A latent class analysis of lifetime potentially traumatic event exposure and the buffering role of social support on anxiety during the COVID-19 pandemic

Abstract

This study examined relationships among lifetime trauma exposure, perceived social support, and anxiety during the first year of the COVID-19 pandemic. The sample included 246 diverse adults (59.8% women, M age = 44.39 [SD = 12.96]) living in Ohio who completed questionnaires on trauma exposure, perceived social support, and anxiety symptoms. Latent class analysis identified four distinct classes based on patterns of potentially traumatic event exposure: Low, Moderate-Low, Moderate, and High. Further analyses using the BCH method demonstrated that the High Exposure class was associated with significantly higher anxiety than other classes. Perceived social support buffered anxiety symptoms for the Low Exposure and Moderate-Low Exposure classes. This study highlights how trauma exposure impacts psychological and mental health in the context of a public health crisis and underscores the role of social relationships as a protective factor, with implications for trauma-informed mental health planning and programing to address disparities.

Keywords

anxiety latent class analysis moderation analysis social support trauma

Introduction

In March 2020, the COVID-19 pandemic dramatically disrupted daily life as the virus quickly spread globally, posing threats to physical and emotional wellbeing. Following the declaration of the pandemic as a public health emergency, research emerged that demonstrated negative effects on mental health (Wu et al., 2021), disproportionately impacting historically oppressed and marginalized groups, such as Black, Indigenous, other People of Color (BIPOC) and low-income communities (Raifman and Raifman, 2020). However, much of this research neglected to take into consideration the role of lifetime potentially traumatic events, despite studies illustrating the salience of prior trauma in influencing distress during subsequent, highly stressful events (Stroud, 2020) and research that demonstrates how environmental and social conditions stemming from structural inequalities are associated with increased exposure to potentially traumatic events (Benjet et al., 2016). Therefore, the present study explored associations among lifetime exposure to potentially traumatic events, social support, and anxiety during the first year of the COVID-19 pandemic.

Variation in potentially traumatic event exposure

In their lifetime, most people are exposed to an event that poses actual or extreme potential threat of physical, psychological, or emotional harm (Benjet et al., 2016), known as potentially traumatic events (Substance Abuse and Mental Health Services Administration [SAMHSA], 2014). Potentially traumatic events include single, discrete events like experiencing a car crash, natural disaster, or mass shooting, or ongoing, chronic events such as child maltreatment or experiencing long-term intimate partner violence (Cook et al., 2005). However, risk of exposure to different forms of potentially traumatic events is not distributed equally among different social identities and positions–that is, race, gender identity, sexual orientation, and socioeconomic status (SES) (Benjet et al., 2016; Rich et al., 2005). Further, studies have demonstrated that individuals with a history of exposure to potentially traumatic events are more likely to have experienced multiple types of events than single events, referred to as polyvictimization (Benjet et al., 2016).

Structural violence theory (Galtung, 1969) and a Social Determinants of Health (SDOH) framework (Centers for Disease Control and Prevention [CDC], 2022; World Health Organization [WHO] and Calouste Gulbenkian Foundation, 2014) lend an understanding to the variation in exposure to potentially traumatic events. Structural violence theory posits that violence manifests in the ways that political, economic, and social structures and systems–influenced by racism, sexism, heterosexism, and classism–establish unequal advantage, power, and resources among groups of people and interpersonally (Farmer, 2004; Galtung, 1969). These inequitable systems influence the conditions in which someone develops prenatally, is born, matures and ages, works, and otherwise lives their life, known as SDOH (CDC, 2022; WHO, 2014). These inequitable and unjust conditions give rise to social risk and circumstances in which different groups are more vulnerable to different forms of potentially traumatic events, which have been associated with adverse physical and mental health outcomes (Benjet et al., 2016).

Differences in exposure to potentially traumatic events arise along race, ethnicity, sexual orientation and gender identity, as well as SES. For example, Black and Hispanic individuals, compared to White individuals, are more likely to witness intimate partner violence (Roberts et al., 2011; Turner and Lloyd, 2004). Black individuals have reported higher exposure to assault (Roberts et al., 2011; Turner and Lloyd, 2004). BIPOC are at risk to experience discrimination based on their race and/or ethnicity, and women of all races and ethnicities are more likely to be exposed to victimization, sexual assault and child sexual abuse, domestic or intimate partner violence, and gender discrimination (Benjet et al., 2016; Valentine et al., 2019). Sexual minority, including transgender, individuals are more likely to experience traumatic events than heterosexual individuals, and sexual minorities are susceptible to discrimination based on their sexual orientation and gender identities (Blondeel et al., 2018). Finally, research consistently demonstrates that people who occupy lower SES tend to report higher rates of exposure to traumatic events than those with higher SES (Benjet et al., 2016; Hatch and Dohrenwend, 2007). Women of all racial and ethnic identities experience higher rates of poverty than men; moreover, among women, those who are Indigenous, Black, Asian or Pacific Islander are more likely to live in poverty (Bleiweis et al., 2020), putting them at disproportionate risk to experience potentially traumatic events.

Potentially traumatic events, COVID-19, and anxiety

While not everyone remains, or even becomes, traumatized due to experiencing a potentially traumatic event, research over the last several decades supports the notion that the effects of these experiences can influence subsequent mental health outcomes, such as anxiety (Forman-Hoffman et al., 2012; SAMHSA, 2014). A dose-response relationship–initially evidenced by the landmark adverse childhood experiences study (Felitti et al., 1998)–has been underscored in the research literature: experiencing potentially traumatic events, particularly early in life, can lead to greater anxiety compared to those who do not report these experiences (Mersky et al., 2013), and people who experience multiple potentially traumatic events are more likely to develop symptoms of anxiety (Bridges-Curry and Newton, 2022). The stress sensitization model (Monroe and Harkness, 2005; Post and Weiss, 1998) illustrates the consequences of exposure to multiple potentially traumatic events; it suggests that prior exposure to extreme stressors–such as potentially traumatic events–may lead to increased neurobiological responsiveness during the experience of subsequent stressors, even if the stressors are lower in severity. Thus, someone exposed to lifetime potentially traumatic events may experience more severe anxiety symptoms during a public health crisis, such as the COVID-19 pandemic.

