A longitudinal analysis of Veteran well-being using the Canadian Longitudinal Study on Aging

Abstract

Introduction:

Understanding well-being trajectories in Canada’s aging Veteran population is critical, as most research is cross-sectional, U.S.-centric, and male-focused. Comparing Canadian male and female Veterans to non-Veterans over eight years, this study longitudinally assessed social integration and life satisfaction in relation to depression.

Methods:

Data from the Canadian Longitudinal Study on Aging (CLSA) (N = 51,338; 3,215 Veterans) across three time points (2011, 2015, 2018) were analyzed. A multi-group cross-lagged panel model (CLPM) was applied using structural equation modelling (SEM) after establishing measurement invariance.

Results:

The overall model structure was invariant between male and female Veterans, suggesting a shared psychosocial architecture shaped by military service. Significant variance emerged when comparing Veterans to same-sex civilians. For female Veterans, life satisfaction rather than social integration was the primary protective factor against depression, a pathway distinct from non-Veteran females. For male Veterans, the relationship between social integration and depression was complex and shifted over time, a dynamic not observed in their civilian counterparts.

Discussion:

This study addresses a key gap by longitudinally including both male and female Veterans. Longitudinal analyses suggest well-being trajectories differ between Veterans and non-Veterans, highlighting Veteran status as salient in aging models. The protective role of social integration appears less direct for female Veterans, while for male Veterans, its interplay with depression is more dynamic. This heterogeneity demonstrates that military service has lasting, sex-specific effects on well-being, rendering generalized models of aging insufficient for the military community and underscoring the critical need for tailored support for Canada’s aging Veterans.

Lay Summary

As Canada’s Veteran population grows older, it is important to understand how their well-being changes over time. Using data from the Canadian Longitudinal Study on Aging, in which researchers followed over 51,000 adults (with over 3,200 Veterans) for eight years, this study investigates how two aspects of well-being — social connections and life satisfaction — relate to depression as people age. Our findings show that military service creates unique pathways to well-being that differ significantly from civilians. For women Veterans, a strong sense of life satisfaction was the most critical factor in protecting against depression, a pattern distinct from their civilian peers. For male Veterans, the link between social connections and depression was more complex and shifted as they aged. A key insight from our study is that the shared experience of military service appears to create a common psychological foundation for Veterans, making the overall relationship between social connections, life satisfaction, and depression function similarly for both men and women who have served. However, within that shared foundation, their paths to well-being diverge from their civilian counterparts.

Introduction

The well-being of aging Veterans reflects a lifetime of unique challenges and strengths. As of 2021, there are over 460,000 living Canadian Veterans, and their median age continues to rise.¹ Two-thirds (66%) are now aged 55+ years, nearly half (42%) are 65+ years, and about 7% are 85+ years, with projections suggesting one-third of Canadian Armed Forces (CAF) Veterans will exceed 70 years in 2026.² This upward shift in average age identifies an urgent need to explore how aging trajectories and service-related exposures jointly influence long-term health and well-being of Canada’s aging Veteran population. Yet despite aging’s established impact on well-being³ and the recognized need for empirical data on aging Veterans,⁴ comparative longitudinal research on well-being determinants between aging Canadian Veterans and non-Veterans remains limited.

The complexities of aging in the Veteran population

Aging in Veterans is distinctly shaped by military service. Alongside potential “healthy soldier” and “military capital” benefits,⁵ service can entail exposures to physical/psychological trauma, arduous activity, and toxins. These exposures often lead to increased risks of chronic conditions, accelerated biological aging, and frailty.^2,6 CAF Veterans have higher rates of back pain, arthritis, depression, anxiety, and PTSD than non-Veterans.² Notably, the number of Veterans living with three or more concurrent mental health disorders rose from 3.1% in 2002 to 21.9% in 2018.² Such comorbidities complicate care and heighten risks for functional decline, long-term care needs, and mortality.⁷ Socially, aging Veterans face heightened risks of isolation, with nearly 1 in 4 Canadian Veterans living alone — a factor known to worsen health outcomes in older adults.⁸ The vital role of friends and families, who at times act as unpaid caregivers, is a key feature of the social and health domains in Veteran well-being.⁸

This picture is further complicated by a well-documented, yet paradoxical, trend in which the prevalence of some mental health disorders appears to decrease with age; older Veterans often report a lower burden of PTSD and depression symptoms compared to younger cohorts.^9–11 This phenomenon is multi-faceted, with several proposed explanations. These explanations include survivor bias, which posits that individuals who survive into older age are a more resilient group, as severe mental illness can be associated with premature mortality.^9,10 Additionally, many older Veterans demonstrate significant psychological resilience and a capacity for successful aging, with a majority perceiving themselves to be aging well despite past adversities.¹² Clinical factors also might play a role; PTSD symptoms might manifest differently in later life, leading to under-recognition, and older cohorts might be less likely to seek mental health services or disclose psychological distress because of stigma.^2,13 Acknowledging this complexity is essential for avoiding a monolithic view of Veteran morbidity and for accurately contextualizing the health needs of the aging Veteran population.

Heterogeneity and equity in Veteran aging

The Veteran population is far from homogenous.¹⁴ Varied service eras, deployment types, rank, length of service, military sexual trauma, and discrimination contribute to diverse health trajectories.^15,16 The needs of equity-deserving groups (women, Indigenous, racialized, and 2SLGBTQI+ Veterans) have been historically understudied, particularly from an aging perspective.^2,17 For example, women Veterans face unique challenges, such as health care access barriers, social isolation, economic insecurity, and heightened mental health risks potentially linked to military sexual trauma or post-service integration difficulties.^17,18 Similarly, 2SLGBTQI+ Veterans endured systemic discrimination (i.e., the LGBT Purge), with lasting impacts on health, social integration, and service access.² Yet how these distinct life experiences specifically shape the aging process and influence well-being in later life for diverse Veteran groups remains insufficiently explored. Addressing these questions requires an intersectional approach that recognizes and empirically examines the compounded effects of multiple forms of marginalization on aging outcomes. While this study tackles some of these aspects, given the epidemiological data set and quantitative methods applied, much of these contextual factors could not be fully addressed.

