Influenza pandemics rank among the most devastating infectious disease events in human history, with the ‘Spanish flu’ of 1918–1919 alone estimated to have claimed 50 million lives worldwide.1 The subsequent pandemics of the 20th century, although less severe, still exacted a heavy toll, causing an estimated 1–4 million deaths each.2 In the 21st century, the 2009 H1N1 ‘swine flu’ pandemic demonstrated the speed with which a novel influenza virus can spread in an interconnected world, even if its virulence turned out to be comparatively mild.3

The pandemic threat posed by influenza A viruses stems from their high mutability and their segmented RNA genomes, which facilitate the rapid evolution of novel strains through mutation and reassortment.4 Aquatic birds are the natural reservoir for influenza A viruses, harboring a vast diversity of subtypes from which pandemic strains can emerge (Fig. 1a).5 Zoonotic transmission of avian influenza viruses from these wild birds to domesticated birds and other livestock is a key first step in this process, providing opportunities for the virus to spread and adapt to human hosts and acquire the ability to spread efficiently from person to person. This is particularly troubling given that H5N1 has become widespread in US commercial and backyard poultry, with continued detections reported by the US Department of Agriculture (USDA; Fig. 1b).6 Recent spillovers into US mammals, including dairy cattle, underscore the expanding host range and the opportunities for selection of mammal-adaptive changes, reinforcing the need for integrated animal–human surveillance.

Among avian influenza subtypes, A(H5N1) has been a top concern for pandemic preparedness since it first caused human infections in Hong Kong in 1997.7 The virus has since become enzootic in poultry across multiple countries in Asia and Africa, leading to sporadic but often severe human infections. From January 2003 to December 12, 2024, the WHO reported 954 confirmed human cases of H5N1, resulting in 464 deaths.8 This corresponds to a case fatality rate (CFR) of ∼49%, far exceeding that of seasonal influenza viruses (<0.1%)9 and even surpassing the 1918 pandemic virus (2–3%).1 However, this aggregate estimate spans multiple goose/Guangdong-derived lineages and is sensitive to case ascertainment. Reported severity has varied over time, and the currently dominant clade, 2.3.4.4b, has been associated with comparatively fewer reported fatal human cases compared with earlier waves. Accordingly, we treat historical CFR as an upper-bound indicator of potential severity rather than a stable property of contemporary US viruses. Most recently, in January 2025, a death attributed to H5N1 was confirmed in Louisiana, marking the first H5N1-related human death in the USA.10

However, the true extent of human H5N1 infections might be greater than these figures suggest. Seroprevalence studies indicate that subclinical or mild H5N1 infections occur more frequently than previously recognized, particularly among populations with high exposure to poultry. A meta-analysis of H5 seroprevalence surveys in China estimated a seroprevalence of up to 3.2% in poultry workers.11 Similarly, a recent Centers for Disease Control and Prevention (CDC) study found detectable H5N1 antibodies in 7% of US dairy workers exposed to infected cattle, with half of the seropositive individuals reporting no symptoms.12 These findings suggest that the actual H5N1 infection fatality rate, although still high, is lower than the observed CFR.

Despite this broader spectrum of disease, the risk of H5N1 sparking a pandemic remains a serious concern. The segmented genome of H5N1 allows it to reassort with human influenza viruses, potentially acquiring gene segments that enhance transmissibility or alter antigenicity.13 Furthermore, the high mutation rate of the virus generates a diverse viral population that can rapidly adapt to new host environments.14 Several experimental studies have identified H5N1 strains with mutations that enable transmission via respiratory droplets or aerosols between ferrets, the gold-standard animal model for influenza transmissibility.[15], [16]

To assess the pandemic risk posed by zoonotic influenza viruses, such as H5N1, the CDC developed the Influenza Risk Assessment Tool (IRAT),17 while the WHO created the Tool for Influenza Pandemic Risk Assessment (TIPRA).18 However, these tools have limitations: they heavily rely on epidemiological and clinical data, which might be scarce during the early stages of an outbreak; they do not provide clear quantitative risk estimates; and their assessments are not routinely updated as new evidence emerges.

In this context, the ViTAP framework was developed as a novel, complementary approach for assessing the pandemic potential of zoonotic viruses.19 ViTAP quantifies the likelihood of a virus causing a pandemic based on its intrinsic characteristics, such as genomic structure, receptor binding, transmission mode, immune evasion, and virulence. Unlike other tools, ViTAP deliberately excludes external factors, such as pre-existing immunity in human populations, the presence of therapeutics or vaccines, and geographical or environmental conditions. Therefore, we view ViTAP as complementary to IRAT/TIPRA: trait-based scores can be combined with geographical spread, exposure intensity, and population immunity to inform a fuller situational risk assessment. By focusing on traits inherent to the virus itself, ViTAP enables risk assessments even when epidemiological or clinical data are limited. Critically, ViTAP is designed to be a dynamic framework, with scores updated as new virological and epidemiological evidence becomes available.

Here, we present the first comprehensive assessment of the pandemic potential of H5N1 avian influenza using the ViTAP framework. We conducted an extensive literature review to gather data on the key traits of the virus and assign initial ViTAP scores. These scores were then refined through structured elicitation of input from leading experts in influenza virology, epidemiology, and public health. We assessed the sensitivity of the final scores to different assumptions about the relative importance of each trait. Finally, we compared the ViTAP score of H5N1 to those of previous pandemic influenza viruses to provide context for the current level of risk. As a next step, applying the same influenza-adapted ViTAP rubric to other contemporary zoonotic influenza A viruses (e.g., H9N2 in poultry and enzootic swine IAV lineages) will help benchmark discrimination within influenza A.

Our findings aim to inform global efforts to prepare for, and mitigate the risk of, an H5N1 pandemic. By identifying the key viral traits that drive the pandemic potential of H5N1, we pinpoint crucial knowledge gaps and surveillance priorities. Importantly, we demonstrate how a rigorous, yet accessible tool, such as ViTAP, can provide actionable insights for pandemic preparedness, even for a virus as complex and long-studied as H5N1. Given that the threat of pandemic influenza persists, integrating such structured assessments into decision-making will be crucial for safeguarding human lives and livelihoods worldwide.