Study findings

This section contains discussion of the findings of the study. For comparisons of the study results with other estimates of mortality in Iraq, please see this section.

1: What were the main study findings?

The following is a summary of the main findings of the MIT/Bloomberg mortality study.

Total mortality has increased substantially

The study found that 650 thousand excess deaths have occurred in the period after the invasion in March 2003 to June 2006.

Excess deaths means these deaths would not have occurred if the mortality rate had not increased from its pre-invasion level. Absent other influences on the mortality rate, they therefore are attributable to the invasion and its aftermath.

The number of excess deaths is estimated statistically, using a technique similar to opinon polls, and the true number is 95% certain to lie in the interval 390-940 thousand (the "95% confidence interval"). The true number is likely to be near the centre of this interval than at the extremes, and it is equally that it is higher than 650 thousand and lower than 650 thousand.

The study found a pre-invasion mortality rate was 5.5 per 1,000 people, compared to 13.3 per 1,000 people in the period March 2003-June 2006.

Most excess deaths were caused by violence

Overall deaths were split relatively evenly between violent and non-violent causes, but excess deaths were overwhelmingly attributable to violent causes.

Of violent deaths, more than half (56%) were from gun shots. Car bombs, other explosion/ordnance, and air strikes accounted around 13% of deaths each, with the remaining deaths attributable to unknown causes or accidents. The study sample was not large enough to establish with statistical precision the number of deaths attributable to a particular cause, and the true proportion of deaths from a particular cause therefore potentially could be either higher or lower within a relatively wide range.

Just under one third of deaths (31%) were attributed to coalition violence (95% confidence interval: 26-37%).

There was a trend of increasing violence and mortality

The study also found that death from violence has been increasing dramatically each year since 2003. The first after the invasion was the least violent in the period, with an excess mortality rate of 2.6 in 2003-2004 (corresponding to just over 100,000 deaths until September 2004). The mortality rate increased to 5.6 and 14.2 in 2004-2005 and 2005-2006, respectively.

Deaths were concentrated among adult men

Three quarters of the deaths (77%) were in males. Just under ten times as many men as women had died from violence , compared to 3.4 times as many men dying overall (from all causes). Men in the 15-44 age group had particularly pronounced increase in mortality, approaching the mortality rates in the 60+ age group.

Like the causes of deaths, these findings are very uncertain. Estimates of number of deaths in a particular sub-group of the sample (e.g., males of a particular age groups) are imprecise, meaning that a wide range of true mortality rates are consistent with the findings.

2: Is the estimated pre-war mortality rate consistent with other information?

Summary: the calculation of excess deaths depends on an accurate estimate of pre-war mortality. The figure in the Lancet study is consistent with other recent estimates for Iraq, as well as estimates for other countries in the region.

Some commentators have expressed concern that the pre-war crude mortality rate of 5.5/1000 found by the study may not be consistent with other estimates. If the number estimated were lower than the true number, the number of excess deaths would be over-estimated, while a higher number would lead to an under-estimate of excess deaths. For details of how excess deaths are calculated, please see this FAQ

Fred Kaplan raised these issues in in an article in Slate:

"The Lancet study's base-line number is dubious ... according to data from the United Nations, based on surveys taken at the time, Iraq's preinvasion mortality rate was 10 per 1,000. ... It is also noteworthy that, if Iraq's preinvasion mortality rate really was 5.5 per 1,000, it was lower than that of almost every country in the Middle East, and many countries in Western Europe.

This criticism raises four main issues:

First, the statement implies that the figure is implausible because it is lower than rates in Western Europe. However, this is based on a misunderstanding of the variable measured. The figure refers to the crude mortality rate, or total number of deaths per 1,000 population, which is sensitive not only to health but also to the demographic structure of the population. For example, Sweden has a "high" crude death rate of 10.3/1,000, reflecting the fact that one-sixth of the population is over 65 years of age and old people are more likely to die than young people; by contrast, (much poorer) Algeria has a death rate of 4.6/1,000 and 4.8% of the population over 65 years of age).

Second, the 5.5/1,000 mortality figure is in fact similar to other estimates of mortality rates. These include:

There clearly is uncertainty about the exact pre-war crude morality rate in Iraq, but other estimates range between 5 and 6, and none is as high as 10/1,000.

Third, it is reasonable to check that the estimated crude mortality rate is similar to that for other countries in the region with similar age structure and income levels. This is borne out by data; for example, countries including Egypt, Jordan, and Syria all have a death rate of 5 in the US Census Bureau Report.

Fourth, while it is useful to evaluate the mortality finding against collational evidence, it is important that the collational evidence itself is reliable. The UN data are not in fact based on "surveys taken at the time", as stated. Instead, most demographic information in the UN report cited above is based on data from a census carried out in 1987, along with some modelling and projection of likely trends (link. No specific source is stated for the crude mortality rate figure, and the source therefore cannot be evaluated. By contrast, the Lancet study is a rare example of a study explicitly designed to measure crude mortality in Iraq, employing a methodology developed for this purpose.

3: How does the study attribute deaths to the coalition?

The study found that just under one-third (31%) of violent deaths were attributed to the coalition. The true number is 95% certain to be in the range 26-37%.

These figures were derived by asking respondents about the causes of deaths, and classifying those that were "directly attributed to coalition forces or to air strikes" as coalition violent deaths. The proportion of violent deaths attributed to the coalition was much the same over the study period, although the total number of deaths increased sharply over the period, implying that the tota number of deaths from coalition violence increased each year from 2003.

These figures are correct to the extent that respondents were able to identify the cause of death accurately. This may be uncertain in some cases. In particular, there is likely to be more uncertainty about the cause of death than about the incidence of death in the first place.

The study made no attempt to define categories such as "civilian", "insurgent/combatant", or "military" deaths.

4: How many of the excess deaths were civilian?

The survey used for the study recorded the cause of death but made no attempt to separate the deaths of civilians from military personnel, insurgents, or other potential classifications. Unlike some other estimates, the MIT/Bloomberg study records total deaths.

5: If Iraq is very violent, why is there not more emigration?

Some commentators have suggested that the study findings are incompatible with migration patterns:

...if the violence really was as bad as the Lancet studies say, there should be massive refugee flows outward, but the UNHCR numbers do not show this. link

This contention does not appear to be borne out by the data, which show that a very large number of people have been displaced within Iraq, consistent with the very violent environment found by the study.

A UNHCR briefing note from 13 October 2006 states that:

"The government of Iraq, UNHCR and its partners estimate there are now more than 1.5 million people displaced within Iraq itself" ...

We estimate that up to 1.6 million Iraqis are now outside their country, most of them in Jordan and Syria ...

Inside Iraq, the government estimates that up to 50,000 people are leaving their homes every month

Similar observations have been made by the International Organization for Migration, which states that "Numbers of Displaced Rise Inexorably"

The Iraqi government also has stressed that the number of internally displaced persons as well as refugees to other countries is large:

Some 890,000 Iraqis have moved to Jordan, Iran and Syria since Husseinís fall, [Iraqi Immigration Minister Abdul-Samad] Sultan said. More than 300,000 have fled to other parts of Iraq, he added.

The authors of the study discuss how such factors may affect the estimates of the study:

Large-scale migration out of Iraq could affect our death estimates by decreasing population size. Out-migration could introduce inaccuracies if such a process took place predominantly in households with either high or low violent death history. Internal population movement would be less likely to aff ect results appreciably. However, the number of individual households with in-migration was much the same as those with out-migration in our survey. (p. 1427)