Conceptualizing well-being in aging Veterans: frameworks and key determinants

Traditional healthy aging models prioritize low disease probability, high physical/cognitive function, and active engagement.¹⁹ However, this strong emphasis on physical health can be limiting when considering Veterans with service-related disabilities, including mental health conditions, potentially leading to exclusionary definitions.²⁰ Consequently, recent research advocates for a broader, inclusive perspective on aging Veteran well-being.^14,21 This perspective maintains the value of the “active engagement” component in traditional models¹⁹ while explicitly integrating determinants like resilience, purpose, social connectedness, and autonomy, especially when biomedical health is compromised. It also recognizes that Veteran well-being relies on leveraging psychosocial strengths against health challenges and unique service experiences.

To guide this holistic understanding, several well-being frameworks have been created.^22–24 Veterans Affairs Canada’s (VAC) Well-being Surveillance Framework,²⁵ a foundational model for measuring and assessing Veteran well-being in post-service life, informs programs for all Veterans by conceptualizing well-being as multi-faceted and contextual. This model contains seven domains: health, purpose, finances, social integration, life skills, housing/physical environment, and culture/social environment (Figure 1). The extensive application of this model in studies of CAF Veterans offers a strong foundation for its further validation, including longitudinal validation, to clarify how its domains interrelate and shape well-being trajectories in Canadian Veterans and strengthen its utility.²⁵

Figure 1.

VAC Well-being Surveillance Framework

Source: Adapted from the VAC Well-being Surveillance Framework.²⁵

Social integration, characterized by supportive relationships, community engagement, and a sense of belonging, is a robust determinant of positive aging outcomes across populations,^17,26 and it is thus a core domain of the VAC framework. For Veterans, strong social networks can be particularly crucial in navigating post-service life, buffering against service-related health challenges, such as PTSD and chronic pain, and mitigating the heightened risks of loneliness and isolation often documented in this group.^17,27,28 However, the transition from the structured community of military life to civilian settings can also disrupt these networks. Veterans may face distinct barriers to forming new connections because of health issues, stigma, difficulties in readjustment, or a sense of otherness from civilian life.^16,29

Similarly, life satisfaction, a person’s subjective assessment of their overall quality of life, functions as a key determinant of well-being and is closely tied to meaningful engagement, purpose, and social participation,^21,30 all of which are acknowledged in the VAC framework. While these factors are important for all older adults, life satisfaction for Veterans is often intertwined with successful community reintegration, the ability to maintain valued roles, and positive perceptions of their military service’s impact on their life.²¹ Studies show that psychosocial resources, such as a sense of purpose and resilience, can sustain life satisfaction even amid physical health challenges or disabilities, underscoring that well-being is not solely determined by the absence of disease.²¹

Conversely, depression poses a significant threat to well-being, and its prevalence is notably higher among Veterans, often co-occurring with PTSD, anxiety, and chronic physical conditions.^2,21,31 Depression diminishes life satisfaction and psychological health, and impairs social functioning, leading to social withdrawal and exacerbated isolation, thereby creating a detrimental cycle.^31,32 The interplay between depression and a history of military service might not only affect health as a domain of well-being but also further complicate life satisfaction. For instance, depression may prevent Veterans from benefiting from the resilience-building aspects of their service or compound challenges related to unaddressed trauma or a reluctance to seek treatment.³¹

Knowledge gaps and limitations in comparative research on Veteran well-being

Despite important insights from existing research, including frameworks that offer valuable conceptual guidance, substantial knowledge gaps persist regarding how well-being evolves for aging Veterans compared to aging non-Veterans, particularly in the Canadian context where dedicated research is still developing.^4,33,34 Existing research on Veteran well-being, largely U.S.-based and reliant on cross-sectional designs,^2,5 consistently shows Veterans experience unique strengths (e.g., greater health care access, higher occupational status) and vulnerabilities (e.g., elevated rates of mental health conditions, chronic diseases, functional limitations) compared to non-Veterans. However, the predominance of cross-sectional studies with primarily male samples provides only limited temporal perspectives, thereby hindering understanding of well-being trajectories, impacts of key life transitions, or the nuanced experiences of women and other equity-deserving Veteran sub-groups.^17,35

In Canada, initiatives like the Canadian Veteran Health Survey (CVHS) have recently provided the most comprehensive national estimates to date, enabling direct comparisons between Veterans and non-Veterans across a range of health and social domains.³⁶ The CVHS has illuminated the high prevalence of chronic conditions and critical social determinants of health, particularly among older Veterans.⁴ While the survey provides valuable insights, its cross-sectional design offers a snapshot in time. This snapshot, combined with a sample that, despite efforts toward greater inclusivity, still largely reflects the historical demographic of male Veterans, somewhat constrain the survey’s ability to fully illuminate the life-course dynamics and diverse trajectories of diverse Veterans.^4,17 Earlier efforts, such as the Life After Service Studies (LASS) — a key Canadian initiative examining Veterans released since 1998 — while valuable for providing insights into Veteran well-being, and having evolved from primarily cross-sectional analyses to incorporating longitudinal components in later iterations (e.g., LASS 2016 and LASS 2019^4,25), are similarly constrained by specific aspects of their design and demographic coverage when examining long-term determinants of well-being across the broader aging spectrum.⁴

The foregoing review identifies important factors influencing the well-being of aging Veterans and highlights significant gaps in the existing literature. Specifically, there is a pressing need for longitudinal research that compares key determinants of well-being between Canadian Veterans and their non-Veteran counterparts, particularly attending to potential differences in aging trajectories for men and women, respectively. Moreover, while the VAC framework provides a valuable multi-dimensional lens, there is a recognized opportunity for empirical investigation to understand how the constructs within such frameworks evolve over time and contribute to well-being trajectories, potentially advancing the framework’s practical applicability for informing policy and program development.

The present study

The purpose of this study was to longitudinally examine and compare key determinants of well-being — social integration, life satisfaction, and depression — among aging Canadian Veterans and aging non-Veterans of the same sex over an eight-year period. This objective also contributes to the empirical understanding of latent psychological constructs relevant to the VAC framework, focusing on domains that are highly salient to Canadian Veterans and broadly recognized in well-being research, while also explicitly analyzing how these constructs manifest and evolve differently for Veteran and non-Veteran male and female groups.

