Premorbid Traumatic Stress and Veteran Responses to the COVID-19 Pandemic

Abstract

The COVID-19 pandemic has had unprecedented effects on lifestyle stability and physical and mental health. We examined the impact of pre-existing PTSD, alcohol use disorder (AUD), and depression on biopsychosocial responses to the pandemic, including psychiatric symptoms, exposure to COVID-19, and housing/financial stability, among 101 U.S. military Veterans enrolled in a longitudinal study of PTSD, a population of particular interest given their trauma histories and defense readiness training. Participants (83.2% male; 79.2% white; mean age = 59.28 years) completed pre-pandemic clinician-administered psychiatric diagnostic interviews and a phone-based assessment between May and September 2020 using a new measure, the Rapid Assessment of COVID-19-Related Experiences (RACE), which assessed COVID-19 pandemic responses and its effects on mental and physical health. COVID-19 diagnosis and testing were also extracted from the electronic medical record. Multivariate regressions showed that, controlling for demographics, pre-pandemic PTSD (β = .332; p = .003) and AUD symptoms (β = .228; p = .028) were associated with increased pandemic-related PTSD symptoms. Pre-pandemic AUD was associated with increased substance use during the pandemic (β = .391; p < .001) and higher rates of self-reported or medical record-based COVID-19 diagnosis (β = .264; p = .019). Minority race was associated with pandemic-related housing/financial instability (β = −.372; p < .001), raising concerns of inequities in the population. Results suggest that pre-existing PTSD and AUD are markers for adverse pandemic-related psychiatric outcomes and for COVID-19 illness. These findings carry implications for the importance of targeting prevention and treatment efforts for those at greatest risk.

In late 2019, the first cases of the coronavirus disease 2019 (COVID-19) from the SARS-CoV-2 virus were reported in China. COVID-19 rapidly grew to the scale of a global pandemic, and by March 2020, there was widespread community transmission necessitating masking, social distancing, quarantines, school closures, and other major lifestyle changes to mitigate exposure and disease spread. As the only global pandemic since the Spanish Flu of 1918–19, COVID-19 has had an unprecedented impact on society, the economy, and healthcare. The effects of COVID-19 on physical health and mortality have been devasting for the global population, especially for the elderly, racial/ethnic minorities, and those with preexisting chronic health conditions (Shadmi et al., 2020). The pandemic also poses risks to mental health and wellbeing (Pfefferbaum & North, 2020), including anxiety concerning threats to health and safety, depression and loneliness relating to social isolation and loss, and the potential for pandemic-related increases in substance misuse and housing and financial instability. There is also stress associated with the need for increased vigilance (e.g., to avoid coronavirus exposure) and adaptability (e.g., to follow changing guidelines and community regulations).

Numerous studies have demonstrated the detrimental effects of the COVID-19 pandemic on mental health. In a self-report study of 2,485 college students, those who lived in areas with higher rates of COVID-19 infection experienced greater symptoms of PTSD and depression compared to students living in less affected areas (Tang et al., 2020). Similarly, in a sample of 7,143 undergraduates, those who had family or friends who contracted the coronavirus reported greater symptoms of anxiety compared to those without proximal exposure to the virus (Cao et al., 2020). Increases in anxiety were also reported among students who experienced pandemic-related disruptions to daily life or economic standing (Cao et al., 2020). Thus, indirect effects of the pandemic, such as employment or financial instability, may also contribute to worsened mental health (Mimoun et al., 2020). In light of this, identification of individuals at greatest risk carries important public health implications, such that it could inform prevention and intervention efforts so as to mitigate the deleterious long-term effects of the pandemic.

Individuals with pre-existing mental health conditions are likely among those at increased risk for detrimental outcomes in response to the pandemic. Pre-existing PTSD may be a vulnerability, given that exposure to new onset stressors (e.g., the COVID-19 pandemic) have been shown to increase PTSD symptom severity associated with prior traumatic events (Bramsen et al., 2006; Schock et al., 2016). In addition, the omnipresent threat and mandates to social distance stemming from the COVID-19 pandemic may reinforce or even exacerbate symptoms of PTSD, such as emotional disconnect from others, hypervigilance about safety, behavioral avoidance of crowds and other social situations, and maladaptive cognitions about the basic safety of the world. In a web-based self-report study using a large, cross-sectional sample of UK Veterans with pre-existing mental health conditions (i.e., PTSD, anxiety, depression, anger difficulties, alcohol misuse), Murphy et al. (2020) found that the pandemic exacerbated pre-existing psychiatric symptoms, especially symptoms of depression, anxiety, and PTSD. This effect was moderated by social support such that worsening of pre-existing mental health conditions was greater among individuals with low levels of social support (Murphy et al., 2020).