Research suggests that individuals who were exposed to potentially traumatic events prior to the COVID-19 pandemic experienced higher levels of anxiety during the pandemic (Jakubowski et al., 2023; Prout et al., 2020; Russo et al., 2022). Yet, there is a need for more research that examines the anxiety outcomes of individuals during the COVID-19 pandemic who have been exposed to different forms and combinations of lifetime potentially traumatic events. In addition, there is a need to examine factors that may be protective against anxiety outcomes, such as perceived social support.

Perceived social support

Social support is conceptualized as an important SDOH for promoting positive mental health outcomes (WHO, 2014) and has been shown to mitigate the effects of experiencing a potentially traumatic event on poor mental health outcomes, including anxiety (Marie-Mitchell and Kostolansky, 2019). Research suggests that perceived social support may serve as a protective factor against the anxiety stemming from experiences of the COVID-19 pandemic (Liu et al., 2020).

The current study

Despite increased research on mental health during the pandemic, further studies are needed on the links between potentially traumatic events, anxiety, and the potential protective role of social support. Guided by structural violence theory (Galtung, 1969) and a SDOH framework (CDC, 2022; WHO, 2014), the present study used latent class analysis (LCA) to identify patterns of potentially traumatic event exposure among participants and examine predictors of membership in identified latent classes. Subsequent analyses tested the moderating role of perceived social support on anxiety outcomes. Research questions and hypotheses are as follows:

Research Question 1: Are there distinct classes of shared potentially traumatic event exposure among participants?

Hypothesis 1a: The LCA will reveal a distinct polyvictimized class.

Hypothesis 1b: The LCA will reveal a distinct class representing low exposure to potentially traumatic events.

Research Question 2: What is the relationship between potentially traumatic event exposure classes and anxiety?

Hypothesis 2: Classes with greater potentially traumatic event exposure will have statistically significantly higher levels of anxiety compared to classes with less potentally traumatic event exposure.

Research Question 3: To what extent does perceived social support moderate the relationship between potentially traumatic event exposure class membership and anxiety?

Hypothesis 3: Perceived social support will buffer the relationship between class membership and anxiety, where higher perceived social support will be associated with lower anxiety.

Methods

Study sample and procedures

The COVID-19 Pandemic, Social Inequalities, and Mental Health study is a longitudinal panel study of adults living in Cuyahoga County, Ohio, approved by the Institutional Review Board at Case Western Reserve University. Data were collected between October 2020 (baseline; N = 324) and May 2021 (follow-up; N = 264). Participants were recruited via paper flyer postings at community-based centers, virtual flyers posted on virtual community boards and sent to email mailing lists, and social media advertisements. Adults (at least 18 years of age) who spoke English living in Northeast Ohio were eligible for this study. Quota sampling was employed to obtain a more representative sample among Black/African American and White racial identities, as well as female/woman and male/man identities. United States Census data were used to determine participant quotas based off an initial target of 500 study participants, so that the study sample better reflected the population of the geographic area in which the survey was distributed. Once a quota was hit for a particular demographic group (e.g. White women, n = 164 or 32.8%), individuals who identified with that group were no longer eligible to participate. After providing written informed consent, participants completed demographic questions and were asked about their experiences with COVID-19, lifetime trauma exposure, social connections, and mental health. The current study used 6-month follow-up data of participants with complete responses, resulting in a final analytical sample that included 246 participants.

Measures

Potentially traumatic events

The Brief Trauma Questionnaire (BTQ; Schnurr et al., 1999) is a 10-item self-report questionnaire that assessed exposure to potentially traumatic events (i.e. events that present life threat or serious injury). Two discriminatory experiences items were added to the questionnaire for this study: Have you experienced rejection, violence, or discrimination, such as experiencing poor treatment from others, being called names or insulted, or being threatened, harassed, or physically assaulted because of your (1) race or ethnicity and (2) gender identity or sexual orientation? Participants indicated whether or not they had ever experienced each type of event (1 = yes, 0 = no). Previous research demonstrates high interrater reliability for the BTQ (κ = 0.60–1.00; Koenen et al., 2009; Lancaster et al., 2011), and adequate reliability between the questionnaire and a trauma-focused interview (Schnurr et al., 2002).

Anxiety

The six-item anxiety subscale of the Brief Symptom Inventory (BSI) assessed psychological symptoms of anxiety (Derogatis and Melisaratos, 1983). Participants were asked, indicating on a 5-point scale ranging from 1 (not at all) to 5 (extremely), how much they were bothered by a particular problem (e.g. nervousness or shakiness inside, being suddenly scared for no reason, feeling tense or keyed up) in the past 7 days. Scores were summed (possible range 6–30), with higher scores indicating higher anxiety. The BSI anxiety subscale demonstrates adequate internal consistency (α = 0.79, Derogatis, 2001), as well as test-retest reliability (Derogatis and Melisaratos, 1983).

Perceived social support

Perceived social support was measured using four questions from the Lubben Social Network Scale (Lubben, 1988; Lubben et al., 2006). Examples of questions include: How often do you see or hear from the relative or friends with whom you have the most contact? (0 = less than monthly, 5 = daily) and How many relatives or friends do you feel at ease with that you could talk about private matters? (0 = none, 1 = one, 2 = two, 3 = three or four, 4 = five to eight, 5 = nine or more). Total scores were calculated (possible range 0–20), with higher scores indicating higher levels of perceived social support. The Lubben Social Network Scale has demonstrated adequate internal consistency, including with research during the COVID-19 pandemic (α = 0.84, Rumas et al., 2021).

Sociodemographic covariates

Educational attainment was used to represent socioeconomic status, as it strongly influences employment and income and is relatively constant for adults (Oakes and Kaufman, 2017). Highest degree or level of education earned was self-reported by participants and was treated as interval-like using the following values: (1) less than 8th grade/8th grade completed, (2) some high school completed/high school diploma or equivalent, (3) some college completed/trade school or vocational education completed/associate’s degree, (4) bachelor’s degree, (5) master’s degree/another professional degree, (6) doctorate.

