–900,903)	11 (9.8–12.3)	4.4
2015	1,449,484 (1,287,987–1,624,407)	19.6 (17.4–22)	3.7	680,908 (605,634–762,934)	9.21 (8.19–10.3)	4.5	847,800 (753,536–950,001)	11.5 (10.2–12.9)	4.2
2016	1,536,119 (1,365,397–1,723,349)	20.5 (18.3–23)	4.6	723,495 (643,677–811,577)	9.67 (8.61–10.8)	4.3	898,828 (799,115–1,008,293)	12 (10.7–13.5)	5.0
2017	1,637,360 (1,454,998–1,838,349)	21.6 (19.2–24.3)	5.4	772,965 (687,455–867,886)	10.2 (9.08–11.5)	5.8	958,323 (851,764–1,075,902)	12.7 (11.3–14.2)	5.5
2018	1,740,995 (1,545,368–1,955,208)	22.7 (20.2–25.5)	5.1	823,757 (731,766–925,324)	10.8 (9.56–12.1)	4.7	1,019,104 (904,760–1,144,481)	13.3 (11.8–14.9)	5.9
2019	1,833,451 (1,625,512–2,061,331)	23.7 (21–26.6)	4.4	869,657 (771,517–977,935)	11.2 (9.97–12.6)	4.5	1,073,514 (951,916–1,206,932)	13.9 (12.3–15.6)	3.7
2020	8,785,810 (8,293,683–9,286,498)	112 (106–119)	372.6	4,197,442 (3,964,125–4,433,370)	53.7 (50.7–56.7)	286.3	4,196,387 (3,945,787–4,451,542)	53.7 (50.5–57)	379.5
2021	14,735,778 (14,034,478–15,432,449)	187 (178–196)	67.0	7,041,909 (6,706,883–7,374,212)	89.3 (85.1–93.5)	63.3	6,919,227 (6,571,564–7,267,889)	87.7 (83.3–92.2)	66.3

^aAnnual rate of change is calculated based on the hospitalisation outcome rate per 100,000 population; for example, the annual rate of change from 2010 to 2011 for in-hospital deaths is 9.8/9.42–1 = 4.1%

The onset of the COVID-19 pandemic in 2020 marked a sharp escalation in clinical severity across the general population. Viral LRTD hospitalisations—now dominated by SARS-CoV-2—drove ICU need to 112.0 (95% UI 106.0–119.0) per 100,000 population in 2020 and 187.0 (178.0–196.0) in 2021. IMV need rose to 53.7 (50.7–56.7) and 89.3 (85.1–93.5) per 100,000 population, and in-hospital mortality spiked to 53.7 (50.5–57.0) and 87.7 (83.3–92.2) per 100,000 population, respectively. These represent five- to eight-fold increases over pre-pandemic levels, and during the pandemic years, rates were 15–20 times greater than historical peaks.

Global Clinical Outcomes of Viral LRTD by Aetiology

Globally, for all ages, influenza-related ICU need rose by 60% from 8.6 (95% UI 7.6–9.6) to 13.8 (12.3–15.5) per 100,000 population from 2010 to 2019, with IMV need increasing by 62% from 4.7 (4.2–5.3) to 7.6 (6.8–8.6) per 100,000 population. SARS-CoV-2 ICU need admission rates were 101 (96.7–105.0) and 180.0 (173.0–186.0) per 100,000 population in 2020 and 2021, respectively. SARS-CoV-2 IMV needs reached 48.1 (46.2–49.9) and 86.0 (82.9–89.0), and in-hospital death rates were 46.8 (45.0–48.6) and 83.7 (80.7–86.6) per 100,000 population in 2020 and 2021, respectively—far exceeding the 0.92–4.18 per 100,000 population range for other pathogens (Fig. 2). For counts of the global clinical outcomes by aetiology, please see Supplementary Materials Table S8.

Fig. 2.

