It used to be contained and sporadic, but this pandemic is happening at a scale never seen before
In February 2024, dairy farmers in Texas began noticing their cows were producing less milk and that the milk was discoloured, yellow and thicker than it should be. First reported by news outlets as a ‘mystery illness’, what was happening to these cows was unheard of – and worrying.
In February 2024, dairy farmers in Texas began noticing their cows were producing less milk and that the milk was discoloured, yellow and thicker than it should be. News outlets reported on a mystery illness in cows. Then, in March, H5N1 avian flu was diagnosed as the cause of the illness. It was unheard of for cows to get bird flu, so the diagnosis was not suspected at first. The epidemic spread rapidly across the US, affecting 14 states by October, including the biggest dairy producer in the country, the state of California.
In August, as I was about to depart for the US for a pandemic war game I had designed for military stakeholders at the United States Indo-Pacific Command, I received an invitation from the Los Alamos National Laboratory (of Oppenheimer fame) to speak at a special symposium they were holding in Washington DC. The meeting had been convened around the spread of bird flu in the US and the risk this posed for a human pandemic.
Bird flu used to be contained and sporadic, but there is a pandemic occurring in birds and animals at a scale we have never seen before.
In late 2020, a specific new variant of H5N1 called clade 2.3.4.4b emerged and has continued to spread across the past four years, moving into more and more bird and mammalian species. It has infected more than 100 types of wild birds that are not waterfowl and that migrate on different routes from the traditional routes that avian influenza spreads. These include the eagle, pheasant, penguin, bar-tailed godwit, Pacific golden plover, bristle-thighed curlew, great sand plover, eastern curlew, Asian dowitcher, black-tailed godwit, broad-billed sandpiper, grey-tailed tattler, terek sandpiper, grey knot, red knot, ruddy turnstone and others.
This variant has shown some worrying features, including severe neurological signs in infected birds and animals. Other unusual presentations have included conjunctivitis, so it’s not the classic severe respiratory illness we would expect. A strain found in eagles in the US has been shown to be severely invasive to the brain and neurological system. If this strain mutated to become a pandemic strain, we may see severe effects on the brain in addition to the lungs.
H5N1 2.3.4.4b has caused 70 human cases as of March 2025 in the US alone but does not transmit easily between humans. It remains a virus adapted to birds. Most cases were farm workers who had close contact with infected cattle or poultry. However, one person in Missouri developed H5N1 without any known risky exposures. Several people around this person also had symptoms, including health workers who treated them and family contacts, but only one family contact tested positive.
The risk to humans from mutation of H5N1 is higher than ever simply because of the unprecedented spread, which increases the statistical probability of a mutation that will cause a pandemic. If the probability of the dreaded mutation is related to the rate of infection in wild birds, farmed and domestic animals, and the mixing of humans and such animals, then there are far more chances for this to occur today than at any time in the past.
Influenza viruses must bind to a suitable receptor in the cells of humans, birds or animals to invade the body and cause infection. The key event that could cause a human pandemic is a mutation that switches the affinity of the virus from bird to human. Such a switch means the virus can bind to receptors in the human respiratory tract and invade the body. Humans have different receptors in the upper respiratory tract from birds and this is why the avian strains do not easily spread between humans. Bird flu preferentially binds to the throats of birds and does not attach easily to the human nose and throat.
Some mammals, however, such as ferrets and pigs, have similar respiratory tract receptors to humans. That is why ferrets are used in influenza research in the lab. Pigs are a genetic mixing vessel because they have both bird and human receptors, so flu viruses can mutate in a pig to cause a human pandemic.
Many new mammalian species have been infected since 2021, and some of these animals may be more like humans and therefore suitable genetic mixing vessels to create a human pandemic strain. Also, the infection of terrestrial wild animals that live close to human communities, such as red foxes and squirrels, increases the risk of infection of domestic animals such as cats and dogs, which can bring the virus into households. This is a possible new route for a human pandemic.
In the past, H5N1 epidemics in birds were sporadic and would die down after culling of infected poultry. Since 2021, the pattern has changed and it has not gone away or subsided but steadily increased and infected a wider range of birds and animals. This is a completely new pattern for avian influenza. I have been following H5N1 since 1997 and the current situation is unprecedented and extremely worrying.