Racial and ethnic identity were self-reported by participants. Racial identity categories included Asian, Black or African American, Middle Eastern, Native American, Alaska Native or Indigenous, Native Hawaiian or Pacific Islander, White, and/or other race. Ethnic identity categories included Mexican, Mexican American or Chicano, Puerto Rican, or other Hispanic, Latinx, or Spanish origin. Racial and ethnic identities were collapsed into White only, Black and/or African American only, and other racial and ethnic identities due to low endorsement of other racial or ethnic identity categories and overall sample size.

Gender identity was self-reported by participants, using the following response options: female or woman (1), male or man (2), or additional gender identity (3). Female or woman and additional gender identity were collapsed into one category representing minoritized gender identities, due to extremely low frequencies for additional gender identity (n = 1), resulting in two categories for analysis: female/woman or other (0), and male/man (1).

Control variables

Control variables were selected based on a review of the literature that demonstrates associations among variables of interest and preexisting mental health issues, physical health challenges (Hao et al., 2020; Wu et al., 2021), and quarantining and/or social distancing (Kim and Laurence, 2020). Previous mental health services were included as a dichotomous variable (1 = yes, 0 = no). General health was entered on a scale 1–5, ranging from poor to excellent. Current sheltering in place (i.e. quarantining or remaining home with the exception of leaving for essential reasons) was included as a dichotomous variable (1 = yes, 0 = no).

Analytic approach

Descriptive analyses

Preliminary analyses using SPSS version 28 included examining descriptive statistics, normal distribution based on established criteria, visual inspection of histograms, assessing multicollinearity, examining bivariate scatterplots for linearity and outliers, and assuming homoscedasticity.

Latent class analysis (LCA) and auxiliary variable analysis

Statistical analysis was conducted iteratively using Mplus version 8. First, LCA was used to identify meaningful groups with similar patterns of prior trauma exposure. LCA is a person-centered approach that identifies underlying subgroups, capturing heterogeneity within and between groups (Gunzler et al., 2021; Muthén and Muthén, 2000). A one-class model was first fitted, followed by additional classes to find the best-fitting number of latent classes. The Akaike information criterion (AIC), Bayesian information criterion (BIC), and sample size–adjusted Bayesian information criterion (Adjusted BIC), Likelihood Ratio Test (LRT), Bootstrap Likelihood Ratio Test (BLRT), and the Vuong-Lo-Mendell-Rubin Likelihood Ratio Test (VLMR LRT) were used to comparatively assess model fit. Lower values on each of these indices suggests better data fit. Entropy was examined to determine how accurately each model separates classes (Wang et al., 2017), where an entropy value of above 0.8 is acceptable (Celeux and Soromenho, 1996). Optimal latent class structure was determined based on the cited indices, interpretability, and model parsimony.

To estimate trauma exposure class associations with anxiety, the BCH method of modeling with covariates was used, which maintains class membership based on weights (McLarnon and O’Neill, 2018). This approach allows a multigroup model to be estimated where class membership is held constant but adjusted based on weights that reflect class membership uncertainty (McLarnon and O’Neill, 2018). A model test was performed to identify potential statistically significant different mean scores of the outcomes between classes.

Perceived social support moderation analysis

Perceived social support’s buffering role on anxiety for each trauma exposure class were examined using the steps outlined above. Overall and within-class linear regression of anxiety on perceived social support and the covariates were specified to examine potential class-specific perceived social support relationships. As with the previous step, a model test was performed to examine potential differences in mean scores of the outcomes between classes.

Results

Descriptive statistics and bivariate correlations

Most participants (n = 171, 69.5%) reported exposure to at least one potentially traumatic event, and 127 participants (51.6%) experienced two or more potentially traumatic events. The most frequently reported form of potentially traumatic event was accident/injury (n = 90, 36.6%), followed by physical abuse (n = 84, 34.1%) and racism (n = 75, 30.5%). The majority of the sample identified as having a historically minoritized race (n = 138), with 31.3% and 24.8% of participants identifying as Black/African American or other race, respectively. Most participants identified as female/woman or other gender (N = 147, 59.8%). The mean level of education was some college/trade school/associate degree (M = 3.58). Reliability for study measures ranged from good to excellent (social support, α = 0.827; BSI anxiety α = 0.911), and the BTQ demonstrated stability through bootstrapping and tests of entropy. No evidence of violations of normality was identified. Appendix Tables A1 and A2 present descriptive of study variables and bivariate correlations, respectively.

Latent class analysis of potentially traumatic event exposure (Research Question 1)

A series of LCA models were fit, each adding one class; the four-class solution was deemed optimal based on fit indices for each solution (Table 1) and unveiled distinct classes of potentially traumatic event exposure. Three classes showed varying probabilities of exposure, with one high-exposure (i.e. polyvictimized) class and one low-exposure class, supporting hypotheses 1a and 1b.

Table 1.

Summary of fit statistics for latent classes based on potentially traumatic event exposure indicators (N = 246).

Total classes	AIC	BIC	Adjusted BIC	Entropy	Adjusted LRT	BLRT	VLMR LRT
1	2251.36	2282.90	2254.37	–	–	–	–
2	2073.60	2140.20	2079.97	0.76	194.23**	197.76**	197.76**
3	2063.32	2164.98	2073.05	0.73	29.73	30.27*	30.27
4	2057.21	2193.92	2070.29	0.85	25.65*	26.12*	26.12**
5	2063.20	2234.96	2079.63	0.85	13.91	14.16	14.16

AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; LRT = LRT Likelihood Ratio Test; BLRT = Bootstrap Likelihood Ratio Test; VLMR LRT = Vuong-Lo-Mendell-Rubin Likelihood Ratio Test.

Significant at the p < 0.05 level compared to prior model. **Significant at the p < 0.01 level compared to prior model.

Boldface indicates final model.