Health-related anxiety about the virus and isolation resulting from social distancing also has potential to exacerbate other disorders that are often comorbid with PTSD, such as depression and alcohol use disorder (AUD). Given the association between loneliness and depression (Aylaz et al., 2012; Barg et al., 2006), it is possible that pandemic-related social isolation may worsen depression. Similarly, drinking alcohol has been shown to be a coping mechanism for stress (Hasking et al., 2011; Woolman et al., 2015) and trauma (Hawn et al., 2020), and a recent study demonstrated that COVID-related stress is associated with increased alcohol use in the general population (Rodriguez et al., 2020). Thus, for those who have pre-existing depression and AUD, especially in conjunction with PTSD, pandemic-related stress may worsen these symptoms.

Pre-existing mental health conditions may also increase risk for poorer physical health outcomes related to the pandemic. A review by Bailey et al. (2021) posits that pre-existing AUD might increase risk of contracting COVID-19 and increase disease severity. This was supported by the results of a retrospective study of 73,099,850 patients (12,030 with a COVID-19 diagnosis and 7,510,380 with a substance use disorder [SUD] diagnosis) in an electronic health database of records from hospitals and health providers across the United States, which demonstrated that patients with SUD were at increased risk of contracting COVID-19 (Wang et al., 2021). Additionally, in a retrospective study of 7,348 adult COVID-19 patients, schizophrenia spectrum disorders were associated with an increased risk of mortality (Nemani et al., 2021), further demonstrating the potential impact of psychiatric conditions on COVID-19 illness risk and severity.

Though numerous studies have examined the effects of both the COVID-19 pandemic broadly and the viral illness specifically on mental health, there has been relatively less research to date concerning how a range of pre-existing psychiatric diagnoses and symptoms might predict biopsychosocial responses to the pandemic including, psychiatric symptoms, exposure to the disease, and financial and housing stability during the pandemic. The studies that do exist have been limited by cross-sectional and retrospective designs and the use of web-based self-report data. Temporal separation of predictors and outcomes using a longitudinal design and clinician-based assessments may offer a more valid and nuanced approach to evaluating mental health predictors of psychiatric symptoms in response to the pandemic. This would allow for targeted prevention, monitoring, and intervention to be made available to those at greatest risk. Our group had the opportunity to address these questions as we were in the midst of a longitudinal study of Veterans with PTSD and comorbid psychiatric conditions when the pandemic began. The primary aim of this study was to examine how baseline (pre-pandemic) clinician-assessed traumatic stress (PTSD, AUD, depression) predicted three biopsychosocial responses to the pandemic: (a) changes in mental health symptoms during the COVID-19 pandemic; (b) exposure to the SARS-COV-2 virus and associated illness; and (c) changes in housing and financial stability. We developed a new measure, the Rapid Assessment of COVID-19-Related Experiences (RACE), to evaluate these outcomes, and thus a secondary aim was to conduct a preliminary evaluation of the measure’s psychometric characteristics (reliability and factor structure). To our knowledge, no other measure to date assesses changes in a range of pandemic-related experiences among those with pre-existing trauma and psychiatric symptoms.

We examined these issues in a longitudinal sample of 101 trauma-exposed U.S. military Veterans. Trauma-exposed Veterans are a particularly unique population to study for these purposes because social distancing mandates and the threat of COVID-19 may reinforce or exacerbate pre-existing PTSD symptoms, while at the same time, Veterans have unique military and disaster preparedness training that might also influence responses to the pandemic. We hypothesized that greater psychiatric symptoms at baseline would predict worse pandemic-related psychiatric and housing/financial outcomes. We also hypothesized that baseline psychiatric disorders, specifically AUD and PTSD, would increase risk for COVID-19 illness, given that pre-existing SUDs have been shown to increase risk for COVID-19 (Wang et al., 2021) illness and that PTSD is associated with adverse physical health outcomes (Wolf & Schnurr, 2016). Given that demographic characteristics are differentially associated with PTSD (e.g., minorities, women, and younger individuals tend to have greater symptoms; Frueh et al., 2007; Ditlevsen & Elklit, 2010; Koenen et al., 2003) and that the pandemic is disproportionately affecting minority populations (e.g., in terms of job insecurity, risk for COVID-19-related death; Boserup et al., 2020; Couch et al., 2020), we simultaneously examined age, sex, education, and minority race and ethnicity in association with biopsychosocial responses to the pandemic.

