The Iraq deaths study was valid and correct
Last week, the medical journal The Lancet published the findings of an important study of deaths in Iraq. President George Bush and Prime Minister Howard were quick to dismiss its methods as discredited and its findings as not credible or believable. We beg to differ: the study was undertaken by respected researchers assisted by one of the world's foremost biostatisticians. Its methodology is sound and its conclusions should be taken seriously.
Professor Gilbert Burnham and colleagues from the Johns Hopkins Bloomberg School of Public Health in Baltimore and Al Mustansiriya University School of Medicine in Baghdad measured deaths in Iraq between January 2002 and July 2006. They surveyed 12,801 individuals in 1849 households in 47 representative clusters across the country.
Their study is important in providing the only up-to-date, independent, and comprehensive scientific study of mortality after the 2003 invasion and occupation of Iraq. The study found that mortality had risen alarmingly since March 2003 and continues to rise. The number of conflict-related excess deaths, above and beyond those that would normally occur, was estimated at 655,000. While precision about such figures is difficult, we can be confident that the excess deaths were above 390,000, and may in fact be as high as 940,000. The vast majority (92 per cent) of the excess deaths were due to direct violence.
The cross-sectional household cluster sample survey method used is a standard, robust, well-established method for gathering health data. A copy of a death certificate was available for a high proportion (92 per cent) of deaths. Conservative assumptions were made about deaths of uncertain cause and about the small areas not sampled.
Except in situations of highly reliable, well-maintained, comprehensive vital statistics collection ?- clearly not the case in Iraq at present ?- such surveys have been repeatedly demonstrated to be the best method for establishing population rates for key health indicators such as deaths, disability and immunisation coverage. Where passive information collection (such as death counts in morgues or hospitals) are incomplete, as is the case in Iraq today, population-based survey methods can be expected to find higher rates ?- often considerably higher ?- but that more accurately reflect the true situation.
Conducting such a rigorous study within the constraints of the security situation in Iraq is dangerous and difficult, and deserves commendation. We have not heard any legitimate reason to dismiss its findings. It is noteworthy that the same methodology has been used in recent mortality surveys in Darfur and Democratic Republic of Congo, but there has been no criticism of these surveys.
link
