How Deadly is H1N1 Influenza?

On January 15, the CDC released their estimate for the number of Americans who died from the H1N1 pandemic. They calculate that between April and December 12, 2009 the virus killed around 11,160 Americans and infected about 55 million.

Media have widely reported that seasonal flu epidemics kill about 36,000 Americans a year, so I was surprised to see this low death count for the 2009-2010 H1N1 pandemic. The CDC’s flu-tracking website shows that pediatric death rates from H1N1 are double to triple the rates from seasonal flu earlier this decade, and this year’s number of doctor visits and hospitalizations for flu is higher than it has been in previous years. While I didn’t expect a huge increase, I certainly wasn’t expecting 25,000 fewer flu deaths this year!

Why is the 2009-2010 H1N1 pandemic less deadly than prior flu pandemics? Is it a statistical fluke, given how difficult it is to estimate every single person who died from flu and flu-related complications? Or is the H1N1 pandemic truly less lethal than prior epidemics?

Some might argue the lower death rate is because the epidemic has not yet run its course, and many more people will die this spring. However, we have seen a peak and cases are now declining dramatically. So while early reporting might inaccurately exclude a few thousand later deaths, the CDC’s estimates are not entirely premature for the purpose of comparison to peaks in prior years.

One possible factor lowering the death rate is the surprising finding that H1N1 spares the elderly. Flu typically strikes people over the age of 65 the hardest, but not this year. The majority of H1N1 cases are in people younger than 65, and deaths are unusually skewed towards younger patients. In most years, 90% of flu-related deaths are in people older than 65, but this year, they only account for 10% of the deaths.

One study found that about a third of people older than 60 already had antibodies to H1N1, compared to only 6% of young adults. This suggests adults were exposed to a virus similar to H1N1 in the 1950’s and retain lifelong immunity, thus limiting the overall mortality. A large portion of the missing 25,000 deaths could be immune older adults who survived the flu season without a sniffle.

Another possibility to explain the lower death count is the vagaries of number-crunching. It’s difficult to estimate the number of people who die from flu. There is no accurate and complete nationwide flu-tracking system, the CDC uses data from only ten states to develop their estimates. Educated guesswork accounts for underreporting, and data from these states is extrapolated the entire nation.

Flu cases and deaths are under-reported, as it’s difficult to define what a flu-related death actually is. If someone is weak from the flu, falls out of bed striking their head, and dies from a bleed in their brain, does that count as a flu-related death? What about a person who catches the flu and has a heart attack? (Flu is known to increase the risk of heart attacks, just like everything else.) A doctor might only record heart attack on the death certificate, unaware that the patient had been suffering from the flu. Assuming the doctor does report the patient died of flu, this information can be recorded in different places—for example, the death certificate or the patient’s chart.

Given these difficulties, statisticians have access to imperfect data and have to make several assumptions. The data used and assumptions made will vary from study to study. The study that determined 36,000 annual deaths from flu used data from the National Center for Health Statistics, which compiles death certificate information from previous years but not the current year. The current CDC estimate of 11,000 deaths uses rawer data from the Emerging Infections Program. Given the different methodologies, perhaps comparing these two numbers is like comparing apples to oranges.

Comments