TORONTO — Are you confused by the H1N1 numbers? Wondering why public health officials are making such a fuss about a virus that has so far killed so few people?
You aren’t alone.
After all, we’re told seasonal flu kills between 4,000 to 8,000 Canadians and between 250,000 and 500,000 people worldwide each year. Yet as of late last week, seven months into this outbreak, H1N1 had killed 161 Canadians and an estimated 6,260 people around the globe.
Critics of Canada’s pandemic response point to the discrepancy between those sets of numbers and question the full court press.
But the thing is, as tempting as it is to compare those two sets of figures and conclude that H1N1 is much ado about nada, you can’t do it. Those two sets of numbers count different things, experts say.
“You might as well compare the number of flu deaths with the number of Subarus sold in Canada,” says Jordan Ellenberg, an associate professor of mathematics at the University of Wisconsin who explained the problem in an article published online in Slate Magazine earlier this year.
“If you want to compare the number of confirmed deaths to seasonal flu to the number of confirmed deaths from H1N1, OK, you can do that,” he says in an interview. “But what you can’t do is compare the number of certified deaths on one side to the best estimate of the full number of deaths on the other side.”
Confirmed H1N1 death tallies capture the blessedly few times someone who caught this bug died from it after testing positive for it. The seasonal flu numbers are estimates, mathematical calculations aimed at capturing all the deaths influenza had a hand in.
The frequent attempts to equate the two are driving Dr. Kumanan Wilson bonkers.
Wilson is an expert in public health policy as well as an internal medicine physician at the Ottawa Health Research Institute. He readily admits he never sees anyone die of seasonal flu — a common claim that drives infectious diseases experts crazy.
Wilson is, however, seeing the destructive power of this strain of influenza.
Wilson says it is “disingenuous” to criticize the response to this pandemic by comparing the low death toll to the substantially higher estimates of seasonal flu deaths. “I feel it under plays the significance of this.”
But what is the difference between counting confirmed flu deaths one by one and estimating seasonal flu death tolls? The answer lies in the way flu kills and the way statisticians try to capture that effect.
Influenza does kill some people directly. We see it sometimes with this strain of flu; some victims develop an aggressive and ultimately fatal viral pneumonia.
But generally that number is small. A 2007 study of looking at 10 years worth of influenza deaths in Canada reported that deaths directly attributed to flu made up only eight per cent of all influenza-related deaths in a given year.
In most cases, influenza contributes to or hastens death. A bout of flu can trigger a heart attack in someone with heart disease.
But the cause of death is likely to be listed as a heart attack.
Other times flu weakens the immune system of say an 80 year old. Bacteria seize the opportunity and a pneumonia develops. The cause of death may be bacterial pneumonia, but the pneumonia wouldn’t have occurred if the person hadn’t caught the flu.
A lot of seasonal flu doesn’t get captured in official reports of deaths and hospitalizations. Not everyone dies in hospital. Hospitals don’t always test for influenza. And even if they do, if the test isn’t done early enough in the infection, the test may come back negative.
The 2007 study, led by Dena Schanzer of the Public Health Agency of Canada, estimated there were on average 12.5 deaths attributable to flu for every certified flu death from 1990 to 1999.
Public health researchers come up with figures like that by plotting curves of annual flu outbreaks against what is known, from long-term study, about how many people on average each week of the year. When you lay the two curves over each other, you see deaths rise when flu hits.
Some of the excess mortality, as it is called, may be due to other factors, like poor winter driving conditions. But some of the excess is due to influenza. And over the years, public health researchers have worked out ways to calculate the portion of those extra deaths for which flu is to blame.
“We try to isolate the amount that we can attribute to influenza, using our mathematical models,” explains Dr. Jeff Kwong, of Toronto’s Institute for Clinical Evaluative Sciences.
Some people question the tactic. But Ellenberg says it’s a valid approach.
“If somebody gets in a car crash and they bring them to the hospital and they die, you would say ’Hey, that person was killed in a car crash,”’ he explains. “And if somebody was like ’No, when they were in the car they were alive. They died of blood loss in the hospital,” that would be a little ridiculous, right?“
Often, though, there can be considerable lag time between an outbreak and the attempts to quantify its impact. It takes time for vital statistics to be processed. Kwong says it may be 2011 or so before Canadian researchers can come up with a good estimate of what has happened this year with H1N1.
The U.S. Centers for Disease Control decided not to wait.
Last week they released new estimates of the toll H1N1 has taken in the United States. Their calculations, the methods for which are explained on the CDC’s website (http://www.cdc.gov/h1n1flu/estimates—2009—h1n1.htm), nearly quadrupled their estimate of how many Americans have died from the virus. They now peg it at 3,900.
Dr. Anne Schuchat, director of the center for immunization and respiratory diseases, says the agency knew focusing on lab confirmed cases was painting a “very incomplete” picture of the problem.
Canada’s Chief Public Health Officer, Dr. David Butler-Jones, doesn’t think the discrepancy will be as big here. “No test will capture 100 per cent (of cases). But certainly we believe we’re much closer to the actual number here in Canada.”
But influenza expert Dr. Allison McGeer says she isn’t sure how many cases are being missed. She knows the tests aren’t sensitive enough to pick up all the cases. But are they missing 15 per cent or 35 per cent? She can’t tell.
Harvard epidemiologist Marc Lipsitch, a modelling expert, isn’t sure how much it matters at this point.
“If all this effort goes to save hundreds or a few thousand lives, that’s still hundreds or a few thousand lives that have been saved,” he says. “If these are preventable deaths, which I think is pretty clear many of them are, we should try to prevent them.
———