Would you estimate the percentage of movie fans in the world by polling at a cinema? Would you calculate the number of overweight Americans by taking a survey at Weight Watchers locations? Not if you were aiming for any sort of statistical accuracy; such methods would not give you samples at all representative of a larger population.

Regrettably, just this sort of sampling bias may have lead to huge errors in avian flu mortality numbers. While stony-faced newscasters somberly report that more than 50% of people infected with avian flu ultimately die, the death rate may be much lower.

A study published in the January 9th issue of Archives of Internal Medicine found that as many as 600 to 750 people in Vietnam may have come down with a mild variant of the avian flu—one that does not carry lethal consequences. This suggests that current mortality estimates, which are derived from only the most severely ill patients, are biased. We may be underestimating the virus’ transmission rate while overestimating the deadliness of avian flu.
Thorson’s work suggests that the WHO statistics do not account for all cases of H5N1 infection in humans, because many cases go unreported by patients who do not seek medical assistance for milder symptoms.

Christian Sandrock, a physician and an assistant professor at the University of California Davis Health System, as well as a specialist in public health issues, says Thorson’s study brings public attention to a hotly-debated question in the medical community: Is there a mild or asymptomatic version of avian flu?

Sandrock explains that, without serology studies to confirm a patient’s infection with H5N1, Thorson’s study does not fully answer that question.

"It is not at all conclusive," he said.

However, Sandrock thinks that the WHO’s 50% mortality figure will eventually be revised downward.

"There’s going to definitely be some people with the disease that were either asymptomatic or subclinical, that really didn’t come up on radar," Sandrock said. "I think 50% is going to be the high end. How much it’s going to drop, I don’t know. What we’d like to see is that it goes from 50% to 3%. But, I think what we may see is that it goes from 50% to 40%, or 50% to 30%, which is still a phenomenally high mortality."