Weapons injuries during and after periods of conflict: retrospective analysisBMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7120.1417 (Published 29 November 1997) Cite this as: BMJ 1997;315:1417
- David R Meddings, epidemiologist ()a
- Accepted 10 November 1997
Objective: To assess the relative frequency of weapon injuries during conflict and after periods of conflict in the absence of disarmament.
Design: Retrospective analysis of a database of war wounds.
Setting: Region with a protracted conflict between rival combatant groups and a subsequent transition to the uncontested military authority of a single power.
Subjects: 2332 people who received weapons injuries during the conflict or post-conflict periods and were admitted to hospital within 24 hours of injury.
Main outcome measures: Percentage change in mean monthly admission rate by weapon type between conflict and post-conflict periods; annual incidence of injury by weapon type during conflict and post-conflict periods; percentage change in annual incidence by weapon type between conflict and post-conflict periods.
Results: Mean monthly admission rates for injuries from fragmentation munitions decreased by 8% between conflict and post-conflict periods and by 23% for injuries from mines and 32% for gunshot injuries. The decline in admissions for all injuries was 23%. After adjustment for population growth over the study period, declines in annual incidence were 22% for fragmentation munitions injuries, 34% for mine injuries, and 40% for gunshot injuries. The decline in incidence for all injuries was 33%. In-hospital mortality from weapons related injuries increased from 2.5% to 6.1% (P<0.001) between conflict and post-conflict periods.
Conclusions: In this setting, continued availability of weapons is associated with increased mortality and a level of injuries from weapons that is only somewhat reduced from that observed during a period of conflict.
The availability, redistribution, and recirculation of weapons during and after armed conflicts is generating increasing unease and policy debate.1 2 In particular, there is growing concern that the continued availability of weapons in societies that have undergone armed conflicts is an important risk factor for sustained violence.3 This analysis examined the incidence of weapons injuries in a region that did not undergo disarmament but did experience a transition to control by a single military faction.
The region is situated within a country which, under the Eiseman classification,4 would be described as having an ongoing minor conventional war, and which has received massive amounts of weapons over much of the past 20 years.5 By March 1995 this region had come under the uncontested control of a faction that subsequently asserted control over most of the rest of the country. This transition was accompanied by an increasing number of returning refugees, a number of humanitarian organisations starting programs of support, and a shift in the zones of active conflict to areas hundreds of kilometres away. There was clearly no attempt at mass disarmament.
The International Committee of the Red Cross has provided surgical care to people from this region wounded in war since 1983. The principles of neutrality and impartiality by which the International Committee of the Red Cross operates preclude identifying the location of this region; disclosure of this information may have security implications for activities of the International Committee of the Red Cross in the country concerned.
Subjects and methods
Rates of weapons injury were compared between the 50 month period of January 1991 to March 1995 and the 18 month period of September 1995 to March 1997. These periods were designated as “conflict” and “post-conflict”—terms that are open to interpretation but none the less provide labels for referring to two time periods separated by an important military event. The rationale for the 6 month gap between the periods was to incorporate a transition phase between the two.
Subjects had all been injured by weapons and been admitted within 24 hours of injury during either conflict or post-conflict periods to one of two hospitals. Hospital A was an International Committee of the Red Cross hospital established in 1983 and situated adjacent to the region of interest in a neighbouring country. Hospital B, located some 200 kilometres from hospital A in the regional capital, began to receive support from the International Committee of the Red Cross in 1995 when the regional military situation stabilised. After hospital B became functional in May 1996, admissions to hospital A were phased out, ending in July. Both hospitals had about 200 inpatient beds. Hospital B was closer to the site of injury and had staff whose skills were less well developed.
An unknown proportion of people injured by weapons die before reaching hospital, and some facilities not supported by the International Committee of the Red Cross also treated people injured by weapons during both periods, but surveys indicate that hospitals A and B probably cared for at least 75% of injured people who reached hospital.
Data were collected from all patients admitted to either hospital as part of the wound database that was started by the International Committee of the Red Cross in 1991.6 Data include age, sex, type of weapon injury, delay between wounding and admission, in-hospital mortality, and variables relating to operative treatment and classification of the wound.7
A total of 2332 people were injured by weapons during conflict or post-conflict periods and admitted to hospital A or B within 24 hours of injury. The time of admission was restricted to reduce the possibility of including injured people who came from outside the region.
The wound database classifies injuries as those caused by guns, fragmentation munitions (bombs, shells, and grenades), and mines (antipersonnel or anti-tank mines). The primary analysis was a comparison of the mean admission rate during conflict and post-conflict periods for these three categories of weapons injuries and for all weapons injuries.
The annual incidence for injuries was calculated to adjust for population growth, using the population of the three provinces most likely to have been the site of injury for people who reached hospital within 24 hours. Population estimates were calculated from unpublished figures from the United Nations High Commissioner for Refugees. These figures provide numbers of refugees living in neighbouring countries by province of origin, along with annual returns by province. The projected population assumes an annual growth of 2.2%, based on historical demographic data.
Table 1 provides descriptive data pertaining to the study population. The proportion of gunshot injuries in hospital B was higher than for hospital A and mortality rose for all types of injury across periods.
