3272

PREVENTING DEATHS IN THE CANADIAN MILITARY

Homer C.N. Tien, MD, MSc, FRCSC, FACSac, Sanjay Acharya, MD, FRCPCc, Donald A. Redelmeier, MD, MS, FRCPCb

American Journal of Preventive Medicine
March 2010
Volume 38, Issue 3, Pages 331-339 (March 2010)

http://www.ajpm-online.net/article/PIIS0749379709008940/fulltext

Combat fatalities are reported by the media as a frequent cause of military deaths, yet they may not reflect the most common and preventable ways that soldiers die.

Purpose

The purpose of this study was to quantify the leading causes of death in the military and to identify modifiable behaviors that potentially contributed to death.

Methods

This was a retrospective chart review of all Canadian Forces members who died during the past quarter century (January 1, 1983, to December 31, 2007) and included autopsy reports, death certificates, coroner reports, hospital records, military reports, and other miscellaneous sources. Underlying cause of death and modifiable behaviors potentially contributing to death were determined.

Results

A total of 1889 individuals died during the study period, and a cause of death was identified for 1710 cases (91%). Traumatic injuries caused 57% of deaths, and medical disease was responsible for 43%. The four leading specific causes of death were motor-vehicle crashes (384 deaths, 22%); neoplasms (374 deaths, 22%); suicide (289 deaths, 17%); and cardiovascular disease (285 deaths, 17%). Combat deaths accounted for less than 5% of all deaths (70 deaths). Approximately 35% of all deaths were attributable to potentially modifiable behaviors, which included suicide (219 non–alcohol-related deaths, 13%); smoking (159 deaths, 9%); and alcohol use (186deaths, 11%).

Conclusions

Public attention focuses on combat fatalities, yet most military members die from other causes. Avoiding future deaths requires targeting suicide, smoking, and alcohol consumption, in addition to trauma care for combat injuries.

Introduction

The profession of arms is dangerous. Ongoing media coverage1, 2, 3 of the conflicts in Iraq and Afghanistan documents the mortal threats that military members face during war. What may be overlooked, however, is that inadvertent injuries, natural diseases, and suicides may cause more deaths than does combat.4, 5, 6, 7 If so, disease prevention and lifestyle modification programs might help to substantially reduce all-cause mortality.5

Developing prevention programs requires an understanding of the mortality burden attributable to external, modifiable factors (actual causes of death).8 In 2004, it was reported9 that tobacco, poor diet, inadequate physical activity, excessive alcohol consumption, and motor-vehicle crashes were the leading actual causes of death in the U.S. No such analysis has been performed for the military.

The purpose of this study was to determine causes of death in an entire military force—the Canadian Forces. Primary attention was focused on distinguishing between medical and traumatic causes of death and on determining behaviors contributing to death. The goal was to help reduce future deaths in military members serving their country.

Methods

The Canadian Forces consists of 62,000 regular force members and 25,000 reservists.10Health records of all active-duty members who died between January 1, 1983, and December 31, 2007, were retrospectively reviewed. Active-duty members were identified as those receiving a salary from the Department of National Defense at the time of their death.

The Directorate of Casualty Support and Administration for the Canadian Forces identified all relevant cases. These cases were deterministically linked via a unique social insurance number or military service number to a military human resource database for demographic information (Human Resources Management System, PeopleSoft Inc.). This database was also used to obtain the total population size of the Canadian Forces for each study year. This study was approved by the Surgeon General of the Canadian Forces Health Services and the institutional ethics board at Sunnybrook Health Sciences Centre.

The health records of all study patients were retrieved from the National Archives of Canada. These records included autopsy and coroner reports, death certificates, hospital records, and military reports investigating the death. Also included were the periodic health examinations and diagnostic testing obtained for each patient during their military career. Upon enlisting, all military members undergo a screening health examination with a full history and physical exam, baseline BMI measurements, laboratory testing, and chest radiography. These screening procedures are then repeated every 5 years until age 40 years, when the frequency increases to every 2 years.

