Each year, millions of patients in the USA are hospitalized following traumatic injury or illness (Rui, Kang, & Albert, Reference Rui, Kang and Albert2013), many of whom are at risk for posttraumatic stress disorder (PTSD), depression, and anxiety (Carlson, Palmieri, & Spain, Reference Carlson, Palmieri and Spain2017; Grant, Beck, Marques, Palyo, & Clapp, Reference Grant, Beck, Marques, Palyo and Clapp2008). A recent study found that 25.1% and 11.5% of patients receiving emergency care after a motor vehicle crash (the most common trauma in industrialized countries) met criteria for 3-month PTSD and depression, respectively (Ziobrowski et al., Reference Ziobrowski, Kennedy, Ustun, House, Beaudoin, An and van Rooij2021). Risk for posttraumatic symptoms has also been found to differ by racial and ethnic groups. Prior studies of hospitalized injury survivors showed that Latinx and Black patients had significantly elevated risk for developing early and 12-month PTSD symptoms relative to White patients (Santos et al., Reference Santos, Russo, Aisenberg, Uehara, Ghesquiere and Zatzick2008; Zatzick et al., Reference Zatzick, Rivara, Nathens, Jurkovich, Wang, Fan and Mackenzie2007). Another USA population-based study also found that Black patients hospitalized following an acute injury had significantly higher risk of early PTSD symptoms than White patients (Stephens et al., Reference Stephens, Sue, Roy-Byrne, Unützer, Wang, Rivara and Zatzick2010). In contrast, a recent study found no racial/ethnic differences in PTSD symptoms within 3-months after trauma, but lower posttraumatic anxiety and depression symptoms in Black patients relative to White patients, which were accounted for by prior trauma exposure and childhood emotional abuse (Harnett et al., Reference Harnett, Dumornay, Delity, Sanchez, Mohiuddin, Musey and Ressler2022). There is limited research examining whether additional risk factors prior to, at the time of, and post-injury or illness might explain racial/ethnic differences in acute and longer-term posttraumatic symptoms.
Prior research has suggested that racial/ethnic differences in posttraumatic stress may result from differences in sociodemographic and clinical factors, the type and number of traumatic exposures, and support from family and friends (Alegría et al., Reference Alegría, Fortuna, Lin, Norris, Gao, Takeuchi and Valentine2013). Social support, including negative social reactions in response to trauma disclosure and social constraints that lead the trauma survivor to feel unsupported or misunderstood, has also been identified as one of the most robust predictors of posttraumatic stress (Zalta et al., Reference Zalta, Tirone, Orlowska, Blais, Lofgreen, Klassen and Dent2021). A prior study found that variations in social support explained differences in PTSD between Black and White inner-city women in the USA (Gaffey et al., Reference Gaffey, Aranda, Burns, Purim-Shem-Tov, Burgess, Beckham and Hobfoll2019). Racial/ethnic identity may also be a proxy for experiences of minoritized groups that confer greater risk for mental health problems (Holzer & Copeland, Reference Holzer, Copeland, Paniagua and Yamada2013). For example, US racial/ethnic minority populations are more likely to live in high-stress environments (Jackson, Knight, & Rafferty, Reference Jackson, Knight and Rafferty2010), and pre-trauma life stress has been found to be strongly associated with posttraumatic stress (Carlson et al., Reference Carlson, Palmieri, Field, Dalenberg, Macia and Spain2016). Racial/ethnic disparities in posttraumatic stress could also contribute to socioeconomic disadvantages and cycles of violence, so that responses to trauma can lead to additional trauma exposure (Eitle & Turner, Reference Eitle and Turner2002).
