Re: Civilian deaths from weapons used in the Syrian conflict
Thank you for your comments. A number of the stated concerns seem to be based on mistaken assumptions that we hope to clarify here and that are addressed in our article and the published Appendices. We answer the comment point-by-point:
Price et al. start their commentary with the following assumption albeit incorrect: “First, the article apportions responsibility for casualties to particular perpetrator organizations.” The methodology that we adopted most definitely does not attribute responsibility for casualties to any specific perpetrator. The potential attribution of perpetrators from our data was discussed extensively during the peer-review process of the paper and judged unfeasible. In one case, a specific form of aerial bombardment - barrel bombs dropped by helicopter - is attributed in our text to government forces based on cited evidence and because, as of the date of the study, helicopter-delivered barrel bombs were not a weapon type used by non-government forces in the Syrian conflict but were used by the government. We took particular care that our analysis did not over-reach the data by attributing responsibility to perpetrator groups.
Price et al. go on to write that this is “based on a single snapshot of territorial control that ignores the numerous (and well-documented) changes in this phenomenon over time” and that this is based on a map of territorial control published by the BBC. The map is presented early in the paper with the sole intention of showing the main parties involved in the conflict. We recognise that control of areas is fluid, as we illustrate in the example of the map, and our selection of the term non-state armed group (NSAG) recognizes the fact that there are multiple, and shifting, NSAGs acting and interacting within areas of the conflict. Although we hope to see systematically territory-adjusted and time-adjusted Syrian casualty analyses in the future by Price et al., and hope that they will also examine deaths from different weapon types among different civilian demographic groups, we also expect that any attribution of deaths to specific perpetrators will be moderated by recognition of the many biases and data gaps that are known to occur in perpetrator attribution.
Price et al. express concern about the potential effects of underreporting, particularly in government controlled areas, citing as we did sources that indicate casualty reporting in Syria can be affected by who has control over a territory. In our paper we address the issue of underreporting specifically as it relates to our odds ratios findings. In the case of execution deaths, we describe the limitations affecting reporting and do not include executions in government controlled areas in models, as noted in the footnote of Table B2, because the very low reported numbers do not support statistical analysis for comparisons between children, men and women. We do report the odds of death by execution for children vs. men and vs. women in the NSAG controlled areas, which had substantial numbers of reported executions, as we do for all areas with sufficient data on deaths from a weapon type.
The commentators miss an important point here. One of the most consistent and key aspects of our analysis that seems to be overlooked by Price et al. is the similarity of the odds ratios for civilian deaths by weapons used across both NSAG controlled and government controlled areas (i.e., shells, ground level explosives and shooting, table B3 in the supplement). This is not a study to estimate total number of civilian deaths from the conflict but a study to understand the relative likelihood of death by weapon type. As such the point is not whether underreporting exists, but rather whether associated bias may exist. We show that, even if under-reporting might have been more frequent in government controlled areas, the odd ratios of civilian deaths by similar weapons provided results consistent with those in NSAG controlled areas. This reinforces the main point of our paper, which is that explosive weapons used in populated areas are extremely dangerous for civilians, especially for children and women. We acknowledge that more research should be done on this area and we hope that Price et al., and other researchers, will soon re-analyse this topic in-depth.
Whilst International Humanitarian Law (IHL) recognises that non-combatants might be killed or injured in a conflict, there is a duty on combatants, whoever they are, to minimise such death and injuries. Drawing attention to the impact of certain weapons on vulnerable population groups, for example the relative dangers of explosive weapons to children in a war, should stimulate a debate on whether sufficient attention is given to this aspect of IHL. Seeking to use our paper to attribute responsibility misses the point - the paper highlights the high cost that non-combatants are paying. Whilst we leave it to others to consider whether this is so high that it breaches IHL, we believe that our main recommendation, that further thought is given to the use of explosive weapons in built up areas, is not undermined by the comments by Price et al.
As physicians and epidemiologists, we strongly reject the suggestion that we should await future political and legal accountability efforts before drawing attention to the ongoing serious impact of weapons on civilians. We also reject the suggestion that we should not draw attention to inappropriate suffering in case of its misinterpretation in what the commentators vaguely describe as a "post-conflict transition" or because it might "mislead important questions regarding culpability, reparation, reconciliation, and ultimately justice". To us, seeking to prevent the death and injury of civilians today takes priority over future legal processes which may or may not occur. Given that there is no tangible political settlement to the conflict in sight, every effort should be made to highlight and continuously monitor the human costs and abuses committed by all sides.
Jose M Rodriguez-Llanes
Fouad M Fouad
Competing interests: No competing interests