Death Stats For Flu Can Be Misleading, CDC Report Shows

  • Share
  • Read Later

The Centers for Disease Control and Prevention (CDC) are revising their estimates of the number of people who die each year from flu.  We need good death estimates before we can judge how well our infection-control policies are really working, researchers say — but, according to a new report released this week, earlier flu stats have failed to convey important, “substantial variability in mortality.” 

That failure might lead to wrong-headed policy — or it can simply confuse the public. All throughout last year’s swine flu pandemic, for example, the media reported that flu usually kills some 36,000 people in the U.S. every year. That number, in fact, is roughly how many people died of influenza-associated respiratory and circulatory causes in the year 1999, the CDC report shows. But that doesn’t actually mean influenza kills that many people every year. Instead, the new report shows, there’s really no such thing as a typical flu year.

Flu virulence can change a lot from year to year because the flu virus itself mutates so rapidly. (This frequent mutation is also why we can catch a flu many times in our life, and why we need a new flu vaccine every year.) This means that the number of deaths from flu can fluctuate wildly from one year to the next. The new CDC report this week gives the estimated number of influenza deaths each year from 1976 up to 2007. Back in the winter of 1986-87, in the flu season with the fewest deaths during the 30-year period, just over 3,000 people died. But in the year with the most deaths, 2003-04, there were more than 48,000 people killed. It’s a difference of nearly a factor of 16 — huge variation to see over the course of just a few years.

Over the entire 30-year period, the average number of deaths from flu was somewhere around 23,600 deaths per year — many fewer than was generally reported last year during the flu epidemic. But the point of the new report is that, for most policy or research purposes, that average doesn’t matter very much. What matters instead is the range of probable outcomes. A single estimate doesn’t cut it, the report concludes.

In any given year, the death toll seems to depend to a large extent on the strain of the virus that dominates. In years when A(H3N2) strains were common, death rates were far higher than in years when other influenza sub-types  like H1N1 dominated, the report shows.