Characteristics and Correlates of Ten-year Trajectories of Posttraumatic Stress Symptoms in Older U.S. Military Veterans

Abstract

Objectives:

To examine the nature and correlates of 10-year trajectories of posttraumatic stress disorder (PTSD) symptoms in older U.S. military Veterans.

Design and Setting:

A nationally representative web-based survey of older U.S. Veterans who participated in the National Health and Resilience in Veterans Study over 5 waves between 2011 and 2021.

Participants:

1,843 U.S. Veterans aged 50 and older (mean age=67).

Measurements:

PTSD symptoms were assessed using the PTSD Checklist. Self-report measures at baseline assessed sociodemographic characteristics; trauma exposures; psychiatric and substance use disorders; mental, cognitive, and physical functioning; and psychosocial factors including expectations of aging. Latent growth mixture modeling identified the nature and correlates of 10-year PTSD symptom trajectories.

Results:

Most of the sample had no/low PTSD symptoms (88.7%), while 6.0% had consistently subthreshold symptoms, 2.7% consistently high symptoms, and 2.6% increasing symptoms. Relative to the no/low symptom group, the subthreshold and high symptom groups reported more medical conditions and cognitive difficulties, with younger age and more lifetime traumatic events additionally linked to the high symptom trajectory. Relative to the no/low symptom group, Veterans with increasing symptoms were more likely to report functional disability and lifetime nicotine use disorder, cognitive difficulties, negative expectations regarding physical and emotional aging, and traumatic events over the study period.

Conclusions:

Despite high rates of trauma exposure, most older Veterans do not evidence symptomatic PTSD trajectories; however, about 11% do. Results underscore the importance of assessing PTSD symptoms in this population and considering longitudinal trajectories as well as associated risk and protective factors.

INTRODUCTION

By later adulthood, most individuals will have experienced at least one potentially traumatic event.¹ Some traumatic events may be more likely to occur at younger ages—such as combat and motor vehicle accidents, while others are more likely at older ages—such as life-threatening illness and the unexpected death of a loved one.² Although most older adults do not have PTSD,³ greater traumatic exposure over the lifecourse has been associated with higher rates of posttraumatic stress disorder (PTSD).³

The prevalence of PTSD in older adults is relatively low, ranging from 1–3%^4,5 but is higher in older women^6,7 and those with combat trauma,⁸ with a pooled prevalence of 8% for older Veterans in a meta-analysis.⁹ In cross-sectional studies of older adults, PTSD is associated with cognitive decline and dementia,¹⁰ as well as worse physical health,¹¹ impaired functioning,¹² and comorbid psychiatric and substance use disorders.⁶ Feelings of social detachment and estrangement are a symptom of PTSD,¹³ which may partially explain why loneliness is associated with PTSD in older adults.¹⁴ The relationship between loneliness and PTSD is concerning given that loneliness and social isolation are risk factors for late-life suicide¹⁵ and mortality.¹⁶ Rates of subthreshold or partial PTSD (i.e., meeting some, but not all, diagnostic criteria) in older adults are higher than full PTSD, estimated at 17% for older Veterans who utilize services through the U.S. Veterans Affairs Healthcare System.³ Like full PTSD, subthreshold PTSD is associated with worse cognitive, physical, and mental health, including suicidal ideation and attempts.³

Numerous case reports¹⁷ and empirical studies^18–20 have described the occurrence of exacerbated, emergent, or re-emergent PTSD symptoms with aging. In one of the largest studies (N=1,450), a 25-year follow-up in a national sample of Vietnam Veterans, 16% of Veterans reported an increase in PTSD symptoms and 8% reported a decrease.²¹ The long length of time between the two data collection points, however, may not accurately capture potential intricacies of discrete symptom trajectories over time.

Some longitudinal studies indicate that older persons may experience increased symptoms of PTSD in later life, associated with cognitive changes¹⁸ and experiences of new traumas or substance use disorder,¹⁹ while other longitudinal studies suggest persistent symptom and no-symptom (resilient) trajectories over time.^19,22 The Later-Adulthood Trauma Reengagement (LATR) model²³ frames emergent and re-emergent PTSD symptoms in late-life as reflecting normative processes of reminiscence about personal memories in the context of age-associated role changes (e.g., retirement), illness (e.g., cancer), and social loss (e.g., spousal bereavement). In the LATR model, unresolved negative cognitions about the meaning of traumatic events as one ages may lead to emotional distress, including PTSD symptoms. This model is consistent with studies describing how negative age stereotypes were associated with greater risk for PTSD,²⁴ pointing to how maladaptive meaning-making about aging may influence distress. Alternatively, adaptive meaning-making and engagement with social supports may lead to optimization of gains and mitigation of losses as described in theories of selective optimization with compensation in aging.²⁵

To further understanding of the longitudinal course of PTSD in older adults over multiple measurement periods, data from a contemporary, nationally representative sample of older U.S. military Veterans were analyzed to address two aims: (1) characterize predominant 10-year PTSD symptom trajectories utilizing multiple waves; and (2) describe and quantify the relative importance of sociodemographic, military, psychosocial (including negative age stereotypes) and trauma characteristics associated with PTSD symptom trajectories. Based on prior studies, we hypothesized we would find: (1) evidence of 4 longitudinal PTSD symptom trajectories¹⁹; and (2) association of elevated PTSD symptoms with cognitive changes¹⁸, lower social support²⁶, trauma exposure³, and negative age stereotypes²⁴.