Historically, the epicentre of bird flu epidemics was Asia, particularly China, Indonesia and Vietnam, with Egypt also affected a decade ago. Since 2021, the global hotspots have shifted from Asia to Europe, the Americas and Africa, where H5N1 clade 2.3.4.4b has spread in an unprecedented manner to an increasing array of wild birds, wild mammals and farmed animals. In the past, wild birds would largely infect farmed poultry, but we are now seeing unprecedented epidemics in farmed cattle and even goats.
Wild animals that have been infected for the first time include seals, sea lions, red foxes, coyotes, raccoons, mountain lions, skunks, squirrels and others. Infections have also been documented in domestic cats. New pathways for human pandemic emergence that fly in the face of traditional thinking (which is that wild birds infect pigs or poultry with bird flu, which then mutates to infect humans and cause a pandemic) are now open.
Another new route is newly infected species of wild birds that have never been the carriers of avian influenza in the past. Birds use specific flyways that are like airline routes, and in the past the spread of bird flu has been restricted to the flyways of ducks, swans and geese (waterfowl). So, we need to think beyond the traditional flyways of waterfowl as the routes of avian influenza spread.
This matters for Australia, which has been spared H5N1. In 2024, it was the only continent free of the virus. The waterfowl flyways that spread this virus around the world bypass Australia because it has a boundary called the Wallace Line, which separates the fauna of Asia and Australia. H5N1 has not crossed that line.
However, H5N1 reached Antarctica in 2024, so there are now new flyways through which the virus can reach Australia. The impact on our farming and our unique native wildlife, some of which are endangered, could be devastating. In the US, the outbreaks in farms also pose a threat by increasing the likelihood of H5N1 mixing with human or animal influenza strains to cause a pandemic. The more mixing there is between humans and infected animals or poultry, the greater the risk of a pandemic.
The current epidemics in dairy farms in the US are not easily explained. It may be due to a combination of spread by wild birds, terrestrial animals, trade of livestock and agricultural practices. Our research suggests the virus was introduced into US farms by wild birds, but that subsequent spread has been due to domestic agricultural practices as well as two-way spread between cattle and poultry.
Some interstate spread has been linked directly to the importation of cattle from affected states. Cattle feed may include poultry litter, which includes feathers, faeces and other refuse from chickens. This practice is banned in Australia and many other countries but not in the US. The spread of the epidemic across farms may be due in part to the use of poultry litter for cattle feed. There have been poultry outbreaks of this virus in the US preceding the cattle outbreak, which supports this hypothesis.
Other farming practices that may have spread the virus include the milking machines, as a very high concentration of virus has been found in the milk. The process of using and cleaning those machines results in widespread aerosolisation of the milk. There is now genetic evidence of spread from cattle back to poultry, thus completing a never-ending cycle of farm infections.
The US dairy farm outbreaks began sometime in February 2024, with the first case of transmission from a cow to a human documented in Texas in late March that year. The strain of H5N1 isolated in the farm worker in Texas was an avian virus that had some genetic mutations but did not show adaptation for human transmission.
However, the reason for concern is the increased opportunities for mutation of the virus from farm animals or poultry in proximity to humans. This may include people working on farms, but also contamination of the food supply with H5N1. In fact, traces of H5N1 have been confirmed through polymerase chain reaction testing in the commercial milk supply in a high proportion of cartons sampled. PCR testing is the most common way of testing for viruses and identifies fragments of RNA of the virus but it cannot tell you if the virus is live and infectious.
To date, a live virus has been found only in raw milk; however, the first publicly available data was from independent scientists who tested milk and milk products off the supermarket shelves. They found that 38 per cent of samples tested had H5N1 viral fragments detected, which indicates that the epidemic in dairy farms is more widespread than reporting would suggest.
The US Department of Agriculture has been guarded and concerned about the economic impact on dairy farms. It was slow in sharing H5N1 genetic data through the public platform Global Initiative on Sharing All Influenza Data, which provides free public access to influenza virus genomic data. The US Food and Drug Administration, meanwhile, has assured the public that pasteurisation guarantees the milk is safe to drink.
Theoretically, pasteurisation should kill viruses and bacteria; however, there is a growing alternative lifestyle movement that prefers unpasteurised, raw milk, which provides fertile ground for a pandemic to emerge. There are also anecdotal reports of many humans suffering influenza-like illness at the same time as cattle are being infected but refusing to be tested.