Accordingly, we first sought to determine if our longitudinal modelling differed between Veterans and their same-sex non-Veteran counterparts. We predicted that military service would influence the relationships between psychosocial factors and depression, resulting in significantly different trajectories for Veterans and same-sex non-Veterans. Our second objective was to examine whether these pathways differed by sex within the Veteran cohort. We hypothesized that sex would not influence the relationship between psychosocial factors and depression, resulting in significantly different trajectories between male and female Veterans, postulating that the shared experience of military service would have sex as a less pronounced differentiator among Veterans compared to the general population.

Methods

The CLSA consists of a national, stratified, random sample of 51,338 participants aged 45 to 85 years as of 2011 (mean = 61.94, SD = 9.85) intended to be representative of the aging Canadian population. Participants for the tracking cohort were recruited from all 10 provinces. Participants for the comprehensive cohort were recruited from within 25-50 km of 11 data collection sites located in seven provinces.^37–39 It is important to note that while the CLSA sampling weights are designed to ensure the sample is representative of the broader aging Canadian population, recruitment was not stratified to be representative of the Canadian Veteran population itself. The VAC framework was used as a guide to relevant aspects of well-being in relation to Canadian Veterans. Recognizing the dynamic nature of well-being, VAC designed the framework with flexibility to add indicators of well-being domains,²⁵ allowing researchers to adapt indicators to specific data sources and research questions.

In the present study, items from the CLSA were systematically mapped to these seven domains via a two-stage approach. First, conceptual alignment was conducted by reviewing the descriptions and thematic content of each domain.²⁵ Based on this review, the CLSA data set was then queried for items that conceptually aligned with these domain descriptions, guided by the theoretical underpinnings of well-being in the psychological literature^23–24 and existing research within the Canadian military community.³⁴ Organized in Table 1 are relevant CLSA variables that were in line with the VAC framework domains that served as the basis for this study.

Table 1.

VAC domains and determinants of well-being and relevant CLSA variables

Well-being domains	VAC domain themes	CLSA potential indicators
Health	Physical, mental, social, spiritual	General and mental health
Purpose	Sense of meaning	Hobby, life satisfaction
Finances	Income/financial security	Statistics Canada poverty threshold
Social integration	Friends, family, community	MOS Scale
Life skills	Life management, coping, education	Alcohol, smoking, education
Housing/physical environment	Built and natural environment	Independent living
Culture/social environment	Values, beliefs, attitude of population	Connectedness to local area

VAC = Veterans Affairs Canada; CLSA = Canadian Longitudinal Study on Aging; MOS = Medical Outcomes Study.

Second, the identified CLSA items were evaluated for their suitability in our approach that relied on the representation of latent psychological constructs as determinants of well-being. This step was critical for ensuring that the selected indicators could be validly and reliably used in our longitudinal models. This revealed that three of the seven VAC domains — health, purpose, and social integration — had potential CLSA indicators that were amenable to being modelled as latent constructs. However, the health domain, as represented by available CLSA items, was captured by a single general health item as opposed to a validated scale. Therefore, the Center for Epidemiologic Studies Depression Scale Reduced 10 (CESD-R-10) was used in this study as an indicator of mental health, as it has been used successfully across wide age ranges, including older adults.⁴⁰

Therefore, our longitudinal model was ultimately designed to examine two core psychosocial domains of the VAC framework: purpose and social integration, and only a facet of the health domain. While the VAC framework is a holistic model that identifies seven key domains, our focus on latent psychological constructs precluded the inclusion of all domains. Several other domains, such as housing and life skills, are also operationalized with single categorical indicators (e.g., homelessness status) or disparate behaviours (e.g., smoking, educational attainment) that would inject potential testing bias because of re-running modelling for each subset category of interest.

It is important to clarify that depression is not conceptualized as a direct proxy for the broader health domain. Rather, its inclusion is theoretically relevant given the population of interest and depression’s strong relationship with the two domains assessed.²¹ Common symptoms related to depression, such as social withdrawal, can directly impact capacity for social integration while diminishing sense of meaning and life satisfaction. Given the higher prevalence of depression among Canadian Veterans and its strong comorbidity with physical health conditions, modelling the trajectories of purpose and social integration while accounting for depression’s influence is useful and provides insights relevant to the health domain.³³

This systematic process of conceptual mapping, followed by an assessment of analytical suitability guided our selection of variables. Two core VAC-aligned well-being domains were selected: social integration and purpose. Social integration was operationalized as the perceived availability and multidimensional nature of an individual’s social support via the Medical Outcomes Study (MOS) Social Support Survey,⁴¹ while purpose was measured as life satisfaction using the Satisfaction with Life Scale (SWLS).⁴² These domains, along with depression, form the basis of the longitudinal model proposed in Figure 2.

Figure 2.

Modelling purpose and social integration with depression

Grouping categories

Veteran status was the primary category attended to in this study. A Veteran was identified as any participant who held a role within a military organization as either an officer or as a non-commissioned member. As the CLSA included Veterans with foreign service, analyses were run twice: once including only Veterans who served in the CAF (i.e., without the Veterans who served exclusively in another military) and once with all Veterans. Assessment of the significance and directionality, considering the inclusion or exclusion of Veterans with foreign service, revealed no substantive differences. Therefore, the analyses presented include all Veterans.

Sex and Gender Equity in Research (SAGER) research guidelines were also considered from the outset of the study. Accordingly, sex and gender were initially identified as categorizing variables. Yet, given the non-static nature of gender across the data set, gender could not provide a stable grouping measure. In other words, there were participants that reported different genders over time, highlighting the fluidity of gender. Thus, categorization is split across sex (males and females) and not gender (men and women). While this approach does not fully reflect gender, research does support cross-sex-gender comparisons in the general population, given that most participants’ sex overlap with their gender identity.⁴³

Psychological constructs

Satisfaction with Life Scale (SWLS)

The SWLS is one of the most widely used psychometric assessments of well-being and is a specific indicator of the purpose domain in the VAC framework. As seen in Table 2, the SWLS consist of five items on a seven-point Likert scale anchored at 1 (strongly disagree) and 7 (strongly agree). Participants’ scores are the mean of the five items.⁴²

Table 2.

