Police-Related Stress and Carotid Intima-Media Thickness Among African American Women

Abstract

Objective:

Anti-Black police violence and harassment have been identified as public health issues. However, studies have primarily focused on direct and/or vicarious police encounters. A dearth of studies has also examined vigilance related to future police encounters, and to our knowledge, no studies have examined all three kinds of police-related stress among African American women.

Methods:

We employed a latent class analysis (LCA) approach to identify classes of African American women (N = 422), aged 30–46, based on the patterning of various forms of self-reported police-related stress: direct, vicarious, police-related vigilance for self, and police-related vigilance for children. We then examined associations between latent class membership and carotid intima-media thickness (IMT), a marker of cardiovascular risk.

Results:

We identified three latent classes of police-related stress: High Child Vigilance-High Personal Exposure, No Child Vigilance-High Personal Exposure, and Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure class. Findings from the fully adjusted model reveal that the No Child Vigilance-High Personal Exposure class had lower common carotid artery (CCA) IMT than the High Child Vigilance-High Personal Exposure and Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure classes.

Conclusions:

Vigilance for children’s future police encounters may be associated with increased cardiovascular risk in African American women. Agency, system, and policy-level solutions may be needed to reduce anti-Black police violence and improve the cardiovascular health of this high-risk population.

Introduction

Cardiovascular disease (CVD) is the leading cause of death in women in the United States. However, African American women are burdened with higher CVD morbidity and mortality rates relative to women of other races/ethnicities (1, 2). African American women aged 35–44 have rates of CVD that are higher than those of African American men, White men, and White women (3). Socioeconomic and traditional clinical risk factors do not completely account for this excess risk (3, 4), and CVD prevention and awareness campaigns have been less effective among African American women as compared to White women (5). Thus, the American College of Cardiology Cardiovascular Disease in Women Committee and the American College of Cardiology Health Equity Task Force have issued a call to action to address the crisis of poor cardiovascular outcomes among African American women (6). Indeed, achieving cardiovascular health equity is one of the quintuple aims of the American College of Cardiology (7).

Police brutality, which is a broad spectrum of routine manifestations of structural racism that is intended to control the movements, behaviors, and prospects of African Americans (e.g., verbal abuse, lethal and non-lethal abuse, hyper-policing in African American neighborhoods (Bowleg et al., 2022), has been identified as a public health issue and barrier to achieving health equity (8–10). African Americans are significantly more likely to be stopped, searched, handcuffed, and arrested relative to any other racial/ethnic group (11, 12), are more likely to be the targets of non-lethal use of force (e.g., taser, pepper spray; 13–15), and are 2.8 to 3.5 times more likely than Whites to be killed by police officers while unarmed, even after controlling for crime rates (16–18). Notably, the extant literature on policing primarily focuses on the experiences and outcomes of African American men and adolescent boys, rendering African American women’s police-related experiences and health outcomes invisible.

We posit that the interlocking systems of sexism and racism engender distinct forms of police-related stress for African American women. Specifically, African American women (e.g., Breonna Taylor, Sandra Bland) are personal victims of police brutality (#SayHerName). African American women are approximately 1.4 times more likely than White women to be killed by police officers (19) and account for almost 20 percent of women fatally shot by the police (20). Moreover, African American women are exposed to fatal and nonfatal vicarious police encounters (e.g., community members, family members), which have been linked to shorter telomere lengths among African American men (21), poorer self-rated physical health among urban-born adolescents (22), and vigilance around future police encounters among youth (23). Critically, African American women’s vigilance regarding police-related exposures likely extends beyond their own experiences with law enforcement to include the experiences of their loved ones, especially their children (24–25).

As of this writing, there are no studies examining police encounters and cardiovascular health in African American women. In the current investigation, we adopt an intersectional, person-centered—rather than variable-centered—approach to identify classes or profiles of African American women based on the patterning of co-occurring police encounters. Previous studies examining linkages between police exposure and health typically focus on single types of police-related stress separately, instead of identifying the patterning or clustering of police-related exposures that exist along various dimensions. We used a latent class analysis (LCA), a type of mixture model, to identify profiles or classes of African American women based on the constellation of various dimensions of police-related stress (direct, vicarious, police-related vigilance for self, and police-related vigilance for children). We then examine whether classes of police-related stress have distinct associations with carotid intima media thickness (IMT) among African American women. Carotid IMT is an accessible, reliably measured subclinical marker of atherosclerosis that has also been shown to be a robust predictor of myocardial infarction and stroke in both clinic and community samples (26–28), in African Americans (29, 30), and during midlife (31). Psychosocial stress is a well-established contributor to the development and progression of atherosclerosis, operating through mechanisms such as increased inflammatory responses, autonomic imbalance, dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis, and endothelial dysfunction (32–34). Chronic exposure to psychosocial stressors, including racialized stress, gendered stress, and racism-related vigilance, has been shown to accelerate increased carotid IMT—which is typically higher among African Americans relative to Whites—and coronary artery calcification, even after controlling for traditional CVD risk factors (35–41).We hypothesized that classes characterized by greater personal and vicarious police-related stress and police-related vigilance (for self and child(ren)) would be associated with greater carotid IMT, even after adjusting for sociodemographic factors, risk factors, and depressive symptoms.