The dairy farm outbreaks were detected because the infected cows were producing less milk and the quality of the milk was visibly different – yellow, thick and viscous compared with normal milk. No such obvious signal would be present among beef cattle, and beef farmers are reluctant to test their herds. Eating beef, especially a rare steak, may well be a risk in the US, but the highest risk is dairy as the concentration of virus is greatest in milk.
There also have been outbreaks in farmed goats in the US, so the infection may be widespread in farmed animals. If farmers are not compensated financially, they will not test and report H5N1. Unless the government substantially compensates farmers and expands surveillance to other farmed animals, this situation is unlikely to subside and will increase the risk of a H5N1 pandemic in humans. The case fatality rate in humans to date since H5N1 first emerged in 1997 is around 50 per cent, but only one of the human cases in the US has been fatal.
The good news is that we are better prepared for an influenza pandemic than we were for the Covid-19 pandemic because influenza is a highly researched virus. There has been much more research done on it than on coronaviruses when SARS-CoV-2 first emerged. We also have seasonal flu vaccines and the same technology can be used to create a pandemic vaccine that is an exact match to a new pandemic strain.
The Holy Grail of influenza vaccines is a universal vaccine that will protect against any influenza strain. Although many groups are researching such a vaccine, it remains elusive. Current vaccines target the surface proteins of the virus, the H and N antigens, which are also the parts of the virus that mutate the most. This is why we need an annual seasonal flu vaccine, because each year the virus mutates to be distinct from the previous year’s virus.
We have come a long way in influenza vaccine technology in the past 20 years. For more than 50 years, flu vaccines have been made by growing the virus inside hens’ eggs and extracting the protective proteins from there. Most seasonal vaccines are still made this way.
The added complication with H5N1 is that the virus itself (needed to make the vaccines) kills the eggs required to make the vaccines. We now have many new vaccine technologies, such as recombinant, cell-based and mRNA vaccines, that can be made faster than egg-based vaccines, but how fast we can get these into arms during a pandemic will depend on the agility of regulatory bodies.
Live attenuated influenza vaccines, which use a modified live but harmless influenza virus, are also available in some countries but have a chequered history. There is the possibility of a lab mishap causing illness from live influenza viruses. The 1977 Russian flu pandemic is now accepted as originating from an incompletely attenuated live flu vaccine being developed in China or Russia. This was denied for 30 years, probably because scientists on both sides of the Cold War did not want to inflame political tensions, but there were clear signs it was not a natural virus right from the start. The virus had been extinct for decades and had the characteristic signature of a vaccine strain – sensitivity to temperature.
If a pandemic were to arise today, we have many vaccine technologies to choose from. The likely manufacturers of pandemic flu vaccines will also be the ones that make seasonal or pre-pandemic vaccines. The mRNA vaccines can be made in as short a time frame as six weeks, but getting shots to arms will take longer because of the regulatory process. During Covid-19, we saw emergency authorisation allow vaccines to be made available faster, so a similar process will likely occur during an influenza pandemic. We also have pre-pandemic H5 vaccines that will give partial protection. Getting your seasonal flu shot also can confer a small amount of cross-protection but will not provide proper protection.
We also have antivirals against the flu, such as the neuraminidase inhibitors (Tamiflu, Relenza and intravenous alternatives), which work against all flu strains. At this stage, drug resistance is low, but with widespread use for treatment this may become more problematic. Prevention by vaccines is always better than cure.
We can do even better, though, than waiting for a pandemic to start and then developing vaccines. We can prevent pandemics altogether because they grow exponentially. If we can identify very early signs of a pandemic, it can be stopped in its tracks. For example, if the Covid pandemic had been detected in Wuhan early, further spread could have been prevented by identifying cases, isolating them, tracing their contacts and quarantining their contacts.
Governments conduct surveillance for many infections to enable quick action if there is an uptick. This is done through laboratories and doctors reporting cases of diseases that are mandated as notifiable. In Australia, more than 70 infectious diseases are notifiable, including measles, whooping cough, HIV and other serious or vaccine-preventable infections.
Surveillance for H5N1 or pandemic influenza requires testing of animals, birds and humans and rapid reporting of infections. It also requires enablers and incentives for farmers to test and report, as well as sharing genetic sequence data as soon as possible.
In early March 2024, a human case of H5N1 occurred in a two-year-old child in Victoria. They had acquired the infection in India and apparently had no close contact with birds, animals or sick people. Public disclosure of the case did not occur until May 22, 2024, more than two months later. This kind of delay is not ideal and shows that delays in reporting can occur anywhere.