According to the sociocultural framework for mental health disparities (Alegría, Pescosolido, Williams, & Canino, Reference Alegría, Pescosolido, Williams, Canino, Pescosolido, Martin, McLeod and Rogers2011), competing needs and discrimination may also explain racial/ethnic differences in trauma responses. Competing needs are basic social and economic resources; when unmet, they can impair recovery after trauma by creating barriers to addressing mental health needs – including transportation, financial, and information barriers (Davis, Ressler, Schwartz, Stephens, & Bradley, Reference Davis, Ressler, Schwartz, Stephens and Bradley2008). Discrimination in the form of major, acute experiences and daily, low level microaggressions is also related to racial/ethnic disparities in mental health (Torres & Taknint, Reference Torres and Taknint2015; Williams & Mohammed, Reference Williams and Mohammed2009). Discrimination in criminal justice, political, and immigration contexts may also lead to negative expectations and feelings of hopelessness (Utsey & Ponterotto, Reference Utsey and Ponterotto1996). Recent evidence demonstrates that racial discrimination increased the risk of developing future PTSD symptoms following traumatic injury among Black patients (Bird et al., Reference Bird, Webb, Schramm, Torres, Larson and deRoon-Cassini2021).
Using a sample of patients hospitalized after emergency care following severe injury or illness, we examined whether there were racial/ethnic differences in acute and longer-term PTSD, depression, and anxiety symptoms, and whether pre-, peri-, and post-trauma risk factors explained such differences. Severely injured or ill patients were included since the ICD-11 and DSM-5 diagnostic systems both specify injury and life-threatening illness as potential traumatic stressors. Using a theoretical framework delineating the risk factors associated with posttraumatic symptoms (Carlson, Dalenberg, & Muhtadie, Reference Carlson, Dalenberg, Muhtadie, Reyes, Elhai and Ford2008) in combination with the sociocultural framework for mental health disparities (Alegría et al., Reference Alegría, Pescosolido, Williams, Canino, Pescosolido, Martin, McLeod and Rogers2011), we examined how pre-, peri-, and post-trauma risk factors interacted with race/ethnicity to reflect mechanisms of action that might confer vulnerability or resilience to traumatic stress, exacerbate distress in response to trauma, or foster rapid recovery after exposure. We hypothesized that Latinx and Black patients would report higher levels of PTSD, depression, and anxiety symptoms than White patients, and that these differences would be accounted for by pre-, peri-, and post-trauma risk factors that could influence trauma response.
Method
Participants and procedure
Participants were 1320 adult patients (18–89 years old) admitted after emergency care following severe injury (n = 661) or illness (n = 658) at hospitals in Palo Alto, CA, Akron, OH, and Baltimore, MD. Injuries included motor vehicle crash, being hit by a car, physical assault, falling, work injury, knife or gunshot, sport injuries, burns, and animal bites. Illnesses included heart, breathing, gastrointestinal, or liver problems; sepsis or bad infections; brain, nerves, vision, or speech problems; problems with blood sugar and diabetes; sickle cell disease; kidney problems; and cancer. Using self-reported information, patients were coded into mutually exclusive groups as ‘Asian, Native Hawaiian, or Other Pacific Islander’ (‘Asian’; n = 91), ‘Hispanic or Latino’ (‘Latinx’; n = 196), ‘non-Hispanic Black’ (‘Black’; n = 310), ‘multiracial’ (n = 60), ‘non-Hispanic White’ (‘White’; n = 654), ‘American Indian or Alaska Native’ (n = 7), and ‘other-race’ (n = 2). Due to small sample size, American Indian or Alaska Native and other-race patients (n = 9) were excluded, and so was the only White patient reporting their gender as ‘other’. Thus, our final analytical sample included 1310 patients.
Patients were identified via electronic records and approached in their hospital rooms soon after admission (mean = 4 days, median = 3 days, mode = 2 days) between June 2018, and February 2021 (time 1). Eligible patients had cognitive capacity to answer study questions and were able to provide contact information for follow-up assessments. Patients admitted for psychiatric treatment were excluded. After providing informed consent, patients completed all study measures at time 1 using tablet computers, paper questionnaires, or orally. Bilingual staff recruited patients who spoke or read English (n = 1216), Spanish (n = 87; all but one Latinx patients), or Mandarin (n = 7; all Asian patients). Repeated measures of posttraumatic symptoms were collected in two follow-up assessments at 2 and 6 months post-admission (times 2 and 3). Trauma responses did not differ at any time point between patients admitted after injury or illness. We also found no differences in trauma responses between Latinx patients who completed their assessments in Spanish or English, between Asian patients who completed their assessments in Mandarin or English, or between patients admitted before or during the COVID-19 pandemic (i.e. before v. on or after March 2020). There were racial/ethnic differences in follow-up response rates [χ2(12) = 109.68, p < 0.01]. Analyses of missing data showed that within racial/ethnic groups, patients who completed all follow-up assessments were statistically similar to patients who missed at least one follow-up assessment, which made plausible assuming that data were missing at random. Thus, we applied multiple imputation methods following a prior study of repeated patient-reported measures with high rates of missing data (Biering, Hjollund, & Frydenberg, Reference Biering, Hjollund and Frydenberg2015; see details in online Supplementary material).