SWLS questions

In most ways, my life is close to my ideal.

The conditions of my life are excellent.

I am satisfied with my life.

So far, I have gotten the important things I want in life.

If I could live my life over, I would change almost nothing.

The MOS Social Support Survey

Two sub-scales from the MOS Social Support Survey were used to assess perceived social integration using a Likert scale anchored at 1 (none of the time) to 5 (all of the time), as seen in Table 3.⁴¹ The scale begins with the following prompt: “People sometimes look to others for companionship, assistance, or other types of support. How often is each of the following kinds of support available to you if you need it?”

Table 3.

MOS sub-scales

Emotional/informational support sub-scale
Someone you can count on to listen to you when you need to talk.
Someone to give you good advice about a crisis.
Someone to give you information to help you understand a situation.
Someone to confide in or talk to about yourself or your problems.
Someone whose advice you really want.
Someone to share your most private worries and fears with.
Someone to turn to for suggestions about how to deal with a personal problem.
Positive social interaction sub-scale
Someone to have a good time with.
Someone to get together with for relaxation.
Someone to do something enjoyable with.

Note: Given only two sub-scales of the overall MOS scale were used, this change in scale application required construct validation to identify a single factor was appropriate, as shown via parallel analysis.

CESD-R-10

The CESD-R-10⁴⁴ is broadly used to screen for depression symptoms in primary care using a four-point Likert scale anchored at 1 (rarely or none of the time) to 4 (most or all of the time). Studies have confirmed that it has strong psychometric properties, including predictive accuracy in community populations (see Table 4).^45,46

Table 4.

CESD-R-10 items

I was bothered by things that don’t usually bother me.

I had trouble keeping my mind on what I was doing.

I felt depressed.

I felt everything I did was an effort.

I felt hopeful about the future.*

I felt fearful.

My sleep was restless.

I was happy.*

I felt lonely.

I could not “get going.”

^*Indicates reverse coded items.

Parallel analysis

Parallel analysis was conducted to verify the single factor loading for each scale, which was especially needed for social integration as this included only two sub-scales of the overall MOS scale. Additionally, this process also validated these scales broadly across Canadians. Parallel analysis scree plots identified unidimensionality of factors, as expected.

Results

Presented here are the results of the longitudinal analysis, beginning with data cleaning and descriptive statistics. This eight-year cohort comprised 51,338 participants, and given the longitudinal approach assessing change across constructs of interest, missing data were assessed across each time point, and participants who did not respond to any items in a given scale were removed. From baseline in 2011, n = 8,393 participants did not respond to any one scale in full and were subsequently removed from analysis. An additional n = 7,877 from the first follow-up in 2015 and another n = 4,825 in the 2018 were also removed from analysis for the same reason. Data were then examined across each wave, and there was no clear systemic pattern in demographics from participants removed from analysis in comparison to those that were included.

Data imputation

Participants with no responses on a given scale were removed via listwise deletion, as imputing their scores would have yielded unreliable estimates. Following this step, partially missing data were then imputed using random forest independently across each time point, as these data likely contained both missing-at-random and missing-not-at-random data.⁴⁷ Imputation was run three times (once per time point).

Descriptive statistics

Analysis was conducted on n = 38,502 Canadians, of which n = 3,125 (8.12%) self-reported as Veterans (12.19% female, 87.81% male). Overall, n = 27,390 (71.14%) were married or in a common-law relationship, n = 10,605 (25.95%) reported complete retirement, n = 2,781 (7.22%) reported partial retirement, and n = 11,305 (29.36%) stated they were not retired. Table 5 presents additional categorical demographic information, whereas Table 6 provides details concerning the assessed latent constructs.

Table 5.

Categorical demographics

Variable	Category	Total (%)	Civilians (%)	Veterans (%)
Income (Individual)	<20,000	5,500 (14.28%)	5,236 (14.80%)	264 (8.45%)
	20,000-50,000	13,797 (35.83%)	12,675 (35.83%)	1,122 (35.90%)
	50,000-100,000	12,536 (32.56%)	11,355 (32.09%)	1,181 (37.79%)
	100,000-150,000	3,261 (8.50%)	2,940 (8.31%)	321 (10,27%)
	>150,000	1,799 (4.67%)	1,655 (4.68%)	144 (4.61%)
	Don’t know/Refused	1,724 (4.18%)	1,516 (4.29%)	93 (2.98%)
Income (Household)	<20,000	1,661 (4.31%)	1,560 (4.40%)	101 (3.23%)
	20,000-50,000	8,132 (21.12%)	7,510 (21.22%)	622 (19.90%)
	50,000-100,000	13,301 (34.55%)	12,078 (34.14%)	1,223 (39.14%)
	100,000-150,000	7,184 (18.65%)	6,577 (18.60%)	607 (19.42%)
	>150,000	6,032 (15.67%)	5,595 (15.82%)	437 (13.98%)
	Don’t know/Refused	2,192 (5.71%)	2,057 (5.81%)	135 (4.32%)
Ethnicity	White	36,765 (95.49%)	33,793 (95.52%)	2,972 (95.10%)
	South Asian	272 (0.71%)	259 (0.72%)	13 (0.42%)
	Black	259 (0.67%)	239 (0.68%)	20 (0.64%)
	Chinese	240 (0.62%)	232 (0.66%)	8 (0.26%)
	Latin American	115 (0.30%)	104 (0.29%)	11 (0.35%)
	Arab	80 (0.21%)	69 (0.20%)	11 (0.35%)
	Southeast Asian	54 (0.14%)	52 (0.15%)	2 (0.06%)
	Filipino	50 (0.13%)	46 (0.13%)	4 (0.13%)
	Japanese	47 (0.12%)	47 (0.13%)	0 (0.0%)
	West Asian	28 (0.07%)	17 (0.05%)	11 (0.35%)
	Korean	14 (0.04%)	12 (0.03%)	2 (0.06%)
	Other	578 (1.50%)	507 (1.43%)	71 (2.27%)
Veterancy	Reserves	980 (2.54%)	—	980 (31.36%)
	Army	857 (2.22%)	—	857 (27.42%)
	Air force	705 (1.83%)	—	705 (22.5%)
	Navy	431 (1.11%)	—	431 (13.80%)
	Other	147 (0.38%)	—	147 (0.05%)

Table 6.

