Deaths Revealed by Wikileaks Are Tip of Iceberg

The documents on the U.S. War in Iraq published by Wikileaks contained data on 15,000 Iraqis killed in incidents that were previously unreported in the Western media or by the Iraqi Health Ministry and, therefore, not counted in compilations of reported Iraqi war deaths by Iraqbodycount.org. The Western media are dutifully adding these 15,000 deaths to their so-called "estimates" of the total numbers of Iraqis killed in the war.

This is deceptive. What the unreported deaths really demonstrate is that the passive methodology of these body counts is a woefully inadequate way to try and estimate the number of deaths in a war zone. These 15,000 deaths are only the tip of an iceberg of hundreds of thousands of unreported Iraqi deaths that had already been detected by more serious and scientific epidemiological studies. Unfortunately, the U.S. and British governments have successfully suppressed these studies by confusing the media and the public about their methods and accuracy.

There is nothing unusual about such large numbers of deaths being unreported in a war zone. It bears out the experience of epidemiologists working in war zones around the world that "passive reporting" of war deaths generally only captures between 5 to 20 percent of actual deaths. This is partly a result of the changed nature of modern war: about 86 percent of the people killed in the World War I were uniformed combatants, whose identities were meticulously recorded, while 90 percent of the people killed in recent wars have been civilians, making counting and identifying them much more difficult.

In my book, Blood On Our Hands: the American Invasion and Destruction of Iraq, I discuss the various efforts to count the dead in Iraq. The interim Iraqi government's Health Ministry started collecting civilian mortality figures from hospitals in 2004 and, in June that year, started separating the figures for people killed by resistance forces from those killed by U.S. and other occupation forces. Knight Ridder correspondent Nancy Youssef was given the figures for the period between June 10 and September 10, 2004 and covered them in an article on September 25, 2004 that the Miami Herald titled "U.S. attacks, not insurgents, blamed for most Iraqi deaths."

During this three-month period, the Health Ministry counted 1,295 Iraqis killed by the occupation forces and 516 killed in what the ministry called terrorist operations, but it agreed with hospital officials who told Youssef that these figures only captured part of the death toll. The Centcom press office refused to provide her with an alternative estimate, although it admitted that the U.S. command did have one, while the International Committee of the Red Cross told her it didn't have sufficient staff in Iraq to compile such information.

Youssef questioned whether some of the Iraqis counted as killed by the occupation forces might have been resistance fighters, but Dr. Shihab Jassim of the Health Ministry's operations section told her the Ministry was convinced that nearly all were civilians, because a family member wouldn't report it to the occupation-controlled Health Ministry if his or her relative died fighting for the Mahdi Army or other resistance forces. This view was corroborated by Dr. Yasin Mustaf, the assistant manager of al-Kimdi Hospital in Baghdad: "People who participate in the conflict don't come to the hospital. Their families are afraid they will be punished. Usually, the innocent people come to the hospital. That is what the numbers show."

Dr. Walid Hamed, another Health Ministry official told Youssef, "Everyone is afraid of the Americans, not the fighters. And they should be." Another doctor she spoke to had lost his three-year-old nephew in a checkpoint shooting and a doctor at the Baghdad morgue told her about a family of eight who were all killed by a helicopter gun-ship after they went up to sleep on their roof to escape the summer heat. Overall, officials attributed the high numbers of civilians killed by occupation forces primarily to air strikes rather than to shootings by ground forces.

Also in September 2004, an international team of epidemiologists, led by Les Roberts and Gilbert Burnham from Johns Hopkins School of Public Health and Drs. Lafta and Khudhairi of Al Mustansiriya University in Baghdad, conducted the first of two more scientific studies of mortality in Iraq. This one covered the first 18 months of the war. Roberts had worked with a joint team from the Center for Disease Control and Doctors Without Borders in Rwanda in 1994 and had conducted similar studies in war zones around the world. Mortality estimates he produced in the Democratic Republic of Congo (DRC) in 2000 were widely cited by American and British leaders and the UN Security Council drafted a resolution demanding the withdrawal of all foreign forces from the DRC following that report.

In Iraq, the epidemiologists found that, "Violent deaths were widespread...and were mainly attributed to coalition forces. Most individuals reportedly killed by coalition forces were women and children.... Making conservative assumptions, we think that about 100,000 excess deaths or more have happened since the 2003 invasion of Iraq. Violence accounted for most of the excess deaths and air strikes from coalition forces accounted for most violent deaths." Their report was published in the Lancet, the British medical journal, in November 2004.

There was nothing surprising in their conclusions in light of the already existing evidence that "coalition" air strikes had killed thousands of civilians, both during and after the invasion. However, their report was quickly dismissed by the American and British governments. The American media, following their tradition of deference to U.S. officials, took their cue from the government and more or less ignored the study. Following the publication of the epidemiological team's second study in 2006, which garnered a bit more media attention, President Bush said only, "I don't consider it a credible report."