METHODS

Participants

Participants were 1,843 Veterans aged 50 and older (mean=66.7, SD=9.3, range=50–93) who participated in the 2011–2021 National Health and Resilience in Veterans Study (NHRVS). The NHRVS sample was drawn from KnowledgePanel^®, a probability-based, online survey panel maintained by GfK Knowledge Networks, Inc. (now Ipso), which includes over 50,000 households representing approximately 98% of the U.S. adult population. Panel members who endorsed a history of military service were asked to participate in the NHRVS. Wave 1 of the NHRVS was completed in 2011 by 3,157 Veterans and consisted of an anonymous, 60-minute online survey. A total of 2,157 Veterans (68.3% of the Wave 1 cohort) completed a Wave 2 survey in 2013, 1,538 (71.3% of the Wave 2 cohort) completed a Wave 3 survey in 2015, 1,310 (85.2% of the Wave 3 cohort) completed a Wave 4 survey in 2018, and 1,007 (76.9% of the Wave 4 cohort) completed a Wave 5 survey in 2021. Participants in the current study included 1,843 Veterans aged 50 and older at the 2011 assessment (Wave 1) who had PTSD symptom data at a minimum of two follow-up assessments (mean number of follow-ups=3.0 (SD=0.80, range=2–4). The Institutional Review Board of the VA Connecticut Health Care System approved the NHRVS; all participants provided informed consent prior to participation.

Variables

PTSD Symptoms.

The PTSD Checklist-Specific (PCL-S)²⁷ assessed past-month PTSD symptoms according to the Diagnostic and Statistical Manual for Mental Disorders-IV (DSM-IV) in Wave 1 (17 items, range 17–85, α=0.95). The PTSD Checklist-5 (PCL-5;20 items, range 0–80) was used in Waves 2, 3, 4, and 5 (α’s=0.95 to 0.96); to harmonize the two versions of the PCL, scores were standardized at each timepoint and analyzed as continuous variables. Symptoms were rated in response to self-reported ‘worst’ traumatic event on the Trauma History Screen (if no traumatic event was endorsed, a score of 0 was assigned). A positive screen for PTSD at Wave 1 was operationalized as a score ≥46, which corresponds to a cut score of ≥31 on the PCL-5²⁸; these thresholds were used to label symptom trajectories as above or below a PTSD threshold at each timepoint.

Independent Variables.

The Trauma History Screen (THS)²⁹ assessed the occurrence of 13 potentially traumatic life events (PTE) at any point in life at the Wave 1 assessment. In the NHRVS, life-threatening illness or injury was added. At Waves 2, 3, 4, and 5, the Life Events Checklist for DSM-5³⁰ assessed potentially traumatic events (PTE) over the preceding time periods.

Demographic (age, gender, race/ethnicity, education, household income, marital and employment status), military (enlisted vs. drafted, combat status, years served), health (medical, disability, cognitive, lifetime history of psychiatric and substance use disorders), and psychosocial variables (protective psychosocial traits, social connectedness, altruism, religiosity/spirituality, negative age stereotypes, active lifestyle, history of mental health treatment), measured at Wave 1, were examined as predictors of PTSD symptom trajectories (Supplemental Table 1).

Data Analysis

Latent growth mixture modeling (LGMM) identified 10-year PTSD symptom trajectories using Mplus version 8.8. The relative fit of 1-to-6 class solutions were examined using Akaike’s Information Criterion (AIC), Bayesian Information Criterion (BIC), Lo–Mendell–Rubin adjusted likelihood test, Entropy values, and bootstrap likelihood ratio tests, as well as parsimony (i.e., selecting solution with fewer classes), clinical and theoretical considerations of each solution (i.e., selecting solution consistent with hypothesized trajectories based on prior work), and size of smallest class (i.e., >2% of the sample). Chi-square analyses and analyses of variance (ANOVAs) were used to compare trajectories on Wave 1 variables, as well as the number of PTE since the Wave 1 assessment. Variables that differed between PTSD symptom trajectory groups were incorporated into conditional LGMMs to identify those that differentiated symptomatic PTSD trajectories. When composite variables (e.g., number of medical conditions) were significantly associated with trajectories, post-hoc analyses were conducted to examine which specific components were independently associated with the trajectory. Relative importance analyses (RIAs) were then conducted to determine the relative variance in symptomatic trajectories of each significant correlate identified in the multivariable model. These analyses decompose the overall variance explained (R²) by regression models into proportional contributions while accounting for intercorrelations among these variables, thus quantifying the relative significance of each independent variable. We utilized p<.05 as a criterion across analyses to consider a wide range of potential correlates of PTSD symptom trajectories. In interpreting and discussing results we emphasize those variables with the highest RIA values. Post-stratification weights based on geodemographic data from the most contemporaneous U.S. Census Bureau Current Population Survey were applied in inferential data analyses to permit generalizability to the general U.S. veteran population.