Means and standard deviations across latent constructs

Variable (mean [SD])	Time 1 (2011)	Time 2 (2015)	Time 3 (2015)
SWLS	5.63 (1.27)	5.70 (1.26)	5.63 (1.39)
MOS Social Support	4.28 (0.71)	4.29 (0.72)	4.32 (0.74)
CESD-R-10	1.64 (0.50)	1.63 (0.50)	1.60 (0.50)

SWLS = Satisfaction with Life Scale; MOS = Medical Outcomes Study; CESD-R-10 = Center for Epidemiologic Studies Depression Scale Reduced 10.

Analysis

Modelling was conducted using R Studio, version 4.3.1. Multi-group SEM was applied, as it benefits the CLPM structure applied in this study.⁴⁸

Measurement invariance

To ensure that there was change across time rather than response shifts,⁴⁸ MG-CFA was used to assess measurement invariance of the constructs. It was found that the constructs were invariant across time, as shown in Table 7. This result implies that modelling measures actual change rather than a shift in response style across time. Thus, the measures of social integration, satisfaction with life, and depression were understood by the participants in the same manner across each assessment point from 2011 to 2018.

Table 7.

Comparative fit indices assessing measurement invariance across time

	Goodness-of-fit indices
Measurement	df	x2=	CFI	TLI	RMSEA	SRMR
Satisfaction with life
Configural	15	3,522.691	0.984	0.969	0.078	0.022
Metric	23	3,645.443	0.984	0.979	0.064	0.025
Scalar	31	3,850.166	0.983	0.984	0.057	0.025
Residual	41	10,405.535	0.954	0.966	0.081	0.040
Social integration
Configural	105	75,875.743	0.909	0.883	0.137	0.045
Metric	123	76,117.042	0.909	0.900	0.127	0.046
Scalar	141	76,737.542	0.908	0.912	0.119	0.046
Residual	161	77,870.808	0.907	0.922	0.112	0.047
Depression
Configural	105	24,853.159	0.894	0.864	0.078	0.041
Metric	123	25,019.851	0.893	0.883	0.073	0.042
Scalar	141	26,538.844	0.887	0.892	0.070	0.044
Residual	161	233,492.134	0.880	0.899	0.067	0.048

Notes: All tests were tested and found invariant over time when compared (configural-metric, metric-scalar, scalar-residual); all χ² were significant at the p < 0.001 level because of large sample size. CFI, TLI, and SRMR primarily showed good fit. While RMSEA in several cases was above the 0.07 threshold, this result is unlikely to be diagnostic of fit problems.⁴⁹

CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

Primary analysis was conducted by comparing measurement invariance across groups. As shown in Table 8, male and female Veterans showed model invariance. Yet male and female Veterans showed model variance in comparison to their same-sex civilian counterparts.

Table 8.

Comparative fit indices assessing measurement invariance across groups

	Goodness-of-fit indices
Group comparisons	df	x 2	CFI	TLI	RMSEA	SRMR
Male/female Veterans
Configural	5,204	17,839.216	0.992	0.991	0.039	0.050
Metric	5,314	19,474.326	0.991	0.990	0.041	0.052
Females (Veteran/non-Veteran)
Configural	5,204	97,889.713	0.990	0.989	0.042	0.043
Metric	5,314	100,212.772	0.990	0.989	0.042	0.043
Males (Veteran/non-Veteran)
Configural	5,204	87,459.264	0.991	0.990	0.041	0.045
Metric	5,314	89,536.511	0.991	0.990	0.041	0.046

Notes: There was no difference in nested model comparison between configural and metric slope constraints for Veterans in comparison to their civilian same-sex counterparts. Thus, differences are due to the model applying differently to each group.⁵⁰

CFI = comparative fit index; TLI = Tucker-Lewis index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

Longitudinal pathways

Given that the posited model fit differently when comparing same-sex Veterans and non-Veterans, it allowed for the interpretation of the cross-lagged regression pathways for these two models as shown in Table 9. Across all groups, the autoregressive paths for social integration, life satisfaction, and depression were strong and significant, indicating high stability in these constructs over time. These results are detailed below and presented visually in Figures 3 and 4.

Table 9.

Regression pathways for Veterans compared to same-sex non-Veterans

	Veteranfemale		Non-Veteran female		Veteranmale		Non-Veteran male
	Est.	SE	Est.	SE	Est.	SE	Est.	SE
Social integration (2011-15)
Social integration	0.719	0.014	0.692	0.003	0.700	0.007	0.682	0.003
Life satisfaction	NS	NS	0.097	0.003	0.161	0.009	0.115	0.004
Depression	NS	NS	0.021	0.006	0.059	0.017	NS	NS
Life satisfaction (2011-15)
Life satisfaction	0.744	0.031	0.742	0.005	0.800	0.015	0.739	0.006
Depression	NS	NS	-0.195	0.008	-0.066	0.025	-0.219	0.011
Depression (2011–15)
Life satisfaction	−0.063	0.020	0.742	0.005	−0.074	0.007	−0.092	0.004
Depression	0.728	0.048	−0.195	0.008	0.571	0.019	0.688	0.009
Social integration (2015-18)
Social integration	0.875	0.019	0.688	0.003	0.667	0.007	0.681	0.003
Life satisfaction	NS	NS	0.065	0.003	0.050	0.009	0.078	0.004
Depression	NS	NS	−0.032	0.007	−0.092	0.014	NS	NS
Life satisfaction (2015-18)
Life satisfaction	0.702	0.031	0.702	0.005	0.775	0.013	0.731	0.006
Depression	−0.261	0.038	−0.259	0.007	−0.325	0.022	−0.275	0.009
Depression (2015-18)
Life satisfaction	−0.078	0.019	−0.126	0.003	−0.115	0.008	−0.113	0.008
Depression	0.530	0.032	0.566	0.006	0.686	0.020	0.614	0.003

Notes: NS = not significant; all other listed regressions were significant at p < 0.0001.

Figure 3.

Veteran females compared to non-Veteran femalesV = Veteran; C = civilian (non-Veteran); grayscale double-headed arrows represent fitted covariance.