Study measures
Good internal consistency (Cronbach's α) was observed for all measures included in the present study across all racial and ethnic groups (a complete breakdown is presented in online Supplementary Table S1).
Outcomes: PTSD, depression, and anxiety symptoms at times 1–3
Symptoms were rated on a six-point scale from 0, ‘none of the time/not at all’, to 5, ‘all or most of the time/four or more times a day’. The time reference was since the time of hospitalization at time 1 and in the past week at times 2 and 3. A total symptom severity score was derived by summing all items, with higher scores representing worse symptoms. At time 1, measures of depression and anxiety were shortened to reduce burden on acutely injured or ill patients based on full measures collected among the first 518 patients. These shortened versions were used at all time points in the current analyses. PTSD symptoms were evaluated using the Screen for Posttraumatic Stress Symptoms (Carlson, Reference Carlson2001), a 20-item self-report assessing PTSD symptoms according to DSM-5 criteria (range = 0–100). Depression symptoms were assessed using six of the nine items of the Patient Health Questionnaire-9 (PHQ-9) (Kroenke & Spitzer, Reference Kroenke and Spitzer2002), a self-report assessing symptoms of DSM-IV depressive disorders (range = 0–30, r = 0.97 with full PHQ-9). Anxiety symptoms were assessed with two of the seven items of the Generalized Anxiety Disorder-7 (GAD-7) (Spitzer, Kroenke, Williams, & Löwe, Reference Spitzer, Kroenke, Williams and Löwe2006) – a self-report assessing GAD symptoms according to DSM-IV diagnostic criteria – plus three novel items assessing anxiety independent of a diagnosis and using updated language developed for the present study (range = 0–25, r = 0.98 with full GAD-7 plus three novel items).
Pre-trauma risk factors from time 1
Sociodemographic data included gender (female, male), age (18–89 years old), marital status (never married, married/cohabitating, separated/divorced/widowed, other), years of education (0–11, 12, 13–15, 16+), and household income (<$35 K, $35 K to <$75 K, $75 K+). Two early-life risk factors highly predictive of mental health outcomes (Carlson et al., Reference Carlson, Palmieri, Field, Dalenberg, Macia and Spain2016) were assessed using measures of childhood home life (How was your home life growing up?, 1 ‘unhappy’ to 5 ‘very happy’) and caretaker dysfunction (sum of affirmative responses about whether any caretaker ever stayed in a hospital for emotional or psychiatric reasons, attempted or completed suicide, abused drugs or alcohol, or was arrested). Two items for number of lifetime sudden, terrible events, and how many of these events were upsetting for a month or longer assessed past trauma exposure. Both items were censored at the 98th percentile (50 and 25 events, respectively) to reduce the effect of extreme outliers, and averaged to create a summary score (range = 0–37.5).
The nine-item Everyday Discrimination Scale (EDS) was used to capture everyday experiences with unfair treatment that are chronic or episodic but generally minor (0 ‘never’ to 5 ‘almost every day’, range = 0–45) (Kim, Sellbom, & Ford, Reference Kim, Sellbom and Ford2014; Williams & Mohammed, Reference Williams and Mohammed2009). Extreme discrimination was assessed using the Race-Related Events Scale (RRES), a brief screening of lifetime exposure to stressful and potentially traumatizing experiences due to race/ethnicity, gender, or religion (Waelde et al., Reference Waelde, Pennington, Mahan, Mahan, Kabour and Marquett2010). A count of 12 RRES items not overlapping with the EDS were included (e.g. verbal conflict, physical fight/assault, being chased). Financial stress was measured with three items from past epidemiological research (Alegria et al., Reference Alegria, Vila, Woo, Canino, Takeuchi, Vera and Shrout2004) about whether patients would say they have enough money to meet their needs (0 ‘more than I need’ to 2 ‘not enough’), whether it was hard to pay their monthly bills (0 ‘not at all hard’ to 3 ‘very hard’), and how often in the past 12 months they did not have enough money to buy food (0 ‘never’ and 3 ‘often’). Total scores were computed summing all items (range = 0–8). Past mental health problems were measured with one item asking how much of the time in the past had feeling anxious, down, or depressed kept the patient from enjoying life (0 ‘not at all’ to 5 ‘all or almost all the time’).