Method

Participants

Participants (n = 101) were drawn from a cohort of Veterans who were enrolled in an ongoing longitudinal study of PTSD and common comorbid psychiatric conditions, including depression and AUD, prior to the start of the COVID-19 pandemic. To be eligible for the longitudinal study, Veterans (age 18+) had to endorse a history of trauma exposure. Veterans were excluded if they were at immediate risk of self- or other-harm or if they were under the acute influence of substances. To be eligible for the follow-up COVID-19-related phone assessment, participants had to have completed the most recent follow-up assessment prior to the onset of the pandemic (N = 175), and this assessment was used as the baseline data for these analyses. Of the 175 eligible participants, 102 of them were reached by telephone for this assessment; of these, one declined to participate. Participant characteristics are provided in Table 1. The resulting study sample was primarily male (83.2%), white (79.2%), and middle-aged (M = 59.28; SD = 12.92). The most common types of traumas the participants experienced were combat or warfare (43.6%), unwanted sexual contact as an adult (11.9%), and an assault by an acquaintance or stranger (5.9%). Most participants had experienced multiple types of traumatic experiences (M = 7.01, SD = 4.40). Analyses revealed no significant differences with respect to age, sex, minority status, education, or PTSD, AUD, and depression diagnosis and severity among eligible participants who did versus did not participate in the COVID-19 phone assessment (smallest p = .159).

Table 1.

Participant Characteristics (N=101)

Variable	M (SD) or n (%)
Age	58.28 (12.92)
Sex (Male)	84 (83.2)
Hispanic or Latinx	7 (6.9)
Race	21 (20.8)
White	82 (81.2)
Black or African American	16 (15.8)
American Indian or Alaska Native	6 (5.9)
Native Hawaiian or other Pacific Islander	1 (1.0)
High School Education or More	78 (77.2)
Pre-pandemic PTSD diagnosis	44 (43.6)
Pre-pandemic MDD diagnosis	24 (23.8)
Pre-pandemic AUD diagnosis	19 (18.8)

Note. M = mean; SD = standard deviation; PTSD = posttraumatic stress disorder; MDD = major depressive disorder; AUD = alcohol use disorder. Categories for race were not mutually exclusive.

Procedure

Eligible Veterans were contacted by phone and asked to participate in a brief phone-based assessment relating to the COVID-19 pandemic. Elements of informed consent were discussed at the start of the phone call. Phone assessments took place between May and September 2020. On average, the assessments occurred 522.05 days (SD = 234.17; range: 138 – 1005) after the participant’s most recent in-person study evaluation. The pre-pandemic assessment was completed in-person and involved an approximate 4.5-hour protocol that included self-report assessments, structured diagnostic interviews, a brief neuropsychological assessment, and a blood draw. The diagnostic assessments were videotaped, and approximately 30% of them were coded by a second rater to determine diagnostic reliability. Participants had scheduled breaks for food and were able to take additional breaks as needed. The protocol was approved by the local institutional review board. Participants received $135 for their participation in the pre-pandemic in-person assessment and $20 for their participation in the COVID-19 phone assessment. In addition to asking participants about whether they had been diagnosed with COVID-19 as part of the phone assessment, we also had permission to view the VA electronic medical record (EMR) to determine COVID-19 testing and diagnoses. Medical record information was evaluated as of April 13, 2021.

Measures

Rapid Assessment of COVID-19-Related Experiences (RACE; Wolf & Fein-Schaffer, 2020).

This questionnaire was newly developed for the purposes of this study to efficiently measure a range of both adverse and adaptive biopsychosocial responses to the COVID-19 pandemic, including change in psychiatric symptoms; to date, no other measures have been established for this purpose. The measure was rationally derived after reviewing the literature concerning mental health questionnaires developed in response to prior SARS viruses. The RACE consists of 26 items across five subscales that capture recent self-reported pandemic-related changes in: (1) housing/financial stability (two items); (2) mood/anxiety symptoms (three items); (3) substance use (three items); (4) and PTSD symptoms (eight items). The substance use subscale included questions about alcohol use, non-alcohol substance use, and prescription drug use. A fifth subscale assessed personal and friend/family COVID-19 exposure (five items). In addition, the RACE includes descriptive items assessing social distancing habits (three items), concern over contracting COVID-19 (one item), and a rating of change in sense of personal resilience during the pandemic (one item). The personal resilience item was included to ensure that the RACE covered both adaptive and maladaptive psychiatric responses, as there is evidence that resilience and psychiatric symptoms are at opposite ends of the same underlying construct (Wolf et al., 2018).