Data were abstracted using a standardized data sheet with the underlying cause of death categorized using one of 17 ICD-9-CM diagnostic categories.11 The primary analysis compared the proportion of deaths from trauma relative to medical causes. Medical deaths corresponded to codes 001–799. Traumatic deaths corresponded to one of the external codes (E800–999). The exceptions were heat-induced illness (E900) and exertion-related illness (E927), which were classified as medical deaths. Combat deaths (E979, E990–E999) were also identified, which were any fatal injury caused intentionally by foreign (nonallied) paramilitary or military forces. The final determination of each cause of death for this study was made by the author. Full details of the methodology are described elsewhere.12 The anticipated sample size of 1500 was designed to provide sufficient statistical power to identify point estimates with a standard deviation of less than 1.25%.

Data on demographic factors, cause of death, circumstances surrounding death, as well as clinical details from the most recent periodic health examination preceding death were collected. Double data entry was performed on 5% of the primary data and yielded a kappa statistic of 0.75. In response, 100% of all entries were double-checked with abstraction sheets to minimize data-entry errors. Five percent of records were also checked for reliability for cause of death by an independent physician assessor, yielding a kappa statistic for this reliability check of 0.95.

Individual behaviors contributing to death9 were analyzed. These included smoking, alcohol use, physical inactivity and poor diet, certain sexual behavior, suicide, illicit drug use, not using a life-jacket in water-transport deaths, and seat-belt usage in motor vehicle–related deaths. In some cases, a death was deemed completely attributable to a modifiable behavior.13, 14 For most medical diseases, however, only a fraction of deaths can be attributed to behaviors. Details of the methods used to calculate the fraction of deaths that can be attributed to each behavior are described in Appendix A, available online at www.ajpm-online.net and elsewhere.15, 16

Smoking-Attributable Deaths

In this study, the number of ischemic heart disease and respiratory cancer deaths attributable to smoking was directly calculated using study data, annual smoking prevalence data in the Canadian Forces Health and Lifestyle Information Surveys,17, 18, 19 and estimates of the total Canadian Forces population for each year (Appendix A, available online at www.ajpm-online.net). Former smokers were considered to have no lingering health risks associated with smoking, and those with unknown smoking status were considered nonsmokers. Smoking-related fire deaths were classified as 100% smoking-attributable.14 For all other smoking-attributable diseases, published population-attributable fractions were used to calculate smoking-attributable mortality.20

Alcohol-Attributable Deaths

Alcohol-attributable deaths were identified and calculated for acute and chronic conditions. As with smoking, the number of alcohol-attributable deaths was directly calculated using study data related to ischemic heart disease deaths.21 The prevalence of high-risk drinking was considered constant throughout the study period.19 To avoid overestimation, former drinkers, occasional drinkers, and those with unknown alcohol consumption were presumed to have no lingering detrimental health effects. Alcohol-induced cirrhosis was considered to be 100% attributable to alcohol.13 For all other chronic conditions, published population-attributable fractions were used to calculate the number of alcohol-attributable deaths.22 For all acute injuries, a death was identified as alcohol-attributable if the blood alcohol level exceeded 10 g/dL at time of death.13

Other Behavior-Attributable Deaths

Physical inactivity and poor diet are behaviors associated with obesity.9 Obesity is defined as a BMI of ≥30.23 Although obesity increases the risk of developing many chronic diseases,24, 25 the population-attributable fractions of deaths from obesity are small when taking into account other confounding risk factors such as diabetes and smoking.24 Therefore, in this study, the number of obesity-attributable deaths for each chronic disease was not calculated. Instead, the total number of obesity-attributable deaths was calculated using published population-attributable fractions.24 There is also considerable controversy regarding the relationship between being overweight (BMI of 25–29.9) and the risk of premature mortality.26, 27 Because of these uncertainties, deaths associated with being overweight were not analyzed.