Global viral LRTD clinical outcome rates per 100,000 population, all ages, by aetiology and year (2010–2021). Trends over time are broken out into individual sub-panels for influenza, RSV, hMPV, SARS-CoV-2, and a residual other viruses category. Annual rates are plotted for three distinct severe clinical endpoints: intensive care unit need, invasive mechanical ventilation need, and In-hospital deaths from 2010 through 2021. Note that the y-axis scales vary across the individual pathogen sub-panels to maximize visual legibility across highly disparate absolute disease burdens

Prior to 2020, the highest rates for ICU need were observed among the oldest age groups. Adults aged 85 years and older experienced the greatest burden: influenza-related cases reaching as high as 734.8 (95% UI 643.7–830.7) per 100,000 population, and in-hospital mortality exceeding 440 per 100,000 population. Overall, adults aged 65 and older accounted for disproportionately high ICU need and mortality, with influenza and RSV hospitalisations generating intensive care utilisation rates 10–20 times greater than those in middle-aged adults. Neonates also showed higher rates; among early neonates, RSV-associated IMV needs reached 17.4 (15.1–19.6) per 100,000 population. hMPV, RSV, and “other” viral aetiologies consistently showed a U-shaped distribution across ages, with highest severity among the youngest and oldest populations. All estimates referenced here can be found in Supplementary Materials Figure S3.

Discussion

This analysis offers a comprehensive assessment of the global burden of viral LRTDs from 2010 to 2021, underscoring both the relative stability and the dramatic shifts that have occurred over the past decade. Between 2010 and 2019, viral LRTD episode rates remained relatively consistent, primarily driven by common pathogens like influenza and RSV. However, hospitalisation rates for viral LRTD increased by 40% during this period, suggesting increasing severity of illness, possibly due to factors such as changes in healthcare access, improvements in diagnostic capabilities, shifts in viral virulence, or increasing prevalence of comorbid conditions, which may negatively impact clinical outcomes. The emergence of SARS-CoV-2 in 2020 brought about a profound transformation in the epidemiology of viral LRTD. In 2020 and 2021, episode and hospitalisation rates surged by orders of magnitude due to the pandemic, with COVID-19 causing a substantial increase in both clinical severity and healthcare demand.

Importantly, this study provides the first internally consistent, globally comprehensive assessment tracking five distinct viral aetiologies and three severe clinical endpoints (i.e., ICU need, IMV need, in-hospital deaths) simultaneously across 204 countries over a 12-year window. Before this analysis, pathogen-specific global estimates tracking shifting clinical severity over time were severely limited. This unified framework enables the quantification of these essential estimates for structural health system planning and resource allocation, which can be iteratively updated as additional data become available.

Geographically, the burden of viral LRTD was not evenly distributed. In the years leading up to the pandemic, regions like South Asia had the highest rates of viral LRTD episodes, while North America had the highest hospitalisation rates, likely influenced by healthcare infrastructure and reporting standards [10, 11]. The pandemic, however, caused significant shifts in these patterns. Eastern Europe including Russia, for instance, saw some of the greatest increases in both episode and hospitalisation rates during 2020–2021, outcomes likely due to weaker healthcare systems and delayed vaccination campaigns that were notably exacerbated by widespread misinformation and disinformation regarding COVID-19 vaccines across the region [12]. In contrast, regions like East Asia and Australasia showed more contained surges, attributable to earlier and more stringent public-health responses [13, 14]. The comparatively muted rise in documented hospitalisations across many low- and middle-income countries in other regions likely reflects barriers to care access and diagnostic capacity constraints rather than lower disease burden. These findings highlight the crucial role of timely, evidence-informed public-health responses and healthcare system preparedness in mitigating the impact of viral LRTD outbreaks. Estimates of viral LRTD hospitalisations and clinical outcomes in our analysis were generally lower (often two to three times lower) than those reported in other studies, likely reflecting differences in study design, including geographical coverage and case definitions (e.g., GBD LRTD versus broader respiratory infectious disease categories) [15–17].

The demographic patterns observed in this study further reinforce the critical need to protect vulnerable populations. Older adults (≥ 65 years) were the most consistently affected group across all clinical outcomes, experiencing significantly higher rates of hospitalisations, ICU need, and in-hospital mortality. This demographic remains highly vulnerable to severe outcomes from viral respiratory infections, primarily due to age-related immune senescence and the high prevalence of comorbidities in this group [18, 19]. As health systems worldwide face increasing pressure from ageing populations and rising emergency-care demand, strengthening prevention and treatment strategies for respiratory infections in older adults will be essential to mitigate future strain. The high incidence of IMV need among neonates further brings attention to the heavy clinical burden on the youngest age groups, which warrants further research and public health interventions. These findings also point to the importance of age-specific strategies, including vaccination and early detection, as part of a comprehensive approach to reducing morbidity and mortality from viral LRTD. In this context, vaccine hesitancy—particularly among older adults—may limit coverage and reduce the potential population-wide impact of vaccination, further contributing to the observed burden [20].