Methods

Participants

Participants were from the Mechanisms Underlying Stress and Emotions (MUSE) in African American Women’s Health Study (42–43). Women were identified via consumer residential lists and voter registration records from a range of census tracts in the greater Atlanta, GA metropolitan area. An introductory brochure was sent via mail, followed by a telephone call for prescreening. Eligibility criteria included: self-identification as a Black/African American woman; aged 30–45 (at screening); pre-menopausal; not pregnant or lactating; and having a uterus and ≥1 ovary. Ineligibility criteria included: history of clinical CVD (e.g., myocardial infarction, angina, intermittent claudication, cerebrovascular disease, or coronary revascularization); chronic illness known to influence atherosclerotic disease (e.g., HIV, kidney disease); current treatment for psychiatric disorders; and current illicit drug use or alcohol abuse. The final cohort consisted of N = 422 women (42–43).

Procedure

Between 2016 and 2019, participants completed an in-person visit, where study staff collected height, weight, resting blood pressure, and additional clinical data. Carotid ultrasound to assess IMT was then performed. Following this, demographic and psychosocial characteristics were assessed via face-to-face interview. All procedures were approved by the Emory University Institutional Review Board, and all participants provided written, informed consent.

Intima Media Thickness Outcomes

Certified vascular sonographers at Emory University obtained duplex ultrasound bilateral carotid images using a Mindray M7 system equipped with a linear transducer (L12–4s). Bilateral clips of the following 10 mm segments were obtained: the common carotid artery (CCA) just proximal to the bifurcation at 3 angles (anterior, lateral and posterior), the carotid bulb from the plane where the thickest far wall was seen (anterior, lateral or posterior), and the internal carotid artery (ICA) from the plane where the thickest far wall was seen (anterior, lateral or posterior). Each clip included four QRS complexes. The clips were saved and streamed to the University of Pittsburgh Ultrasound Research Lab for later reading. Vascular Research Tools 5 (Medical Imaging Applications, LLC, Coralville, IA) was used to measure IMT of the near and far walls of the CCA, the far wall of the bulb and the far wall of the ICA. All IMT measurements were made by electronically tracing the lumen-intima interface and the media-adventitia interface across all segments from images obtained in end-diastole (upstroke of R wave). The mean IMT of the near and far walls of the CCA segment was used in analyses.

Assessment of Police-Related Stress

The MUSE cohort included a wide range of psychosocial stress questionnaires, designed to assess racism-related exposures specifically. Although none of the scales were explicitly designed to assess police-related stress, several of these scales included items that referenced police contact.

Direct Police-Related Stress

The women’s direct experiences of police-related stress were assessed with a single item from the Race-Related Events Scale (44), a 22-item scale that assesses lifetime exposure to both direct and vicarious experiences of racism. Specifically, the women responded “yes” or “no” when asked whether they had ever been harassed by police or a security guard.

Vicarious Police-Related Stress

The women’s vicarious experiences were via a single item from the Psychiatric Epidemiology Research Interview (PERI) Life Events Scale (45) that had been modified for midlife women (46): “In the past 12 months, have you experienced the following…and if you have, indicate how upsetting it was: A family member had legal problems or a problem with police?” Participants responded on a 5-point Likert scale ranging from 0 (no) to 4 (Yes, very upsetting and still upsetting).