RESULTS

Trajectories of PTSD Symptoms

Table 1 shows fit statistics of 1- to 6-class unconditional latent growth mixture models of PTSD symptoms over the 10-year study period. Based on fit statistics, theoretical and clinical considerations, and class sizes, the 4-class model was selected as the best-fitting solution. Mean class probabilities were high, averaging 0.934 (range=0.872–0.988) suggesting that the average likelihood of trajectory assignments was high.

Table 1.

Fit indices for one- to six-class unconditional latent growth mixture models of PTSD symptoms over the 10-year study period.

	BIC	SSA-BIC	AIC	Entropy	LMR LRT p value	Bootstrapped LRT p value	% of sample in smallest class
1-class	14774.828	14743.058	14719.636	--	--	--	100%
2-class	13591.461	13663.210	13591.461	0.985	<.0001	<.0001	5.3%
3-class	13277.317	13226.486	13189.011	0.958	0.006	<.0001	3.0%
4-class	13074.785	13014.423	12969.921	0.959	0.54	<.0001	2.6%
5-class	12754.162	12684.269	12632.741	0.952	0.13	<.0001	0.9%
6-class	12718.476	12639.030	12580.476	0.941	0.26	<.0001	0.8%

Note. BIC=Bayesian Information Criterion; SSA-BIC=sample size-adjusted Bayesian Information Criterion; AIC=Akaike Information Criterion; LMR LRT=Lo-Mendell-Rubin Likelihood Ratio Test.

Bolded text indicates model selected as the best fit to the 10-year PTSD Checklist data based on fit statistics, parsimony, theoretical and clinical considerations, and size of smallest class.

Degrees of freedom for LMR LRT and Bootstrapped LRT = number of classes - 1.

As shown in Figure 1, the majority of the sample had no/low PTSD symptoms (88.7%), while 6.0% had consistently subthreshold symptoms, 2.7% had consistently high symptoms, and 2.6% had increasing symptoms. Intercept and slope estimates for each trajectory are shown in the Figure 1 footnote.

Figure 1. Trajectories of PTSD symptoms over the 10-year study period.

Note. Percentages are weighted; error bars represent 95% confidence intervals for slope estimates.

Black dashed line represents cut score for positive screen for PTSD (Score ≥46 using DSM-IV version of the PTSD Checklist [Wave 1 or Year 0] and ≥31 on the DSM-5 version of the PTSD Checklist²⁷ [Waves 2, 3, 4, and 5 or Years 2, 4, 7, and 10).

No/low symptoms: intercept (standard error [SE])= −0.244 (0.018); slope (SE)=0.002 (0.003); mean class probability=0.872

Increasing symptoms: intercept=0.392 (0.194); slope=0.325 (0.059); mean class probability=0.906

Subthreshold symptoms: intercept=1.748 (0.231); slope= −0.083 (0.023); mean class probability=0.969

High symptoms: intercept=4.018 (0.317); slope= −0.135 (0.057); mean class probability=0.988

Traumatic Life Events by PTSD Symptom Trajectory

Table 2 shows PTE by PTSD symptom trajectory. Across the total sample 86.1% reported exposure to at least one potentially traumatic event during their lifetime. Results revealed significant differences in endorsements of all PTE assessed, with the high symptom group having the highest prevalence of PTE. The high symptom group was also more likely than the increasing and subthreshold symptom groups to report an assaultive potentially traumatic event (See Table 2 footnote) as their index trauma at the Wave 1 assessment. Examination of individual index traumas revealed significant group differences (χ²(42)=136.75, p<0.001), with Bonferroni-corrected comparisons indicating that the increasing symptom group was significantly more likely than the no/low symptom group to report being hit or kicked hard enough to injure during adulthood as their index event (5.7% vs. 0.7%); and that the high symptom group was more likely than the no/low symptom group to report being attacked with a gun, knife, or weapon (9.3% vs. 2.5%), and less likely than the no/low, subthreshold, and increasing symptom groups to report a sudden death of close family member or friend (7.0% vs. 34.8%, 31.2%, and 40.0%, respectively). The high and subthreshold symptom groups were also more likely than the no/low symptom group to report a military trauma as their index event (30.2% and 16.1% vs. 6.9%).

Table 2.

Lifetime potentially traumatic life events by PTSD symptom courses over the 10-year study period