The cynicism of these official dismissals was eventually exposed by yet another set of leaked British documents. On March 26, 2007, the BBC published a memo from Sir Roy Anderson, the chief scientific adviser to Britain's Ministry of Defence, in which he described the epidemiologists' methods as "close to best practice" and their study design as "robust." These documents included memos sent back and forth between worried British officials saying things like, "Are we really sure the report is likely to be right? That is certainly what the brief implies." Another official replied, "We do not accept the figures quoted in the Lancet survey as accurate," but added, in the same e-mail, "the survey methodology used here cannot be rubbished, it is a tried and tested way of measuring mortality in conflict zones."

The methodology that the British officials were referring to was a "cluster sample survey," the same type of study that Les Roberts had conducted in the Democratic Republic of Congo in 2000. Prime Minister Blair had publicly cited that study's figures to the 2001 Labour Party Conference to justify British policy in Africa, but he dismissed the study in Iraq, telling reporters in December 2004, "Figures from the Iraqi Ministry of Health, which are a survey from the hospitals there, are in our view the most accurate survey there is." This was interesting in light of Youssef's report. Blair dismissed the overall numbers in the Lancet report, but avoided the even more sensitive question of who killed all these people, on which the Health Ministry and the epidemiologists were in total agreement.

The Western media widely cited the Iraqi Health Ministry and Iraqbodycount.org as sources for civilian mortality figures, but these both used a passive methodology to count deaths, essentially adding up deaths that had already been reported either in hospital records or in Western media accounts. Epidemiologists working in other war zones over the past 20 years have typically found that such passive methods only capture between 5 percent and 20 percent of actual deaths. That is why they have developed the cluster sample survey method to obtain a more accurate picture of the deadly impact of conflicts on civilians.

Cluster Sample Details

The cluster sample survey method used in war zones was adapted from epidemiological practice in other types of public health crises, surveying a representative sample of a population by clusters to estimate the full extent of a health problem that affects the whole population. As Roberts pointed out, "In 1993, when the U.S. Centers for Disease Control randomly called 613 households in Milwaukee and concluded that 403,000 people had developed Cryptosporidium in the largest outbreak ever recorded in the developed world, no one said that 613 households was not a big enough sample. It is odd that the logic of epidemiology embraced by the press every day regarding new drugs or health risks somehow changes when the mechanism of death is their armed forces."

In Iraq in September 2004, the epidemiological teams surveyed 988 households in 33 clusters in different parts of the country, attempting to balance the risk to the survey teams with the size needed for a meaningful sample. Michael O'Toole, the director of the Center for International Health in Australia, said, "That's a classical sample size. I just don't see any evidence of significant exaggeration.... If anything, the deaths may have been higher because what they are unable to do is survey families where everyone has died."

Beyond the phony controversy in the media regarding the methodology of these epidemiological studies, there was one significant question regarding the numbers in the 2004 study. This was the decision to exclude the data from a cluster in Fallujah due to the much higher number of deaths reported there (even though the survey was completed before the final assault on the city in November 2004). Roberts wrote in a letter to the Independent, "Please understand how extremely conservative we were: we did a survey estimating that 285,000 people have died due to the first 18 months of invasion and occupation and we reported it as at least 100,000."

The dilemma they faced was this: in the 33 clusters surveyed, 18 reported no violent deaths (including one in Sadr City), 14 reported a total of 21 violent deaths, and the Fallujah cluster reported 52 violent deaths. This last number is conservative for the reason Michael O'Toole highlighted. As the report stated, "23 households of 52 visited were either temporarily or permanently abandoned. Neighbors interviewed described widespread deaths in most of the abandoned homes but could not give adequate details for inclusion in the survey."

Leaving aside this last factor, there were three possible interpretations of the results from Fallujah. The first, and indeed the one the epidemiologists adopted, was that the team had randomly stumbled on a cluster of homes where the death toll was so high as to be totally unrepresentative and therefore not relevant to the survey. The second possibility was that this pattern among the 33 clusters, with most of the casualties falling in one cluster and many clusters reporting zero deaths, was an accurate representation of the distribution of civilian casualties in Iraq under "precision" aerial bombardment. The third possibility, which effectively incorporated the other two, was that the Fallujah cluster was atypical, but not sufficiently abnormal to warrant total exclusion from the study, so that the real number of excess deaths fell somewhere between 100,000 and 285,000.