While the ageing global population likely contributed to the overall increase in severe clinical outcomes over the past decade, the COVID-19 pandemic accelerated these trends dramatically. By 2021, SARS-CoV-2 was responsible for more than double the ICU need rates and more than triple the mortality compared to pre-pandemic levels. The observed rise in severity is likely multifactorial, with factors such as the emergence of more virulent variants of the virus and shifts in clinical practice (e.g., lower thresholds for ICU need) all playing significant roles [21, 22]. In addition, the limited availability of effective therapeutic options for hospitalised and severely ill patients—particularly during the early phases of the pandemic—likely contributed to poorer clinical outcomes, pointing to persistent gaps in treatment strategies for severe viral LRTDs. The variable availability of vaccines across regions also compounded these disparities, further exacerbating the global burden [23]. Collectively, these findings reinforce the necessity of equitable vaccine distribution alongside continued innovation in therapeutics and preparedness for future viral LRTD outbreaks, including both vaccine development and strategies for rapid deployment in diverse global settings.

An important consequence of the COVID-19 pandemic was the redistribution of the global viral LRTD burden. During the peak years of the pandemic, viral LRTD episodes and hospitalisations attributed to other pathogens, including influenza, RSV, and hMPV, decreased by nearly 50%. This redistribution likely reflects the success of public health measures designed to control the spread of SARS-CoV-2, such as travel restrictions, social distancing, and increased hygiene practices, which simultaneously reduced the transmission of other respiratory viruses [24, 25]. The reduction in the burden of other viral pathogens emphasises the dynamic nature of viral epidemiology and suggests that interventions for one pathogen can have widespread effects on the transmission of others. As these patterns continue to evolve, evaluating post-pandemic trends will be critical to understanding the re-emergence of pathogens, potential changes in virulence, and shifting patterns of population susceptibility. This demonstrates the need for a broader, more integrated approach to viral surveillance, one that considers the interactions between multiple respiratory pathogens and the full range of public health measures. These findings also establish a critical benchmark, essential for understanding post-pandemic re-emergence patterns, shifts in population susceptibility, and evolving aetiology-specific dynamics.

Our estimates which demonstrate marked regional variations in viral LRTD burden may reflect the complex interplay of epidemiological, structural, and economic factors. While the HAQ Index and SDI were significant predictive covariates in our models quantifying population-level unmet need for preventive and therapeutic interventions, we cannot identify whether this unmet need stems from a lack of therapeutic interventions, access barriers, or staffing constraints. In high-income countries, existing healthcare infrastructure may have facilitated the management of surges in cases, including the provision of intensive care and mechanical ventilation. By contrast, in low- and middle-income countries, more limited healthcare resources and infrastructure, including delays in vaccine distribution, could have contributed to heightened vulnerability during the pandemic [26, 27]. These patterns may also reflect differences in access to diagnostic tools, vaccines, and therapeutic interventions [23]. Together, these observations stress the importance of strengthening healthcare systems globally, with particular attention to intensive care capacity, robust viral surveillance, and equitable access to healthcare resources [23, 27]. They also underline the need to prioritise global health equity to ensure that all populations, particularly those in resource-constrained settings, can access preventive and therapeutic measures to mitigate the burden of viral LRTDs.

The rapid changes in viral LRTD epidemiology brought about by the COVID-19 pandemic also emphasise the importance of timely and accurate data collection. Surveillance systems must be able to quickly capture and track emerging trends in viral infections, allowing for a swift public health response. Global health systems must be better equipped to handle the complexity of tracking multiple pathogens simultaneously, as the pandemic has shown how interwoven global viral dynamics can be [28]. This requires a holistic approach to surveillance, one that includes data on aetiology-specific viral infections, clinical outcomes, and healthcare infrastructure capacity, among other factors. Real-time surveillance data would enable more precise forecasting and resource allocation, ensuring that public health responses are appropriately tailored to meet the demands of an outbreak. Clear and timely communication of public health threats is also critical to ensure that surveillance findings translate into effective action, with targeted public health messaging and sustained community education helping to mitigate the impact of misinformation during outbreaks. In addition to strengthening surveillance, there is an urgent need to enhance in-hospital management strategies and intensive care capacity, as rising trends in ICU and IMV needs may exceed the limits of existing healthcare systems. At the same time, improved prevention and management of chronic conditions associated with severe LRTD outcomes may reduce the size of vulnerable populations during future outbreaks, complementing investments in acute care capacity Addressing these challenges will require investment in workforce training and medical education, expansion in intensive care resources, and the development of context-appropriate clinical guidelines to ensure that healthcare systems are resilient and able to adapt to future surges of severe LRTD cases.