Psychiatric symptom items and the personal resilience item on the RACE anchored response options to the past two-weeks and followed a standard structure with Likert-like response options. For example, the item assessing pandemic-related increases in PTSD-associated nightmares was as follows: “In the past two weeks, to what extent has the COVID-19 pandemic affected your nightmares about past trauma?” Response options were on a 5-point scale ranging from “I am much less bothered by trauma nightmares than usual” (1) to “I am a lot more bothered by trauma nightmares than usual” (5) with no change in symptoms at the mid-point of the scale. The COVID-19 exposure items (e.g., “Have you been diagnosed with COVID-19 by a healthcare professional?”) and questions about social distancing (e.g., “Were you able to practice social distancing starting around mid-March?”) used dichotomous (yes/no) response options. Affirmative responses to the initial COVID-19 items related to personal and friend/family COVID-19 illness were followed by additional questions assessing degree of illness (e.g., to determine history of hospitalization, intubation, etc.). For the purposes of this paper, we refer to self-exposure to COVID-19 illness as exposure (based on self-report and medical record data) while exposure of a participant’s family or close friend to COVID-19 is referred to as proximal exposure. The psychiatric symptom and resilience response options were keyed so that higher scores indicated more pathological outcomes. The housing/financial subscale was keyed so that higher scores indicated greater stability. The mood/anxiety and PTSD symptom subscales demonstrated adequate (α=.636) to good (α=.831) internal consistency, respectively. Reliability was not assessed for the other subscales, as they were either comprised of a single item or items that were not expected to covary because they were assessing distinct phenomena. Items were read to participants over the phone along with the possible response options. The RACE is included in the supplementary materials.

Clinician-Administered PTSD Scale for DSM-5 (CAPS; Weathers et al., 2013).

The CAPS is the gold standard diagnostic tool for PTSD. The CAPS was administered by interviewers, ranging from trained bachelor’s level psychology technicians to licensed clinical psychologists, during the pre-pandemic in-person assessment and was used to assess current (i.e., past month) PTSD diagnosis and symptom severity. PTSD symptoms were anchored to the participant’s self-identified pre-pandemic worst traumatic experience. The diagnostic reliability based on a subset of about 30% of the interviews from the initial data collection demonstrated good inter-rater reliability (kappa for PTSD diagnosis = .78; intra-class correlation coefficient for PTSD severity: r = .78).

Structured Clinical Interview for DSM-5 (SCID; First et al., 2015).

Sections of the SCID (modules for major depression, substance use disorders, generalized anxiety disorder, panic disorder, agoraphobia, and antisocial personality disorder from the SCID-PD [First et al., 2016]) were administered at the pre-pandemic in-person assessment following standard SCID administration rules. For this study, we examined common diagnoses on the SCID (major depressive disorder and alcohol-use disorder; see Table 1) and symptom summary scores on these modules as predictors of COVID-19-related outcomes. The diagnostic reliability from the pre-pandemic data collection demonstrated good inter-rater reliability for SCID-based diagnoses (kappa for depression diagnosis = .94; kappa for AUD diagnosis = .93).

Data Analysis

As the RACE was newly developed for the purposes of this study, we first conducted a confirmatory factor analysis (CFA) of items included in the mood/anxiety and PTSD subscales using the weighted least squares estimator (WLSMV) to account for the categorical nature of the response options. The CFA was conducted using the Mplus 8.5 statistical modeling software (Muthén & Muthén, 1998–2020) and evaluated using standard fit indices and guidelines (Hu & Bentler, 1999), including root mean square error of approximation (values < .06 suggest good fit), standardized root mean square residual (values < .08 are indicative of good fit), and the comparative and Tucker-Lewis fit indices (values ≥ .95 are consistent with good model fit). These fit statistics were evaluated together, such that a single fit statistic that fell outside these guidelines in a model that otherwise demonstrates good fit to the data would not negate the overall acceptability of the model. We also tested a competing model based on the initial CFA results and compared the fit using a nested chi-square test, adjusting for the use of the WLSMV estimator using the ‘DIFFTEST’ function in Mplus.