In each case, however, these figures were only the mid-point of a statistical range, leaving considerable uncertainty over the actual number of deaths. The epidemiologists found, with 95 percent certainty, that the excess number of deaths as a result of the war, excluding the 3 percent of the country represented by the cluster in Fallujah, was somewhere between 8,000 and 194,000. In itself, this was hardly a solid or satisfactory conclusion. However, it was very unlikely that the actual number of dead was close to either of those extremes and there was a 90 percent likelihood that it was more than 44,000.

The Fallujah cluster, statistically representing the most devastated 3 percent of the country, reported 52 of the 73 total violent deaths in the survey. Even if this was not a perfect representation of the distribution of violent deaths, these parts of the country by definition suffered considerably worse than other areas, and yet the published estimate of about 100,000 violent deaths effectively counted zero violent deaths in these areas. The survey team that visited Fallujah reported that "vast areas of the city had been devastated to an equal or worse degree than the area they had randomly chosen to survey," so that the area chosen did in fact appear to be representative of many severely bombed areas. One could therefore arrive at the estimate of "about 100,000 excess deaths or more" by looking at the survey data in a number of different ways, which made the authors very confident in their interpretation. There were other conservative biases built into the study, such as ignoring empty and bombed-out houses, as Michael O'Toole pointed out, but no serious criticisms were made that would account for a significant over-estimate of deaths resulting from these methods. The main criticism made by politicians and journalists was that these studies produced higher estimates than passive reporting, but that is exactly what one would expect.

One larger survey that did produce lower civilian mortality figures was the Iraq Living Conditions Survey (ILCS). This survey was conducted by the Coalition Provisional Authority's Ministry of Planning and Development Cooperation in April and May 2004 and it was published in May 2005 by the UN Development Program. The "UNDP" imprimatur and the large sample size gave credence to its reassuringly low figure of about 24,000 "war deaths."

However, its estimate of war deaths was derived from a single question posed to families in the course of a 90-minute interview on living conditions conducted by officials of the occupation government. By contrast, the mortality studies published in the Lancet were designed with the sole purpose of obtaining accurate mortality figures, and included extensive precautions to guarantee the anonymity of the respondents and to reassure them of the independence of the survey teams.

Jon Pederson, the Norwegian designer of the ILCS, said that its mortality figures were certainly too low. Survey teams that returned to the same houses and inquired only about child deaths found almost twice as many as in the main survey. This suggested precisely the reluctance to report violent deaths that Roberts and his colleagues sought to overcome by stressing their impartiality. In April or May 2004, a question regarding "war-deaths" could still be interpreted to refer only to the invasion itself, as opposed to the long guerilla war that followed it. This interpretation is supported by the fact that more than half the deaths reported in the ILCS were in the southern region of Iraq, which bore the brunt of the invasion but was later more peaceful than other regions.

In January 2005, the health ministry provided the BBC with a summary of its hospital survey for the previous six months which painted a similar picture to the one given to Youssef in September. It counted 2,041 civilians killed by U.S. forces and their allies, and 1,233 killed by so-called insurgents. After the BBC broadcast these figures all over the world, it received a call from the Health Minister of the occupation government claiming that his ministry's report had been misrepresented and that the number of deaths attributed to the occupation forces was not accurate. The BBC issued a retraction, and the Health Ministry stopped providing breakdowns of its figures that attributed any responsibility for civilian deaths to the occupation forces.

Another actual nationwide count of civilian deaths was published by a group called Iraqiyun on July 12, 2005. Iraqiyun was an Iraqi humanitarian group headed by Dr. Hatim Al-Alwani and affiliated with the political party of Interim President Ghazi Al-Yawer. It counted 128,000 actual violent deaths, of whom 55 percent were women and children under the age of 12. The report specified that it included only confirmed deaths reported to relatives, omitting significant numbers of people who had simply disappeared without trace amid the violence and chaos. It was highly unlikely that an effort like this to actually count every one of the dead could result in anything but a significant undercount, for the reasons already discussed.

Update in 2006

Between May and July 2006, Roberts, Burnham, and Lafta led a second epidemiological study in Iraq to update their estimate of at least 100,000 deaths between March 2003 and September 2004. They increased their sample size to 1,849 households, comprising 12,801 individuals in 47 clusters. They were now surveying the results of 40 months of war. These factors enabled them to narrow the statistical range of their results. This time they were able to say with 95 percent certainty that between 426,000 and 794,000 Iraqis had died violent deaths as a consequence of the war. Their best estimate was that there had been about 655,000 excess deaths, of which about 600,000 were violent deaths. The finding of the earlier survey that at least 100,000 Iraqis had been killed by October 2004 was validated, with a new estimate of 112,000 excess deaths for that period. This also validated the conservative assumption that the Fallujah sample was unusual but not irrelevant.