Police-Related Vigilance for Self

The women’s personal police-related vigilance was assessed using two items (i.e., scenarios) from the “expectations” subscale of the Race-Based-Rejection Sensitivity Questionnaire (47), which inquires about expectations of racism across a variety of situations. The first situation reads as follows: “Imagine you’re driving down the street, and there is a police barricade just ahead. The police officers are randomly pulling people over to check drivers’ licenses and registrations. I would expect that the officers might stop me because of my race/ethnicity.” The second situation reads as follows: “Imagine you have just finished shopping for new clothes, and you are leaving the store carrying several bags. It’s closing time, and several people are filing out of the store at once. Suddenly, the alarm begins to sound, and a security guard comes over to investigate. I would expect that the guard might stop me because of my race/ethnicity.” Responses to both items were scored on a 6-point Likert scale ranging from 0 (Very Unlikely) to 5 (Very Likely), with higher scores being indicative of personal police-related vigilance.

Police-Related Vigilance for Children

The women’s police-related vigilance for their children was assessed using two items from the 8-item Concern for Children Scale, a subscale of the Telephone-Administered Perceived Racism Scale that was designed for epidemiological use among African American women (48–49). Specifically, the women were asked: “How concerned are you about your children being harmed or harassed by the police because they are Black?” and “How concerned are you about your children getting stopped in a predominantly White area because they are Black?”. Participants responded on a scale of 0 (Not at all concerned) to 3 (Very concerned). However, for 113 of the women in our cohort, these items were not applicable, as they reported having no children. Thus, responses were re-coded such that responses were: 0 (No children, not applicable), 1 (Not at all concerned), 2 (Mildly Concerned), 3 (Moderately Concerned), and 4 (Very Concerned).

Covariates

Covariates were chosen based on prior research (50, 51–52). Age and partner status (married/living with partner vs. not) were self-reported. Education was self-reported in years and as highest degree status and then categorized into: high school or less, some college or occupational training, and college or higher. Family income was self-reported into predefined ranges and collapsed into: <$35,000, $35,000–$49,999, $50,000–$74,999, ≥$75,000/year; a fifth category was created to capture volunteered responses of “Refused/ Don’t know.” Family size was reported as the number of individuals in the household. Body Mass Index (BMI) was calculated as weight/height (kg/m2) and modeled continuously. Current smoking and antihypertensive use were self-reported as yes/no. Clinic systolic blood pressure (SBP) was assessed and modeled continuously. Depressive symptoms were assessed with the 21-item Beck Depression Inventory (BDI) (53) and modeled continuously. Neighborhood safety was assessed via the 3-item Neighborhood Safety subscale (α = .83 for present study) of the modified Neighborhood Health Scale (54), was modeled continuously, and was included as a covariate to account for neighborhood crime, which has been linked to police activity as well as CVD (55–56). Finally, the women reported on whether they had medical insurance coverage (yes/no).

Statistical Analyses

LCA is used to classify subpopulations of participants within the larger group of participants. We ran a series of LCA models within Mplus 8.10 (57) to identify classes of women based on personal and vicarious police exposure as well as police-related vigilance for themselves and for their children. We used full information maximum likelihood (FIML) estimation to account for the missing (less than 2%) CCA IMT data (58). The LCA was conducted using a total of 5 continuous and one categorical police encounter/exposure indicators from the various measures (described above). Once the best fitting model and number of classes were identified and saved, the class number was exported to IBM SPSS Version 29. We then performed a series of analysis of covariance (ANCOVA) models to examine class-based differences in CCA IMT. Eight women (1.90%) did not have CCA IMT data. Missingness at this stage was handled via listwise deletion, and ANCOVA analyses included 414 women.

First, to identify the best-fitting LCA model, we ran an iterative series of unconditional LCA models starting with a one-class solution and increasing in the number of classes. We ran these models as unconditional LCA models, as evidence indicates that class enumeration with unconditional models provides the most accurate class specification across diverse causal patterns within population models (59). In line with best practices for unconditional models, we recorded and compared fit statistics among consecutive models, including the Bayesian Information Criterion (BIC) and the adjusted BIC (ABIC), as well as the Lo-Mendell-Rubin test (LMR) and bootstrapped likelihood ratio test (BLRT), which assess whether the addition of another latent class significantly improved model fit (60). Significance for the LMR and BLRT p-values indicates that the additional class significantly improves the model. We report the entropy, which ranges from 0 to 1, wherein larger values indicate better classification (61). In addition to these fit indices, we considered the conceptual and theoretical relevance of each potential class solution (62, 63). Thus, we selected the model that was both supported by fit statistics and consistent with a theoretically relevant outcome.

After identifying the optimal unconditional LCA model, a series of ANCOVA models were conducted to examine class-based differences in CCA IMT. The initial, minimally adjusted model examined the class-based differences in CCA IMT, after controlling for age. The second model additionally adjusted for marital/live-in partner status, education, family income, family size, and self-rated neighborhood safety (we wanted to assess the influence of multidimensional police-related stress above and beyond the influence of neighborhood violence and crime). In the subsequent model, we also adjusted for BMI, clinic SBP, use of antihypertension medicine, current smoking, and insurance status. Finally, the fully adjusted model additionally included depressive symptoms.