	No/low Symptom n=1,634 (weighted 88.7%) 1	Increasing n=51 (weighted 2.6%) 2	Subthreshold n=117 (weighted 6.0%) 3	High n=41 (weighted 2.7%) 4	Overall test of difference χ2	Pairwise contrasts
	N (weighted %)	N (weighted %)	N (weighted %)	N (weighted %)
Sudden death of close family member or friend	967 (59.7%)	33 (66.7%)	99 (81.7%)	36 (88.4%)	31.49***	4,3>1
A hurricane, flood, earthquake, tornado, or fire	571 (33.6%)	16 (26.2%)	52 (41.5%)	22 (52.4%)	9.79*	NS
Saw someone die suddenly or get badly hurt or killed	568 (32.8%)	21 (31.0%)	70 (52.2%)	33 (83.7%)	59.75***	4>3>1
Life-threatening illness or injury	493 (30.9%)	25 (36.6%)	56 (50.5%)	24 (66.7%)	37.49***	3,4>1; 4>2
During military service - saw something horrible or was badly scared	398 (23.3%)	21 (46.3%)	67 (52.7%)	32 (83.7%)	118.08***	4>3>1
A really bad car, boat, train, or airplane accident	286 (18.0%)	13 (22.0%)	30 (16.1%)	15 (46.5%)	23.03***	4>1,3
Attacked with a gun, knife, or weapon	254 (16.1%)	14 (21.4%)	44 (30.9%)	25 (65.1%)	77.11***	4>3>1
Sudden abandonment by spouse, partner, parent, or family	220 (14.8%)	12 (16.7%)	48 (39.8%)	22 (66.7%)	103.51***	4>3>1,2
Hit or kicked hard enough to injure – as a child	205 (12.4%)	14 (19.0%)	23 (17.4%)	17 (42.9%)	34.39***	4>1,3
Sudden move or loss of home and possessions	167 (11.5%)	9 (14.6%)	40 (36.2%)	20 (59.5%)	114.72***	4>1,2
Hit or kicked hard enough to injure – as an adult	168 (10.1%)	10 (19.0%)	33 (22.3%)	23 (57.1%)	94.99***	4>3>1
A really bad accident at work or home	158 (9.0%)	10 (12.2%)	32 (22.3%)	23 (60.5%)	121.97***	4>3>1
Forced or made to have sexual contact – as a child	83 (5.3%)	2 (5.1%)	16 (11.8%)	5 (11.6%)	9.23*	NS
Forced or made to have sexual contact – as an adult	34 (2.4%)	2 (4.8%)	3 (2.2%)	6 (11.9%)	14.88**	4>1
Other Criterion A-qualifying trauma	106 (6.0%)	10 (14.3%)	36 (33.0%)	24 (55.8%)	191.85***	4>1,2
Assaultive index trauma	100 (6.2%)	6 (11.8%)	4 (3.2%)	9 (19.0%)	14.09**	4>1,3

Note. Statistically significant group difference:

^*p<0.05;

^**p<0.01;

^***p<0.001.

Assaultive index trauma=attacked with a gun, knife, or weapon; hit or kicked hard enough to injure – as a child; hit or kicked hard enough to injure – as an adult; forced or made to have sexual contact – as a child; forced or made to have sexual contact – as an adult.

Degrees of freedom for each χ² test =3.

Demographic, Military, Health, and Psychosocial Characteristics by PTSD Symptom Trajectory

Table 3 shows demographic, military, health, and psychosocial characteristics by PTSD symptom trajectory. Group differences were observed for all assessed variables except current employment status, enlistment status, years of military service, engagement in altruistic activities, and three active lifestyle indices (sports/exercise, reading, and writing).

Table 3.

Demographic, military, health, and psychosocial characteristics of 10-year courses of PTSD symptoms in older U.S. military Veterans

