Mental Illness–Related Stigma in Canadian Military and Civilian Populations: A Comparison Using Population Health Survey Data

Data were from the 2013 Canadian Forces Mental Health Survey (CFMHS) (28) and the 2012 Canadian Community Health Survey–Mental Health (CCHS-MH) (29). Both surveys employed a sampling framework resulting in representative samples of CAF personnel and the Canadian general population, respectively.

The CFMHS was a stratified random sample of all serving CAF Regular Force personnel as well as reservists who were deployed in support of the mission in Afghanistan, although only the Regular Force population was included in the study reported here (N=6,696; overall response rate of 80%) (28). The CCHS-MH used a cluster sampling approach (N=25,113; overall response rate of 70%) targeting the population of noninstitutionalized individuals ages 15 and older living in private dwellings in Canada’s ten provinces, excluding full-time members of the CAF and persons living on reserves and other Aboriginal settlements (29). Participants in both surveys were recruited via a telephone call from Statistics Canada interviewers. For both surveys, the primary reasons for nonresponse were noncontact and refusal to participate. Interviews were conducted face to face with 100% of military personnel and 87% of civilians, with additional civilian interviews conducted via telephone. Data for both surveys were collected by using a computer-assisted interview, and the wording of all overlapping content across surveys was identical (28,29).

We followed procedures from a recent military-civilian comparison (30) to restrict the civilian sample in order to approximate the military population. Our restricted civilian sample (N=8,394) excluded those who were not full-time employed, not ages 17 to 60, had recently immigrated (such persons would not have been eligible for CAF service), and had reported any chronic conditions that would typically preclude military service (for example, heart disease or severe obesity). This restricted sample was used for all analyses.

All respondents were asked, “Have you ever received treatment for an emotional or mental health problem?” and whether this was during the past 12 months.

Respondents who reported mental health care seeking in the past 12 months completed the Mental Health Experiences Scale (15), which assesses the experience and impacts of stigma. Respondents were first asked the following question regarding perceived stigma: “During the past 12 months, did you feel that anyone held negative opinions about you or treated you unfairly because of your past or current emotional or mental health problem?”

Respondents who endorsed perceived stigma also rated how much these perceptions of negative opinions or unfair treatment affected them over the past 12 months in each of six domains: family relationships, romantic life, work or school life, financial situation, housing situation, and health care for physical health problems. Each question was rated on a scale of 0, not affected, to 10, severely affected. These scores were then summed into an overall impact score. Possible overall scores ranged from 0 to 60, with higher scores indicating greater impact of stigma. For this study, we considered those who did not report perceived stigma to have no impact of stigma. In our analyses, we examined the mean overall impact score and mean score in each impact domain.

Sociodemographic variables included sex, age category (<25, 25–34, 35–44, and >44), race (white versus nonwhite), marital status (single, married or common law, and separated or divorced), household income category (<$50,000, $50,000–$99,999, $100,000–$149,999 and ≥$150,000), and highest educational attainment (less than secondary, secondary only or some postsecondary, and postsecondary completion). These groupings were chosen on the basis of the survey content, the distribution of respondents, sample size in each of the subgroups in the two populations, and conceptual considerations.

We used several measures common to both surveys to adjust for differences in the need for mental health care in the two populations.

SAS version 9.3 and Stata version 13.1 were used for data analysis. We used modified Poisson regression to examine the effect of population (military versus civilian) on the likelihood of mental health care seeking and perceiving stigma, and we used linear regression to examine the effect of population on the level of stigma impact (overall and in each domain). For all analyses, we examined the effect of population in an unadjusted model, followed by a model controlling for sociodemographic characteristics (partially adjusted model) and a model controlling for sociodemographic characteristics as well as variables related to the need for care (fully adjusted model).

Listwise deletion was used for missing data in all regression models, which resulted in the exclusion of cases representing between .5% and 4.7% of the population. The CFMHS and CCHS-MH include survey weights provided by Statistics Canada that ensure representativeness of the samples and account for nonresponse. Variance estimates for all analyses were calculated with bootstrap methods using 500 replicate weights provided by Statistics Canada (28).

The relevant policy and review committees within Statistics Canada reviewed and approved all aspects of data collection (including ethical aspects) and all conditions of data access and use. Participation in the surveys was voluntary, and participants provided informed consent.

This study had a number of strengths. For instance, we used data from two recent, contemporaneous, population-based surveys with high response rates, highly comparable methods, and identical survey items. Also, we employed robust methodological procedures, including sample restriction to create a sample of Canadian civilians that was more comparable to the military population as well as adjustments for key sociodemographic characteristics and variables related to the need for mental health care.

Several limitations must also be acknowledged. First, the stigma items were asked only of individuals who reported accessing mental health services in the past year. That is, individuals with mental health problems who did not seek treatment but who may nevertheless have experienced stigma were not reflected in the analysis. The population burden of perceived stigma may therefore have been systematically underestimated in both populations. However, individuals may avoid seeking treatment in order to avoid being stigmatized (label avoidance) (36), which may mean that those who do not seek care are actually less likely to report stigma and which may suggest that we, in fact, overestimated the proportion of individuals who perceived stigma. Second, the two surveys did not measure all of the same mental disorders, and thus we were able to include only three past-year disorders in our adjustments. Importantly, the WHO-CIDI posttraumatic stress disorder (PTSD) module was not included in the civilian survey; this disorder likely plays a large role in stigma experiences in the military (18,35). We did, however, adjust for additional markers of need for care (for example, psychological distress) to compensate for the non-inclusion of the PTSD module for civilians. Given that the prevalence of some disorders is greater among military personnel than among comparable civilians (30), our approach likely undercontrolled for this excess prevalence. However, the essence of our findings remained unchanged after we additionally controlled for a self-reported diagnosis of PTSD as a chronic health condition and for perceived unmet need for care. Third, the amount of variance explained by models predicting the impact of stigma was relatively low, and we acknowledge the possibility of residual confounding in these analyses. Fourth, we could assess only perceived stigma experiences, and we had no knowledge of the factual circumstances behind those perceptions. Finally, our findings may not generalize to other nations and their military forces.

In this study, we found a much greater perceived burden of stigma in the military, a population that proved to be more likely to seek care. This apparent discrepancy suggests that the relationship between stigma and care seeking may differ between military personnel and civilians. It is possible that in military samples, factors other than stigma attenuate any negative effect of stigma on care seeking (37). For example, CAF mental health literacy efforts that enhance recognition of the need for mental health care (25) might have more than counterbalanced any negative effect of greater stigma. Indeed, recognition of need is one of the pillars of the CAF’s program, and there is strong evidence from 2002 CAF data that failure to recognize need was a leading barrier to care (7). Our findings thus point to the potential power of interventions other than the direct destigmatization efforts that have been the focus of recent efforts to neutralize the effects of stigma on care seeking in military and other populations.

A second and less encouraging hypothesis is that in the military, stigma is a consequence of care seeking, not an antecedent. Mental health literacy and destigmatization efforts may have led not only to improved attitudes toward mental illness (22,38) but also to greater care seeking (37), and those seeking care may be much more likely to perceive stigma after seeking care. This coheres with our findings on the disproportionate impact of stigma experiences in the military, especially in the workplace. If stigma and its impacts are indeed more often a consequence than an antecedent of care seeking in the CAF, then there could be opportunities to intervene and minimize negative consequences. If the underlying experiences reflect unfair application of medical standards, or if unfair treatment occurs in other domains (for example, promotion), there might also be opportunities to obtain more favorable results—in military organizations and in other workplaces.