Results

Participant Characteristics

Participant sociodemographic characteristics for the overall sample and for the latent classes are presented in Table 1. Women ranged in age from 30 to 46, with the average age being 37.92. Thirty-six percent of the women were partnered or married, and the average household size was 3.6 persons, although the range was 1 to 12. Participants were generally well educated, with nearly half of the sample (47.4%) reporting having a college degree or higher. The cohort was socioeconomically diverse, with annual household incomes ranging from < $35,000 to >$75,000. On average, the women neither agreed nor disagreed that their neighborhood was safe. Scores on the BDI (M = 6.04) were skewed towards the low end of the distribution. BMIs were generally in the obese range. Moreover, 10% of participants were current smokers and 83.6% reported having medical insurance coverage. Average clinic SBP was 119.01 mmHg and ranged from 80 mmHg to 166.4 mmHg. Approximately 16.8% of participants reported antihypertensive use. Finally, the mean CCA IMT value was .61 mm.

Table 1.

Demographic, Psychosocial, and Risk Factor Characteristics (N = 422)

	Overall Sample N=422	Class 1 n=235	Class 2 n=115
N (%)
Marital/partnered status
Yes	152 (36.0%)	107 (45.7%)	98 (86.0%)
No	270 (64.0%)	127 (54.3%)	16 (14.0%)
Education
HS or less	133 (31.5%)	82 (34.9%)	26 (22.8%)
Some college	88 (20.9%)	50 (21.3%)	19 (16.7%)
College or higher	200 (47.4%)	103 (43.8%)	69 (60.5%)
Income
<35k	103 (24.4%)	61 (26.0%)	23 (20.4%)
35k-50k	87 (20.6%)	44 (18.7%)	24 (21.2%)
50k-75k	95 (22.5%)	49 (20.9%)	31 (27.4%)
>75k	135 (32.1%)	77 (32.8%)	31 (27.4%)
Antihypertensive medication use
Yes	71 (16.8%)	44 (18.7%)	15 (13.0%)
No	351 (83.2%)	191 (81.3%)	100 (87.0%)
Insurance Status
Yes	353 (83.6%)	200 (85.1%)	90 (78.3%)
No	69 (16.4%)	35 (14.9%)	25 (21.7%)
Current Smoking
Yes	42 (10.0%)	26 (11.1%)	10 (8.7%)
No	380 (90.0%)	208 (88.9%)	105 (91.3%)
Harassed by police
Yes	96 (22.7%)	62 (26.4%)	27 (23.7%)
No	326 (77.3%)	173 (73.6%)	87 (76.3%)
M SD
Age in years	37.92 (4.25)	38.40 (4.25)	37.15 (4.45)
Family size	3.57 (1.80)	4.05 (1.62)	2.17 (1.44)
Self-rated neighborhood safety	2.47 (1.03)	2.49 (1.04)	2.44 (0.92)
Body Mass Index (kg/m2)	32.76 (8.35)	33.02 (8.26)	31.83 (8.16)
Clinic systolic blood pressure	119.01 (14.58)	119.06 (14.77)	119.87 (14.61)
Clinic diastolic blood pressure	80.87 (11.95)	80.80 ( 12.10)	81.23 ( 11.91)
Depressive symptoms (BDI)	6.04 (6.92)	6.42 (6.86)	6.67 (7.44)
Concern for Children (harmed/harassed by police)	2.19 (1.75)	3.81 (0.40)	0.00 (0.00)
Concern for Children (stopped in White neighborhood)	2.21 (1.76)	3.76 0(.49)	0.00 (0.00)
Family - Problems with police	0.51 (1.11)	0.68 1.25)	0.39 (0.94)
Expect - Police	2.41 (1.57)	2.66 (1.52)	2.32 (1.65)
Expect - Security Guard	2.38 (1.47)	2.67 (1.45)	2.31 (1.46)
Mean of CCA IMT	0.61mm (0.08)	0.62mm (.08)	0.59mm (0.07)

Nearly a quarter (22.7%) of the African American women reported having been harassed by the police or a security guard. Most women did not report having a family member who had legal problems or a problem with the police. Generally, the African American women believed it was “somewhat likely” that they would be stopped by the police officer and security guard in the two hypothetical scenarios. Finally, the African American women reported being moderately concerned or vigilant about their children being stopped, harassed, or harmed by the police.

