is the direct responsibility of the U.S. Army Corps of Engineers. The Corps has as powerful computers as anyone, but in a recent 10-year period, the Sacramento River experienced a 100-year flood three times, devastating parts of California—despite the fact that each time the Corps raised the standard for a 100-year flood. The point is not that the river exceeded the 100-year level three times in 10 years. Random chance could account for that. The point is that it did so even though the Corps changed the definition of a 100-year flood. A senior Corps official confessed that the Corps simply did not have enough data to know what a 100-year flood was.

We may be in a similar situation with influenza. We have had only three pandemics in the 20th century. That is not a good base on which to build models. Indeed, the Centers for Disease Control and Prevention’s model of what would happen in the United States should another pandemic strike predicts that the most likely death toll would fall between 89,000 and 207,000. Yet the actual death tolls of two of the three pandemics fell well outside the predicted range. Adjusted for population, 1968 deaths were somewhat fewer than the best case scenario, and 1918 nearly 800 percent worse than the worst case. (In 1918, antibiotics would likely have lessened this gap, but the increased population of those with impaired immune systems would somewhat balance that benefit, and increase deaths.)

In addition, we have not taken advantage of the data that we do have. Several presentations at this conference demonstrate that fact—some on the plus side, by deriving findings of value by reviewing records from 1918, but also on the negative side, by making certain assumptions about 1918 that conflict with actual data.

A careful review of old data would also prove valuable. Studying 1889 (and enough data can be found, possibly from earlier pandemics as well), 1918, 1957, and 1968 might tell us whether each followed the same patterns, which in turn could help us to devise strategies for the use of antivirals and vaccines.

Virtually every expert on influenza believes another pandemic is nearly inevitable, that it will kill millions of people, and that it could kill tens of millions—and a virus like 1918, or H5N1, might kill a hundred million or more—and that it could cause economic and social disruption on a massive scale. This disruption itself could kill as well.

Given those facts, every laboratory investigator and every public health official involved with the disease has two tasks: first, to do his or her work, and second, to make political leaders aware of the risk. The preparedness effort needs resources. Only the political process can allocate them.