	No/low Symptom N=1,634 (weighted 88.7%) 1	Increasing N=51 (weighted 2.6%) 2	Subthreshold N=117 (weighted 6.0%) 3	High N=41 (weighted 2.7%) 4	Overall test of difference F or χ2	df	Pairwise contrasts
	Weighted mean (SD) or N (weighted %)	Weighted mean (SD) or N (weighted %)	Weighted mean (SD) or N (weighted %)	Weighted mean (SD) or N (weighted %)
Demographic characteristics
Age	67.5 (9.6)	63.8 (9.0)	65.1 (10.0)	60.3 (7.0)	10.71***	3, 1475	1,3>4
Male sex	1,538 (94.9%)	45 (92.9%)	99 (88.3%)	34 (86.0%)	12.67**	3	1>3
Caucasian race/ethnicity	1,428 (82.4%)	44 (75.6%)	94 (73.4%)	28 (57.1%)	21.46***	3	1>4
Some college or higher education	1,410 (69.6%)	48 (80.5%)	102 (74.5%)	38 (92.9%)	13.38**	3	4>1
Married/living with partner	1,322 (78.2%)	41 (76.2%)	84 (71.0%)	28 (54.8%)	14.80**	3	1>4
Household income ≥ $60K/year	908 (48.4%)	26 (31.7%)	48 (34.0%)	16 (34.9%)	13.82**	3	1>3
Currently employed	553 (32.4%)	24 (40.5%)	38 (27.7%)	9 (16.3%)	7.17	3	--
Military characteristics
Enlisted in military	1,359 (83.9%)	45 (90.5%)	100 (86.0%)	34 (74.4%)	4.52	3	--
Combat exposure	546 (31.2%)	18 (31.7%)	50 (38.7%)	28 (74.4%)	36.82***	3	4>1,2,3
Years served in military	7.1 (7.9)	7.6 (7.7)	7.7 (7.5)	9.8 (10.5)	1.60	3, 1475	--
Physical and cognitive health
Number of medical conditions	2.8 (1.8)	2.8 (1.7)	3.8 (1.9)	5.9 (2.0)	46.05***	3, 1475	4>1,2,3; 3>1,2
Any ADL disabilityb	37 (2.5%)	6 (11.9%)	13 (9.7%)	8 (14.0%)	37.22***	3	2,3,4>1
Any IADL disability	122 (7.3%)	16 (28.6%)	38 (36.6%)	22 (59.5%)	195.08***	3	4>2>1; 3>1
Cognitive difficulties	6.5 (8.5)	17.0 (14.3)	22.5 (18.5)	43.2 (34.9)	201.81***	3, 1475	4>2,3>1
Trauma and psychiatric history
Number of lifetime traumas	2.8 (2.3)	3.6 (3.3)	5.1 (3.0)	8.6 (2.9)	96.42***	3, 1475	4>3>1,2
Traumas since baseline	2.6 (2.7)	4.6 (3.7)	4.6 (4.0)	7.8 (5.5)	54.92***	3, 1475	4>2,3>1
Lifetime major depressive disorderc	156 (8.7%)	12 (14.3%)	51 (33.0%)	28 (69.0%)	184.58***	3	4>3>1
Substance use disorder history
Lifetime alcohol use disorderc	634 (39.8%)	23 (40.5%)	70 (62.8%)	28 (64.3%)	27.99***	3	4,3>1
Lifetime drug use disorderc	155 (10.0%)	9 (19.0%)	30 (20.4%)	18 (39.5%)	45.53***	3	4,3>1
Lifetime nicotine use disorderd	299 (20.3%)	16 (48.8%)	38 (25.5%)	16 (34.9%)	24.46***	3	2>1,3
Protective psychosocial traits
Resiliencee	30.2 (6.3)	27.5 (6.0)	28.2 (8.1)	21.5 (13.0)	25.24***	3, 1475	1>3>4
Dispositional gratitudef	6.2 (1.0)	6.0 (1.0)	6.0 (1.4)	5.1 (1.7)	17.18***	3, 1475	1,2,3>4
Purpose in lifeg	22.0 (3.9)	21.3 (4.7)	20.5 (6.2)	14.7 (5.9)	42.14***	3, 1475	1,2,3>4; 1>3
Dispositional optimismh	5.0 (1.3)	5.0 (1.4)	4.6 (1.9)	3.7 (1.9)	12.89***	3, 1475	1,2,3>4
Curiosityi	5.2 (1.3)	4.8 (1.3)	5.3 (1.6)	4.0 (2.0)	12.37***	3, 1475	1,2,3>4; 1>3
Community integration	4.5 (1.6)	4.3 (1.3)	3.7 (2.1)	2.2 (1.4)	32.31***	3, 1475	1,2,3>4; 1,3>4
Social connectedness
Number of close friends/relatives	9.3 (9.5)	8.6 (13.9)	7.4 (10.1)	2.8 (3.5)	6.81***	3, 1475	1,2>4
Secure attachmentj	1,305 (79.4%)	31 (60.0%)	54 (56.4%)	7 (14.3%)	121.35***	3, 1475	1>2,3>4
Social supportk	20.1 (4.5)	19.1 (5.8)	17.6 (5.9)	12.8 (5.3)	38.59***	3, 1475	1,2,3>4
Loneliness	4.0 (1.5)	4.7 (1.8)	5.6 (2.2)	7.6 (1.6)	94.26***	3, 1475	4>3>2>1
Altruism
Volunteer on a weekly basis	694 (42.7%)	14 (24.4%)	37 (24.7%)	9 (14.6%)	27.63***	3	1>3,4
Altruism more than 10 times/yearl	683 (40.0%)	18 (31.0%)	55 (42.6%)	19 (54.8%)	5.43	3	--
Religiosity / spirituality
Frequency religious servicesm	3.1 (1.8)	2.3 (1.7)	3.0 (1.9)	2.0 (1.4)	6.99***	3, 1475	1>2,4; 3>4
Frequency private spiritual activitiesm	3.5 (2.2)	2.6 (2.1)	3.9 (2.2)	2.9 (2.0)	4.10**	3, 1475	3>2
Intrinsic religiositym	10.2 (3.8)	9.0 (4.1)	10.7 (4.1)	9.0 (4.1)	3.03*	3, 1475	NS
Negative expectations regarding aging
Physical aging	2.1 (0.7)	2.5 (0.6)	2.2 (0.7)	2.0 (0.9)	5.78***	3, 1475	2>1,4
Emotional aging	0.9 (0.8)	1.3 (0.8)	1.2 (1.0)	1.1 (0.9)	8.32***	3, 1475	2,3>1
Cognitive aging	1.8 (0.7)	1.7 (0.9)	1.3 (1.1)	1.3 (1.1)	7.49***	3, 1475	1,2,3>4
Active lifestyle
Number of days/week sports/exercise	2.6 (2.3)	1.8 (2.2)	2.6 (2.6)	2.1 (2.5)	1.85	3, 1475	--
Number of days/week reading	4.3 (2.7)	3.3 (2.5)	4.3 (2.8)	3.2 (2.9)	3.86**	3, 1475	NS
Number of days/week writing	1.2 (1.9)	1.5 (1.8)	1.1 (1.8)	0.8 (1.9)	0.72	3, 1475	--
Number of days/week use computer	6.0 (1.9)	6.1 (1.7)	6.2 (1.5)	4.5 (2.7)	8.76***	3, 1475	1,2,3>4
Ever received mental health treatment	244 (14.7%)	17 (26.2%)	60 (40.9%)	27 (76.2%)	139.86***	3	4>3>1

Note. Statistically significant group difference:

^*p<0.05;

^**p<0.01;

^***p<0.001.

df=degrees of freedom

Multivariable Predictors of PTSD Symptom Trajectories

Increasing vs. No/Low Symptom.

