If you’ve been putting off your flu shot until the season really gets going, wait no longer. It’s already here—and it’s looking like it’s going to be a doozy. Influenza viruses quietly circulate year-round in the US, but every winter they go big, triggering a seasonal epidemic of sniffles, sweats, and sore throats. And this year it’s come earlier than usual, just in time for a potential peak over the holidays.
Normally, infections start to increase around the beginning of January, peak in February, and then come down by the end of March. But according to data released by the Centers for Disease Control and Prevention on Friday, the flu is already widespread in 12 states, including California and New York. “We’re seeing flu activity increasing around us,” says Brendan Flannery, an epidemiologist in the CDC’s influenza division. “The uptick does indeed seem to be here.”
Why the spike is ahead of schedule is anyone’s guess. Influenza viruses are wildly unpredictable, and so far nothing—not weather patterns or demographic changes or social media data—has explained what drives an early season versus a late one.
Not that scientists haven’t tried. For the past five years, the CDC has hosted a flu forecasting challenge, asking universities, companies, and independent labs to submit their best predictions for the viruses’ spread. At the end of each season the agency evaluates every model; this year there are 35. And right now they’re pretty evenly split. About half think the peak will come the last week of the year, the rest predict max infection rates in January and February.
Matt Biggerstaff, another epidemiologist in the CDC’s influenza division, says they’ll have to wait until July, when they’ve got all the season’s surveillance data, to see how well each one performed. But in general, he says, most of them can get pretty accurate for near-term forecasts—a week or two out. Anything farther than that and the signals start to fall off. That’s because there just isn’t that much data for the models, many of which are based on machine learning, to learn from.
“We only go back to the ’90s for some of these systems,” says Biggerstaff. “It’s not like weather forecasting, where you have stations collecting real-time measurements from thousands of locations all over the country. We just don’t have that kind of resolution.” But they’re trying to get there. This will be the first year that CDC is supplying state-level data to the modelers. Previously they only got as granular as the 10 reporting regions defined by the department of Health and Human Services.
But even more data would be better—say at the county or city level—says Roni Rosenfeld, a computer scientist at Carnegie Mellon University who leads one of the modeling groups that has submitted projects to the CDC. His team’s systems took the top two spots in last year’s challenge. “Influenza may be able to cross the globe in matter of weeks, but its dynamics are very local,” says Rosenfeld. “And we just don’t have good enough data for local conditions.”
To make up for it, Rosenfeld’s group has sought out supplementary sources from around the web. Combining digital surveillance data from things like flu search queries on Google, CDC page visits, and even sick-sounding tweets with input from CDC’s conventional surveillance system, they trained a machine learning model to look forward and make predictions about future flu activity. “The CDC’s forecasting initiative has done so much to drive this field forward in a short amount of time,” says Rosenfeld. “It’s been less than a decade since anyone thought this was even a possibility.”
Biggerstaff and Flannery hope that with more time and more research they’ll be able to project a month or two into the future. That would actually allow hospitals to get ahead of the spike to bulk up their staffing and provide some urgency to public health officials pushing vaccine campaigns on the ground. “Now that only happens after we’ve seen the first signals coming through the surveillance data,” says Biggerstaff. In other words, a bunch of people had to get sick for you to start seeing scary headlines and getting calls from your doctor’s office. In a world where forecasting worked, you could have been getting that news before all your coworkers started dropping like flies. “If we could move that timeline back a few weeks we could get a lot more people treated right at the start of flu activity instead of waiting for it to spread and intensify.”
- Tara Haelle
What's Inside a Flu Shot? Formaldehyde and Chicken Eggs
- Sarah Zhang
Scientists Get One Step Closer to a Universal Flu Vaccine
- Anna Vlasits
The Clue to Next Year's Flu Strain Could Be Inside You
More on the Flu
It would have been useful this year. With the peak potentially falling during the holidays, many doctor’s offices, outpatient clinics, and hospitals will be running skeleton crews, even as the worst is poised to hit. Medical researchers have raised concerns over how effective this year’s flu vaccine is against the most severe strain: H3N2. Because influenza mutates so quickly into so many different strains, each year’s vaccine is unique—designed to target only the strains that are most likely to cause the most damage in the next season. According to CDC data, H3N2 is currently the most prominent strain out there, responsible for 90 percent of infections since October. And it’s the one most likely to send kids and the elderly to the hospital.
The same strain also dominated in Australia, which experienced record influenza rates this past season. Down under, the vaccine was only about 10 percent effective against H3N2—and the vaccine composition, which protects against three other flu types in addition to H3N2, is the same one being used in the US. For now, it’s the best thing there is, and considering how many tricks the influenza virus has got up its lipid envelope, it’s remarkable scientists have created one that works at all. Most years the vaccine cuts your chance of catching the flu in half; the odds have been stuck at parity for decades.
But that hasn’t stopped researchers from going after something even bigger: a vaccine that can fight all types of the virus. And they’re now closer than ever before. In October, scientists in UK began enrolling 2,000 adult patients for a two-year clinical trial to test a universal vaccine—the first time any such treatment has progressed beyond phase one testing. The vaccine, developed by Oxford University and a spin-out biotech company called Vaccitech, will recruit 500 British subjects this season, and the rest during the 2018/19 season. If it works, it would be the kind of shot you’d only need every few years.
Until that happens though, you might want to knock out out that pesky little prick before you head home. Just make sure you call first. You’re not going to be the only one wanting it.
Related Video
Science
What’s Inside a Flu Shot?
Each year the World Health Organization determines which of the thousands of influenza variants are most likely to circulate. Flu vaccines then use these variants at the starting point when developing the annual flu shot. Find out what’s inside the influenza-warding shot.