The patterns of potentially traumatic event exposure that characterize the four classes are presented in Figure 1. “Low Exposure” (Class 1) consisted of just over half of the sample (n = 132). “Moderate Exposure” (Class 2) was characterized by a very high probability of having experienced the violent death of a loved one (close family or friend), a relatively moderate probability of endorsing physical abuse, witnessing a traumatic event, and racism, relatively low probability of experiencing sexual abuse, a life-threatening illness, and sexism or gender discrimination, and a very low probability of endorsing accident/injury or disaster/war. “High Exposure” (Class 3), accounted for 21% of the sample (n = 51) and was characterized by high probability of endorsement of all potentially traumatic events with a very high probability of accident/injury, and a relatively high probability of experiencing physical abuse, as well as moderate probability of endorsing racism, the violent death of a loved one, sexual abuse, witnessing a traumatic event, and almost a 25% chance of endorsing a life-threatening illness and sexism or gender discrimination. “Moderate-Low Exposure” (Class 4) accounted for 17% of the sample (n = 42) and had high probability of physical abuse endorsement, as well as moderate probability of racism, sexual abuse, accident/injury, and relatively low probability of endorsing sexism or gender discrimination, disaster/war, and witnessing a traumatic event, and a very low probability of endorsing the violent death of a loved one and life-threatening illness.

Potentially traumatic event exposure and anxiety outcomes (Research Question 2)

Relationships among anxiety and the covariates were preliminarily examined within classes to determine if covariates should be permitted to vary by class in the final model. No differential relationships with anxiety by class on gender or racial identity were identified. Evidence for varying effects of education on anxiety was found within class and, as such, the effect of education was allowed to vary by class in the final model.

Analyses for the final model demonstrated class-specific moderation of the effect of education on anxiety. In the Moderate Exposure class, higher education was associated with higher anxiety (b = 1.60, p = 0.026); in the Moderate-Low Exposure class, higher education was associated with lower anxiety (b = −1.45, p = 0.005). There were no moderating relationships evident in the Low Exposure and High Exposure classes.

In addition, other racial identity had an overall effect on the outcome (b = 1.730, p = 0.40), suggesting that anxiety scores are statistically significantly higher for those of other racial identities compared to White; however, this relationship did not differ by class. Currently sheltering in place (b = 1.99, p = 0.006) and use of previous mental health services (b = 1.65, p = 0.016) each had a statistically significant positive relationships with anxiety, where endorsement of sheltering in place and use of previous mental health services were associated with higher anxiety. Physical health did not have a statistically significant relationship with anxiety (b = −0.65).

Table 2 shows the estimated mean anxiety scores and standard errors across classes. The overall Wald χ² was statistically significant (p < 0.001); post-hoc pairwise testing indicated statistically significant differences (p < 0.05) between the High Exposure class and all other classes on mean anxiety scores. Closer examination of pairwise comparisons demonstrated that compared to the High Exposure class, the anxiety scores were 4.15, 4.23, and 4.81 points lower in the Low Exposure, Moderate Exposure, and Moderate-Low Exposure classes, respectively.

Table 2.

Estimated means of anxiety scores across latent classes (N = 246).

Mean Score and Standard Error	Class 1N = 132 (54%)	Class 2N = 21 (9%)	Class 3N = 51 (21%)	Class 4N = 42 (17%)	Overall Wald χ²(3)
Mean	8.71_a	8.63_a	12.86_b	8.05_a	17.42**
Standard error	1.34	1.38	1.40	1.33

Anxiety possible range 6–30, where higher scores indicate higher anxiety. Estimated means are adjusted for the effects of sheltering in place due to COVID, physical health, prior mental health service use, gender, race, and education. The effect of education was also allowed to vary by class. Means with different subscripts are statistically significantly different at p < 0.05 based on Tukey post-hoc tests. Class 1 = Low Exposure, Class 2 = Moderate Exposure, Class 3 = High Exposure, Class 4 = Moderate-Low Exposure.

Significant at the 0.05 level (2-tailed); **Significant at the 0.01 level (2-tailed).

Moderating effects of perceived social support (Research Question 3)

In the Low Exposure and Moderate-Low Exposure classes, perceived social support moderated the relationship between class and anxiety, where higher perceived social support was associated with lower anxiety (Low Exposure class b = −0.18 p = 0.05; Moderate-Low Exposure class b = −0.41 p = 0.016; Appendix Figure A1). No relationships were found in the Moderate Exposure or High Exposure classes, partially supporting hypothesis 3. Because education demonstrated varying effects with anxiety in preliminary analyses, it was allowed to vary by class. Results indicated statistically significant relationships with anxiety and education in the Moderate Exposure class (b = 2.27, p < 0.001; Appendix Figure A2). Given this finding, a three-way interaction was tested between education and anxiety, but results did not suggest evidence of such a relationship. Other racial identity (b = 2.182, p = 0.042) and currently sheltering in place (b = 2.04, p = 0.022) were statistically significant in the model. The omnibus Wald χ² indicated statistically significant differences between classes (p < 0.001; Table 3), and post-hoc pairwise testing indicated statistically significant differences between the High Exposure class and all other classes (p < 0.001). Closer inspection of the pairwise comparisons showed the Low Exposure class had a 4.49-point lower mean anxiety score than the High Exposure class, the Moderate Exposure class had a 4.37-point lower mean anxiety score than the High Exposure class, and the Moderate-Low Exposure class had a 5.42-point lower mean anxiety score than the High Exposure class (Appendix Figure A3).

Table 3.

Estimated means of anxiety scores across latent classes in the final model, further adjusted for social support (N = 246).

Mean Score and Standard Error	Class 1N = 132 (54%)	Class 2N = 21 (9%)	Class 3N = 51 (21%)	Class 4N = 42 (17%)	Overall Wald χ²(3)
Mean	7.83_a	7.95_b	12.32_a	6.91_c	18.19**
Standard error	1.91	1.81	1.76	1.91

Anxiety possible range 6–30, where higher scores indicate higher anxiety. Estimated means are adjusted for the effects of sheltering in place due to COVID, physical health, prior mental health service use, gender, race, education, and social support. The effect of education was also allowed to vary by class. Means with different subscripts are statistically significantly different at p < 0.05 based on Tukey post-hoc tests. Class 1 = Low Exposure, Class 2 = Moderate Exposure, Class 3 = High Exposure, Class 4 = Moderate-Low Exposure.

Significant at the 0.05 level (2-tailed); **Significant at the 0.01 level (2-tailed).