Figure 4.

Veteran males compared to non-Veteran malesV = Veteran; C = civilian (non-Veteran); grayscale double-headed arrows represent fitted covariance.

Female Veterans versus non-Veteran females

In the initial period (2011-15), the pathways for female Veterans differed notably from their civilian counterparts. For non-Veteran females, higher social integration significantly predicted higher subsequent life satisfaction (0.097), and higher life satisfaction was a significant protective factor against future depression (-0.195). In contrast, neither of these pathways was significant for female Veterans. The only significant cross-lagged effect for female Veterans in this period was the negative path from depression to subsequent life satisfaction (-0.063).

In the subsequent period (2015-18), a protective effect of life satisfaction against depression emerged for both groups, with nearly identical path estimates for female Veterans (-0.261) and non-Veteran females (-0.259). This result indicates that as this cohort aged, life satisfaction became an equally effective buffer against depression for all females. However, the influence of social integration remained distinct; it continued to predict higher life satisfaction for non-Veteran females (0.065) but remained non-significant for female Veterans.

Male Veterans versus non-Veteran males

For males, the relationship between latent constructs was more complex. During the 2011-15 period, higher social integration was a significant predictor of higher subsequent life satisfaction for both male Veterans (0.161) and to a lesser degree for non-Veteran males (0.115). Life satisfaction, in turn, was protective against later depression for both groups, although this effect was substantially stronger for non-Veterans (-0.219) than for Veterans (-0.066). Notably, a counterintuitive finding emerged for male Veterans in this period: higher social integration was associated with a slight increase in subsequent depression (0.059), a pathway that was not significant for non-Veteran males.

This dynamic shifted in the 2015-18 period. Higher social integration became protective against depression for male Veterans, with the pathway reversing to become negative and significant (-0.092). This result suggests a temporal shift in how social connections relate to mental health for male Veterans as they age. In this later period, life satisfaction remained a protective factor against depression for both male Veterans (-0.325) and non-Veteran males (-0.275). Finally, depression consistently predicted lower subsequent life satisfaction for all groups across both time periods.

Discussion

The present study longitudinally investigated key well-being determinants — specifically social integration, life satisfaction, and depression — among aging Canadian Veterans and their non-Veteran counterparts, analyzed separately by sex over an eight-year period using data from the CLSA. Guided by the VAC framework, this research sought to provide an empirical understanding of how these determinants, as latent psychological constructs, evolve and interact differently within these distinct population groups. The two key findings of this study are that 1) the applied measurements modelled similarly between male and female Veterans — which points to a shared psychosocial architecture shaped by military service — and 2) significant variance emerged when comparing Veteran groups to their same-sex non-Veteran peers. This result indicates that the longitudinal dynamics of social integration, life satisfaction, and depression differ based on Veteran status, highlighting unique pathways to well-being for those who have served in the military.

Interpretation of key findings

The central finding of model variance between same-sex Veterans and non-Veterans suggests that military service and the subsequent transition to post-service life fundamentally shape the long-term trajectories of the psychosocial constructs assessed. The regression pathways (Table 8) reveal these differing dynamics.

For female Veterans, a notable finding in the initial 2011-15 period was the non-significant direct predictive pathways from baseline social integration to subsequent life satisfaction and depression. This finding contrasts with non-Veteran females, where social integration positively predicted later life satisfaction. Such a divergence could imply that for female Veterans early in this aging cohort, the nature, source, or impact of social integration might be qualitatively different, or perhaps less directly influential on these specific outcomes, in the short term. This outcome could be influenced by unique post-service reintegration challenges, experiences of military-specific trauma, or different social network compositions, as highlighted by literature on women Veterans facing distinct barriers and risks.^2,17 However, over the longer term (2015-18), social integration for female Veterans showed strong stability, although direct paths to life satisfaction and depression from prior social integration remained non-significant, suggesting other factors might be more salient or that the benefits of social integration manifest differently. The strong negative path from life satisfaction to subsequent depression in the 2015-18 period for female Veterans (-0.261) stresses the protective role of life satisfaction, a finding consistent with literature emphasizing purpose and meaning in Veteran well-being.^21,30

For male Veterans, social integration consistently predicted subsequent life satisfaction in both observed periods (2011-15: 0.161; 2015-18: 0.050), a pattern also seen in non-Veteran males, although the strength of this association appeared somewhat stronger for Veterans initially. This finding aligns with the importance of camaraderie and social bonds often emphasized in military culture and post-service life.²⁵ Interestingly, for male Veterans, baseline social integration also had a modest predictive path to higher subsequent depression in the 2011-15 period (0.059), which was not observed for non-Veteran males. This counterintuitive finding warrants further exploration. While social support is a well-established protective factor against depression in Veterans,⁵¹ one potential explanation is that certain types of social networks for male Veterans, while present, might not always be functionally supportive. Indeed, some have questioned whether networks populated by individuals who are also in distress might place members at further socio-emotional risk.⁵² Such networks could become sources of stress through mechanisms such as co-rumination on shared trauma. This temporal dynamic might also reflect the unique challenges of military-to-civilian transition. This conclusion is particularly plausible given that the MOS Social Support Survey measures the perceived availability of social support, not necessarily its functional quality during interaction.⁴¹ A Veteran may accurately report having a network to use, but the act of social participation itself could become a source of stress if those individuals are unable to understand their military experiences — a phenomenon known as reverse culture shock.⁵³ Finally, the longitudinal nature of this finding supports a healthy help-seeking effect. Research consistently shows that Veterans face significant barriers to care, such as stigma and a military culture of self-reliance, which often delay the acknowledgement of mental health problems.⁵⁴ A strong social network can act as a catalyst for gradual encouragement, where partners or friends help the Veteran overcome these barriers over the subsequent period. This situation can lead to a greater willingness to acknowledge and report depressive symptoms later, making the statistical association appear as though social integration leads to depression when it may be leading to more honest self-reporting. The shift in the 2015-18 period, where higher social integration for male Veterans was linked to lower subsequent depression, suggests a more conventionally protective role as they age further. This reversal might indicate that over time, male Veterans successfully renegotiate their social networks, strengthening supportive bonds and aligning with the established role of social support as a critical protective factor for well-being.²⁶ Similar to female Veterans, higher life satisfaction was strongly protective against subsequent depression for male Veterans across both timeframes.