Peri-trauma risk factors from time 1
Two items assessing patients' perception of how terrible and out of control the event that brought them to the hospital seemed (0 ‘not at all’ to 4 ‘a lot’) were used to measure subjective trauma severity. Total scores were calculated as the sum of both items (range = 0–8).
Post-trauma risk factors from time 1
Expected social support was assessed with six items from the Medical Outcomes Study Social Support Survey, a self-report of recent thinking about four dimensions of social support: tangible, affectionate, emotional/informational, and positive social interaction (Sherbourne & Stewart, Reference Sherbourne and Stewart1991). Patients were asked how much support they think they would get – if needed – after leaving the hospital (0 ‘none of the time’ to 5 ‘all or most of the time’, range = 0–30). Social constraints were assessed with seven items from the Leopore Social Constrains Scale, a self-report of perceived friends and family constraining behaviors on expressing trauma-related thoughts and feelings (e.g. friends or family act uncomfortable when you talk about the event that brought you to the hospital; 0 ‘none of the time’ to 5 ‘all or most of the time’, range = 0–35) (Lepore & Ituarte, Reference Lepore and Ituarte1999).
Data analysis plan
We first described differences in risk factors at time 1 between racial/ethnic minority and White patients. Using individual growth curve models, we then examined racial/ethnic differences in PTSD, depression, and anxiety symptoms at the time of hospitalization (time 1) and 2 and 6 months post-admission (times 2 and 3). These models also examined racial/ethnic differences in the rate of PTSD, depression, and anxiety symptoms change over time. Symptoms at times 1–3 were separately modeled as the dependent variable. Individual growth curve models allowed us to (1) account for the correlation between repeated measures from the same patient, (2) account for racial/ethnic differences in responses at each time point, and (3) examine how responses changed over time within patients and how that change was related to a patient's race/ethnicity. The final equation was as follows:
where y ij represents patient i's symptoms at time j, time ij represents follow-up time (coded 0, 2, and 6 at times 1, 2, and 3, respectively), race/ethnicity i indicates patient i's racial/ethnic identity, and risk factors i are a vector of pre-, peri-, and post-trauma risk factors at time 1. The parameter b 0 represents symptoms at time 1 for the referent group (White patients), and the vector b1 measures differences in symptoms at time 1 between racial/ethnic minority and White patients. The parameter b 2 measures symptoms change over time for White patients, and the vector b3 measures differences in the rate of change between racial/ethnic minority and White patients. We estimated two sets of models, with and without adjustment for risk factors. In post-hoc analyses, we expanded the model from Equation (1) to examine whether risk factors at time 1 partially explained racial/ethnic differences in both outcome scores at each time point and outcome score change over time. To avoid model overspecification, we estimated separate models considering the role of each risk factor in explaining racial/ethnic differences. Analyses were conducted using the Stata software version 15 (StataCorp, 2017) and the RStudio software version 1.4.1717 (RStudio Team, 2021; see online Supplementary Material for additional details).
Results
Racial/ethnic differences in risk factors
There was racial/ethnic variation for most pre-, peri-, and post-trauma risk factors at time 1, and specific patterns varied across risk variables (Table 1). Compared to White patients, Asian patients were the most likely to report being married or cohabitating, having the most education and income, and the lowest levels of caretaker dysfunction and financial stress. Latinx patients were the most likely to report not having a high school degree, the lowest levels of everyday discrimination, and the highest levels of subjective trauma severity and social constraints. Black patients were the most likely to report never being married and not being in the highest household income category; they also had the second highest levels of everyday discrimination, extreme discrimination, and financial stress, and the second lowest levels of expected social support. Multiracial patients were the most likely to report not having a college degree, being in the lowest household income category, being unhappy about their childhood home life, having the highest levels of caretaker dysfunction, everyday discrimination, extreme discrimination, and financial stress, and the lowest levels of expected social support.