An earlier study9 attributed all HIV deaths to sexual behavior. The same methodology was used in this study. However, “sexual behavior—HIV” was distinguished from “sexual behavior—other.” A death was classified as being from suicide behavior only if it was not already ascribed to alcohol use. A death was classified as being from illicit drug use only if the death was deemed to be inadvertent and was not already ascribed to alcohol use. A death was classified as being from seat-belt usage in motor-vehicle crashes only if it was not already ascribed to alcohol use and only if the coroner or autopsy report specifically recorded that the military member was not using seat belts at the time of the crash. A death was classified as being from life-jacket usage in water-transport deaths only if it was not already ascribed to alcohol use and only if the coroner or autopsy report specifically recorded that the member drowned and was not wearing a life jacket. Data were analyzed using SAS, version 8.2.

Results

A total of 1889 deaths were identified, of which 1738 charts (92%) were available, located, and reviewed. Cause of death was not determinable in 28 cases, leaving 1710 cases (91%) on which further analyses were conducted for cause of death. A periodic health examination was located for 1436 cases (84%) and was used for determining smoking, alcohol, and BMI status. Details of death were determined from autopsy reports in 26% of cases (n=447); coroner reports in 3% (n=50); hospital records in 37% (n=627); death certificates in 9% (n=162); and military investigations in 10% of cases (n=178). Other additional documentation was used for 15% (n=246).

Most deaths occurred outside of normal working hours (n=1369), and only a few occurred on duty while on overseas missions (n=118). Major trauma caused 57% of total deaths (n=968), and medical diseases caused 43% (n=742). Motor vehicle–related deaths caused 22% of deaths and were the leading single cause of death (n=384), of which about one in seven (n=57) occurred on duty. Neoplasms caused 22% of deaths overall (n=374); suicide caused 17% (n=289); and cardiovascular diseases caused 17% (n=285). Combat caused 4% of total deaths (n=70; Figure 1). About 84% of combat deaths (n=59) occurred during the last 2 years of the study, during which time members of the Canadian Forces were deployed to southern Afghanistan.

Firearms and Blast

Firearms and blast were the underlying mechanism causing death in about one in 12 deaths (n=147); however, about 50% of these were due to inadvertent training incidents or suicide rather than actual combat. Of the 289 military suicides, 21% (n=60) involved firearm use. Of ten murders, three involved firearm use. There was one death from a firearm from an inadvertent incident during a hunting trip, and six inadvertent deaths from firearms that were related to military training. Of 70 combat deaths, six were from firearms. Therefore, a total of 76 deaths were attributable to firearm use.

Discussion

Military members are almost equally at risk of dying from medical diseases as from traumatic injuries. Although traumatic deaths are common, the majority reflect inadvertent roadway crashes or suicide. Indeed, most deaths occurred during off-duty hours away from military bases. Combat deaths were rare, and caused less than 5% of overall deaths. As a result, only limited opportunities are available to substantially reduce future premature military deaths by improving the quality of acute care provided by the military healthcare system. Greater opportunities remain for primary prevention and medical risk factor management.

Almost one quarter of all military deaths can be attributed to individual behaviors, the three major ones being suicide, alcohol consumption, and tobacco use. These results differ from the general population because of the reduced risks attributable to physical inactivity and poor diet in young, healthy military members, for whom trauma remains the leading cause of death.28 Incidents involving firearms and munitions caused 147 deaths (9% of deaths overall), yet many of these were related to training or suicide.

These results argue for more disease prevention, mental health care, and safety programs to complement the traditional military focus on trauma. During times of war, trauma care is critical and saves lives.29 However, future gains in survival are likely to be small.30 Prevention measures will likely have a greater impact than improved trauma care on reducing future combat deaths, as the majority of deaths on the modern battlefield are nonsurvivable.31 However, efforts to improve body armor or develop blast countermeasures rely on organizational commitment rather than individual behavior.