Across all groups, depression showed considerable stability (i.e., autoregressive paths). The negative impact of depression on subsequent life satisfaction was also consistently observed, emphasizing the pervasive detrimental effect of depression on broader well-being, a point that is well supported by literature indicating higher depression prevalence in Veterans and its link to diminished life satisfaction.³¹

The established invariance of the constructs over time confirms that the study captured genuine change in these psychological states rather than shifts in how participants understood the measures. The invariance between male and female Veterans suggests that despite differing life experiences, these fundamental constructs operate similarly within the Veteran population when comparing sexes. However, the variance when compared to non-Veterans clearly identifies that military service is a profound life-course experience that alters these fundamental psychological dynamics.

Connection to broader literature

These findings longitudinally support interconnected domains within the VAC framework, showing how social integration and purpose (via life satisfaction) dynamically interact with depression. Divergences between Veterans and non-Veterans align with literature on unique Veteran stressors (e.g., service exposures, transition challenges, differing social support) and life trajectories.^6,15,34

Distinct psychosocial pathways for Veterans align with the broader literature demonstrating that comparisons of health and well-being between Veteran and non-Veteran populations defy simplistic narratives of linear improvement or decline for either group. Veteran health varies by age and well-being domain, and while some studies note higher rates of diagnosed conditions in Veterans, self-rated health may not mirror these objective measures.⁵⁵ Recent reviews reinforce this point, arguing that aging actively is possible even with comorbidities and that the aging experiences of Veterans are highly varied, necessitating a departure from one-size-fits-all models.¹⁴ Furthermore, our longitudinal findings reinforce the value of viewing well-being as a dynamic process rather than a static state.⁵⁶ For instance, even day-to-day experiences can significantly influence indicators of successful aging, highlighting that trajectories of well-being are continually shaped by ongoing life events and psychosocial factors.⁵⁷ Our finding of distinct psychosocial pathways for female Veterans, where life satisfaction plays a more central protective role, resonates with research highlighting their unique challenges and the need for Veteran-sensitive approaches. For example, access to Veteran-centred spaces and support groups has been found to be critical for improving the mental health outcomes of older women Veterans, suggesting that the quality and nature of social integration might be uniquely important in shaping Veteran well-being trajectories.⁵⁷

Veteran population heterogeneity (e.g., combat status, service era) further complicates comparisons.⁵ Veterans may show simultaneous strengths and vulnerabilities, making comparisons multi-faceted.⁵⁵ For example, higher Veteran depression prevalence^2,7,21,34,35 likely contributes to distinct trajectories, as depression impacts perceptions of life satisfaction.³¹ This point is illustrated by post-9/11 U.S. Veterans reporting poorer health yet sometimes better health behaviours/access.⁵⁸

The differing pathways related to social integration are particularly salient. While military service is known for fostering strong camaraderie, the transition to post-service life can disrupt these networks. Making meaningful, new civilian connections can be challenging,²⁶ potentially leading to loneliness or perceived lower social support.⁵⁹ This challenge is widespread. In a national survey of U.S. Veterans, more than half (56.9%) reported feeling lonely, which is linked to adverse mental health outcomes and a more than threefold increase in suicidal ideation.⁶⁰ Consequently, older Veterans often express an interest in reconnecting with military peers, valuing the camaraderie and shared identity that civilian relationships might not provide.^14,61 Our finding that social integration’s predictive power toward depression differs between Veterans and non-Veterans, and even between initial and later periods for Veterans, provides needed complexity. Quality and nature of support over time, not just presence, are likely critical, especially for Veterans who have experienced trauma.⁵¹

Life satisfaction, reflecting purpose, is consistently linked to positive outcomes, and for Veterans, it is often linked to successful reintegration and positive service appraisal.²¹ This study reinforces life satisfaction as a strong protective factor against depression in aging Veterans, aligning with successful aging models emphasizing purpose and psychosocial strengths amid adversity.^14,21 In fact, results from a U.S. national study show that over 80% of older Veterans report aging successfully despite health challenges, attributing this result to protective factors such as resilience, gratitude, and a strong sense of purpose in life.¹²

Strengths and limitations

Notable strengths include the longitudinal design (three CLSA waves) — an advance over prevalent cross-sectional Veteran research. A sex-stratified, non-Veteran comparison group provides important context on Veteran aging. Rigorous methodology, including established scales and robust analyses also enhances confidence. Finally, grounding constructs within the VAC framework anchors the research in a theoretical and policy-relevant Canadian context.

Several limitations warrant acknowledgement. First, demographic constraints precluded intersectional analyses. The female Veteran sample was underpowered for granular sex- or gender-based analysis. Furthermore, insufficient representation of both 2SLGBTQI+ Veterans and ethnically diverse Veterans (including Indigenous and racialized groups) within the data set prevented meaningful analysis of their distinct well-being pathways. This is a critical gap, as equity-deserving Veteran groups often navigate unique challenges profoundly impacting their well-being trajectories,² and these challenges may be informed by experiences and effects of discrimination based on sex, gender, race, sexual orientation, or military sexual trauma. Such challenges, including gaps in sex- and gender-responsive and/or culturally safe care, and difficulties accessing services, particularly in remote regions, can significantly affect their well-being and trust in support systems. Ultimately, these unaddressed gaps can translate into significant unmet health and well-being needs, heightening vulnerabilities for diverse Veterans and adversely impacting their mental and social well-being as they navigate the complexities of aging and transition into later life.

Moreover, this study’s categorization based on sex (male/female) rather than gender (man/woman) constitutes another limitation. This approach was adopted because self-reported gender changed across the longitudinal data set, and while sex provided stable groups for analysis, it may not fully reflect well-being needs and experiences tied to gender identity and expression. Also, given that analysis was conducted only with participants who chose to complete scales to some degree, this study has potential bias in findings toward those who had some reason to complete all survey waves across the years assessed.