Police-Related Stress Classes

Model fit indicators for models of classes 1–5 can be found in supplemental material. The ABIC and BIC decreased up to the 5-class model. Although the LMR and BLRT p-values remained significant for the 4-class and 5-class solutions, the standard error was not generated for one of the LCA indicators, suggesting that the models became unstable due to the extraction of too many classes. Moreover, one of the classes in the 5-class solution comprised only 3.8% of the sample, contradicting recommendations that LCA classes should comprise at least 5% of the sample (61). Thus, we identified the 3-class model as our optimal class solution. The entropy for this model was .98, indicating clear delineation of the classes (61).

The largest class, the High Child Vigilance-High Personal Exposure class (55.69% of the sample, n=235), included African American women who, on average, reported being moderately to very concerned about their children being stopped in a predominantly White neighborhood because they are Black and/or being harassed or harmed by the police. Among these women in the largest class, 26.4% reported being personally harassed by police or security. Figure 1a depicts the percentage of women who were harassed by the police or security guards across all three classes. Figure 1b depicts the z-scores for the latent class indicators. Z-scores within .5 standard deviation (SD) were considered average whereas z-scores that exceeded .5 SD were considered high or low. Based on z-scores, the women in this class scored almost one SD above the sample mean for the two Concern for Children items.

Figure 1a.

Percent of Women Harassed by the Police or Security Guards

Figure 1b.

Summary of Z-scores for Continuous Indicators

The next largest class, the No Child Vigilance-High Personal Exposure class, comprised 27.49% (n=115) of the sample. This class included the 113 African American women (97.39% of the latent class) who reported having no children as well as two African American women who reported not being concerned about their children being stopped in a predominantly White neighborhood because they are Black and/or being harassed or harmed by the police. In terms of raw scores, relative to the High Child Vigilance-High Personal Exposure class (M =3.81), the No Child Vigilance-High Personal Exposure class (M <.001) had significantly lower scores on the first, F(2, 419) = 3791.51, p <.001, η²_p = .948, and second Concern for Children item (M =3.76 vs. M < .001), as would be expected given their motherhood status, F(2, 419) = 2361.61, p <.001, η²_p = .919. The No Child Vigilance-High Personal Exposure class was also significantly less likely than the High Child Vigilance-High Personal Exposure class to expect (M =2.67 vs. M =2.31) that they would be (personally) stopped by the security guard in the hypothetical scenario, F(2, 417) = 18.27, p < .001, η²_p = .081. Among these women, 23.5% reported being harassed by police or security. Based on z-scores, the women in this class scored almost 1.5 SDs below the sample mean for the two Concern for Children items.

The smallest class, the Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure class (17.06%; n=72), comprises African American women who, on average, reported being moderately concerned about their children being stopped in a predominantly White neighborhood because they are Black and/or being harassed or stopped by the police. They reported significantly less concern for their children than the High Child Vigilance-High Personal Exposure class on the first (M =2.03) and second (M =2.19) Concern for Children item but more concern than the No Child Vigilance-High Personal Exposure class. Among these women, 9.40% reported being harassed by police or security. Based on the z-scores, the women have average scores (less than .5 SD below the sample mean) for having a family member who had legal problems or a problem with police and being concerned about their child(ren) being stopped in a predominantly White area because they are Black. Moreover, the women scored .5 SD below the mean for their expectations around being pulled over by the police in the hypothetical scenario.

There were class-based differences on age, education, family size, and depressive symptoms. Specifically, High Child Vigilance-High Personal Exposure class (M_age = 38.37) was older than the No Child Vigilance-High Personal Exposure class (M_age = 37.11), F(2, 414) = 3.65, p = .027, η²_p = .017. The No Child Vigilance-High Personal Exposure class was significantly more likely to have a college degree or higher compared to the other two classes, X² (1, N = 421) = 11.71, p = .020. Additionally, the No Child Vigilance-High Personal Exposure class (M =2.17) reported a smaller family size than the Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure (M_FamilySize =4.30) and High Child Vigilance-High Personal Exposure (M_FamilySize = 4.05) class, F(2, 414) = 62.28, p < .001, η²_p = .231. Finally, the Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure class (M = 3.80) reported lower depressive symptoms than the High Child Vigilance-High Personal Exposure (M = 6.42) and No Child Vigilance-High Personal Exposure (M = 6.61) class, F(2, 414) = 4.525, p = .011, η²_p = .021.