Focusing solely on combat deaths may result in missed opportunities to save lives. Combat soldiers are more likely to start smoking to cope with the stress of deployment32 and more likely to abuse alcohol after returning from deployment.33 Combat veterans are also more likely to suffer from mental health problems.34 Further, combat veterans are more likely to die from inadvertent injury after deployments.35, 36 Therefore, even in times of war, prevention strategies targeting seat-belt usage, tobacco use, excessive alcohol consumption, and mental health may have the greatest impact on reducing future military deaths.

The Canadian military already offers comprehensive health promotion programs, and future prevention efforts should build on this tradition. For example, the Canadian Task Force on Preventive Health Care (CTFPHC)37 reports good evidence to support smoking-cessation counseling and using nicotine-replacement therapy as an adjunct, and fair evidence to support referrals to smoking-cessation programs. Consequently, the Canadian Forces already offers a smoking-cessation program called Butt Out38 that has met with substantial success. The number of smokers within the Canadian military has decreased more rapidly than in the general population.19 One area for improvement, however, is smoking initiation, as many recruits in various military organizations start smoking during or after basic training.19, 39 Military organizations should target future smoking-prevention efforts at new recruits during induction.

The Canadian Forces now also routinely screens military members for alcohol problems during their periodic health examination and offers an alcohol rehabilitation program to those with identified alcohol dependence.38 Unfortunately, very few interventions have been shown to reliably treat alcohol dependence. Further, alcohol use and alcohol abuse remain an important part of off-duty military culture. It is hoped that ongoing efforts to limit the availability of alcohol in the military and to encourage responsible alcohol consumption will reduce future alcohol-related deaths.

Motor vehicle–related fatalities can be prevented by increasing seat-belt usage.40 The CTFPHC41 reports good evidence to support the use of legislation to increase seat-belt usage, and fair evidence to support the role of individual counseling. Both U.S. and Canadian military regulations require seat-belt usage by on-duty personnel. As a substantial number of fatal motor-vehicle crashes occur during on-duty times, military commanders are uniquely positioned to further reduce crash fatalities. Commanders can dispel misperceptions and actively support and enforce these regulations.42 In addition, future prevention efforts could focus on individual counseling at the time of periodic health examinations to reinforce the need for seat-belt usage.

With regard to the mental health of military members, the popular press has reported43, 44 increased numbers of suicides in young combat veterans. Unfortunately, few interventions have been reliably shown to prevent suicide. The CTFPHC45 reports fair evidence for the positive impact of only physician education programs on suicide prevention. Despite the lack of evidence to support the use of suicide prevention interventions, the Canadian Forces has proactively established a comprehensive program to attempt to reduce suicides in its returning combat veterans. Physicians are deployed to prepare soldiers for reintegration, and the Canadian Forces has established operational stress centers to help returning soldiers reintegrate and to reduce the risk of suicide.38 However, further research is required in this area.

This study also found that the proportion of Canadian soldiers using firearms to commit suicide (21%) was substantially lower than that in the U.S. military. It was found46 that approximately 60% of U.S. military members use a firearm to commit suicide. This difference is likely accounted for by more stringent Canadian gun-control laws. However, it is not clear whether this difference caused a reduced rate of suicide or merely affected the proportion of suicides that were firearm related.

Limitations

Cause-of-death studies are subject to many biases; in particular, suicides tend to be under-reported and cardiovascular causes over-reported.47, 48, 49 In this study, these biases were minimized by using and reviewing multiple sources of data, including autopsy reports. However, approximately 8% of charts were not located, and these tended to be for younger men who were probably less likely to have died from medical causes. Hence, this study may have underestimated the frequency of trauma deaths relative to natural deaths.