The number of refugees returning each year and estimated year end population (Table 2) were used to calculate the annual incidence of injuries per 100 000 population (Table 3). Mean monthly admissions to hospital declined by 8% for injuries from fragmentation munitions, 23% for mine injuries, and 32% for gunshot injuries. Overall, mean admissions for weapons injuries declined by 23%.
Adjusting for population growth provides the declines given in the final column of Table 3. This adjustment is subject to several limitations. Population estimates assumed that historical rates of population growth remained constant, injuries were underascertained, and figures from the United Nations High Commission for Refugees account for returns but not departures from the region over the study period. Notwithstanding these limitations, these figures provide an estimate of temporal changes in weapons injuries, adjusted as well as possible for population growth. Such figures provide a means to compare weapons injury rates in different settings but are rarely published because the limitations described here almost always apply.
Table 1 shows an increase in mortality between the two periods for all injury types, and particularly for mine injuries. In fact, the observed increase in mortality was of such magnitude that, despite the overall decline in weapons injuries, more people died of weapons injuries per month during the post-conflict period than before. Several factors that may account for some of this increase are shown in Table 4. In general, patients treated at hospital B were more likely to arrive within 6 hours of injury, more likely to be less than 10 years old, and less likely to receive blood transfusions. Reduced chance of receiving blood transfusion and weapons injuries occurring in children can easily be seen to affect subsequent mortality. Rapid access to hospital is associated with higher mortality because severely wounded people who might have died during a longer evacuation survive long enough to die in hospital. Indeed, the hospital situated in the area under study had higher mortality for gunshot wounds (5.5% v 3.6% in hospital A) and mine injuries (12.2%v 7.1%) (P<0.05).
The extent of the increase in mortality due to mine injury during the post-conflict period raises the question of whether the factors discussed here account for all of the increase in mortality or only a portion. None of these factors apply to Hospital A, in which mortality for mine injury tripled during the post-conflict period. Other potential explanatory mechanisms include the laying of new mines with a higher explosive charge and the resettlement of new areas where mines of higher explosive charge have been previously laid.
This study quantifies the decline in rates of weapons injury observed in a region that experienced a transition to military stability and in which disarmament did not occur. Declines were in the range of 10-30% depending on injury type, and 20-40% after adjustment for population growth.
Limitations, potential biases, and terminology
Several limitations and sources of potential bias should be considered. A more sophisticated analysis might have been carried out on these data. For example, trend analysis or spline regression might have been used to examine whether trends of injury or admission rates over time existed between the two periods, or whether important non-linearities existed in incidence over the two periods. These methods, however, might have obscured the simpler message that was the aim of the present paper. In addition, restricting the analysis to comparing injury rates during the conflict period with those from the final 6 months of the post-conflict period (September 1996 to February 1997) resulted in smaller declines than those seen in table 3 for all categories of weapons with the exception of mine injuries.
There are three important sources of bias in the present analysis. Firstly, one might expect to see increased numbers of hospital admissions over the two time periods because of the opening of a second hospital during the post-conflict period. Secondly, some individuals in the dataset were evacuated by air to hospital B from areas of active conflict outside the region of interest. This number is thought to be small, and not all would have arrived within less than 24 hours. Finally, the increased proportion of patients arriving within 6 hours of injury at hospital B is a bias, since some probably would have died during the longer evacuation to hospital A during the conflict period.
These biases can be addressed by repeating the analysis presented in Table 3 on data for hospital A alone. This shows an increase in the declines observed for gunshot and fragment injuries (from 32% to 63% and 8% to 10%, respectively) and a decrease in the decline observed for mine injuries (from 23% to 21%).
Weapons, societies, and conflict
This analysis has shown that the risk of weapons injury remains high in the area studied. Though this might seem self evident and merit little further consideration, there are several reasons why such a reaction may be misguided.
Firstly, it is important to note that much of this injury is not related to landmines and required an act of volition on the part of the weapon user. The implication is that the continued availability of weapons provides a means to resolve differences. This has profound implications for how a society will function. Indeed, a convincing case has been made that weapons availability is an especially important risk to the social future of cultures attempting to emerge from the aftermath of conflict.8
Secondly, the end of the Cold War and the waning of restraint by the superpowers has led to an increase in regional and communal conflict.9 Traditional notions of conflict are being replaced by a complex picture of skirmishes, mass killings, and largescale criminal activity. All of these are being carried out by armed elements within populations that may well not exist within any formal chain of command.
Thirdly, the international community is increasingly confronted by popular demand to mount some form of response to such situations. Experiences in settings where such efforts have been undertaken have shown the importance of disarmament, demobilisation, and reintegration of former combatants into post-conflict society.10 Such undertakings are costly, difficult, and require a long term commitment that often extends beyond the time frame of popular concern.
The availability of weapons is increasingly being implicated as a factor in social destabilisation; the problem is increasingly passing out of the framework of the nation state; and efforts to respond to it are complex, costly, and prone to failure.2 Although there are no simple solutions, the trade in arms is clearly the overarching issue.1 Viewed in the context of these considerations, the index of sustained violence observed here should give rise to some sobering thought, particularly among policy makers and those in government who are concerned with the issue of weapons transfer and its implications.
I thank Virginia Gamba and Michael Klare for their knowledgeable and thoughtful comments.
Funding: No additional funding.
Conflict of interest: None.