Table 4 shows results of multivariable analyses of PTSD symptom trajectories. Relative to the no/low group the increasing symptom group was younger (mean age 63.8), more likely to have an impairment in one or more instrumental activities of daily living (IADL) disability and lifetime nicotine use disorder, and reported more cognitive difficulties, negative expectations regarding physical and emotional aging, and PTE over the study period.

Table 4.

Results of multivariable analysis of sociodemographic, health and psychosocial variables associated with symptomatic PTSD symptom trajectories in older U.S. military Veterans

	Increasing vs. No/Low	Subthreshold vs. No/Low	High vs. No/low	High vs. Subthreshold
Nagelkerke R2	0.27	0.37	0.80	0.73
	RRR (95%CI)	RRR (95%CI)	RRR (95%CI)	RRR (95%CI)
Age	0.94 (0.90–0.98)**	--	0.89 (0.81–0.97)*	--
Male sex	--	--	--	--
Caucasian race/ethnicity	--	--	--	--
Some college or higher education	--	--	--	--
Married/living with partner	--	--	--	--
Household income ≥ $60K/year	--	--	--	--
Combat exposure	--	--	--	11.57 (1.55–86.52)*
Number of medical conditions	--	1.17 (1.02–1.34)*	2.59 (1.73–3.87)***	2.28 (1.49–3.48)***
Any ADL disability	--	--	--	--
Any IADL disability	2.83 (1.09–7.33)*	2.41 (1.28–4.54)**	--	--
Cognitive difficulties	1.04 (1.01–1.08)**	1.05 (1.03–1.90)***	1.09 (1.03–1.14)***	1.04 (1.01–1.09)*
Number of lifetime traumas	--	--	1.34 (1.08–1.66)**	1.40 (1.05–1.87)*
Assaultive index trauma	--	--	--	8.40 (1.98–35.71)**
Traumas since baseline	1.17 (1.06–1.29)**	1.16 (1.07–1.25)***	--	--
Lifetime MDD	--	2.15 (1.11–4.15)*	--	--
Lifetime alcohol use disorder	--	1.77 (1.01–3.15)*	--	--
Lifetime drug use disorder	--	--	--	--
Lifetime nicotine use disorder	3.56 (1.70–7.43)***	--	--	--
Resilience	--	--	--	--
Dispositional gratitude	--	--	--	--
Purpose in life	--	--	--	--
Dispositional optimism	--	--	--	--
Curiosity	--	--	--	0.54 (0.31–0.94)*
Community integration	--	--	0.49 (0.31–0.78)**	--
Number of close friends/relatives	--	--	--	--
Secure attachment	--	--	--	--
Social support	--	--	--	--
Loneliness	--	1.42 (1.23–1.64)***	2.51 (1.68–3.75)***	--
Volunteer on a weekly basis	--	0.30 (0.16–0.57)***	--	--
Church few times monthly+	--	--	--	--
Private spiritual activities per week+	--	--	--	0.48 (0.30–0.77)**
Negative expectations of physical aging	2.50 (1.25–5.00)**	1.70 (1.09–2.70)*	--	--
Negative expectations of emotional aging	1.72 (1.10–2.70)*	--	--	--
Positive expectations of cognitive aging	--	--	--	--
Number of days/week use computer	--	--	0.53 (0.38–0.72)***	0.65 (0.45–0.94)*
Ever received mental health treatment	--	--	--	--

Note. RRR=relative risk ratio; 95% CI= 95% confidence interval; degrees of freedom for each comparison=1.

IADL – instrumental activities of daily living. ADL = activities of daily living. MDD = major depressive disorder.

Results of the RIA revealed that cognitive difficulties (24.6% relative variance explained [RVE]), negative expectations of emotional aging (23.4%), and number of PTE over the study period (17.7%) explained the majority of the variance in predicting the increasing symptom trajectory, with negative expectations of physical aging (12.4%), IADL disability (11.8%), age (6.8%), and nicotine use disorder (3.3%) explaining the remainder of the variance in this outcome.

Subthreshold vs. No/Low Symptom.

Relative to the no/low group, the subthreshold symptom group reported more medical conditions, cognitive difficulties, and negative expectations of physical aging, and were more likely to endorse an IADL disability, and lifetime history of major depressive disorder (MDD) and alcohol use disorder (AUD). They also scored higher on a measure of loneliness, were less likely to be engaged in volunteerism, and reported more PTE over the study period. Planned secondary analyses revealed that the subthreshold symptom group was more likely to have been diagnosed with sleep disorder (44.7% vs. 14.4%; Wald χ²(1)=40.30, p<0.001; OR=2.47, 95%CI=1.32–4.61), respiratory disease (24.5% vs. 8.0%; Wald χ²(1)=15.61, p<0.001; OR=3.30, 95%CI=1.66–6.56), hypertension (72.3% vs. 57.7%; Wald χ²(1)=5.75, p=0.017; OR=2.07, 95%CI=1.09–3.93), and liver disease (7.4% vs. 1.8%; Wald χ²(1)=4.41, p=0.036; OR=3.41, 95%CI=1.01–11.55).