Discussion

This study used LCA to identify patterns of lifetime potentially traumatic event exposure and examined their relationships with anxiety during the COVID-19 pandemic. It also explored how perceived social support moderated anxiety outcomes among these exposure classes. Findings demonstrated that those with more complex trauma histories (i.e. those who endorsed greater exposure to potentially traumatic events) had higher anxiety during the pandemic. In addition, perceived social support moderated the effect of anxiety for two classes.

Potentially traumatic event classes

Results yielded four distinct latent classes. In the present study, three of the latent classes were characterized by a variety of potentially traumatic event exposures, and the High Exposure class stood out as having at least a moderate probability of endorsing of all potentially traumatic event exposures, including a very high probability (1.00) of accident/injury, a relatively high probability of experiencing physical abuse, as well as a moderate probability of endorsing a number of other exposures. In addition, a Low Exposure class emerged, characterized by a very low probability of experiencing accident/injury, disaster/war, life-threatening illness, racism, sexual abuse, sexism or gender discrimination, and witnessing serious injury or death. Together, these findings support hypotheses 1a and 1b.

The potentially traumatic event exposure classes that emerged align with nascent research examining classes or profiles of trauma exposure that typically finds three or four classes with one low exposure class and one or more polyvictimized or complex trauma class (e.g. Bridges-Curry and Newton, 2022). However, much of the existing literature utilizes samples of children or focuses on traumatic events that occurred in childhood. The present study uniquely contributes to the body of literature by focusing on adults and capturing potentially traumatic events that have occurred across the lifespan. Studying adults who have experienced potentially traumatic events can provide insights into the impact of these experiences on family systems and relationships. For example, histories of experiencing potentially traumatic events can affect behavioral and mental health for caregivers which, in turn, can adversely impact intrafamilial dynamics and children’s functioning (Roubinov et al., 2021), such as heightened stress during the COVID-19 pandemic (Hagan et al., 2022).

Potentially traumatic event classes and mental health

In line with stress sensitization theory, the High Exposure class reported significantly elevated anxiety compared to other classes, even after including control variables, supporting hypothesis 2. This is consistent with research examining the relationship between potentially traumatic event exposure and higher anxiety during the COVID-19 pandemic (e.g. Jakubowski et al., 2023; Prout et al., 2020; Russo et al., 2022).

The findings for the Moderate-Low Exposure class aligns with other research illustrating that higher education is associated with lower risk of experiencing potentially traumatic events (Benjet et al., 2016; Hatch and Dohrenwend, 2007). These findings underscore the complexities of interactions among life circumstances and opportunities that influence mental health outcomes. Notably, in the Moderate Exposure class, higher education was associated with higher anxiety, and in the Moderate-Low Exposure class, higher education was associated with lower anxiety. The Moderate Exposure class also uniquely demonstrated a high probability (1.00) of experiencing the death of a loved one. Previous research has shown that the unexpected death of close family or friends may be the most distressing potentially traumatic event experienced by some, even for those who have faced significant lifetime potentially traumatic events (Keyes et al., 2014). Combined with findings that showed those with higher levels of education may lead to greater exposure to information about the COVID-19 pandemic (Yu et al., 2020), and the life-threatening nature of COVID-19, having previously experienced the death of a loved one may have exacerbated anxiety symptoms during this time.

The moderating effect of perceived social support

Perceived social support demonstrated a class-specific moderating effect on anxiety for the Low Exposure and Moderate-Low Exposure classes, partially supporting hypothesis 3. In these classes, higher perceived social support was associated with lower anxiety. This aligns with other research showing that perceived social support can buffer anxiety and other poor mental health effects of COVID-19 experiences for those exposed to potentially traumatic events (Fu et al., 2023; Liu et al., 2020; Wang et al., 2022). A key differentiating characteristic of the Low Exposure and Moderate-Low Exposure classes is no, or very little, exposure to death in the form of a loved one dying violently or witnessing death; because death of a loved one has been shown to be one of the most distressing experiences (Keyes et al., 2014), it is possible that social support plays a more salient role in the absence of this experience.

Additionally, the Moderate-Low Exposure class is largely composed of exposure to interpersonal experiences. These class characteristics–absence of death and greater interpersonal potentially traumatic event exposure–suggest perceived social support may be more effective in buffering poor mental health outcomes for those who have mostly interpersonal traumatic experiences, compared to those who have experienced cumulative interpersonal and non-interpersonal (such as exposure to death) potentially traumatic events. Research on relational health illustrates that the presence of supportive, authentic, and empathetic relationships is effective in promoting better mental health outcomes for people who have experienced interpersonal trauma, such as physical abuse and sexual abuse (Ludy-Dobson and Perry, 2010). Further, for those who have extremely complex trauma histories–such as those in the High Exposure class–perceived social support may not be consequential enough of a protective factor to promote better mental health outcomes. Previous research has highlighted the variable associations among history of potentially traumatic event exposure and mental health outcomes. For example, Evans et al. (2013) found that perceived social support from family and friends was associated with fewer symptoms related to trauma for men; however, for women, higher perceived social support from friends, but not family, was tied to decreased trauma-related symptoms. Wilson and Scarpa (2014) found that increased perceived social support from loved ones functioned as a protective factor against post-traumatic stress for those who experienced physical abuse, but not for those who experienced sexual abuse. In addition, these authors found that perceived social support from a significant other increased the risk for post-traumatic stress symptoms for sexual abuse survivors. Findings from the present study suggest a nuanced role for perceived social support in buffering trauma-related anxiety depending on event types and cumulative exposure, underscoring the need for personalized, trauma-informed interventions that take into consideration the unique constellation of one’s life experiences.

Implications for policy and practice

Given the likelihood of another pandemic, policy changes must be implemented that address the impact of potentially traumatic events and the longer-term effects. Effective policy will require both upstream, preventative approaches and downstream, healing-centered approaches and interventions. Large-scale policy changes that reduce the chances of exposure to potentially traumatic events through addressing structural factors that shape inequalities include providing a greater social safety net for low-income families, federal funding to equalize the quality of primary and secondary schools (Thoits, 2010), universal healthcare coverage (Zieff et al., 2020), ensuring equal pay for women (Shriver, 2014), and reparations for African Americans and Indigenous Americans (Darity and Mullen, 2022).