Limitations

This analysis has several limitations. First, bronchitis cases without diagnoses of bronchiolitis or pneumonia are not captured in our estimates of LRTD estimates, as bronchitis alone does not meet the GBD LRTD case definition, potentially leading to underestimation of viral LRTD burden. Some studies suggest that bronchitis represents a relatively modest proportion of lower respiratory infection admissions and the magnitudes differ across age groups and healthcare settings, with estimates ranging from ~ 6% to 15% in observed cohorts, and thus is unlikely to substantially alter overall burden patterns [29]. We were also unable to assess the timing of infection in relation to hospital admission or severity, as these details were not consistently available in the underlying data. Additionally, hospitalisation scalars, although adjusted by HAQ Index for location-specific access, were derived from three high-income settings (e.g., USA, Poland, and Taiwan) and may not fully reflect care-seeking behaviours in low-resource settings.

Data limitations were particularly pronounced in low-income settings, where the burden of viral LRTD may be highest. DisMod-MR 2.1 estimates are contingent on the quality and representativeness of input data; in these locations, modelled estimates relied heavily on covariates such as the HAQ Index and SDI, as well as data from similar geographies, resulting in less precise estimates with wider UIs. While necessary, this modelling approach may lead to under- or overestimation of burden depending on covariate relationships and comparability of diagnostic practices. Differences in test performance and population testing strategies may also introduce bias in the attribution of specific viral aetiologies. Specifically, granular diagnostic practices such as multiplex PCR testing and bronchoscopic sampling are unevenly distributed across clinical settings and geographies, leading to potential underestimation of viral aetiologies. This diagnostic underestimation is likely disproportionately present in low- and middle-income countries, as well as resource-constrained settings with high-income countries where access to advanced molecular testing platforms is limited. The limited availability of post-pandemic data further adds uncertainty, constraining the precision of current burden estimates and future modelling.

Although our methodology aims to harmonise data across sources, some heterogeneity remains. Variability in case definitions (e.g., “medically attended” versus “clinician-diagnosed”) and national reporting requirements may contribute to aetiology-specific or regional biases. We were also unable to analyse more granular viral subtypes, such as influenza A versus B, due to data constraints.

Finally, the use of annual data limited our ability to assess seasonal or short-term trends. Co-infections were counted in multiple aetiology categories, which may have influenced pathogen-specific estimates. In addition, limited testing for less common viruses may have contributed to an underestimation of the burden of LRTD due to “other” viruses. Estimates of ICU and IMV needs were extrapolated from data-rich settings using age-specific ratios, assuming that these ratios are constant across settings and that observed utilisation accurately reflects underlying clinical need. Clinical outcomes by aetiology were stratified by age, but not by sex or year, due to data sparsity.

Strengths

This study has several key strengths. First, it represents the first internally consistent, globally comprehensive assessment of viral LRTD episodes, hospitalisations, and clinical outcomes, including COVID-19, across 233 locations and regions, 26 age groups modelled in the GBD, and multiple demographic subgroups. By leveraging well-established GBD methods, this analysis provides internally consistent and comparable estimates of viral LRTD burden, even in regions with limited data availability.

Another strength of this study is the inclusion of ICU needs, modelled based on healthcare utilisation patterns from data-rich regions. This allows for a global estimation of unmet needs for intensive care services, including mechanical ventilation, regardless of local healthcare infrastructure. The study’s comprehensive approach, encompassing a range of viral aetiologies, both prior to and during the COVID-19 pandemic, offers valuable insights into the impact of the pandemic on global health systems and the need for continued investment in healthcare capacity and viral surveillance. Furthermore, our approach establishes a modelling framework that can be iteratively updated as additional data become available in the future, ensuring access to up-to-date estimates of the burden of viral LRTD to inform global health policy and clinical practice.