We then conducted bivariate correlations in SPSS Statistics 26 to assess associations among baseline psychiatric symptoms, pandemic-related psychiatric symptoms, and COVID-19 exposure. We next ran five linear regression equations to examine how pre-pandemic psychiatric symptom severity predicted pandemic-related changes in psychiatric symptoms (PTSD, mood/anxiety, substance use) and experiences (COVID-19 exposure and housing/financial stability). In each regression, pre-pandemic PTSD, AUD, and depression severity were included as predictors together in the model, controlling for age, sex, race (minority vs. non-minority), and education. For the analyses predicting COVID-19 exposure, we used a variable that reflected those who either self-reported a diagnosis of COVID-19 or who had a positive test result in the electronic medical record. Pre-pandemic PTSD severity was calculated by summing the severity scores for each item on the CAPS per the standard scoring algorithm (Weathers et al., 2013). Pre-pandemic AUD and depression severity were calculated by summing the scores (reflecting threshold, sub-threshold, or negative ratings for each DSM-5 criterion) of the items from the SCID. If a SCID module was discontinued due to lack of initial item endorsement per standard SCID administration guidelines, severity scores of 0 were assigned for unassessed items from that module. Follow-up analyses replaced these three severity scores with pre-pandemic PTSD, MDD, and AUD current diagnoses to see if significant results held when diagnostic determinations were used as the predictors of COVID-19 related outcomes. Additional follow-up analyses were performed to determine if significant effects remained after controlling for proximal COVID-19 exposure, and separately, housing and financial stability, to evaluate if effects attributed to psychopathology were better accounted for by COVID-19-related stress resulting from illness among family and friends or housing instability. There were no missing data in these analyses.

Results

Sample Characteristics with Respect to COVID-19 Exposure and Impact

Sample characteristics with respect to COVID-19 exposure and impact are summarized in Table 2. All participants reported that they were able to socially distance beginning in March 2020. Over a quarter of participants (27.0%) reported a worsened financial situation as a result of the pandemic, while 7.9% reported increased housing instability due to the pandemic. A small percentage (2.0%) self-reported that they were diagnosed with COVID-19 by a healthcare professional, while 23.8% believed they had symptoms of COVID-19 but were not tested. Nearly one third (31.3%) reported that a close family member or friend was diagnosed with COVID-19 by a healthcare professional, and 25.0% had a close family member or friend who thought they had COVID-19 symptoms but was not tested. Over a quarter of the sample (26.7%) reported that a close family member or friend died as a result of COVID-19. Per the electronic medical records (as of April 13, 2021), 6 (5.9%) of the participants tested positive for COVID-19, 28 (27.7%) tested negative, and 67 (66.3%) were not tested.

Table 2.

Summary of Responses to RACE Items Indexing Practical and Medical Health Outcomes of the Pandemic

Variable	n (%)
Able to social distance	101 (100.0)
Frequency of leaving house
Never	15 (15.0)
1–4 times a week	59 (59.0)
Nearly everyday	26 (26.0)
Financial Stability
Worsened financial situation	27 (27.0)
No effect	61 (61.0)
Improved financial situation	12 (12.0)
Faced housing instability	8 (7.9)
Exposure
Diagnosed with COVID-19 per self-report	2 (2.0)
Had symptoms consistent with COVID-19	24 (23.8)
Family/friend diagnosed with COVID-19	31 (31.3)
Family/friend had symptoms consistent with COVID-19	25 (25.0)
Family/friend death from COVID-19	27 (26.7)
COVID-19 Testing per Medical Record
Positive	6 (5.9)
Negative	28 (27.7)
Not tested	67 (66.3)

Note. COVID-19 testing per medical record as of 4/13/2021.

Confirmatory Factor Analysis of RACE Items and Bivariate Correlations

The use of the mood/anxiety and PTSD subscales on the RACE was supported by the results of the confirmatory factor analysis. The two-factor model fit the data well: χ² (43, n = 101) = 70.23, p = .005, root mean square error of approximation = .079, standardized root mean square residual = .061, comparative fit index = .973, Tucker-Lewis index = .966. All items loaded significantly at the p < .001 level on their respective latent variables; the standardized factor loadings for the mood/anxiety items ranged from β = .61 to .86 and the standardized factor loadings for the PTSD items ranged from β = .56 to .92. The two factors were highly correlated with each other (r = .90). Based on this high factor correlation, we compared the two-factor model to a more parsimonious single factor model. We found that the more restrictive single factor model was associated with significantly degraded fit compared to the two factor one (Δχ² = 4.46, Δdf = 1, p = .03). Thus, the two-factor model was preferred.

Correlations revealed expected relations among pre-pandemic psychiatric symptom severity and self-reported pandemic-related changes in psychiatric symptoms. Pre-pandemic PTSD severity was significantly associated with pandemic-related increases in symptoms of PTSD (r = .38, p < .01) and mood/anxiety (r = .33, p < .01). Pre-pandemic AUD severity was significantly associated with pandemic-related increases in PTSD symptoms (r = .32, p < .01), substance use (r = .47, p < .01), and mood/anxiety symptoms (r = .20, p < .05), as well as COVID-19 exposure (r = .27, p < .01). Pre-pandemic depression severity was significantly positively correlated with pandemic-related increases in PTSD symptoms (r = .26, p < .05) and mood/anxiety (r = .33, p < .01). In addition, a sense of reduced personal resilience during the pandemic (as indicated by high scores on this item) was correlated with pandemic-related increases in mood/anxiety (r = .33, p < .01) and PTSD symptoms (r = .26, p < .01), though alterations in sense of personal resilience were not associated with any pre-pandemic variables. Additional correlations among the RACE subscales are shown in Table 3.