Another limitation is the confounding of factors that can occur when attributable fractions are calculated. For example, other cardiac risk factors such as diabetes and hypertension were not controlled for when calculating the fraction of deaths from ischemic heart disease that can be attributed to smoking. Even so, such confounders likely had a minimal impact on the study results, as they are uncommon in otherwise healthy soldiers. In addition, population-attributable fractions were used to calculate mortality attributable to less commonly observed medical diseases in the military, which raises many methodologic difficulties.16, 50, 51 Overall, behavior-attributable mortality was likely underestimated because it was assumed that the behavior was absent if its actual status was unknown.

Based on this study, the authors advocate for prevention strategies to reduce future military deaths. However, the limitation of such an approach is acknowledged, especially with regard to changing behavior. Many of these behaviors leading to death may be perpetuated by organizational issues or culture. For example, although smoking and alcohol consumption remain individual behaviors, the military milieu may predispose its members to misuse because of the reduced costs and availability of cigarettes and alcohol.

Finally, the results of this study can be extrapolated to other settings only with caution. During the study period, the Canadian Forces was involved primarily in peacekeeping missions. Eighty-four percent of all combat deaths occurred during the last 2 years of the study (2006 and 2007) when Canadian soldiers were deployed to Kandahar, Afghanistan.52 Excluding these last 2 years, combat deaths would have been 1% of deaths.

The results of this study are similar to reports on U.S. and British military deaths during periods of relative peace. Specifically, from 1980 to 1993, it was reported5 that inadvertent injuries were the leading cause of U.S. military death, followed by natural causes and suicide. Combat caused only 2% of deaths. Inadvertent injury and diseases were also leading causes of death in British Military Forces from 1980 to 1984, despite the Falkland Islands War.53 Since 2003, however, U.S. and British military members have been actively engaged in combat in Iraq and so have sustained more combat deaths.1, 2, 3, 4 Although Canadian Forces have not been deployed to Iraq, the fatality rate for the Canadian military in Afghanistan is almost four times higher than the fatality rate of the U.S. military in Iraq.54 Therefore, the conclusions from this study are likely to be valid for U.S. and British military forces as well as for the Canadian Forces.

Conclusion

Canadian military members were almost as likely to die from medical diseases as from traumatic causes. About 35% of all deaths were attributable to potentially modifiable behaviors. In addition, combat missions end, whereas soldiers continue to drive, smoke, and drink, making recommendations about improving preventive strategies likely to remain important.



A total of 71 military members died as a consequence of blast injuries, of which 63 were from combat action. The blast mechanisms included land mines, improvised explosive devices, and air-dropped munitions. Seven military members died from blast-related injuries sustained during training exercises: three associated with hand-grenade mishaps and four from a single catastrophic munitions explosion. One military member committed suicide using a hand grenade.

Behaviors Contributing to Death

Of the 1710 deaths, approximately 35% (n=600) were attributable to individual behaviors. Table 2, Table 3, Table 4 list the chronic and acute conditions used to calculate the number of deaths that can be attributed to smoking and alcohol consumption. Suicide behavior (where alcohol was not involved) caused 13% (n=219) of all reviewed deaths. Alcohol potentially caused 11% (n=186) and smoking potentially caused 9% (n=159). Unlike in the general Western population,9 physical inactivity and poor diet (obesity) contributed to few military deaths (Table 5). There were 309 motor vehicle–related deaths in which alcohol use was not a factor. Of these, the autopsy reports in six cases specifically state that seat belts were not used. However, there was no specific mention of seat-belt usage in the majority of the other cases. Likewise, there were 21 cases in which Canadian Forces members drowned during water-transport activities and in which alcohol consumption was not a contributing factor. In four cases, life jackets were not used. In the remainder, no mention was made of life-jacket use at all. The three cases of sexual-behavior deaths unrelated to HIV infection were deaths caused from auto-erotic asphyxiation. There were four deaths from illicit drug overdose; however, these cases were also associated with alcohol use and were therefore classified as alcohol-attributable deaths.