Gauging War's Toll on Iraq's Population

Epidemiologists Gilbert Burnham and Les Roberts found themselves at the center of a media maelstrom last October when their study, published in The Lancet, estimated 654,965 excess Iraqi deaths since the start of the war. The number of "excess deaths," above and beyond the pre-war death rate, overshadowed estimates from all official sources and drew instant ire from politicians and pundits. "I don't consider it a credible report," President George W. Bush announced, holding with the Iraqi government's figure of 40,000 non-military deaths since the start of the U.S.-led invasion. British Prime Minister Tony Blair also dismissed the finding.

Through it all, Burnham, MD, PhD, MSc, remained unfazed. The critiques, he says, come from "people who didn't like the numbers and who found the methodologies difficult to understand." Moreover, he contends that the estimate, derived from a population-based survey, has provided the most realistic picture yet of the war's effects.

Recent news reports lend credence to Burnham's contention: In late March, British newspapers reported that the country's chief scientific advisor to the Ministry of Defence had concluded on October 13 that the survey's methods were "close to best practice" and that the study design was "robust."

Burnham, who is co-director of the School's Center for Refugee and Disaster Response, is quick to point out there was nothing novel about the team's techniques. "Cluster surveys" are employed routinely in developing countries and after disasters in the U.S. to reveal health trends like immunization coverage and utilization of services. They gather specific information from randomly chosen sites, which are then converted to rates and applied to the larger population. The approach presented a viable option for studying civilian mortality amid the chaos of war, when information systems break down and hospital, mortuary and morgue reports become less reliable. Designed in collaboration with the Bloomberg School's Scott Zeger, Shannon Doocy, Elizabeth Johnsonand Stan Becker, and a faculty member from Al Mustansiriya University's School of Medicine in Baghdad, the study involved 47 clusters across Iraq. At each one, interviewers surveyed 40 households, asking respondents to compare the number of family members alive in January 2002 with June 2006. If a family member had died, the researchers asked how the death had occurred and whether the household had a death certificate.

Few households refused to answer questions. "People were gratified somebody was concerned about the level of violence and the consequence to the community," Burnham says.

The real difficulty of the wartime surveys was getting answers without getting the interviewers killed. A team of eight volunteers, all local medical personnel fluent in Arabic, Kurdish and English, strategized carefully to minimize the peril. Some who had participated in an earlier study dropped out because of the increasing danger. (Roberts was in Iraq for the 2004 surveys, but Burnham was cautioned by Iraqi colleagues not to enter the country for the second study because of increased violence.) 

The numbers that turned up shocked Burnham. Ninety-two percent of deaths since the U.S.-led invasion were attributable to violent causes. The surveys also revealed the widespread geographic impact of the war. "Baghdad is actually not the most violent part of Iraq," he says. "The media has access there, and the numbers there get emphasis, but they don't give a picture of the country as a whole." (The worst violence was in the provinces of Diyala and Anbar.) Households attributed 31 percent of deaths to coalition forces. Street violence, assassinations and criminal acts accounted for most other victims. But another surprise was that, in the last year, nonviolent deaths rose above the baseline before the invasion, suggesting a decline in the health infrastructure. In the future, Burnham is interested in studying the exodus of health care workers during wartime and its effects.

The surveys' data point toward the need for better methods to count the dead in times of war. Because governments don't make tallies and groups like the Red Cross have other mandates, it would be useful, he suggests, to establish a neutral, unaffiliated group to monitor non-military casualties.

"War is the ultimate disaster that can happen to a country," Burnham says. "We have to ask: How can we protect people trapped in conflict? From the public health perspective, this begins with reliable data."

In April, Burnham and colleagues made their study data available online to academic and scientific groups for analysis. The goal, Burnham says, was "to advance the understanding of how to best assess mortality during conflicts and to improve the protection of those caught in conflict."