Table 3.

Descriptive Characteristics and Bivariate Associations Among Pre-Pandemic Psychiatric Conditions and Pandemic Related Psychiatric Symptoms and Experiences

Variable	Min-Max	M (SD)	1	2	3	4	5	6	7	8	9	10
1. Pre-pandemic PTSD severity	0–64	22.38 (13.56)	1.00
2. Pre-pandemic AUD severity	0–22	2.43 (5.05)	.247**	1.00	--	--	--	--	--	--	--	--
3. Pre-pandemic MDD severity	0–18	4.55 (6.39)	.575**	.129	1.00	--	--	--	--	--	--	--
4. Housing/financial	2–8	5.59 (1.08)	.018	−.158	−.044	1.00	--	--	--	--	--	--
5. Mood/anxiety sx	8–15	11.35 (1.92)	.326**	.199*	.332**	−.081	1.00	--	--	--	--	--
6. Substance use	6–12	8.29 (.88)	.132	.473**	.031	−.214*	.201*	1.00	--	--	--	--
7. PTSD sx	23–40	28.39 (4.09)	.378**	.315**	.261**	−.039	.686**	.273**	1.00	--	--	--
8. Resilience	1–5	2.95 (1.04)	.077	−.154	.173	.035	.332**	−.105	.258**	1.00	--	--
9. COVID self-exposure	0–1	.27 (.45)	.092	.272**	.030	−.034	.234*	.227*	.311**	−.120	1.00	--
10. Proximal COVID exposure	0–3	.83 (.94)	.016	.364	.566	.011	.092	.062	.289**	−.089	.310**	1.00

Note. Variables not identified as “pre-pandemic” were derived from the RACE, which was administered several months after the start of the pandemic. Min-max is observed min/max. PTSD = posttraumatic stress disorder; AUD = alcohol use disorder; MDD = major depressive disorder; sx = symptoms; min = minimum; max = maximum.

^*p < .05.

^**p < .01

Longitudinal Multivariate Regression Models

The results of the regression models are summarized in Table 4. Significant predictors of pandemic-related changes in PTSD symptoms were pre-pandemic PTSD severity (β = .332; p = .003) and pre-pandemic AUD severity (β = .228; p = .028). Significant predictors of increased substance use during the pandemic included age (β = −.317, p = .002) and pre-pandemic AUD severity (β = .391; p < .001). Pre-pandemic AUD severity also significantly predicted self-reported/EMR-defined exposure to COVID-19 (β = .264; p = .019). Minority race and ethnicity was the only significant predictor of housing and financial instability during the pandemic (β = −.372; p < .001). All effects remained significant (p < .05) after including both proximal COVID-19 exposure and housing and financial stability as covariates (in separate analyses) to account for the possible confounding effects of other COVID-19 related stressors. Models with significant effects for pre-pandemic psychiatric symptoms on pandemic outcomes were re-run using the pre-pandemic diagnostic variables in place of symptom severity scores. The pattern of results was unchanged with respect to significant effects (details available from corresponding author). Sex, education, and pre-pandemic depression did not significantly predict psychiatric outcomes, COVID-19 exposure, or lifestyle stability during the pandemic.

Table 4.

Results of Regressions Examining Pre-Pandemic Psychiatric Symptoms as Predictors of Subsequent Pandemic-Related Symptoms and Experiences

	PTSD		Mood/Anxiety		Substance Use		COVID-19 Self-Exposure		Housing/Financial Stability
Variable	β	p	β	p	β	p	β	p	β	p
Age	−.146	.158	−.095	.365	−.317	.002	−.145	.165	.055	.582
Sex	.140	.175	.068	.520	.032	.747	−.148	.156	.014	.886
Education	.040	.693	−.122	.240	.067	.492	.015	.880	−.048	.620
Minority race/ethnicity	.189	.066	.064	.535	.157	.112	.052	.613	−.372	.000
Pre-pandemic PTSD severity	.332	.003	.198	.091	.101	.353	.077	.525	.021	.856
Pre-pandemic AUD severity	.228	.028	.144	.184	.391	.000	.238	.034	−.029	.793
Pre-pandemic MDD severity	.388	.599	.202	.089	−.128	.245	−.045	.710	−.003	.978