Table 1. Descriptive statistics of risk factors at the time of hospitalization (time 1) by race and ethnicity
a F value of omnibus test comparing non-Latinx White patients and all other racial/ethnic groups.
b Number of lifetime sudden/terrible events were Winsorized to the 98th percentile to reduce the effect of extreme outliers.
c Subjective trauma severity was assessed using two items about how ‘terrible’ and ‘out of control’ the event that brought participants to the hospital seemed (0 ‘not at all’ and 4 ‘a lot’).
Racial/ethnic differences in PTSD, depression, and anxiety symptoms
In unadjusted models, Latinx, Black, and multiracial patients had higher PTSD scores at time 1 than White patients (effect size ‘d’ = 0.27, 0.28, and 0.57, respectively) (Table 2 and Fig. 1a). Two and 6 months post-admission, PTSD symptoms remained higher among Black (d = 0.29–0.34) and multiracial (d = 0.49–0.51) but not Latinx patients. No racial/ethnic differences were observed at either time 1, 2, or 3 in depression symptoms (Table 2 and Fig. 2a). Latinx and multiracial patients had higher anxiety scores at time 1 than White patients (d = 0.21 and 0.27, respectively), which remained higher at time 2 among multiracial (d = 0.21) but not Latinx patients. No racial/ethnic differences in anxiety symptoms were observed by time 3 (Table 2 and Fig. 3a). After adjusting for pre-, peri-, and post-trauma risk factors, most differences in PTSD symptoms at either time point were no longer observed except for Latinx patients, who had lower PTSD scores at time 3 than White patients (d = 0.19) (Table 2 and Fig. 1b). In adjusted models, Black patients also had lower depression and anxiety scores at times 1 and 2 than White patients (d = 0.10–0.27), but they reached the same levels by time 3 (Table 2 and Figs 2b and 3b).
Fig. 1. Unadjusted and adjusted mean PTSD scores at the time of hospitalization (time 1), 2 months post-admission (time 2), and 6 months post-admission (time 3) by race/ethnicity.
Fig. 2. Unadjusted and adjusted mean depression scores at the time of hospitalization (time 1), 2 months post-admission (time 2), and 6 months post-admission (time 3) by race/ethnicity.
Fig. 3. Unadjusted and adjusted mean anxiety scores at the time of hospitalization (time 1), 2 months post-admission (time 2), and 6 months post-admission (time 3) by race/ethnicity.
Table 2. PTSD, depression, and anxiety symptom scores change over time by race/ethnicity in unadjusted models and in models adjusting for risk factors at time 1
PTSD, posttraumatic stress disorder; CI, confidence interval; s.d., standard deviation of the referent group (in parentheses).
a Models adjusted for pre-, peri-, and post-trauma risk factors at Time 1.
*Statistically significant outcome scores change at the 0.05 level across times.
‡ Statistically significant difference at the 0.05 level compared to White patients, conditional on a significant omnibus test to adjust for multiple comparisons.
On average, PTSD, depression, and anxiety symptoms significantly improved between times 1 and 3 among Latinx and White patients, but were stable among Asian, Black, and multiracial patients (Table 2). Symptom scores change over time was neither greater nor lower among racial/ethnic minority compared to White patients except for PTSD scores, which were on average 3.86 points lower between times 1 and 3 among Latinx patients, and this improvement was greater than the average 1.32-point decrease among White patients (d = 0.40). Because risk factors were measured at time 1 only and thus included as time invariant predictors, the rate of symptom scores change over time remained unchanged in adjusted compared to unadjusted models.