Policies that support downstream, trauma-informed and healing-centered interventions should involve both community-level and individual-level approaches. Examples include trauma-informed community-building efforts to strengthen relationships and feelings of social support (Falkenburger et al., 2018) such as those that foster neighbor-to-neighbor interaction, placemaking (such as installment of public art), narrative-making, and storytelling (Falkenburger et al., 2018; Saul, 2013), and establishing affordable, accessible community-based mental health and social service agencies that offer culturally and racially appropriate services (Norris and Alegria, 2005). This may be achieved by incorporating a healing-centered approach, such as healing-centered engagement, into community-based and individual-level interventions. Healing-centered approaches elevate the importance of culture, biopsychosocial conditions, and, imperatively, relationships (Ginwright, 2019).

Limitations and future research

Limitations of this study include a non-representative sample that consists of mostly woman/female-identifying (N = 146) and White participants (N = 108) due to eligibility criteria based off reported trauma history and completion of follow-up survey, despite quota sampling efforts. The study also did not capture many SDOH constructs, such as SES, that may predict potentially traumatic event exposure classes. Future research should use larger samples with nuanced SDOH measures to detect their effects on potentially traumatic events. Additionally, the specific timing of potentially traumatic events was not available in the data; it is possible that education could have occurred after, or during, some potentially traumatic events for participants. Further research should collect and analyze timing of these events. Some forms of potentially traumatic events may not have been captured by the BTQ, such as substance use or medical trauma. Finally, disability status of participants was not available in the dataset, which could impact experiences of potentially traumatic events.

Future research should continue to explore the relationships among social support, SDOH, and potentially traumatic events to further understand the role of social support for people with different experiences of these types of events. This should include examining both individual- and community-level characteristics, as further exploration of different factors may be especially important to consider given low-income and marginalized racial groups are more likely to live in adverse community conditions, and the COVID-19 pandemic had a disproportionate impact on these groups.

Conclusion

The present study informed research on the relationships among potentially traumatic events, social support, and anxiety outcomes during a global pandemic, taking a person-centered approach. Findings point toward targeting social support interventions to buffer anxiety in the context of pandemics, especially among those who may be more vulnerable to negative mental health symptoms due to complex trauma histories and varying SDOH. Continued examination of factors contributing to inequitable potentially traumatic event exposure can help address health disparities through trauma-informed policy and healing-centered practice interventions.

Footnotes

Appendix

Appendix Table A2.

Bivariate correlations among study variables (N = 246).

	1	2	3	4	5	6	78	8	9	10	11	12	13	14	15	16	17	18	19
1. Anxiety	–
2. Social support	−0.19**	–
3. Black/African American	−.04	−0.14*	–
4. Other minoritized race	0.11	0.09	−0.39**	–
5. White	−.06	0.05	−0.60**	−0.51**	–
6. Gender^a	0.06	−.12	−0.16*	−.07	0.21**	–
7. Education	−.11	0.37**	−0.25**	0.14*	0.12	−.06	–
8. Age	−.03	0.06	−0.15*	−.11	0.24**	−0.14*	−.04	–
9. PTE – Accident/injury	0.31**	−.08	0.05	−.01	−.04	0.12	−.02	−.01	–
10. PTE – Disaster/war	0.18**	0.01	−.09	0.13	−.02	0.08	0.03	0.08	0.36**	–
11. PTE – Loved one violent death	0.19**	0.07	0.10	−.03	−.06	−.04	0.02	0.00	0.23**	0.18**	–
12. PTE – Life-threatening illness	0.03	0.03	−.08	−.04	0.11	−.06	0.01	0.30**	0.14*	0.18**	0.16**	–
13. PTE – Physical abuse	0.12	−.03	0.14*	−.08	−.07	−.01	0.00	0.03	0.31**	0.16*	0.34**	0.01	–
14. PTE – Racism	0.22**	0.04	0.14*	0.17**	−0.28**	−.02	0.07	−.08	0.34**	0.09	0.21**	0.01	0.38**	–
15. PTE – Sexual abuse	0.20**	0.10	−.04	0.01	0.03	−.02	0.12	0.06	0.14*	0.12	0.11	0.10	0.24**	0.11	–
16. PTE – Sexism/discrimination	0.31**	0.00	0.09	0.03	−.11	−0.25**	−.04	0.06	0.23**	0.05	0.12	0.06	0.32**	0.27**	0.17**	–
17. PTE – Witnessed injury/death	0.16*	0.03	−.11	0.03	0.08	0.04	0.08	0.01	0.24**	0.16*	0.36**	0.07	0.23**	0.16**	0.25**	0.13*	–
18. Sheltering in place	0.24**	−0.17**	0.09	−.04	−.05	−.02	−0.22**	0.11	0.09	0.09	0.01	−.05	0.07	0.07	0.04	0.11	−.00	–
19. Physical health	−0.28**	0.18**	−.09	0.08	0.01	0.00	0.21**	−.03	−0.15*	−0.18**	−0.15*	−.06	−0.26**	−.08	−.04	−0.23**	−.03	−.09	–
20. Received MH services	0.21**	0.11	−.03	0.02	0.02	−0.18**	0.11	0.04	0.11	0.07	0.11	0.00	0.18**	0.24**	0.16*	0.19**	0.16*	0.04	−.05

Correlation is significant at the 0.05 level (2-tailed). **Correlation is significant at the 0.01 level (2-tailed). ^aWoman/Female or Other = 0, Man/Male = 1. PTE = potentially traumatic event; MH = mental health.

Acknowledgements

The author would like to thank Megan R. Holmes, PhD, Scott Frank, MD, Francisca García-Cobián Richter, PhD, Jennifer A. King, DSW, Nancy Rolock, PhD, and Karen Ishler, PhD for their support, guidance, expertise, and assistance throughout all aspects of this study.

Data sharing statement

The data that support the findings of this study are available upon reasonable request from the corresponding author at discretion of the principal investigator and author’s Institutional Review Board. The data are not publicly available due to privacy and ethical restrictions.

Declaration of conflicting interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Ethics approval

The conduct of the study followed ethical guidelines and was approved by Case Western Reserve University.

Informed consent

Participants provided written informed consent prior to data collection.

ORCID iD

Emily K. Miller

References

Benjet

Bromet

Karam

, et al. (2016) The epidemiology of traumatic event exposure worldwide: Results from the world mental health survey consortium. Psychological Medicine 46(2): 327–343.