Open-source intelligence such as our EPIWATCH system, which uses artificial intelligence to identify serious epidemics, also can help provide early warnings and overcome delays. News agencies report on, and people talk about, unusual or concerning epidemics long before the health department knows about them, so tapping into open-source intelligence can help identify early warnings. This can assist with vaccine development by speeding up the process of identifying an epidemic and then actively going in to test and characterise the pathogen at an earlier stage. For example, the genome sequence for SARS-CoV-2 was released in January 2020 and was required by vaccine manufacturers to enable them to develop a vaccine.
However, there is now ample evidence that patient zero may have been in November 2019 or even earlier, and that epidemic activity was present before the World Health Organisation was notified. By late December, when the WHO was notified, the infection had already spread to Europe and the US. We know this from blood tests done on people in those countries, which showed evidence of exposure to the virus between November and December 2019.
If countries in those regions knew there was a concerning epidemic in China in late 2019, although they could not go into China and investigate, they could have started testing and characterising the virus in their own locations long before it was disclosed by China. Despite censorship of reporting from China, EPIWATCH was able to detect severe unknown pneumonia in Wuhan before the official disclosure date.
Reasons for censorship, delay or lack of reporting need to be understood to increase the chance of early detection of pandemics. In addition to early warning, testing and surveillance, managing the economic impacts for farmers is critical. If there is no financial compensation for farmers, it will result in cover-ups of outbreaks, drive a black-market trade in infected animals and accelerate the risk of a human pandemic.
The widespread H5N1 epidemic in US cattle farms increases the risk of a human pandemic. In Australia, two poultry farms were simultaneously infected with different highly pathogenic avian influenza viruses, H7N3 and H7N9, in May 2024. We had experienced only eight outbreaks before that and never H7N9, which is the virus that emerged in China in 2013 and caused as much concern as H5N1. There are genome sequences in GISAID of H7N9 from Australian wild birds in 2013. At the same time, Western Australia had a poultry outbreak of H9N2, a low pathogenic virus that has been causing severe human infections in China. Three outbreaks in rapid succession in a country normally spared from avian influenza is a warning.
However, with Europe and the Americas the new epicentre of H5N1, these are the likely sites for the emergence of a human pandemic.
Global governance during a pandemic remains a weakness. The International Health Regulations consider disease, trade and economic impacts, and are equally concerned with protecting commerce as they are with health – and so argue against border closure during a pandemic. Many countries closed their border during Covid-19 anyway, demonstrating the unenforceable nature of the IHR.
The Global Health Security Index was launched the year before Covid-19 began but failed to predict which countries would respond effectively. It ranked the US the highest in pandemic preparedness yet we saw severe impacts in New York and other parts of the US early in the pandemic, with massive failures in testing and other aspects of pandemic control.
Our epidemic risk analysis tool called EpiRisk had a similar ranking to the GHSI. However, after Covid-19 and seeing that some low and middle-income countries had fared better than high-income countries, we went back to the drawing board and worked out what parameters were missing from the model to provide a better prediction of preparedness. The model was then tested against Covid-19 responses in a range of countries and adjusted to incorporate other influential factors such as leadership, culture and universal healthcare until it was better able to predict pandemic response. This is the kind of risk analysis that should be routine in pandemic preparedness.
With the dark cloud of H5N1 above us, adequate stockpiling is another action that can mitigate a pandemic. In addition to pre-pandemic vaccines, this would include influenza antivirals, antibiotics to treat bacterial secondary infections, pneumococcal vaccines as a preventive measure against pneumonia and, of course, personal protective equipment. We likely will have forgotten the lessons about masks, as we did after the 2009 influenza pandemic. We will see shortages of masks and health workers being mowed down at the front line because, sadly, the Covid pandemic has resulted in a backlash against many public health measures. We have gone backwards in public health pandemic control since Covid and that will be a setback during a new influenza pandemic. Ultimately, however, risk perception drives human behaviour and tolerance for public health measures, and a H5 pandemic may be much more severe than Covid. When people see friends, family and neighbours dying or becoming seriously ill with pandemic influenza, most will avail themselves of any available protection measures.
This is an edited extract from Vaccine Nation: Science, Reason and the Threat to 200 Years of Progress by leading Australian epidemiologist Raina MacIntyre (NewSouth), available now.
To join the conversation, please log in. Don't have an account? Register
Join the conversation, you are commenting as Logout