The Real Subtypes of the 2025 Flu Season
Have you ever been to a party where you sensed the messiness before it happened?
Bird flu has been blowing up poultry operations across the entire continent of North America for the past two months. Now, human flu season is officially on. The influenza A virus party in the northern hemisphere started early this year and all the hottest subtypes are here: H5N1 poultry outbreaks in both commercial and backyard flocks across the US and Canada, an unusually early seasonal epidemic in Japan, a new H5N5 avian reassortant virus put someone in the hospital in Washington state, and an emergent subclade of H3N2 that isn’t well-matched to the vaccine has popped up when it’s least wanted around the world. Flu season is off to a vigorous start, and although I do love a good party, I don’t like the vibe of this one.
Last year, the flu season in North America was worse than it’s been since the 2009 H1N1 pandemic in terms of illness and mortality. That occurred when the US still had functioning public health infrastructure and hadn’t yet withdrawn from the World Health Organization (WHO), shut down wide swaths of health infrastructure abroad by withdrawing foreign aid and cancelling grants to foreign institutions, decimated the federal workforce, and generally caused destructive chaos in public health. We are no longer in that situation.
The US government, including all the relevant departments and agencies that respond to and manage influenza epidemics in the US, has lost significant terms of its capacity for dealing with it. I don’t like to predict how bad a given infectious disease outbreak is going to be because viruses are always full of surprises, but I have an ominous feeling that this flu season will be worse than the last strictly on the basis of our dysfunctional capabilities. It’s like watching a contrived party on a reality show that you know is going to go off the rails, but you are not sure which drunken Real Housewife is going to catalyze the drama. There is a lot of flu circulating in various species, so it’s worth examining the situation to understand the potential for drama and where it might come from.
Bad time to be a bird (or mammal)
Bird flu season starts before human flu season, since it starts when the birds begin their seasonal migration. Birds normally have low pathogenicity avian subtypes (viruses with H and N that are adapted to birds) that circulate more broadly when the birds flap south along the flyway. Right now they also have a ton of highly pathogenic H5N1. Among wild birds, H5N1 causes different severity disease in different species. It will kill some birds, and many have been found dead, but others, like ducks, fly around and do bird stuff unbothered.
Ducks are a huge problem because H5N1 replicates efficiently in them without causing much disease. They are gregarious creatures and are attracted to places with food, water, and shelter where other birds congregate, like poultry farms. In birds, influenza virus infects the gastrointestinal tract as well as the respiratory tract, so infected birds shit out quite a lot of virus. Wild ducks pulling up to a farm to mingle with the poultry, eat their food, drink their water, and shit on everything is a common way for H5N1 to get into a flock. Farms have made government-subsidized efforts to shore up biosecurity to keep wild birds out, but it is not perfect. When highly pathogenic H5N1 gets into a flock, every bird must be culled to contain the outbreak, prevent it from spreading to other farms, and reduce the risk of transmission to farm workers.
Currently there are 70 active confirmed poultry outbreaks in the US and 49 in Canada (yes, one of those is the wretched ostrich farm, but it should be cleared soon). This equates to millions of dead chickens and turkeys and creates a big economic and agricultural problem. New poultry outbreaks have been detected on a nearly daily basis since September. Last February, Secretary of Agriculture Brooke Rollins raided USDA school and food bank programs to fund an avian flu management strategy (designed to reduce egg prices) that promised to effectively stop wild bird introductions by hiring a handful of epidemiologists to carry out biosecurity audits and subsidizing farm upgrades. This approach has evidently failed, given that outbreaks are steadily increasing.
On the surface, things appear to have improved in America’s dairy farms. Although the US Department of Agriculture (USDA) Animal and Plant Health Inspection Service (APHIS) is not reporting any active livestock outbreaks, I think this is highly misleading. There haven’t been any new herds detected, but it’s unclear how much testing is even occurring in dairy cattle anymore. The USDA’s National Milk Testing Strategy website has not been updated since July 30th, 2025. The APHIS workforce has been devastated by forced retirements, reductions in force (RIFs), and funding cuts. Dairy cows generally don’t get very sick from H5N1, although infection reduces or stops milk production. Undetected circulation on dairy farms is entirely possible in a climate with no testing and a non-functional animal health inspection service.
The situation is not great for wild mammals, either. Over the summer, I threw together a graph showing how many wild mammal detections of H5N1 by USDA I expected based on the mean monthly rate of detections in the first half of the year. I did the same thing with the most recent data (updated through mid-September 2025). Let’s see how wrong I was!
My amateurish projection wasn’t perfect, since I overestimated by about 30 detections. Also, the rate of detections has slowed considerably, which suggests that less testing is occurring or fewer samples are being collected (due to either less sample collection or fewer samples available to collect). I suspect it may be both, but I have no evidence to support that suspicion. This type of data is difficult to assess because it is often collected opportunistically, when someone brings in a dead animal or wildlife officials manage to collect animal samples in the course of outbreak investigation and response. In any case, H5N1 is still infecting plenty of wildlife, whether on land or in the air. There’s a lot of it around.