Police-Related Stress Classes and CCA IMT

To examine the associations between class membership and CCA IMT, we ran a series of ANCOVA models. In the age-adjusted ANCOVA model, the No Child Vigilance-High Personal Exposure (M = .60 mm) class had lower CCA IMT than the High Child Vigilance-High Personal Exposure (M = .62 mm) and Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure (M = .62 mm) class, F(2, 410) = 3.791, p = .023, η²_p = .018. After adjustment for marital/live-in partner status, education, family income, family size, and self-rated neighborhood safety, the differences in CCA IMT were no longer significant, F(2, 402) = 1.995, p = .14, η²_p = .010. In the subsequent model, which adjusted for BMI, clinic systolic blood pressure, use of antihypertension medicine, current smoking, and insurance status, the statistical differences in CCA IMT re-emerged, F(2, 394) = 4.195, p = .016, η²_p = .021. Specifically, the No Child Vigilance-High Personal Exposure (M = .59 mm) class had lower CCA IMT than the High Child Vigilance-High Personal Exposure (M = .62 mm) and Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure(M = .62 mm) class. In the fully adjusted model that additionally included depressive symptoms, the differences in CCA IMT persisted, F(2, 393) = 4.038, p = .018, η²_p = .020, such that the No Child Vigilance-High Personal Exposure class had lower CCA IMT than the High Child Vigilance-High Personal Exposure and Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure class (see Figure 2).

Figure 2.

Adjusted Class Differences in Common Carotid Artery IMT

Supplemental Analyses

Correlation Between CCA IMT and Diastolic Blood Pressure

Given mounting evidence of a feedback loop between carotid IMT and diastolic blood pressure (26, 64), we examined the correlation between CCA IMT and clinic diastolic blood pressure (DBP). We found a small, significant correlation between CCA IMT and clinic DBP among our cohort of African American women (r = .20, p < .001).

Pregnancy History and CCA IMT in the No Child Vigilance-High Personal Exposure Class

Given the known changes in carotid IMT as a function of pregnancy (65–66), we conducted an ANOVA model to determine whether there were differences in CCA IMT among women in the No Child Vigilance-High Personal Exposure class with (N=35) and without (N=80) a history of pregnancy. The 35 women who reported a history of pregnancy (M = .61 mm) did not significantly differ from the 80 women who reported no history of pregnancy (M = .59 mm) in terms of CCA IMT F(1, 109) = 2.33, p = .129, η²_p = .021. Moreover, the 80 women (M = 32.52 ) versus 35 women (M = 30.27) did not differ in terms of BMI, a risk factor for CVD, F(1, 112) = 1.86, p = .176, η²_p = .016.

Police-Related Stress Classes and CCA IMT Among Mothers

We also performed LCA analyses that only included mothers (N=309). As was the case with the full sample, the ABIC and BIC decreased up to the 5-class model. However, the LMR and BLRT p-values became non-significant beginning with the 4-class solution, indicating that the 4-class solution did not improve on model fit over the 3-class solution. Thus, we identified the 3-class model as our optimal class solution. The entropy for the 3-class model is .98.

The largest class, the Low Exposure-Low Vigilance class (58.58% of the sample, n=181), included African American women who scored: .5 SD below the sample mean for having a family member with legal problems or problems with the police, .5 SD below the sample mean for their expectations around being pulled over by the police and being stopped by the security guard in the hypothetical scenarios, and 1.5 SDs below the mean for the two concern for children items. Among the women in this class, 8.3% reported being personally harassed by police or security.

The next largest class, the High Exposure class (22.98% of the sample, n=71), scored: two SDs above the sample mean for having a family member with legal problems. Among the women in this class, 43.9% reported being personally harassed by police or security.

The smallest class, the Moderate Exposure-High Child Vigilance class (18.45%; n=57), scored: .5 SD above the mean for the two concern for children items and 21.0% reported being personally harassed by the police or security.

To examine the associations between class membership and CCA IMT, we ran a series of ANCOVA models. In the age-adjusted ANCOVA model, there were no differences between the classes, F(2, 301) = 1.544, p = .22, η²_p = .010. After adjustment for marital/live-in partner status, education, family income, family size, and self-rated neighborhood safety, there were still no class-based differences in CCA IMT, F(2, 294) = 1.631, p = .20, η²_p = .011. In the subsequent model, which adjusted for BMI, clinic SBP, use of antihypertension medicine, current smoking, and insurance status, there were still no class-based differences in CCA IMT F(2, 287) = 2.131, p = .12, η²_p = .015. In the fully adjusted model that additionally included depressive symptoms, there were also no differences, F(2, 286) = 2.231, p = .11, η²_p = .015.