Bleiweis

Boesch

Cawthorne Gaines

(2020) The Basic Facts About Women in Poverty [Fact Sheet]. Center for American Progress. Available at: https://www.americanprogress.org/article/basic-facts-women-poverty/

Blondeel

de Vasconcelos

García-Moreno

, et al. (2018) Violence motivated by perception of sexual orientation and gender identity: A systematic review. Bulletin of the World Health Organization 96(1): 29–41L.

Bridges-Curry

Newton

(2022) Patterns of trauma exposure, emotion dysregulation, and mental health symptoms: A latent class analysis. Journal of Aggression Maltreatment & Trauma 31(3): 285–303.

Celeux

Soromenho

(1996) An entropy criterion for assessing the number of clusters in a mixture model. Journal of Classification 13(2): 195–212.

Centers for Disease Control and Prevention (2022) Social determinants of health (Healthy People 2030). U.S. Department of Health and Human Services. Available at: https://health.gov/healthypeople/priority-areas/social-determinants-health

Cook

Spinazzola

Ford

, et al. (2005) Complex trauma in children and adolescents. Psychiatric Annals 35(5): 390–398.

Darity

Mullen

(eds) (2022) From Here to Equality: Reparations for Black Americans in the Twenty-First Century, 2nd edn. Chapel Hill: University of North Carolina Press. Available at: https://search.ebscohost.com/login.aspx?direct=true&AuthType=ip,shib&db=cat07006a&AN=cwru.b6524616&site=eds-live&custid=s8481523

Derogatis

(2001) BSI 18, Brief Symptom Inventory 18: Administration, Scoring and Procedures Manual. Minneapolis, MN: NCS Pearson, Inc.

10.

Derogatis

Melisaratos

(1983) The brief symptom inventory: An introductory report. Psychological Medicine 13(3): 595–605.

11.

Evans

Steel

DiLillo

(2013) Child maltreatment severity and adult trauma symptoms: Does perceived social support play a buffering role? Child Abuse & Neglect 37(11): 934–943.

12.

Falkenburger

Arena

Wolin

(2018) Trauma-Informed Community Building and Engagement. Washington: Urban Institute. Available at: https://www.urban.org/research/publication/trauma-informed-community-building-and-engagement

13.

Farmer

(2004) An anthropology of structural violence. Current Anthropology 45(3): 305–325.

14.

Felitti

Anda

Nordenberg

, et al. (1998) Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. The adverse childhood experiences (ACE) study. American Journal of Preventive Medicine 14(4): 245–258.

15.

Forman-Hoffman

Bose

Batts

, et al. (2012) Correlates of Lifetime Exposure to One or More Potentially Traumatic Events and Subsequent Posttraumatic Stress Among Adults in the United States: Results from the Mental Health Surveillance Study, 2008-2012 (CBHSQ Data Review). Substance Abuse and Mental Health Services Administration (US). Available at: http://www.ncbi.nlm.nih.gov/books/NBK390285/

16.

Guo

Zhang

(2023) The associations of pandemic-related difficulties with depressive symptoms and psychological growth among American older adults: Social support as moderators. Journal of Health Psychology 28(5): 417–433.

17.

Galtung

(1969) Violence, peace, and peace research. Journal of Peace Research 6(3): 167–191.

18.

Ginwright

(2019) The future of healing: Shifting from trauma informed care to healing centered engagement. Medium. Available at: https://medium.com/@ginwright/the-future-of-healing-shifting-from-trauma-informed-care-to-healing-centered-engagement-634f557ce69c

19.

Gunzler

Perzynski

Carle

. (2021) Structural Equation Modeling for Health and Medicine. New York: Chapman & Hall/CRC.

20.

Hagan

Roubinov

Cordeiro

, et al. (2022) Young children’s traumatic stress reactions to the COVID-19 pandemic: The long reach of mothers’ adverse childhood experiences. Journal of Affective Disorders 318: 130–138.

21.

Hao

Tan

Jiang

, et al. (2020) Do psychiatric patients experience more psychiatric symptoms during COVID-19 pandemic and lockdown? A case-control study with service and research implications for immunopsychiatry. Brain, Behavior, and Immunity 87: 100–106.

22.

Hatch

Dohrenwend

(2007) Distribution of traumatic and other stressful life events by race/ethnicity, gender, SES and age: A review of the research. American Journal of Community Psychology 40(3-4): 313–332.

23.

Jakubowski

Koffer

Matthews

, et al. (2023) Psychosocial impacts of the COVID-19 pandemic on women with trauma histories: Study of women’s health across the nation (SWAN). Journal of Traumatic Stress 36(1): 167–179.

24.

Keyes

Pratt

Galea

, et al. (2014) The burden of loss: Unexpected death of a loved one and psychiatric disorders across the life course in a national study. American Journal of Psychiatry 171(8): 864–871.

25.

Kim

Laurence

(2020) COVID-19 restrictions and mental distress among American adults: Evidence from corona impact survey (W1 and W2). Journal of Public Health 42(4): 704–711.

26.

Koenen

De Vivo

Rich-Edwards

, et al. (2009) Protocol for investigating genetic determinants of posttraumatic stress disorder in women from the Nurses’ Health Study II. BMC Psychiatry 9(1): 29.

27.

Lancaster

Melka

Rodriguez

(2011) Emotional predictors of PTSD symptoms. Psychological Trauma Theory Research Practice and Policy 3(4): 313–317.

28.

Liu

Zhang

Wong

GTF

, et al. (2020) Factors associated with depression, anxiety, and PTSD symptomatology during the COVID-19 pandemic: Clinical implications for U.S. Young adult mental health. Psychiatry Research 290: 113172.

29.

Lubben

Blozik

Gillmann

, et al. (2006) Performance of an abbreviated version of the Lubben social network scale among three European community-dwelling older adult populations. Gerontologist 46(4): 503–513.

30.

Lubben

(1988) Assessing social networks among elderly populations. Family & Community Health: The Journal of Health Promotion & Maintenance 11(3): 42–52.

31.