But many people say so what? There is no evidence that H5N1 has acquired the ability to transmit efficiently between people, so it’s not a human concern. It might acquire this ability at some point in the unknown future. It may never acquire this ability. That’s true, and as a human, I agree that’s a relief. But even if it doesn’t, it poses a massive ongoing threat to agricultural and food security, the economy, and the environment.
For reasons I don’t completely understand, the USDA-approved Zoetis poultry vaccine doesn’t really seem to be getting used. I am not sure why this is the case. Vaccinated poultry cannot be sold on the international market because the vaccines are not completely sterilizing (they don’t prevent infection altogether, even though they stop disease). That means vaccinated birds could still be infected, so our current trade agreements forbid vaccinated poultry products from being exported. Vaccination does not have an impact on domestic production, however, and should still be considered as part of the US strategy for managing avian flu, considering how much H5N1 is around and how many birds it is killing.
We can’t do much about infected wild birds or animals, but we now have tools to better manage and prevent avian flu in domestic poultry. Doing so should be a priority, considering that reducing the chances for H5N1 making the jump to a pandemic virus depends on reducing the overall incidence of H5N1 infection overall.
Don’t mean to scare anyone but have you ever heard of reassortment?
Reassortment is what keeps most flu people up at night. Influenza viruses have a segmented genome, comprised of 8 pieces of viral RNA. When a host gets infected with two different flu viruses at the same time, these pieces get shuffled up at random and new hybrid flu viruses are created that are called “reassortants.”
Reassortant viruses have very unpredictable properties. Sometimes they are a virological dud and don’t grow very well. Sometimes they acquire new abilities, like infecting new hosts. Recent reassortants have jumped into dairy cows, ostriches, and humans, including the recent case of H5N5 in Washington state. The H5N5 virus infecting this person was previously detected in Canada and it is the result of reassortment between two avian viruses: a highly pathogenic H5N1 virus acquired the neuraminidase gene (NA N5) from a circulating low path North American HxN5 virus. This reassortment probably occurred in a bird and this is still a very avian virus in the sense that it doesn’t contain mutations known to increase viral fitness in mammals. (NOTE FROM ANGIE: Grateful to the colleague who emailed me to let me know I was looking at the wrong H5N5 virus and the one infecting the patient in Washington state was actually not a recent reassortant, although there are a lot of reassortant H5N5 viruses circulating in the US. The risk to humans from H5N1 and H5N5 viruses and from reassortment `doesn’t change much either way. The H5N5 infecting the patient is still a very avian virus and there is no evidence of onward transmission from the patient in Washington to other people. It just has a more old-fashioned non-reassortant Eurasian NA N5—but the risk of reassortment remains even if this virus isn’t the product of it.)
Reassortment happens all the time when there’s a lot of virus circulating and it does result in cross-species jumps. To reduce the risk of a bird flu pandemic, we need to deprive the virus of opportunities to reassort and to jump across species barriers into new hosts. The best way to do this is to reduce the number of infections overall—in birds, livestock, and humans.
Preventing infection in pigs and humans is the most important of all when it comes to pandemic risk. Pigs and humans can be infected with both H5N1 and human seasonal strains in their respiratory tissues. Reassortment in this context is the highest risk for producing a pandemic capable virus. Combining genes from highly pathogenic avian flu with those from a seasonal influenza virus that is already well-adapted to replicating in and transmitting between humans could create a virus with significant pandemic potential. Continued transmission or infection would allow this virus to adapt further to human hosts through mutation and the normal process of evolutionary adaptation.
Another priority should be monitoring for flu in pigs and humans, except unfortunately neither of those things are happening very much. Because of the cuts at USDA/APHIS, the extent of impacts on swine surveillance has been impacted are hard to quantify. Unsurprisingly, it seems that there is less pig surveillance occurring, since the USDA swine surveillance dashboard hasn’t been updated since July 30, 2025. There has only been one known H5N1 infection in pigs, when a genotype D1.2 virus was discovered in a non-commercial farm a year ago in Oregon. But pigs don’t get very sick from many influenza viruses, so we might not find other cases without doing active surveillance using molecular testing. Given the scale of some pig farms in both the US and Canada, where many animals are raised together in close proximity to humans, it’s critical to make sure that swine production only involves producing safe and healthy pork, not novel bird flu reassortants.
In the US, human testing is no longer occurring at the scale it used to. I wrote about this last summer, and nothing has changed since then. Although the National Flu Surveillance system remains intact, the majority of human cases in the US were identified by targeted monitoring and testing. This means specifically identifying and testing people who likely were exposed in the course of interacting with infected animals. Most of the cases identified by this program were dairy and farm workers.
