Tuesday, May 12, 2009

H1N1: pandemics in perspective

My dear critical readers, I know you're a smart bunch. I recognize the need to not freak people out, but seriously: what is wrong with presentation of the following data set?
WASHINGTON (CNN) -- The swine flu virus that has sparked fear and precautions worldwide appears to be no more dangerous than the regular flu virus that makes its rounds each year, U.S. officials said Monday. [That's last week, Monday a week ago, mind you.]

"What the epidemiologists are seeing now with this particular strain of U.N. is that the severity of the disease, the severity of the flu -- how sick you get -- is not stronger than regular seasonal flu," Homeland Security Secretary Janet Napolitano said Monday as the worldwide number of confirmed cases of swine flu -- technically known as 2009 H1N1 virus -- topped 1,080.

The flu has been blamed for 26 deaths: 25 in Mexico and one in the United States, according to the World Health Organization.
First, reporting the raw numbers of deaths in the upswing of an epidemic tells you nothing about the dynamics of the situation. Moreover, "how sick you get" or the severity of the illness is NOT the same thing as how "dangerous" an illness is. I don't know if I'm the only one who thinks this, but I would evaluate the "dangerousness" of an illness as its fatality rate, that is, if you get sick, this is the probability that you will actually die. For instance, the probability you will die in a given skydiving jump is 1 in 100,000 (0.001% fatality). [1] Similarly, "you would have to jump 17 times per year for your risk of dying in a skydiving accident to equal your risk of dying in a car accident if you drive 10,000 miles per year." [1]

The fatality rate for the 1918 H1N1 flu was > 2.5%, as compared to < 0.1% [2]. In comparison, fatality rate is about 0.1% for the Asian and Hong Kong flus,[3]. Let's take a quick, back-of-the-envelope look at the numbers for the 2009 H1N1 virus. It has been in Mexico the longest, so we have the best sample size in terms of latency. 58 deaths divided by 2,282 cases gives us 2.54% (5/12/2009), which looks pretty similar to the 1918 H1N1 virus.

There are a couple ways of looking at this. The fatality rate of swine flu is approximately 1 - fatality of anthrax (in 2-3 days). And given the population of the planet, even anthrax has no chance of ending civilization. And the Black Death, depending on your geographical region, killed 20-80% of the population. (The fatality rate of the bubonic plague is supposed to be in the neighborhood of ~50% in 3-7 days without treatment.) In comparison, the fatality rate of SARS was ~9.6% (globally, with medical attention), even though it varied widely by region. [4]

So... no, it's not the end of the world. 2.54% risk of death is not large in the grand scheme of things (looking at the planet as a whole), but it's pretty high for normal life. And remember, this is risk without potential reward, thrill, or even entertainment. On a personal level, my fatality risk acceptance is ~1% for, say, the magical granting a superpower of some kind (the alternative being 'life continues as normal'). Higher if I get to choose the superpower in question. Maybe about ~0.1-0.01% if I only level up. (Those who know me know that I am generally not at all risk-averse. For instance, when playing video games, when you don't die in real life.)

Seriously, wouldn't you rather go skydiving instead? Yes, go skydiving about 50 times a year for the rest of your natural life (assuming you're 25 now...)? Or go skating on thin ice? If there's a medical team within reach, your risk of death from hypothermia is relatively low.

Don't. Panic. But go wash your hands.

Current music: t.A.T.u. - Белый Плащик

1 comment:

  1. don't you think the 2.54% figure is high? 58 deaths out of 2282 cases that were serious enough for the patient to see a doctor. surely there are a lot of mild cases that don't end up in the talley (although perhaps by this point everyone in mexico with a runny nose is running to the hospital). i'd guess the true death rate is under 1%.

    the 2.5% for 1918 is from 50 million dead (more or less known) divided by 2 billion infected (statistical extrapolation), right? i think the denominator (both then and now) has a pretty large uncertainty.

    that said, even .5% is nothing to sneeze at. (pause for audience groan) i'll wash my hands.