Mortality trends among both serving and released military personnel are of considerable interest to several groups including some federal departments and provincial ministries, health scientists, policymakers and the general public. Participation in both domestic and international operations may subject military personnel to unusual health risks including death. Although programs are in place to assess potential occupational and environmental exposures, it is not possible to know precisely what each military member has been exposed to throughout his/her military career.
In spite of such hazards, comparisons between military populations and the general population usually show that the former have lower rates of all-cause and disease-specific mortality than the latter. These findings are often attributed to a healthy worker effect resulting from the selection process of military personnel that tends to exclude persons with severe disabilities or serious chronic conditions.
This research is part of a larger initiative known as the Canadian Forces Cancer and Mortality Study (CF CAMS), which was initiated to examine mortality and the development of cancer for persons who served in the Canadian military. The project is a collaborative effort between Statistics Canada, the Department of National Defence (DND), and Veterans Affairs Canada (VAC). Included are individuals who enrolled in the Regular Force after January 1, 1972 with follow-up until December 31, 2006 for mortality and December 31, 2007 for cancer. The start date reflects the availability of DND electronic administrative information.
This report focuses specifically on mortality among the dynamic 35-year cohort of Canadian Forces (CF) personnel who enrolled between 1972 and 2006, inclusively. The first objective of this report was to examine causes of death for persons with a history of military service (still serving/released). Additionally, there was an interest from VAC to examine causes of death for persons who were released from the military (see Box 1 – Released Cohort).
Excluded from the study were: individuals who either enrolled before 1972 or enrolled after 2006, those with a missing or invalid date of birth, and those with reserve force service only. The cohort was limited to regular force personnel as historical information on reservists proved to be incomplete in DND administrative datasets.
Death registration is mandatory in Canada, with comprehensive provincial/territorial registration systems that are combined at a national level by Statistics Canada in the Canadian Mortality Data Base (CMDB). Complete death registrations of in-country deaths were available up to December 31, 2006 (see Box 2 – Out-of-Country Deaths). Therefore, all deaths occurring in Canada between 1972 and 2006 were eligible for inclusion in the mortality analysis. A total of 3,969 death records from the CMDB were matched with the entire CF CAMS cohort. Causes of death were grouped using the Tenth Revision of the World Health Organization’s International Classification of Diseases (ICD-10).
Analyses were completed for two groups:
For each cohort, a series of analyses were conducted using standardized mortality ratios (SMR, see Box 3 – Methods and Concepts) to compare all-cause and cause-specific mortality (for both sexes and 5-year age groups) with the general Canadian population. Findings from the SMR analyses for the released CF CAMS cohort prompted further study using a proportional hazards model (see Box 3 – Methods and Concepts) to identify factors associated with the likelihood of committing suicide following release from the CF.
Table 1 gives an overview of the main characteristics of the entire CF CAMS cohort, as well as of the released subpopulation. In the entire CF CAMS cohort, about 87% were males and 80% were non-commissioned members (NCM), their median age at enrolment was 20 years, and 51% had service between the period 1997 and 2006. Individuals in the released CF CAMS cohort joined at younger ages, had shorter periods of service, and only 19% had service between 1997 and 2006. This study followed the released CF CAMS cohort for a median duration of 18 years after release from military service. Reason for release was categorised into voluntary (61%), involuntary (29%), and medical (9%) (see Box 3 – Methods and Concepts).
Of the 188,161 personnel in the entire cohort, there were a total of 3,969 deaths over the 35 year period; only 294 were females, whereas 3,675 were males. In the released cohort (112,225 personnel), there were 2,824 deaths (204 females, 2,620 males). The median age at death was 34 in the entire CF CAMS cohort and 37 in the released CF CAMS cohort.
Table 2 describes the age-adjusted causes of mortality for males and females for the entire CF CAMS cohort. Table 3 describes age -specific analyses for injury and suicide. All reported differences are statistically significant at the 95% confidence level.
Table 4 describes the age-adjusted causes of mortality for the released CF CAMS cohort for males and females. Table 5 describes the age-specific analyses for injury and suicide. All reported differences are statistically significant at the 95% confidence level.
Table 6 describes the factors associated with the likelihood of committing suicide in the released CF CAMS cohort using a proportional hazards model (see Box 3 – Methods and Concepts). The model controlled for several risk factors at the same time including sex, rank, age at release, reason for release, years of service, and period of military service. The model demonstrated that taking into account all the above factors, the risk of suicide was:
Consistent with the findings of other studies, the results show that all-cause and disease-related mortality was lower among individuals with a history of military service compared with the general Canadian population. These findings may be partially accounted for by the “healthy worker” effect resulting from the selection process at enrolment.
Air and space transport accidents were more frequent among males with military service than in the general population. This finding is likely attributable to the higher representation of flying-related occupations in the CF compared with the general population and the intrinsic risks of military aviation.
Elevated risk of dying was noted for females in the military from injuries among those aged 20 to 24, and from suicide for those aged 40 to 44, compared with the same age groups in the general population.
Among the released cohort (military personnel that enrolled and released between 1972 and 2006), there was a higher risk of suicide in males aged 16 to 44, and in women aged 40 to 44, when compared to the general population. These findings prompted the need for a closer examination of factors potentially related to suicide among the released CF CAMS cohort.
Findings from the proportional hazards model showed that the risk of suicide in the released group of military personnel was highest among male non-commissioned members with short periods of service who were released for non-voluntary reasons. Furthermore, the risk of suicide was higher for personnel with military service during the 1972 to 1986 time period.
The reliance of this study on administrative data represents both an advantage and a limitation. On the positive side, it provided a large sample size and an observation period of 35 years that would be difficult to attain with survey data. Unfortunately, interpretation of the study findings was limited as information relating to individuals’ socio-demographic profile, medical and psychological history, deployment to areas of conflict and other potentially relevant variables were not accessible at the time of the study. As it is the case with every record linkage study, the probabilistic linkage of administrative data from two different sources might have also led to the inadvertent inclusion of false positives and false negatives in the final linked database (i.e. individuals still alive who are counted as dead after linkage or, conversely, dead individuals that show up as alive; see Box 3 – Methods and Concepts).
This study was limited in scope by examining mortality trends only among military personnel who enrolled in the Canadian Forces between 1972 and 2006. The lack of available information on individuals who enrolled before 1972 necessitated their exclusion from the CF CAMS cohort resulting in a younger cohort (and subsequent fewer deaths) than expected of a full history of Canadians with Regular Force military service. Consequently, the released component of the cohort was not representative of the entire population that released from 1972 to 2006. The findings of this study therefore apply to a subgroup of the released population and should not be generalized to the overall population of veterans.