They also found some changes in the pattern of violent deaths. Gunfire was now the most common cause of death overall and "the proportion of deaths ascribed to coalition forces had diminished in 2006, although the actual numbers have increased every year." Their overall conclusion, however, was that, "The number of people dying in Iraq has continued to escalate."

This overall trend was extremely disturbing, with each period accounting for more violent deaths than the one before and a proliferation in types of violence over time. Air strikes now accounted for only 13 percent of total violent deaths, but were still responsible for the deaths of about half of all the children killed in Iraq, underlining the inherently indiscriminate nature of air-launched weapons. There had been huge increases in violent deaths among males between the ages of 15 and 44, now accounting for 59 percent of all violent deaths, but the epidemiologists decided not to try to differentiate between combatant and non-combatant deaths. With much of the population now involved in armed resistance to the occupation, they felt that asking questions about this could put the survey teams at greater risk and that responses would not be reliable in any case.

Households attributed 31 percent of violent deaths to coalition forces, which would result in an estimate of at least 180,000 people killed directly by American and other foreign occupation forces. However, the report noted that, "Deaths were not classified as being due to coalition forces if households had any uncertainty about the responsible party; consequently, the number of deaths and the proportion of violent deaths attributable to coalition forces could be conservative estimates." Also, Iraqi forces recruited and trained by U.S. forces and under overall U.S. command played an increasing role in the war, in particular in the reign of terror launched in Baghdad in May 2005. These forces were responsible for the summary executions of thousands of young men and teenage boys, but those deaths were not attributed to "coalition" forces in this survey.

Two more studies of mortality in Iraq were published in January 2008. The first was the Iraq Family Health Survey (IFHS), which was conducted by the same group (COSIT) that conducted the CPA's Iraq Living Conditions Survey in 2004. This study focused exclusively on the death toll, with some cooperation from the World Health Organization and was published in the New England Journal of Medicine. It surveyed deaths only up to June 2006, to provide a comparison with the second survey by Roberts, Burnham, and Lafta. Although it also found evidence of a huge increase in the death rate since the invasion, the IFHS produced a much lower estimate of about 150,000 violent deaths.

Unfortunately, there are several reasons to doubt the accuracy of this lower figure. Like the ILCS in 2004, this survey was conducted by employees of a government that was taking part in the violence it was attempting to quantify. This predictably leads to underreporting. Secondly, its estimate of the pre-invasion death rate for 2002 is about one-third of the official death rate recorded by the World Health Organization. Thirdly, it found no increase in the violent death rate from year to year between 2003 and 2006. Every other data set available, from mortality studies to the Pentagon's statistics on violence in Iraq, showed increases in violence each year. Fourth, it found that only one in six post-invasion deaths was due to violence, compared with a majority of deaths due to violence in the other epidemiological studies and in independent surveys of graveyards.

A fifth factor that surely contributed to the IFHS's low mortality figure was that it was unable to survey mortality in the most dangerous 11 percent of the country. It attempted to compensate for this based on the regional distribution of violent deaths in Iraqbodycount.org (IBC), a record of deaths compiled from international media reports. However, because the unsurveyed areas were also the most dangerous for Western reporters, IBC inevitably undercounted deaths in these same areas. Yet IFHS used this distorted distribution pattern based on passive reporting to estimate deaths in the deadliest parts of the country.

The other survey published in January 2008 was conducted in August and September 2007 by Opinion Research Business (ORB), a British polling firm, in conjunction with Iraq's Independent Institute for Administration and Civil Society Studies. They surveyed 2,414 households and asked them if they had lost a member or members of the household to violence since the invasion. They were unable to survey three provinces (Anbar, Karbala and Irbil) and most of the 8 percent of households who refused to answer were in Baghdad, where death-rates were among the highest. These factors contributed a conservative bias to their estimate. In spite of this, ORB found that about 20 percent of households surveyed had lost at least one member, and estimated that 1.03 million people had died in the war. Without compensating for the conservative biases mentioned above, their data and sample size gave them 95 percent certainty for a number of deaths between 946,000 and 1.12 million.

After the publication of the second epidemiological study in the Lancet, the scale of violent death it revealed was gradually acknowledged among educated circles in the West, including in the United States. The ORB survey provided independent confirmation of the scale of the violence. It also suggested that deaths had continued to increase for at least another year after the publication of the second study in the Lancet and that the death toll probably now exceeded a million violent deaths.

The work of all these researchers showed that the United States and other modern governments could not unleash this kind of violence on another country without eventually facing the consequences of public awareness of the nature and magnitude of its effects. And, although U.S. officials may never publicly acknowledge it, the publication of these studies probably served to restrain some of their most violent impulses in the conduct of the war.

Z

Nicolas J.S. Davies is the author of Blood On Our Hands: the American Invasion and Destruction of Iraq (Nimble Books, 2010). Much of this article is a partial excerpt from his book.