A survey of surveys of Darfur mortality since 2003 in the latest issue of The Lancet, by the reputable analysts Olivier Degomme and Debarati Guha-Sapir, provides the most reliable estimates yet for not only the extent, but the pattern of mortality in Darfur since September 2003.

The authors find that the pattern of deaths “matched well with previously described general mortality patterns in complex emergencies,” giving further confidence in their conclusions. It shows that in the initial period, the majority of deaths were attributable to violence, but that this rapidly fell away, with excess deaths attributable to disease taking much longer to fall, and accounting for about 80% or so of the overall excess death toll.

The total deaths attributable to violence are estimated at 62,305 (95% confidence interval 32,229 to 139,142) and the total, including those attributable to hunger and disease at 211,000-298,000 depending on the baseline. The lower baseline mortality assumption (derived in part from a survey I did in 1986) gives a mean of 298,271 (95% CI: 178,000- 462,000). If a more standard baseline mortality for sub-Saharan Africa is used it is 210,607 (95% CI 91,000-374,000).

The survey is a significant update on earlier surveys insofar as it includes data up to the end of 2008. It finds an increase in general mortality during the period June 2006-September 2007, which explains the increase in the estimated total excess deaths in comparison to earlier estimates. After that period, the death rates drop down to levels that are (depending on baseline assumptions) slightly above or slightly below ‘normal.’

The report includes estimates of the breakdown of mortality rates between north, south and west Darfur, and makes some interesting comparisons between displaced persons and residents. The next stage in analysis, which would be more challenging given the data sources, would be to break it down by ethnic or tribal group. For example, the data for violent fatalities for the last two years show that more Arabs than non-Arabs have died in violence, which was clearly not the case in 2003-04. It would be interesting to investigate if there is a comparable change in patterns of general mortality.