Post-hoc analyses
We examined whether risk factors at time 1 partially explained adjusted differences in PTSD symptoms between Latinx and White patients (online Supplementary Fig. S1) and differences in depression and anxiety symptoms between Black and White patients (online Supplementary Figs S2 and S3). Everyday discrimination, financial stress, past mental health problems, and social constraints were consistently associated with the observed differences. Latinx and White patients with no everyday discrimination, financial stress, past mental health problems, and social constraints had about the same PTSD scores that were stable over time. In contrast, Latinx patients with the highest levels of everyday discrimination, financial stress, past mental health problems, and social constraints had PTSD scores that steadily improved at a faster rate comparted to White patients (online Supplementary Fig. S1). Similarly, Black and White patients with no everyday discrimination, financial stress, past mental health problems, and social constraints had the same depression and anxiety scores. In contrast, Black patients with the highest levels of everyday discrimination, financial stress, past mental health problems, and social constraints had lower depression and anxiety scores at times 1 and 2 than White patients, which reached the same level by time 3 (online Supplementary Figs S2 and S3).
Discussion
We examined whether differences in pre-, peri-, and post-trauma risk factors explained differences in acute and longer-term PTSD, depression, and anxiety symptoms between racial/ethnic minority and White patients hospitalized after emergency care for severe injury or illness. Specific patterns were consistent with the observed differences in posttraumatic symptoms. Expressly, Latinx, Black, and multiracial patients reported higher levels of risk factors that exacerbate trauma response, including lower socioeconomic status (i.e. lower educational attainment, lower income, and higher levels of financial stress) (Alegría et al., Reference Alegría, Fortuna, Lin, Norris, Gao, Takeuchi and Valentine2013; Brewin, Andrews, & Valentine, Reference Brewin, Andrews and Valentine2000), extreme discrimination (Asnaani & Hall-Clark, Reference Asnaani and Hall-Clark2017), and social constraints (Zalta et al., Reference Zalta, Tirone, Orlowska, Blais, Lofgreen, Klassen and Dent2021). Latinx patients also reported an elevated peri-traumatic response (subjective trauma severity), which has been associated with an increased risk of PTSD (Alcántara, Casement, & Lewis-Fernández, Reference Alcántara, Casement and Lewis-Fernández2013). These differences might have contributed to elevations in PTSD symptoms at the time of hospitalization among Latinx, Black, and multiracial patients. Furthermore, Black and multiracial patients reported higher levels of additional risk factors that exacerbate trauma response, such as lower expected social support (Zalta et al., Reference Zalta, Tirone, Orlowska, Blais, Lofgreen, Klassen and Dent2021) and everyday discrimination (Asnaani & Hall-Clark, Reference Asnaani and Hall-Clark2017; Bird et al., Reference Bird, Webb, Schramm, Torres, Larson and deRoon-Cassini2021). These additional cumulative risk factors might explain why elevations in PTSD symptoms persisted at 2- and 6-months post-hospitalization in Black and multiracial patients. The attenuation of racial/ethnic differences in PTSD symptoms once risk factors were accounted for suggested that pre-, peri-, and post-trauma risks factors might play causal roles in observed disparities in mental health following severe injury or illness.
Consistent with two prior studies suggesting an absence of racial/ethnic variability in vulnerability to trauma-related psychopathology other than PTSD (McLaughlin et al., Reference McLaughlin, Alvarez, Fillbrunn, Green, Jackson, Kessler and Alegría2019) or in the association of stressful life events with later depression and anxiety (Turner & Lloyd, Reference Turner and Lloyd2004), we did not observe racial/ethnic differences in depression symptoms. Anxiety symptoms were higher among Latinx and multiracial patients at the time of hospitalization compared to White patients, but these differences faded either at 2 months (Latinx patients) or at 6 months post-admission (multiracial patients). After adjusting for risk factors, a significant difference emerged whereby Black patients had lower depression and anxiety symptoms within 2 months post-admission than White patients, a difference no longer observed at 6 months. A prior study examining racial/ethnic differences in posttraumatic stress within 3 months after trauma also found that Black individuals displayed lower depression and anxiety symptoms than White individuals, which was accounted for by differences in prior trauma exposure (Harnett et al., Reference Harnett, Dumornay, Delity, Sanchez, Mohiuddin, Musey and Ressler2022). Our results suggest that differences in depression and anxiety symptoms between Black and White patients in the first 2 months after trauma were not completely accounted for by pre-, peri-, and post-trauma risk factors.