Conclusions

This analysis provides comprehensive global estimates of incident viral LRTD episodes, hospitalisations, and key clinical outcomes—including ICU need, IMV need, and in-hospital mortality—by viral aetiology, age, and sex from 2010 to 2021. Between 2010 and 2019, global viral LRTD episode rates remained relatively stable at approximately 1200 episodes per 100,000 population, increasing from 82.3 million episodes in 2010 to 94.9 million in 2019, while hospitalisation rates increased by about 40%, indicating rising clinical severity and healthcare demand before the COVID-19 pandemic. During 2020 and 2021, SARS-CoV-2 drove unprecedented increases in viral LRTD burden, with episodes exceeding 1.1 billion and 1.5 billion, and hospitalisations increasing accordingly, while non-COVID viral aetiologies declined during this period. Severe outcomes, including ICU need, mechanical ventilation, and in-hospital mortality, also rose substantially, with older adults experiencing consistently higher rates and notable regional variation in burden. Taken together, these findings highlight the substantial disease burden of several viral LRTD globally, reinforcing the need for sustained aetiology-specific surveillance, strengthened clinical capacity, and equitable access to preventive and therapeutic interventions to improve resilience against future respiratory threats.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Catherine W. Gillespie, Quinn Rafferty, Jiali Lei, Samuel Ewald, Darwin Del Castillo, Maja Pasovic, Ekaterina Maslova, Chengbin Wang, Yao Qiao, Charlotte Lupton, Malin Fageras, and Pratik Sinha were involved in formulating the study questions, study design, and interpretation of the results. Quinn Rafferty and Jiali Lei had full access to the data and were responsible for the data analyses. Quinn Rafferty, Jiali Lei, Jam Suba, and Catherine W. Gillespie drafted the manuscript. Ekaterina Maslova, Chengbin Wang, Yao Qiao, Charlotte Lupton, Malin Fageras, and Pratik Sinha provided critical manuscript feedback. All authors critically reviewed, revised, and approved the final manuscript for submission.

Funding

This study was supported by AstraZeneca PLC. AstraZeneca PLC also provided funding for the journal’s Rapid Service Fee. Quinn Rafferty, Jiali Lei, Jam Suba, Catherine W. Gillespie, and Maja Pasovic are employees of IHME, which received funding from AstraZeneca in connection with the development of this manuscript, and Samuel Ewald and Darwin Del Castillo contributed to this work while employed at IHME. Ekaterina Maslova, Chengbin Wang, Yao Qiao, Charlotte Lupton, and Malin Fagerås are employees of AstraZeneca. Pratik Sinha receives grant funding from the National Institute of Health and the Department of Defense. Pratik Sinha also receives consulting fees from AstraZeneca and Prenosis that are unrelated to the current study.

Data Availability

Global Burden of Disease data are provided by the Institute for Health Metrics and Evaluation (IHME) and used with permission. All rights reserved. For terms and conditions of use, please visit https://www.healthdata.org/data-tools-practices/data-practices/terms-and-conditions. For any usage that falls outside of these license restrictions, please contact IHME Client Services at services@healthdata.org. The other specific input source datasets and granular modelled outputs analysed during the current study may not be publicly available due to third-party proprietary restrictions, data licensing agreements, and institutional data-sharing policies.

Declarations

Conflict of Interest

Pratik Sinha receives grant funding from the National Institute of Health and the Department of Defense and consulting fees from AstraZeneca and Prenosis that are unrelated to the current study. Ekaterina Maslova, Chengbin Wang, Charlotte Lupton, Yao Qiao, and Malin Fageras are employees of AstraZeneca and may hold stock and/or stock options in the company. All other authors; Quinn Rafferty, Jiali Lei, Jam Suba, Maja Pasovic, Samuel Ewald, Darwin Del Castillo and Catherine W. Gillespie declare no conflicts of interest.

Ethical Approval

This analysis used fully deidentified data and did not involve direct interaction with human participants; it was determined to be exempt from formal ethical review and informed consent requirements in accordance with applicable institutional and regulatory guidelines. The authors confirm they had full institutional permission to access and analyse all underlying databases utilized in this study.