From the moment the Iraq war began, the question of who is suffering and in what numbers has been a hotly contested battleground for supporters and opponents of the US decision to invade. Each side has sought to score moral points for their cause by pointing to an alleviation, or growth, in Iraqi suffering.
It's not surprising that a finding this week that 601,000 Iraqis have been killed as a result of the war is controversial. The figure, published in The Lancet, a British medical journal, is 10 times higher than previous estimates. "Not credible,'' President Bush said succinctly following its release Wednesday. "Exceeds reality in an unreasonable way,'' was the assessment of an Iraqi government spokesman.
There is no doubt that Iraq is now an extremely violent place, as Baghdad's official September murder toll of 2,667 makes clear.
But why are the number of Iraqi deaths so difficult to pin down? The short answer is that much of the country is too dangerous for researchers or government officials to travel in search of accurate statistics. The best tally would come from counting every death certificate issued in the country in the three years before and three years since the invasion. But there is no central reporting mechanism for this in the country.
So instead, the researchers, backed up by the Johns Hopkins Bloomberg School of Public Health, relied on the same polling methodology that is used to measure voter preferences or what their favorite TV shows are.
"I loved when President Bush said 'their methodology has been pretty well discredited,' " says Richard Garfield, a public health professor at Columbia University who works closely with a number of the authors of the report. "That's exactly wrong. There is no discrediting of this methodology. I don't think there's anyone who's been involved in mortality research who thinks there's a better way to do it in unsecured areas. I have never heard of any argument in this field that says there's a better way to do it."
Gilbert Burnham, lead author of the paper and a public health professor at Johns Hopkins, adds, "There are several ways of counting, one is simply counting the number of bodies that come to the morgue or reports from hospitals or in newspapers. The problem is those numbers may be valid from the site they're collected from but from nowhere else. So if you want to look at numbers that affect population as a whole, the best way is to do surveys."
The survey relied on face to face questions carried out by Iraqi researchers with members of 2,000 Iraqi families, geographically distributed to best reflect Iraq's demographics. The authors extrapolated from this number that the mortality rate per 1,000 Iraqis was 5.5 in the years immediately before the invasion, and has averaged 13.3 since. That yielded the number of 601,000 murdered and an additional 54,000 who have died of natural causes, likely due to the declining quality of healthcare.
Dr. Burnham says he and his collaborators were acutely aware of the possibility of large errors if their study wasn't well designed. "The possibility of introducing bias in any kind of survey is real, and you spend more time designing the survey to eliminate the sources of bias then actually carrying it out,'' he says. "One of the real risks in this is that people report deaths that don't occur, so we did ask for death certificates. And in 92 percent of cases, they were provided."
To be sure, the researchers of the Lancet study says possible errors leave a range between a low of 392,979 additional deaths and a high of 942,636. The 601,000 figure is the median.
The commonly accepted numbers until now have been much lower than Lancet surveys. The Iraq Body Count, a website that tracks civilian deaths in the war by compiling data from news reports, estimates 48,000 deaths have been reported in the media, while the Iraq Index sponsored by the Brookings Institution in Washington has counted 61,000 civilian deaths. President Bush estimated 30,000 civilian deaths late last year.
The Lancet report follows a 2004 survey published in the same publication and carried out by largely the same group of researchers that found 100,000 excess deaths had resulted in Iraq since the war, which was confronted with the same criticism at the time.
Dr. Burnham says that in the most recent survey, which had twice the sampling size of the original one, the numbers of deaths reported for 2004 mirrored the earlier report.
"Our key message is not to produce numbers but to ask in conflict situations, how can we think more effectively about protecting populations,'' he says. "The general pattern is as time goes along in conflicts that the host population takes more and more of the hits, which is what we're seeing here."
The survey also argues that most of the violence is now Iraqi on Iraqi. It found that 31 percent of violent deaths were claimed to be caused by American forces, a smaller proportion than in the 2004 survey but still a larger absolute number, because of the overall rise in the death toll.
Some critics of the report have charged that the research was politically motivated and that its release was timed to come shortly before upcoming US elections. Dr. Burnham says his "goal was to get this out in July or August, just so people wouldn't say this was tied somehow to elections" but that peer review and other administrative issues slowed up publication.
Dr. Garfield says that critics who charge the release of the paper was politically motivated to "assign to scientists far more conspiratorial ability than we can produce. When would have been a good time to release this? People who don't want to hear this won't ever want to hear it. Even if it came out six months ago, it would still be an inconvenient truth."
He says the most striking result of the survey, to him, is its finding that 2.5 percent of Iraq's population has died as a result of the war. "You can compare that to the civil war, our bloodiest war, in which 1.4 percent of our people died and look at what that meant to the US. Like then, what these numbers are saying is that every family is being touched."
"The numbers don't necessarily argue for an immediate withdrawal, or whatever. The point is we're not going to make good policy if we're not dealing with the true human costs."