Discussion

To the best of our knowledge, this study is the first to examine the association between multidimensional police-related stress and a surrogate measure of subclinical carotid atherosclerosis (IMT) among African American women. In our cohort of early middle-aged African American women from a range of socioeconomic status (SES) backgrounds, we observed that nearly a quarter (~23%) of the African American women reported having been harassed by the police or a security guard. Approximately 20% of the African American women also reported having a family member who had legal problems or a problem with the police, with 17% of these women reporting being at least somewhat upset about this issue currently or in the past. Generally, the African American women were somewhat vigilant about their own future encounters with the police based on their scores for the two hypothetical scenarios involving the police and security guards. Finally, the African American women reported being moderately concerned or vigilant about their children being stopped, harassed, or harmed by the police. From these descriptive statistics alone, it is evident that African American women are burdened with various forms of police-related stress.

Using an LCA approach, we discerned the patterning of the African American women’s police-related stress experiences. We identified three classes of police-related stress: High Child Vigilance-High Personal Exposure, No Child Vigilance-High Personal Exposure, and Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure class. The High Child Vigilance-High Personal Exposure class comprised women with relatively high levels of vigilance centered around their children’s future encounters with law enforcement. The No Child Vigilance-High Personal Exposure class comprised all the women who reported having no children as well as the two mothers who reported not being vigilant about their children’s future encounters with law enforcement. The third class, the Moderate Child Vigilance-Low Self Vigilance-Low Personal Exposure class, comprised women who scored around the sample mean for the concern for children items, who scored below the mean for their expectations around being pulled over by the police in the hypothetical scenario, and who were less likely to have been harassed by a police officer or security guard.

Although the African American women did indeed report direct experiences with the police, vicarious exposure to the police, and self-focused vigilance, the CCA IMT differences between the latent classes seemed to be primarily driven by vigilance centered around their children’s future encounters with the police. After adjusting for all covariates, our findings reveal that the No Child Vigilance-High Personal Exposure class evidenced significantly lower CCA IMT than the other two classes. Our finding provides partial support for our hypothesis and comports with recent qualitative research capturing African American mothers’ concerns about the police-related racial violence that they and their children might experience during their lifetime (67). Evidence from another recent qualitative examination indicates that African American pregnant women reported feelings of anticipation, fear, and stress regarding police brutality towards their children before their children were born (68). This may be warranted, as police officers are 21 times more likely to kill African American versus White teenagers (69). In a third qualitative study (70), a participant noted: “I don’t even know what our children have to do to convince the world that they are children … or even that they are human.”

As this investigation was designed to explore the police-related stress classes for all African American women in our sample, and consistent with open science practices, we did not exclude women who did not report being a mother at the outset. Following our main analyses, we conducted supplemental analyses that focused exclusively on the women who identified as mothers. The results from these supplemental analyses revealed no differences between the three police-related stress classes, indicating that among mothers, there were no differences. As pregnancy induces progressive increases in carotid IMT from the first to the third trimester of pregnancy that is reflective of adaptive remodeling of the arterial wall (65–66), we also conducted additional supplemental analyses that examined whether the women within the No Child Vigilance-High Personal Exposure class who did (N=35) and did not (N=80) have a history of pregnancy significantly differed in terms of CCA IMT and BMI. There were no significant differences, indicating that pregnancy history likely did not drive the differences in CCA IMT between the police-related stress classes.

Given the aforementioned findings and statistics, it is not surprising that youth police stops have been linked to African American mothers’ increased risk for depression- and anxiety-related sleep difficulties and poorer overall health (24–25). These findings, in conjunction with the current findings, provide support for Lewis and colleagues’ (41) conceptual model for expectations of racism and CVD risk. Vigilance, including vigilance specific to future police encounters, is a key component of expectations of racism, which has been linked to IMT (41, 71). Expectations of racism may be linked to atherosclerosis and CVD risk through vigilance, physiological dysregulation (e.g., autonomic, immunologic), and health behaviors (e.g., poor sleep). Our findings indicate that multiple physiological pathways, including inflammatory processes and endothelial damage, potentially explain how police-related stress may contribute to CVD risk. Facets of police-related stress may also activate a recursive pathway between carotid IMT and diastolic changes. We found a small, significant correlation between carotid IMT and clinic DBP among our cohort of African American women, potentially supporting existing evidence for the feedback loop between DBP and carotid IMT. Specifically, elevated DBP contributes to stress on the arterial walls, which promotes endothelial dysfunction and arterial wall thickening (31, 64). Moreover, thickened arterial walls/higher carotid IMT lead to arterial stiffening and increased peripheral vascular resistance (26, 72). Such changes can lead to elevated DBP, as the heart must pump against stiffer vessels. Consistent with recent work on network stressors among African American women (42), expectations of racism extend to members of their close networks, and in the context of police brutality, these expectations may be greatest for their children. Additional research in this area is warranted.

