How does the CDC measure the spread of H1N1?

November 13, 2009 2:01 AM

There's no special place to turn up if you think you've got the swine flu to be tested or otherwise counted--hospitals and clinics tell people to just stay home unless they are having actual health complications. How is the CDC able to say that 22 million people have been infected with H1N1 when if you don't have to be hospitalized, nobody will even test you for it?
posted by autoclavicle to Health & Fitness (18 answers total)

As a current H1N1 recipient, I wondered about this too. The nurse at my doc's office heard my symptoms and said to stay home unless I had complications.
posted by dzaz at 2:35 AM on November 13, 2009

So, your question is about the CDC's method of arriving at its numbers? (Just trying to figure out what's what here).

My niece had H1N1, though she stayed home while sick - her doctor tested her. IT is not too far-fetched to imagine him reporting this data point to the CDC.
posted by From Bklyn at 3:04 AM on November 13, 2009

You've noticed, no doubt, that they're not saying that 22,183, 762 people have been infected. The 22 million is an approximation extrapolated from tests they have done.
posted by jon1270 at 3:08 AM on November 13, 2009

IT is not too far-fetched to imagine him reporting this data point to the CDC.

Yes, the CDC runs the Outpatient Influenza-like Illness Surveillance Network (ILINet) which is a system where sentinel physicians across the country report the number of patients that present with influenza-like illnesses (ILI). So even if you aren't hospitalized, you're still counted.
posted by Rhomboid at 3:13 AM on November 13, 2009

Best answer: It's an estimate based on established epidemiology and statistical inference, i.e. you collect data in a sample group, calculate the infection rate and extrapolate to the general population.

When figures like this are published in studies, they are always qualified with levels of confidence. I.e. figure X has a Y% likelihood of being accurate to within +/- Z. But of course journalists usually just boil it down to one number and focus mainly on the human interest part.

Rest assured that an institution like the CDC will be using the most trusted methods to calculate these numbers.
posted by randomstriker at 3:16 AM on November 13, 2009 [1 favorite]

I was tested. I have asthma and my doctor wanted to be sure I had H1N1 before going any further with treatments, so when I called with my symptoms (because he wanted me to call if I had any flu symptoms), he had me come in after hours to test me.
posted by cooker girl at 3:47 AM on November 13, 2009

Another anecdotal data point: When my son developed flu-like symptoms recently, the first thing the pediatrician did was a nasal swab test for H1N1. Apparently this test is only 70 percent accurate, though (according to the doc), and he said that if it had been positive they would send a sample to the lab for DNA confirmation.
posted by Shoggoth at 5:39 AM on November 13, 2009

Epidemiology, plus estimates of flu-like symptoms outside of the normal influenza period.
posted by gramcracker at 5:57 AM on November 13, 2009

The CDC has run computer models and probably continues to do so fairly often. Computer modeling can be fairly accurate given the right data like population density, locations of deaths, time since introduction as well as a million other parameters.
posted by JJ86 at 6:01 AM on November 13, 2009

Lots of these answers are mentioning people that went to the Dr to get tested, so yes, it is reasonable to assume that those tests are reported and counted, and that data is then used in the statistical models.

But - IME, the folks at my office whose kids got the flu didn't take them to a Dr, or they said the Dr didn't do an actual test to determine which type of flu. They just reported flu symptoms and kept them home. Do the statistics take all of that into account? Will we see statistics later about how many other strains of flu vs H1N1?
posted by CathyG at 6:39 AM on November 13, 2009

Best answer: They're estimates. See this article in the Wash Post. Notably, "The new estimates, drawn from detailed surveillance and record-checking in 10 states...."
posted by inigo2 at 7:08 AM on November 13, 2009

nthing that it's a statistical estimate, not an exact number. Even in an enormous population, you don't need that big of a sample (a few thousand is more than enough) to determine the prevalence of something among the general population (assuming you draw your sample correctly, which I can pretty much guarantee you the CDC does).
posted by Doofus Magoo at 8:09 AM on November 13, 2009

It's a statistical estimate, but don't take that as some indication that it's inaccurate: statistical analysis of population health is what the CDC does.
posted by rokusan at 8:51 AM on November 13, 2009

Best answer: This article has a slightly more detailed explanation of how the estimates are made.
posted by Bangaioh at 8:57 AM on November 13, 2009

Best answer: I work for a health department and we have an epidemiologist. We collect reports of flu-like illnesses (and other kinds of communicable diseases) and report it into the State. The State makes their reports to the CDC. Mefi's are right, these data are estimates. Schools report flu-like illnesses, (FLI), as well as ER's and physicians. There are too many people sick to test them all.

But the data is reliable. And although I am not right in the trenches, it is pretty much H1N1 24/7 these days. And there has been speculation about H5N1 and what that may bring. The CDC is doing a very good job.
posted by chocolatetiara at 9:00 AM on November 13, 2009

They are definitely not doing random sampling of the population - that would be friggin' impossible. Read this.

In short: it's a combination of reported flu-like symptoms and lab-confirmed H1N1-related hospitalizations and deaths.
posted by McBearclaw at 9:01 AM on November 13, 2009

Bah, wrong link. I meant this.
posted by McBearclaw at 9:02 AM on November 13, 2009

Anecdotally, when I went to the doc with flu-like symptoms last week, he told me that the test was not worthwhile as the results would appear after the window for effective use of Tamiflu had closed. That said, he also informed me that he reported my case as "probable" to the CDC.
posted by bluejayway at 9:04 AM on November 13, 2009