It’s not hard to guess why human testing has fallen off a cliff in the US. The CDC has been gutted. Even though the shutdown RIFs were rescinded, CDC’s capacity for carrying out its normal functions, including flu surveillance, had already been shattered. Until last Friday, the CDC had not updated any flu surveillance data since September 20, 2025 (it normally updates every Friday, even during government shutdowns). There’s just not capacity to carry out the work. Furthermore, if I were an American dairy or poultry worker and I thought I had been exposed to H5N1 or had mild symptoms (most cases in farm workers have been mild and even subclinical in some cases), the last thing I would want to do is report it to a government official that might send me to an ICE detention facility for my trouble.
What this all means is that there are a lot of opportunities for H5N1 to reassort in a host that could produce a pandemic capable virus overnight, and we aren’t doing enough to take those opportunities away. There are going to be even more of these opportunities, because human flu season is upon us.
How you, H3N2?
Because the US had not updated FluView, there was not much information about the seasonal strains circulating in the US. Thankfully, I live in Canada, where our public health system is intact, so I checked out the situation up here at FluWatch.
Seasonal influenza is ticking up in Canada (and the US, per FluView). It’s the usual suspects known to be circulating—H1N1 derived from the 2009 pandemic and H3N2. H3N2 has started to take over a little bit, which is never great news. H3N2 is usually thought to be more severe than H1N1 in higher-risk groups (children and older people).
H3N2 has come out of the gate this flu season with a bang, causing a huge early-season epidemic in Japan that resulted in closed schools and businesses in September, which is really, really unusual. There were large late-season H3N2 outbreaks in the southern hemisphere that might have introduced this to Japan so early in the season, but it’s unsettling.
It’s even more unsettling when you take a gander at these viruses themselves. Due to the number and diversity of flu viruses, we classify these viruses into clades and subclades (a type of phylogenetic grouping) based on their genomic sequences. Clade is assigned based on the sequence of hemagglutinin (HA, in this case H3) and subclade is a category of that. There are growing detections of H3N2 subclade K, an emerging variant of H3.
It’s pretty normal for new subclades to emerge. Flu is an RNA virus that acquires mutations every time it replicates, so if it’s replicating a lot, it will continue to evolve and HA will change. What’s alarming about this is not that a new subclade has emerged or is trending toward dominance, but that it is no longer matched well to the H3N2 component of this year’s flu shot. This year’s H3N2 component is from an American J.2 subclade virus. The vaccine targets HA and NA (H and N), so if those change, the vaccine might not work as well.
We can’t tell how effective the flu shot will be until we see what happens during flu season, but we can see how well antibodies it elicits neutralizes circulating viruses. So my colleagues at the National Microbiology Laboratory of the Public Health Agency of Canada in Winnipeg tested it.
Bad news: H3N2 subclade K isn’t neutralized as well as the J.2 viruses that the vaccine was matched to. That means the vaccine raises antibodies that don’t bind well to the subclade K HA. However, that doesn’t necessarily mean the vaccine won’t still work well. Flu immunology is incredibly complicated because we have been exposed to so many different subtypes throughout our lives. The immune system itself is incredibly complex and does not rely solely on antibody neutralization to combat pathogens. This may have no material impact on the severity of this year’s flu season. On the other hand, it may leave even vaccinated people more vulnerable to H3N2 infection, which then creates additional transmission risk as well as higher incidence of severe disease.
This doesn’t mean you should skip your flu shot this year. You should absolutely get it. It will still provide protection, even if it turns out to be relatively weak against H3N2. Flu shots decrease likelihood of being infected and they reduce disease severity, even in years when the vaccine is poorly matched to a dominant circulating subtype. In fact, if these early omens do indeed portend bad times ahead, you will want to have a flu shot more than ever. The flu shot provides some limited protection against the NA (N1) component of H5N1, as well as seasonal H1N1 (which is itself a pandemic H1N1 virus). Flu shots are an easy and safe action we all can take to mitigate the risk of a bad flu season and flu pandemics. But we need to do more than that, and my concern is that we are not doing it.
The bottom line is that while we don’t have evidence yet, I have a bad feeling about this flu season. There is a lot of flu everywhere, infecting a whole lot of different species including humans. There are multiple threats to human and animal health, food security, agricultural production, and the economy. There are many opportunities for reassortment and host adaptation. And we are not doing enough to take these opportunities away.
When I get a bad feeling about some looming infectious threat, I really love to be proven wrong (and people on the hellsite love to tell me that I am WRONG AGAIN!). We have the theoretical ability to do that by seriously investing in surveillance and containing animal outbreaks, increasing flu shot uptake, communicating clearly and transparently with the public about the situation, and proactively identifying human cases. The situation in the US has made this an impossible task, and because of that I fear I may be proven right.
Thanks, Angela.
This is a great write up! It's very informative and so well researched. I also want to say thanks for using a figure from one of my publications :)