Note. Demographic characteristics were included in Step 1 and psychiatric variables in Step 2 of each model. Overall model fit statistics for model predicting pandemic-related PTSD: Step 1: R² = .034, F (4, 95) = 1.873, p = .121; Step 2: Δ R² = .172, Δ F (3, 92) = 6.993, Δ p < .001; Overall fit statistics for model predicting pandemic-related mood/anxiety: Step 1: R² = .035, F (4, 95) = .866, p = .487; Step 2: Δ R² = .133, Δ F (3, 92) = 4.889, Δ p = .003; Overall fit statistics for model predicting pandemic-related substance use: Step 1: R² = .139, F (4, 95) = 3.844, p = .006; Step 2: Δ R² = .138, Δ F (3, 92) = 5.870, Δ p = .001; Overall fit statistics for model predicting COVID-19 self-exposure: Step 1: R² = .046, F (4, 95) = 1.140, p = .342; Step 2: Δ R² = .067, Δ F (3, 92) = 2.308, Δ p = .082; Overall fit statistics for model predicting pandemic-related housing/financial stability: Step 1: R² = .146, F (4, 95) = 4.052, p = .004; Step 2: Δ R² = .001, Δ F (3, 92) = .033, Δ p = .992. PTSD = posttraumatic stress disorder; AUD = alcohol use disorder; MDD = major depressive disorder.

Discussion

Identifying individuals at heightened risk for adverse outcomes during the pandemic is critical for leveraging resources for those with greatest need. In a sample of Veterans who had experienced a range of trauma types including combat and sexual and physical assault, we found that pre-pandemic mental health conditions, especially PTSD and AUD, were associated with psychological and health responses to the pandemic. In particular, our results demonstrated that baseline AUD predicted later exposure to the COVID-19 virus, which is consistent with recent literature (Wang et al., 2021). Given the association between SUD and both increased risk-taking propensity (LaSpada et al., 2020) and reduced harm avoidance (Miller et al., 2003), individuals with SUD may be less able to accurately judge risks with regard to physical health and safety, especially while under the influence. Furthermore, individuals with SUD may be using substances in social settings, which could increase risk of COVID-19 exposure. SUDs also impact immunological responses (Loftis & Huckans, 2013), which may play a role in the increased risk of COVID-19 diagnosis and related symptoms.

Pre-pandemic AUD was also associated with pandemic-related increases in PTSD symptoms, which carries implications for our understanding of PTSD comorbidity and its associations with broad, underlying dimensions representing internalizing (i.e., unipolar mood disorders, anxiety, and somatization disorders) and externalizing (i.e., substance use disorders and antisocial personality disorder) psychopathology (Slade & Watson, 2006). PTSD is often thought of as an internalizing disorder (Slade & Watson, 2006). However, there is evidence that PTSD may arise through genetic liability to either internalizing or externalizing (Wolf et al., 2010) and that it shows phenotypic associations with both psychopathology spectra (Miller et al., 2014; Wolf et al., 2010). Our longitudinal results further suggest that externalizing conditions, such as SUD, are associated with worsening PTSD symptom severity (controlling for baseline PTSD severity). Thus, while PTSD is more strongly associated with internalizing disorders (and in this study, shared more variance in common with internalizing symptoms), the externalizing presentation may be a marker for a particularly unique and problematic PTSD symptom course. The importance of this is further highlighted by our finding that AUD was associated with increased risk of COVID-19 exposure over time.

In addition, our results demonstrated worsening of pre-existing psychiatric symptoms during the pandemic. Pre-pandemic PTSD predicted pandemic-related increases in PTSD symptoms, which is consistent with prior literature concerning PTSD and psychiatric responses to the pandemic (Liu et al., 2021; Murphy et al., 2020). This finding suggests that the COVID-19 pandemic, an acute onset stressor, exacerbates PTSD symptoms related to prior traumas. One possibility is that behaviors that are important for reducing viral exposure during the pandemic, such as quarantining, social distancing, and remaining vigilant about masking, may reinforce existing PTSD symptoms, such as avoidance, estrangement from others, and hypervigilance. In addition, those with pre-pandemic AUD were more likely to report increased substance use during the pandemic. Thus, individuals with PTSD and AUD are at a higher risk for worse pandemic-related psychiatric outcomes, suggesting targeted treatment efforts among these individuals may be particularly useful in mitigating long-term consequences of the COVID-19 pandemic.