Ludy-Dobson

Perry

(2010) The role of healthy relational interactions in buffering the impact of childhood trauma. In: Gil E (ed) Working With Children to Heal Interpersonal Trauma: The Power of Play. New York: The Guilford Press, pp.26–43.

32.

Marie-Mitchell

Kostolansky

(2019) A systematic review of trials to improve child outcomes associated with adverse childhood experiences. American Journal of Preventive Medicine 56(5): 756–764.

33.

McLarnon

MJW

O’Neill

(2018) Extensions of auxiliary variable approaches for the investigation of mediation, moderation, and conditional effects in mixture models. Organizational Research Methods 21(4): 955–982.

34.

Mersky

Topitzes

Reynolds

(2013) Impacts of adverse childhood experiences on health, mental health, and substance use in early adulthood: A cohort study of an urban, minority sample in the U.S. Child Abuse & Neglect 37(11): 917–925.

35.

Monroe

Harkness

(2005) Life stress, the “kindling” hypothesis, and the recurrence of depression: Considerations from a life stress perspective. Psychological Review 112(2): 417–445.

36.

Muthén

(2000) Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism Clinical and Experimental Research 24(6): 882–891.

37.

Norris

Alegria

(2005) Mental health care for ethnic minority individuals and communities in the aftermath of disasters and mass violence. CNS Spectrums 10(2): 132–140.

38.

Oakes

Kaufman

(2017) Methods in Social Epidemiology. San Francisco: John Wiley & Sons.

39.

Post

Weiss

(1998) Sensitization and kindling phenomena in mood, anxiety, and obsessive-compulsive disorders: The role of serotonergic mechanisms in illness progression. Biological Psychiatry 44(3): 193–206.

40.

Prout

Zilcha-Mano

Aafjes-van Doorn

, et al. (2020) Identifying predictors of psychological distress during COVID-19: A machine learning approach. Frontiers in Psychology 11: 586202.

41.

Raifman

(2020) Disparities in the population at risk of severe illness from COVID-19 by race/ethnicity and income. American Journal of Preventive Medicine 59(1): 137–139.

42.

Rich

Gidycz

Warkentin

, et al. (2005) Child and adolescent abuse and subsequent victimization: A prospective study. Child Abuse & Neglect 29(12), 1373–1394. https://doi.org/10.1016/j.chiabu.2005.07.003

43.

Roberts

Gilman

Breslau

, et al. (2011) Race/ethnic differences in exposure to traumatic events, development of post-traumatic stress disorder, and treatment-seeking for post-traumatic stress disorder in the United States. Psychological Medicine 41(1): 71–83.

44.

Roubinov

Luecken

Curci

, et al. (2021) A prenatal programming perspective on the intergenerational transmission of maternal adverse childhood experiences to offspring health problems. American Psychologist 76(2): 337–349.

45.

Rumas

Shamblaw

Jagtap

, et al. (2021) Predictors and consequences of loneliness during the COVID-19 Pandemic. Psychiatry Research 300: 113934.

46.

Russo

Dhruve

Oliveros

(2022) Coping with COVID-19: Testing the stress sensitization hypothesis among adults with and without a history of adverse childhood experiences. Journal of affective disorders reports 10: 100379.

47.

Saul

(2013) Collective Trauma, Collective Healing: Promoting Community Resilience in the Aftermath of Disaster. Milton Park: Routledge.

48.

Schnurr

Vielhauer

Weathers

, et al. (1999) Brief Trauma Questionnaire. Available at: https://doi.org/10.1037/t07488-000

49.

Schnurr

Spiro

III A

Vielhauer

, et al. (2002) Trauma in the lives of older men: Findings from the normative aging study. Journal of Clinical Geropsychology 8(3): 175–187.

50.

Shriver

; Progress, C. for a (2014) The Shriver Report: A Woman’s Nation Pushes Back From the Brink. New York: Macmillan.

51.

Stroud

(2020) The stress sensitization model. In: Harkness

Hayden

(eds) The Oxford Handbook of Stress and Mental Health. 2020-06187-016. New York City: Oxford University Press, pp.349–370.

52.

Substance Abuse and Mental Health Services Administration (2014) SAMHSA’s Concept of Trauma and Guidance for a Trauma-Informed Approach. Washington, DC: U.S. Department of Health and Human Services, pp.1–27.

53.

Thoits

(2010) Stress and health: Major findings and policy implications. Journal of Health and Social Behavior 51(1_suppl): S41–S53.

54.

Turner

Lloyd

(2004) Stress burden and the lifetime incidence of psychiatric disorder in young adults. Archives of General Psychiatry 61(5): 481–488.

55.

Valentine

Marques

Wang

, et al. (2019) Gender differences in exposure to potentially traumatic events and diagnosis of posttraumatic stress disorder (PTSD) by racial and ethnic group. General Hospital Psychiatry 61: 60–68.

56.

Wang

Zhai

Chen

, et al. (2022) A positive legacy of trauma? The role of perceived social support on mental health among earthquake survivors during the COVID-19 outbreak. Frontiers in Psychiatry 13, 845929. https://doi.org/10.3389/fpsyt.2022.845929

57.

Wang

M-C

Deng

, et al. (2017) Performance of the entropy as an index of classification accuracy in latent profile analysis: A Monte Carlo simulation study. Acta Psychologica Sinica 49(11): 1473–1482.

58.

Wilson

Scarpa

(2014) Childhood abuse, perceived social support, and posttraumatic stress symptoms: A moderation model. Psychological Trauma Theory Research Practice and Policy 6(5): 512–518.

59.

World Health Organization and Calouste Gulbenkian Foundation (2014) Social Determinants of Mental Health. Geneva: World Health Organization.

60.

Jia

Shi

, et al. (2021) Prevalence of mental health problems during the COVID-19 pandemic: A systematic review and meta-analysis. Journal of Affective Disorders 281: 91–98. https://doi.org/10.1016/j.jad.2020.11.117

61.

S-C

Chen

H-R

Liu

A-C

, et al. (2020) Toward COVID-19 information: Infodemic or fear of missing out? Healthcare 8(4): 550.

62.

Zieff

Kerr

Moore

, et al. (2020) Universal healthcare in the United States of America: A healthy debate. Medicinar 56(11): 580. doi: 10.3390/medicina56110580