How The CDC Is Using Big Data To Save You From The Flu


[credit provider=”By welivefast on Flickr” url=””]

If you’ve heard it once, you’ve heard it a million times: Get a flu shot.But how come that works? There are countless strains of flu, so unlike vaccines for childhood illnesses, you have to get a flu shot every year. Once is not enough.

It works because of a massive collaboration of doctors, hospitals, and labs that report mega amounts of data to the centres for Disease Control (CDC). The CDC fields some 700,000 reports of patents with flu-like symptoms per week. During flu season, doctors will send more than 5,000 lab tests to the CDC for further analysis.

Using that data, the CDC has created complex flu-tracking systems to determine things like: what strains of flu should go in the annual flu vaccine (which is changed twice a year); is this year’s flu outbreak the kind of flu that will respond well to anti-viral drugs; how deadly the flu is.

The flu season is in full swing right now, according to data analysed by Lynnette Brammer, a flu epidemiologist at the centres for Disease Control and the woman that helped Google create its flu tracking system.

She told Business Insider:

  • The flu started earlier this year and is now quickly on the rise.
  • The flu shot has been proven to protect well against this year’s strains. (So if you haven’t gotten one, you might want to rethink that.)
  • This year’s flu strains respond well to anti-viral meds, so those who get really sick can be helped by seeing their doctor.

If you want to see how bad certified flu cases are in your state, the CDC reports that data weekly on a website called FluView. Meanwhile Google used FluView data to help them build their own flu-tracking website, Google Flu Trends. Flu Trends tracks the spread of the illness based on people searching on flu symptoms. It offers a more instantaneous view of where people are getting sick, but it doesn’t verify if those people really have the flu or just bad head colds. 

The ultimate goal is not to end flu forever, but to keep people from becoming severely ill or dying from the flu.

Here’s a lightly edited transcript of the interview.

Flu map 2012

CDC FluView

Flu map, click for interactive map

Business Insider: How are you saving the world from the flu using technology?Lynnette Brammer: The data I use is the data contained in FluView. None of it comes from Google. There are basically eight different sources of data that go into FluView every week. It includes about 3,000 primary care providers across the country who report every week how many patients they saw that had Influenza-like illness. That gives us information on the proportion of people going to the doctor with flu-like illness.

BI: How many entries a week do you get?

LB: More than 700,000 patient encounters per week. But then you pair that up with other data sources, like our ViralLogic Lab data. That’s the U.S. World Health organisation collaborating lab system. It’s about 85 labs.

[Doctors] can take samples from a subset of their patients, send them off to their state public health lab who will do testing for flu. All the results from the state public health lab come to the ViralLogic Lab.

The majority of labs are reporting to us electronically. When the test results are final, it goes to a folder on their system and then it transfers it over here to the CDC. It’s very timely. It’s a fabulous reporting system. We then report out on a weekly basis. We could analyse that and report daily, but weekly is really fine.

BI: What types of things do you learn with all of this data?

LB: We learn where flu is and in what relative proportion [to the general population], what age groups are getting flu. We do what’s called “antigenic characterization” [of flu strains that made people sick] to see how close they look to the strains used in the vaccine. We do anti-viral resistance testing to make sure the strains are still sensitive to the drugs for flu and we do genetic sequencing on some of them.

We get rates of hospitalizations from lab-confirmed flu to see how much severe disease there is. We have another system that tracks pneumonia and influenza deaths so we can see on the whole population if flu activity is causing more deaths than you would expect for this time of year.

We try and cover the spectrum of where people would come into the healthcare system from flu. We can’t cover people not going to the doctor.

Things like Google Flu trends is where you get that type of data.

BI: How long have you been collecting this much data on the flu?

A lot started with the 2009 pandemic [popularly called Swine Flu], with so many labs reporting. But we have Virtual Logic data from as far back as 1976.

BI: What can you tell us about this year’s flu so far?

LB: Right now, the majority of the viruses are Influenza A, H2 and H3 viruses but there’s a fair amount of influenza B out there. The viruses look like the vaccine strains by and large, so the vaccine ought to work pretty well.

We’re not seeing any anti-viral resistance. There’s more flu going on in the south and southeast than in other parts, but in most of the country, it’s going up pretty quickly.

BI: How does this year compare to other years so far?

LB: So far this year it’s a little earlier. Whether or not it will peak early, I don’t know. It could be one of those years where it finishes by the end of January. Or it could keep increasing steadily for a while and drag on for months.

BI: Is the ultimate goal to end flu?

LB: Influenza strains are changing all the time. And there are animal reservoirs that spread it. Pigs can get flu, birds can get flu. A lot of animals can get flu; dogs, horses, seals, whales … so you are never going to eliminate flu. But we don’t collect data just to know. We collect so we can know enough to pick the strains to go in the vaccine each year. And if we have anti-viral resistance or not and can urge doctors to use more anti-viral.

Our hope is to make better vaccines to reduce the impact of flu and to reduce hospital visits and deaths due to flu.