Last, certain CLSA exclusions (e.g., individuals on First Nations reserves or in long-term care at baseline) limit generalizability to specific segments of the Veteran population. CLSA items also constrained the operationalization of VAC well-being domains, with indicators for only three of seven domains represented by latent psychological constructs. Similarly, while the CLSA sampling weights ensure generalizability to the Canadian population, they are not weighted to reflect the demographic composition of the Canadian Veteran population. Consequently, while our findings are robust given the large cohort of Veterans contained within the CLSA, caution is warranted when generalizing these results to the national Veteran population.

While our findings provide a robust longitudinal analysis of Veteran well-being regarding latent psychological constructs within the domains of purpose and social integration, as well as a facet of the health via depression, a further limitation is that our model represents a partial rather than comprehensive assessment of the VAC framework. The exclusion of domains such as finance, housing, and culture/social environment means our model does not capture the influence of other critical indicators, objective life circumstances, or broader contextual factors on Veteran well-being trajectories. Yet this oversight highlights an important avenue for future research. While our study provides insight into the latent psychological constructs held within the VAC framework, subsequent longitudinal studies within other domains of well-being are needed. Such research in these categorical surveillance domains would complement this research, and it is a next clear step toward providing a more holistic understanding of the Veteran experience over the life course.

Finally, as CLSA does not capture military service characteristics (e.g., component, combat exposure) known to influence well-being,^15,62 this study could not examine specific aspects of military service over time.

Implications for policy, practice, and future research

That Veterans and non-Veterans display differing well-being pathways based on sex suggests the need to support aging given the differing needs of both male and female Veterans. For instance, non-Veteran females reported life satisfaction benefits from earlier social integration, while female Veterans did not. This finding indicates that earlier interventions may be needed to help female Veterans translate social connections into life satisfaction.¹⁷ Bolstering life satisfaction and purpose can have a strong protective effect against depression in all Veteran groups. Consequently, early and sustained attention to mental health, particularly for depression, remains critical because of the consistent and deleterious impact that depression can have on life satisfaction.

This study highlights several important directions for future research. Further investigation into aging Veterans could explore the mechanisms underlying observed pathway differences. Qualitative methodologies could illuminate lived experiences Veterans hold regarding their well-being shaping these trajectories, complementing current data.¹⁴ A more holistic understanding of aging Veteran well-being also requires longitudinally investigating other domains (e.g., finances, life skills, housing) via quantitative approaches such as regression trees.

Additionally, it is critical to prioritize collecting and analyzing data from diverse Veteran sub-groups (Indigenous, racialized, 2SLGBTQI+) to address knowledge gaps.⁴ Examining how specific military (e.g., combat exposure, injury, military sexual trauma) and transition experiences (e.g., securing employment, social adjustment) moderate these longitudinal pathways is also crucial.^6,55,62,63

Conclusion

This longitudinal study, leveraging epidemiological-level data, demonstrates important differences in the temporal effects of social integration, life satisfaction, and depression between aging Canadian Veterans and non-Veterans, with distinct patterns emerging for males and females in comparison to their civilian counterparts. These findings underscore the lasting impact of military service on fundamental psychosocial processes and highlight the necessity of Veteran-centric, sex-specific, and longitudinally informed approaches to understand, monitor, and promote well-being in Canada’s aging Veteran population in order to ensure effective evidence-based support for those who have served.

Biographies

William Van Veen is a Captain in the Canadian Armed Forces employed as a personnel selection officer, having also conducted research for the Director General of Military Personnel Research and Analysis. He continues to serve, with over 20 years in the Royal Canadian Air Force, much of which was as communicator researcher non-commissioned member, where he served a tour of duty in Afghanistan. William holds a master’s in psychology from Carleton University and is currently completing doctoral studies there.

André Lauzon, PhD, is a senior research associate with the applied research team at the Atlas Institute for Veterans and Families and an adjunct professor in the Department of Psychology at Acadia University. His current research focuses on aging Veterans and predictors of flourishing in Canadian Veterans. He previously led the development of a randomized-controlled-trial-tested e-health program and played a key role in the national Caring Forward project. He received his PhD in psychology from the University of Ottawa.

Nicolas Laham is a quantitative researcher with extensive experience in complex statistical methodologies, including advanced regression analyses, structural equation modelling, and psychometrics, as well as holding a broad understanding of a variety of data science techniques. He has completed a master’s in psychology, concentrating on data science, and is in the process of completing a PhD in psychology, concentrating on quantitative methodology, at Carleton University. Nicolas also holds interest in the study of, and teaching in, the philosophy of science and epistemology.

Gray Gaudett is a PhD student in psychology at Carleton University and a research assistant at the Atlas Institute for Veterans and Families. Their research primarily focuses on factors that predict and maintain recovery from an addictive behaviour. Specifically, they are looking at how nostalgic reverie (i.e., sentimental longing) for an addictive behaviour undermines recovery. Other research interests include craving and responsible gambling programming.

Sara Rodrigues, PhD, is an adjunct professor in the School of Psychology at the University of Ottawa, and Director of Applied Research at the Atlas Institute for Veterans and Families. Her current research examines the social determinants of health for Veterans and Families. She has published in European Journal of Psychotraumatology, Human Studies, International Journal of Feminist Approaches to Bioethics, and other journals, and she is co-editor of On the Politics of Ugliness (2018). She received her PhD in social and political thought from York University.

Fardous Hosseiny, MSc, is President/CEO of the Atlas Institute for Veterans and Families. A leading advocate for those impacted by trauma, he supported the first UN event dedicated to mental health as part of Canada’s delegation to UNGA 73. At Atlas, he is a member of the Five Eyes Mental Health Research Collaboration and an executive advisor to GAMHA. Fardous has co-authored over 40 peer-reviewed publications, holds an MSc in the neuroscience of addiction, and is a Certified Health Executive.

Stefania Maggi, PhD, is an associate professor at Carleton University in Childhood and Youth Studies and the Department of Psychology. Her research spans social determinants of youth development, climate change, community resilience, and early career development. Her international collaborations target climate change effects on youth well-being and leverage immersive technology for positive development and environmental action. She founded Mochi4ThePlanet, which promotes climate coping and emotional resilience, and she also actively contributes to several academic and rights-based groups focused on childhood, development, and climate psychology.

Funding Statement

The work of the Atlas Institute is made possible thanks to funding from Veterans Affairs Canada.