Results of the RIA revealed that cognitive difficulties (30.3% RVE) and loneliness (25.0%) explained the majority of the variance in predicting the subthreshold symptom trajectory, with sleep disorder (10.6%), PTE over the study period (8.6%), lifetime MDD (8.5%), hypertension (7.1%), lifetime AUD (3.2%), respiratory disease (2.7%), non-engagement in volunteering (1.7%), liver disease (1.5%), and negative expectations of physical aging (0.8%) explaining the remainder of the variance.

High vs. No/Low Symptom.

Relative to the no/low group, the high symptom group was younger and reported more medical conditions, cognitive difficulties, and PTE. They also scored higher on a measure of loneliness, lower on a measure of community integration, and reported fewer weekly days of computer use. Planned secondary analyses revealed that the high symptom group was more likely to have been diagnosed with sleep disorder (85.7% vs. 14.4%; Wald χ²(1)=42.18, p<0.001; OR=5.45, 95%CI=1.06–28.11), migraine headaches (40.5% vs. 3.6%; Wald χ²(1)=11.98, p<.001; OR=15.51, 95%CI=2.00–120.19), and respiratory disease (35.7% vs. 8.0%; Wald χ²(1)=9.91, p=0.002; OR=3.84, 95%CI=1.66–8.87).

Results of the RIA revealed that cognitive difficulties (23.8% RVE), loneliness (23.0%), and lifetime PTE (15.0%) explained the majority of the variance in predicting the High symptom trajectory, with sleep disorder (11.5%), lower community integration (10.2%), lower computer use (5.4%), respiratory disease (4.4%), migraine headaches (3.7%), and younger age (3.0%) explaining the remainder of the variance in this outcome.

High vs. Subthreshold Symptom.

Relative to the subthreshold group, the high symptom group were more likely to serve in a combat role, reported more medical conditions, cognitive difficulties, and lifetime PTE, and was more likely to endorse an assaultive index event. They also scored lower on a measure of curiosity and reported lower frequency of engagement in private spiritual activities and computer use. Planned secondary analyses revealed that the high symptom group was more likely to have been diagnosed with sleep disorder (85.7% vs. 44.7%; Wald χ²(1)=9.40, p=0.002; OR=5.94, 95%CI=1.90–18.54), migraine headaches (40.5% vs. 6.5%; Wald χ²(1)=6.87, p=0.009; OR=6.36, 95%CI=1.59–25.41), and arthritis (69.8% vs. 46.8%; Wald χ²(1)=5.78, p=0.016; OR=2.56, 95%CI-1.19–5.52).

Results of the RIA revealed that cognitive difficulties (21.6% RVE), lifetime PTE (20.0%), and lower curiosity (12.0%) explained the majority of the variance in predicting the High symptom trajectory, with migraine headaches (11.5%), assaultive index trauma (9.3%), arthritis (8.6%), sleep disorder (7.2%), lower engagement in computer use (3.9%), and private spiritual activities (3.5%), and combat exposure (2.4%) explaining the remainder of the variance.

DISCUSSION

Using longitudinal data across five waves of data collection from a nationally representative sample of U.S. Veterans aged 50 and older, this study examined the nature and associated risk factors for 10-year PTSD symptom trajectories. Four PTSD symptom trajectories were identified. Approximately 11% displayed symptomatic trajectories, including those whose symptoms were consistently at a subthreshold (6.0%) or full-threshold high (2.7%) level, and those with increasing symptoms (2.6%). However, most (88.7%) reported consistently no/low PTSD symptoms.

Lifetime trauma exposure was common, ranging from an average of 3 among those with no/low symptoms to 9 in those with high symptoms; greater trauma exposure was associated with high symptoms in RIA analyses. This rate of trauma exposure is similar to that seen in non-Veteran older adults (Mean=6, SD=6).³¹ Study participants reported non-combat traumatic events, such as unexpected death of a loved one and life-threatening illness,^2,31 as well as combat-exposure. Yet, despite the frequent occurrence of trauma exposure, the majority did not have PTSD symptoms, which is consistent with other cross-sectional and longitudinal studies.^18,19,22 The pattern of overall no/low PTSD symptoms in those aged 50 and older may be explained, in part, by relatively low trauma exposure, benefits from possible prior PTSD treatment, adaptive emotional regulation in aging,²⁵ and/or a selection effect (death) due to the impact of accelerated aging (also known as premature senescence) and multiple morbidities in those with PTSD.³² Longitudinal data spanning multiple decades, including information on those who drop from longitudinal sampling, are needed to clarify potential subgroups of healthy agers.