Depression and anxiety symptom trajectory did not differ by race/ethnicity, but the rate of decline in PTSD symptoms was faster in Latinx than White patients. No differences in symptom trajectory were also found in Harnett et al. (Reference Harnett, Dumornay, Delity, Sanchez, Mohiuddin, Musey and Ressler2022) study of posttraumatic symptoms within 3 months after trauma, including depression, anxiety, and PTSD symptoms (Harnett et al., Reference Harnett, Dumornay, Delity, Sanchez, Mohiuddin, Musey and Ressler2022). Our results suggest that differences in recovery trajectories beyond 3 months might still not occur for depression and anxiety symptoms (although they may occur beyond the 6-month period captured by the current analyses), but they might occur for PTSD symptoms. Although a faster rate of PTSD symptom improvement did not translate into lower symptoms at either 2 or 6 months among Latinx patients, lower 6-month symptoms emerged after accounting for risk factors. The effect of a faster rate of improvement in PTSD symptoms might thus be magnified after accounting for higher levels of risk factors in Latinx patients. This result is in direct contrast to prior research documenting either no differences in risk or higher risk of developing PTSD symptoms following trauma exposure among Latinx compared to White adults (Alcántara et al., Reference Alcántara, Casement and Lewis-Fernández2013; McLaughlin et al., Reference McLaughlin, Alvarez, Fillbrunn, Green, Jackson, Kessler and Alegría2019). A higher risk of PTSD symptoms after trauma might be explained by the fact that prior studies included Latinx adults with high rates of exposure to one of the types of trauma most strongly associated with psychopathology (i.e. physical violence), while we studied Latinx patients experiencing severe injury or illness. Everyday discrimination, financial stress, past mental health problems, and social constraints were four possible mechanisms underlying the lower risk of developing PTSD symptoms 6 months post-trauma. These findings are consistent with prior evidence documenting a significant association between chronic PTSD (i.e. beyond 6 months post-trauma) and racial/ethnic discrimination (Sibrava et al., Reference Sibrava, Bjornsson, Pérez Benítez, Moitra, Weisberg and Keller2019), social disadvantage (Alcántara et al., Reference Alcántara, Casement and Lewis-Fernández2013), psychiatric history, and post-trauma negative social support (O'Donnell et al., Reference O'Donnell, Creamer, Parslow, Elliott, Holmes, Ellen and Bryant2008). However, there is limited research regarding racial/ethnic differences in chronic PTSD, with some studies assessing discrimination in only some racial/ethnic groups or not addressing racial/ethnic variation at all. Our results are also consistent with prior literature suggesting that lower vulnerability to trauma-related psychopathology may be the result of greater psychological preparedness among US minority groups who routinely encounter day-to-day and extreme discrimination, stigma, and structural oppression (Kessler, Mickelson, & Williams, Reference Kessler, Mickelson and Williams1999; Lewis, Cogburn, & Williams, Reference Lewis, Cogburn and Williams2015; Williams, Neighbors, & Jackson, Reference Williams, Neighbors and Jackson2008).