We found that 5.9% of the sample had positive COVID-19 test results, which was lower than the cumulative prevalence of the disease in the northeast region of the United States at the time of study completion. Haderlein et al. (2020) found that Veterans with PTSD were less likely to test positive for COVID-19 than Veterans without PTSD. The authors suggested that those with PTSD were already more socially isolated prior to the pandemic, potentially resulting in lower infection rates in this population (Haderlein et al., 2020). Veterans may also have military training that lends itself to increased hypervigilance and preparedness (e.g., isolating and stocking resources for long periods of time). Thus, the low rates of infection within this clinical Veteran sample could suggest that the characteristics of PTSD (e.g., avoidance, heightened perceived sense of threat) that are associated with increased risk for worsening mental health symptoms are protective with respect to avoiding exposure to the virus. This requires careful clinical consideration in how to prevent COVID-19 exposure among this population without reinforcing PTSD symptoms.

An alarming proportion of participants (26.7%) reported that a close family member or friend died from COVID-19. Murphy et al. (2020) also reported a high proportion (15.1%) of their Veteran sample had known someone who had died from COVID-19. Given that our assessment was administered during the summer of 2020, relatively early in the pandemic, it is particularly troubling that such a high percentage of the sample had experienced a COVID-19-related loss. These losses may meet the definition of a DSM-5-defined traumatic experience, potentially contributing to psychiatric symptoms in response to new traumas. Another concerning trend in the data was that racial and ethnic minorities (a variable included in all analyses along with age, education, and sex) were more likely to lose their housing as a result of the pandemic. This is consistent with known race-related health disparities (Lopez et al., 2021; Tai et al., 2021) and the disproportionate rise in unemployment (Couch et al., 2020) among minority individuals during the pandemic. It is notable that the factors that were associated with changes in psychiatric symptoms during the pandemic were distinct from those that were associated with housing changes. This finding highlights the growing need for programs designed to prevent housing loss among Veterans, such as the U.S. Department of Housing and Urban Development-VA Supportive Housing (HUD-VASH) Program.

Strengths and Limitations

These results should be interpreted in consideration of the study strengths and limitations. The main strengths of the study include its longitudinal design, the use of clinician-administered diagnostic interviews, and the breadth of assessment of both pre-pandemic and pandemic-related mental health. Limitations include the small sample size and that generalizability is limited to primarily male Veterans. An additional limitation is that our pandemic assessment was based on a newly derived measurement self-report tool (the RACE), though this initial examination of its psychometric properties (including internal consistency and the results of the confirmatory factor analysis) supports its use. The RACE was administered relatively early on in the pandemic, before the spike of COVID-19 cases in the U.S. in late 2020 and early 2021, and thus an additional limitation is that we may not have fully captured Veterans’ responses to the worst of the pandemic to date. Additionally, the RACE did not assess mask usage, and this would be a useful addition to future revisions to the measure. Another limitation of the measure is that there are strong demand characteristics when inquiring about social distancing practices, thus it is possible that respondents were reluctant to report that they did not follow these guidelines. Furthermore, due to the small number of participants (5.9%) who either self-reported COVID-19 diagnosis or had a positive test result in the EMR, we considered both participants who were diagnosed with COVID-19 and participants who believed they had COVID-19 symptoms, but were not tested, as having been exposed to COVID-19. It is important to note that the phone assessment was conducted in the initial 6 months of the pandemic, when testing was not easily obtained. In addition, we did not administer the CAPS or SCID over the phone, so our assessment of pandemic-related changes in symptoms was limited by the self-report nature of this evaluation. We also did not assess mental health treatment at the time of the follow-up, so we were not able to control for intervention effects. We did not assess if participants’ exposure to COVID-19 related loss or illness met the DSM-5 definition of a traumatic experience or if they had PTSD symptoms specific to this exposure. In addition, we were underpowered to evaluate potential demographic (or other) moderators of the associations of interest, so it is unclear if associations between pre-pandemic psychiatric symptoms and pandemic-related outcomes might differ by participant characteristics. Finally, though our study benefited from the longitudinal design, we cannot determine causal associations among the data, given the potential for unmeasured, confounding variables.

Conclusions

Overall, our results show that those with pre-existing conditions, particularly PTSD and AUD, are uniquely affected by the pandemic and are at heightened risk for both adverse psychiatric outcomes and exposure to the virus. These findings carry implications for targeting prevention and treatment efforts for these individuals at greatest risk. For instance, brief assessments for PTSD and AUD administered during routine primary care visits followed by provision of brief psychoeducation related to drinking guidelines and risk for COVID exposure or treatment referrals may have clinically significant downstream effects. Just as resources are leveraged for those at increased risk for poor outcomes from the virus itself (i.e., the elderly, those with pre-existing cardiac conditions), targeted prevention and intervention efforts should be made for those at greatest risk for adverse psychiatric outcomes.