Yonatan Zunger’s Blog

An article recently published in the Lancet estimates the number of “excess deaths” in Iraq since the invasion at 655,000, of which 601,000 were due to violence and the balance due to disease, etc. Here’s a Washington Post story; here’s one from the BBC. Unfortunately, I don’t have access to the original journal article; if someone does and could forward me a copy, I’d be very interested to read it.

[In fact: The BBC and WP stories seem to disagree on numbers a good deal, so I’d really like to see the original paper and figure out what’s going on. The WP numbers more closely match what I’ve heard from other media channels, so I’ll use those below]

There’s “controversy,” of course, because the US administration immediately decried the results as false and the methodology as incorrect — which I would personally find a bit surprising, given that poor methodology doesn’t generally get published in top-tier medical journals. The official death toll is less than 1/10th of that.

But there may be a good reason why those two numbers disagree: the official death toll is probably counting very selective types of deaths, e.g. deaths in which US personnel were directly involved, either as combatants or in cleaning up the mess. It’s an attempt to count deaths which came to the attention of US forces. The Lancet study is measuring something else: they did a random survey of 1,849 households across Iraq and asked people about deaths in their family, asking for (and routinely receiving) death certificates to verify the numbers. Based on this, they computed the overall mortality rate in Iraq in deaths per 1,000 people per year, compared this to the known mortality rate prior to the war, and thus computed the number of people who died above the number you would expect to have died had nothing happened. As a sanity check, they noted that their measurement for excess deaths in the time period immediately following the invasion does match the official number for that period fairly closely. Since US forces were more directly involved in everything that was going on then, those numbers ought to match up.

A number measured by these means is both helpful and not: on the one hand, it tells you that there is some total effect going on (which is why this sort of method is very common in epidemiology) but it doesn’t tell you what caused it. However, the second number is pretty surprising: Of the 655k excess deaths, 601k were from violence, primarily from gunshot wounds. Normally in a post-war region, I would expect excess deaths to come mostly from disease, starvation, and the like; the fact that most of these deaths were violent is pretty unusual. Perhaps it means that medicine is improving.

I noticed that a DoD spokesman said “it would be difficult for the U.S. to precisely determine the number of civilian deaths in Iraq as a result of insurgent activity.” This is an attempt to emphasize that people aren’t dying because US forces are killing them, but as a result of the insurgents, which we consider our enemies. This is true but misses the point; on the one hand, nobody was accusing the US forces of killing 655kpeople, and on the other hand, the simple presence of these insurgent forces is a direct consequence of the US invasion. In fact, the relatively low number of non-combat deaths may speak well of US activity on the ground; the absence of the other two of the Four Horsemen bespeaks some good work on keeping food and medicine flowing in a war-torn country. But the high number of overall deaths is directly
attributable to the fact that the US invaded Iraq, and the upper political command has no cover from that.

Edit: I got a copy. It looks like the WP numbers are correct; the 100k number that the BBC cites is the number from previous studies, which this paper means to update. Similarly the number of households surveyed is in fact 1849, not “under 1000.” I’ll read the paper in more detail tonight and update if there’s something interesting in it.