Several limitations are worth noting. We studied patients hospitalized after emergency care following severe injury or illness. Our findings may thus not apply to survivors of other types of trauma exposures. Second, posttraumatic symptoms were assessed using self-reported measures. A prior systematic review suggested that variation in conditional risk for PTSD among Latinx relative to White adults was not significant when clinician-administered diagnostic interviews were used in lieu of self-reports (Alcántara et al., Reference Alcántara, Casement and Lewis-Fernández2013). Future research might thus benefit from assessing posttraumatic symptoms using different methods. Third, we had a small sample of Asian and multiracial patients, which limited our power to detect small differences. Prior research has suggested that Asian adults have a lower risk of developing PTSD symptoms after trauma than White adults (Alegría et al., Reference Alegría, Fortuna, Lin, Norris, Gao, Takeuchi and Valentine2013; McLaughlin et al., Reference McLaughlin, Alvarez, Fillbrunn, Green, Jackson, Kessler and Alegría2019), a finding not observed in our study. Similarly, although no differences in trauma responses by language were observed, we also had limited power given a small sample of patients who responded to the assessments in Spanish or Mandarin. Due to very small sample size, American Indian or Alaska Native patients were excluded from our analyses, but previous research has demonstrated high rates of trauma exposure and PTSD in American Indian and Alaska Native samples (Bassett, Buchwald, & Manson, Reference Bassett, Buchwald and Manson2014). Future work is needed to examine whether the present findings can extend to these minority populations. Data limitations did not allow us to examine racial/ethnic differences by subethnicity and nativity, which have been associated with psychopathology in prior studies (Vilsaint et al., Reference Vilsaint, NeMoyer, Fillbrunn, Sadikova, Kessler, Sampson and Alegría2019). Finally, there were significant rates of follow-up nonresponse and attrition, which may cause bias and reduce efficiency. However, analyses of missing data patterns made plausible assuming that data were missing at random and the choice of multiple imputation viable. Furthermore, a recent study found that for data missing at random, valid multiple imputation reduces bias even when the proportion of missingness is large (Madley-Dowd, Hughes, Tilling, & Heron, Reference Madley-Dowd, Hughes, Tilling and Heron2019).
Several strengths are also worth noting. We considered a variety of risk factors theoretically related to racial/ethnic disparities in mental health among a diverse cohort of patients with longitudinal data. Study procedures systematically recruited a broad range of patients, including patients with any type of injury or illness, Asian and Multiracial patients, and patients who only spoke Spanish or Mandarin. Prior similar studies have included only traumatic injury patients (Harnett et al., Reference Harnett, Dumornay, Delity, Sanchez, Mohiuddin, Musey and Ressler2022) or motor vehicle crash patients (Ziobrowski et al., Reference Ziobrowski, Kennedy, Ustun, House, Beaudoin, An and van Rooij2021). Our findings may thus be applicable to a larger and more diverse population, and provide information about different levels of risk factors potentially related to racial/ethnic differences in posttraumatic symptoms. Lastly, by using imputed data, differences in posttraumatic symptoms were less likely to be associated with race- or ethnicity-related attrition patterns (e.g. Black patients with worse symptoms being more likely to be lost to follow-up than White patients).
The present study showed that although experiencing a potentially traumatic event, such as severe injury or illness, impacted patients' mental health, responses to trauma appeared to differ not by race or ethnicity but by the patients' underlying risk profile. Our results identified pre-trauma exposure to everyday discrimination, financial stress, and past mental health problems and post-trauma exposure to social constraints as salient factors that merit early intervention. Recent work demonstrates that posttraumatic responses in the early aftermath of trauma may be predictive of later trajectories of chronic dysfunction or may map onto other trajectories of gradual recovery and delayed reactions (Bonanno & Mancini, Reference Bonanno and Mancini2012; Galatzer-Levy et al., Reference Galatzer-Levy, Ankri, Freedman, Israeli-Shalev, Roitman, Gilad and Shalev2013; Shalev et al., Reference Shalev, Gevonden, Ratanatharathorn, Laska, van der Mei, Qi and van Zuiden2019). Thus, our findings not only shed light on pre- and post-trauma risk factors associated with racial/ethnic differences in the acute aftermath of trauma, but also highlight risk factors for which early identification could reduce the debilitating effect of traumatic stressors of severe injury or illness. Screening for mental health risk by brief assessment of these risks could afford an opportunity to inform patients about their risk and be an equitable approach to preventive mental health care.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291722002112.
Acknowledgements
The investigators wish to thank the Stanford REDCap platform (http://redcap.stanford.edu), which is developed and operated by the Stanford Medicine Research IT team. The REDCap platform services at Stanford are subsidized by the Stanford School of Medicine Research Office, the National Center for Research Resources, and the National Center for Advancing Translational Sciences, National Institutes of Health, through grant UL1 TR001085.
Financial support
Funding for this study was supported by NIMHD R01MD012273 (MCG, MA, PAP, DAS, MRB, MW, EBC) and NIMH R01MH117247 (MCG, MA). This manuscript represents the views of its authors and not of their employers, the VA, the NIMHD, the NIMH, or the US government.
Conflict of interest
None.
Open access