There are several limitations. First, our findings are cross-sectional, and we are therefore unable to fully determine the temporal associations between police-related stress and IMT. However, the “silent” nature of carotid atherosclerosis may make reverse causality (i.e. those with more disease reporting higher vigilance for children) unlikely. Future studies are needed to identify whether police-related stress is associated with increased carotid atherosclerosis over time as well as the mechanisms that may underlie this longitudinal association. Additionally, we did not utilize a comprehensive measure for policing; however, such measures were virtually nonexistent when these data were collected. Moreover, our cohort was fairly well-educated, potentially limiting the generalizability of the findings to less educated African American women. Yet, it is conceivable that the linkages between police-related stress class membership and CCA IMT may also be documented among African American women who are less educated and who reside in less safe neighborhoods. Previous literature documents the bidirectional relation between policing and criminalized behaviors (55) and that higher education may not be protective against poor cardiovascular health for African American women (73–74) due to the diminishing returns of higher SES in this population. Furthermore, we do not have data on the gender of the women’s children, thereby potentially missing important nuances. Finally, our cohort was recruited from a single city in the Southeastern United States, and while rates of CVD among African Americans are highest in this region (75), our findings may not generalize to other geographic regions.

Despite these limitations, our study also has several notable strengths. The distribution of CCA IMT measurements is concordant with published norms for African American women (30), thereby maximizing the generalizability of the findings. Moreover, we focused on a vulnerable group during a critical life stage wherein they are at elevated risk for stroke and myocardial infarction. Finally, we were also able to assess the patterning of African American women’s police-related stress along multiple dimensions, allowing us to examine a potentially novel contributor to CVD among African American women, a subpopulation facing excessive cardiovascular health inequities.

In conclusion, the current findings indicate that the police-related stressors with which African American women must contend may be distinct from those of African American men. Our findings contribute to the nascent literature on African American women’s police-related stress and health, supporting theoretical assertions that police violence results in historical, collective, intergenerational, and vicarious trauma (76). Despite the parallels between PTSD and expectations of racism, the traumatic nature of police violence, and our initial evidence connecting vigilance for children with greater IMT, we underscore that care should be taken to avoid pathologizing vigilance among African American women, and African Americans more generally. Anti-Black police violence is a public health crisis (8), with the threat of racial violence against African American children—often heightened by the proliferation of news and social media capturing incidents of police brutality against African Americans—constituting a maternal health issue for African American women (67–68, 70). In light of the current findings, interventions that focus on enhancing racial coping among African American mothers amid racial stress and trauma (e.g., EMBRace: 77-78) and disrupting the link between police-related stress and CCA IMT may be imperative. Given the Cardiology Health Equity Task Force’s call to action to address the crisis of poor cardiovascular outcomes among Black women and the American College of Cardiology’s goal of achieving cardiovascular health equity (6–7), we underscore that agency, system, and policy-level solutions (i.e., upstream, structural factors) for reducing racial inequities in policing, particularly anti-Black police violence, are paramount.

Abbreviated Running Title: Policing Stress and Carotid Intima-Media Thickness

ANCOVA

Analysis of Covariance

ABIC

Adjusted Bayesian Information Criterion

BDI

Beck Depression Inventory

BIC

Bayesian Information Criterion

BMI

Body Mass Index

BLRT

Bootstrapped Likelihood Ratio Test

CCA

Common Carotid Artery

CVD

Cardiovascular Disease

FIML

Full Information Maximum Likelihood

HPA

Hypothalamic-Pituitary-Adrenal

ICA

Internal Carotid Artery

IMT

Intima Media Thickness

LCA

Latent Class Analysis

LMR

Lo-Mendell-Rubin Test

MUSE

Mechanisms Underlying Stress and Emotions

PERI

Psychiatric Epidemiology Research Interview

SBP

Systolic Blood Pressure

DBP

Diastolic Blood Pressure

SD

Standard Deviation

SES

Socioeconomic Status