Two symptom trajectories demonstrated consistently elevated symptomatic presentations, although a slight downward trend was indicated for both. Using current population benchmark estimates of 14.2 million total Veterans in the U.S. aged 50 and older,³³ these findings suggest approximately 850,000 Veterans may be experiencing consistently subthreshold PTSD symptoms, and 380,000 Veterans may be experiencing consistently high PTSD symptoms, for a total of approximately 1.2 million older Veterans. It is important to attend to both groups in light of recent findings that subthreshold PTSD conveys similar psychosocial risks as full PTSD,³ and prior work suggesting cut scores for diagnosis of PTSD should be lower for older adults to identify clinically distressing symptoms that may benefit from treatment.³⁴

Relative importance analyses pointed to cognitive difficulties as distinguishing the subthreshold, high, and increasing symptom groups from the no/low group. This finding adds to the literature by demonstrating a relationship between cognition and 10-year symptom trajectories. The relationship of cognition and PTSD is complex, and likely bidirectional, in which cognitive symptoms represent a core symptom of PTSD¹³, and PTSD may elevate risks for later life neurocognitive disorder³⁵.

Relative importance analyses also found that loneliness distinguished subthreshold and high symptom PTSD trajectories relative to the no/low trajectory. This finding is consistent with prior studies that document relationships between PTSD and low social support²⁶. For older adults, this finding is important in the context of increasing attention given to the health risks of loneliness^16,36 and evidence that a lack of belongingness increases risk for late-life suicide, particularly in the context of increased burdensomeness.^15,37 These findings underscore the importance of assessing social connectedness in older Veterans with PTSD, as well as targeting social functioning in older Veterans with PTSD, to enhance well-being and mitigate risks of mental disorders in late life.³⁸

A total 2.6% of the sample, representing approximately 365,000 older Veterans in the U.S, experienced increasing symptoms of PTSD over the 10-year study period. This finding, extends previous work in this area.^18,19,22 It is possible that these symptoms represent a new emergence of symptoms or re-emergence of PTSD symptoms experienced earlier in life.¹⁸

Negative expectations of emotional aging were associated with the increasing PTSD symptoms trajectory. Embracing beliefs such as, “it is normal to be depressed when you are old” - may represent internalized ageism - which has been linked to numerous deleterious aging-related outcomes,³⁹ including PTSD.²⁴ Additional risk factors for increasing symptoms included the experience of recent traumatic events. This is consistent with a 25-year follow-up of Vietnam Veterans indicating that greater past-year stress²⁶ predicted PTSD symptom exacerbation. Thus, mitigating the effects of recent stressors are important targets for efforts designed to prevent and treat PTSD symptoms in older Veterans. The three most prevalent lifetime traumas in the increasing symptom group were sudden death of a close family member and life-threatening illnesses, more common in later life,² and military-related trauma. When also associated with changes in cognitive or other aspects of functioning, these experiences may precipitate engagement with traumatic memories, such as is predicted in the LATR model.²³

Limitations

This study has limitations. First, the use of self-report measures for PTSD, cognition, functional abilities, and other variables, rather than direct clinician assessments, could have led to less reliable symptom reports. Second, the number of individuals in the high and increasing symptom trajectories was small, which likely impacted statistical power and may reduce the potential replicability of results. Third, while 10-year longitudinal data including Veterans with data from at least two time points was utilized, Veterans who dropped out of later study periods may represent those who are more ill or who died. Fourth, the NHRVS focuses on the general population of U.S military Veterans who are predominantly older, male, and white. Consequently, the results here may not generalize to more demographically diverse Veteran subgroups or non-Veterans. Fifth, there were no baseline measurements for the participants prior to age 50, which limits the understanding of temporal links between independent variables and PTSD symptom trajectories.

Conclusions

Results of this study suggest that most older Veterans are emotionally resilient as they age, including those who are experienced traumatic events in their lifetime. However, approximately 11% of older Veterans may present with symptomatic PTSD trajectories, collectively representing approximately 1.6 million U.S. Veterans aged 50+. These results emphasize the importance of trauma-informed care for older Veterans across all health care settings. Some older adults with PTSD symptoms may not meet full diagnostic criteria or may not have been previously diagnosed with PTSD, emphasizing the importance of ongoing assessment of trauma histories and PTSD symptoms, and referral for those who find symptoms troublesome and may benefit from clinical attention. Understanding distinct symptom trajectories and their associated risk factors may be useful in identifying such Veterans and providing treatments that are sensitive to recent and distant traumas, functional status, and opportunities for promoting symptom resolution and resilience.

HIGHLIGHTS.

• What is the primary question addressed by this study?

  What are the 10-year trajectories of posttraumatic stress disorder (PTSD) symptoms in older U.S. military Veterans and what defines those at risk for increasing symptoms with aging?

• What is the main finding of this study?

  No/Low PTSD symptoms over time was most common, followed by consistently high, consistently sub-threshold, and increasing symptoms. Those with increasing symptoms were more likely to report cognitive and functional change, negative expectations of aging, and traumatic events over the study period.

• What is the meaning of the finding?

  Most older Veterans are resilient, but some may have continuing or increasing PTSD symptoms with aging indicating a need to consider age-related changes, internalized ageism, and the occurrence and timing of traumatic events.

Data Statement

The data have not been previously presented orally or by poster at scientific meetings.

Data Availability Statement for this Work

Research data are not available