Adverse policing as a social exposome pathway: Implications for depression, cognition, and dementia risk

Abstract

INTRODUCTION

The social exposome is increasingly recognized as a key determinant for cognitive health. Adverse policing represents an understudied aspect of the social exposome, reflecting persistent encounters with punitive surveillance and law enforcement that accumulate as chronic structural stressors across the life span.

METHODS

Using data from the Health and Retirement Study (2006–2020), we performed exploratory and confirmatory factor analyses to examine the dimensional overlap between depressive symptoms and cognitive performance among older Black adults aged 65 + (n = 265; mean age = 69; 72% women), stratified by adverse policing exposures.

RESULTS

Depressive and cognitive items generally formed separate dimensions, with effort‐related depressive items and cognitive performance items frequently co‐loading, suggesting an intermediate latent factor linked to psychosocial burden.

DISCUSSION

Our findings underscore the importance of integrating social exposome exposures, such as adverse policing, into modelling dementia risk. This will help researchers better characterize how social adversity impacts cognitive health.

Highlights

• Adverse policing practices represent an important and understudied dimension of the social exposome relevant to cognitive aging and dementia risk.

• Among Black older adults, depressive symptoms and cognitive performance generally remain distinct constructs, but overlap emerges in the context of adverse policing exposure.

• Effort‐related depressive symptoms and executive‐function cognitive tasks show consistent cross‐loading under chronic exposure to adverse policing.

• Structural adversity may shape the measurement properties of commonly used depression and cognitive assessments.

• Incorporating criminal‐legal exposures into exposome frameworks may improve equity in dementia research and risk assessment.

1. BACKGROUND

1.1. Social exposome and AD/ADRD risk

Alzheimer's disease and related dementias (AD/ADRD) are complex conditions shaped by environmental exposures that accumulate throughout the life‐course. The exposome provides a framework for characterizing these cumulative exposures, including physical, chemical, biological, social, and structural conditions that influence health. Within this framework, the social exposome encompasses a subset of exposures rooted in political, economic, historical, and structural conditions that shape chronic stress pathways and cognitive health ¹ , ² , ³ . The social exposome captures socially‐patterned exposures that become biologically embedded and contribute to inequities in mental and cognitive health ⁴ , ⁵ , ⁶ , ⁷ , ⁸ . As illustrated in Figure 1, it provides a framework for understanding how exposures like adverse policing relate to depressive symptoms, cognition, and dementia risk ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ . Here, dementia risk refers to pathways and indicators associated with later‐life cognitive impairment or AD/ADRD, rather than a measured dementia diagnosis.

FIGURE 1.

Conceptual model of adverse policing as a social exposome pathway for depression, cognition, and dementia risk. Conceptual model illustrating how socially patterned exposures across the life course may become biologically embedded and contribute to shared neurobehavioral manifestations relevant to depressive symptoms, cognitive dysfunction, and brain health outcomes. The model highlights adverse policing as one structural exposure within the broader social exposome and illustrates how overlapping symptoms such as effort, motivation, fatigue, and executive control may complicate the measurement and interpretation of depressive symptoms and cognitive performance.

1.2. Bidirectional relationships between depression and cognition

Depression is characterized by persistent low mood, anhedonia, and cognitive–affective symptoms affecting motivation and concentration. ³⁰ Depression may reduce cognitive functioning through motivational deficits, impaired concentration, and neurobiological changes. ³¹ , ³² , ³³ Conversely, decreased cognitive control may impair emotional regulation, intensifying depressive symptoms. ³⁴ , ³⁵ , ³⁶ , ³⁷ Depression is associated with increased risk of incident AD/ADRDs, while AD/ADRDs are related to developing depressive symptoms potentially due to neurodegenerative changes and reduced functional autonomy. ²⁰ , ³³ , ³⁸ , ³⁹ , ⁴⁰ This bidirectional relationship complicates efforts to distinguish unique and shared contributions to dementia risk. ⁴¹ , ⁴²

1.3. Pathways linking social exposome framework to mental health and cognition

Structural racism is the ways in which social, economic, and political systems systematically disadvantage individuals based on phenotypic characteristics or perceived ancestry. ⁴³ , ⁴⁴ From a social exposome perspective, structural racism structures the distribution, timing, and intensity of social exposures across the life‐course. These exposures may become biologically embedded through chronic stress, allostatic load, hypothalamic–pituitary–adrenal (HPA)‐axis dysregulation, inflammation, and reduced access to protective resources, thereby contributing to depressive symptoms, cognitive performance, and dementia‐related brain health. ⁴⁵ , ⁴⁶ These exposures are linked to reduced cognitive reserve (i.e., the ability to preserve cognition when there is underlying brain pathology), alterations in neurobiological systems supporting memory and executive control collectively heightening vulnerability to AD/ADRD. ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵²

Adverse policing is one criminal‐legal manifestation of structural racism within the social exposome. Adverse policing practices include persistent surveillance, criminalization, and law enforcement presence in neighborhoods with predominately people racialized as Black and Latinx/a/o. These exposures, whether direct (e.g., stops, arrests, or use of force) or indirect (e.g., vicarious exposure to police violence via media or community experiences), accumulate across the life‐course and impact psychological functioning and neurobiological processes relevant to cognition and dementia risk. ⁵³ , ⁵⁴ Exposure to negative police encounters is associated with nearly double the risk of experiencing depression, anxiety, and suicidal ideation. ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸

1.4. Measurement overlap and the present study

Beyond emotional distress, exposure to structural racism is linked to accelerated biological aging, increased allostatic load, and cognitive decline. ⁵⁹ , ⁶⁰ , ⁶¹ , ⁶² Symptoms like apathy, low motivation, and fatigue may reflect a shared neurobiological response to chronic exposure to structural racism that blurs the distinction between affective and cognitive symptoms, complicating the clinical assessment of dementia risk. ²⁹ , ⁶³ , ⁶⁴ , ⁶⁵ Such measurement overlap may contribute to misclassification, delayed detection, inappropriate symptom attribution, and inequitable referrals; particularly notable given racialized disparities in dementia diagnosis timing. ⁶⁶ , ⁶⁷ , ⁶⁸

Factor analysis allows for examining whether depressive and cognitive indicators cluster as distinct or overlapping dimensions. In the context of structural racism, symptoms like apathy, low motivation, fatigue, and effort may reflect depressive symptom burden, cognitive inefficiency, or a shared symptomatology. A bi‐factor solution would suggest two largely distinct dimensions, while cross‐loadings or three‐factor solutions would suggest overlap or further differentiation among indicators.

Using a social exposomic framework, we assessed how adverse policing (i.e., historical lynching, disproportionate state‐level police violence, and personal negative law enforcement encounters) shape the dimensional structure and measurement properties of depressive and cognitive assessments among participants racialized as Black (henceforth older Black adults) enrolled in the U.S. Health and Retirement Study (HRS, 2006‐2020) with exposure to adverse policing practices. ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² We examined whether depressive symptoms and cognitive performance remained separable or showed dimensional overlap across adverse policing exposure groups. Given the exploratory nature of this study, we did not specify directional hypotheses. Rather, we used the expected conceptual distinction as a baseline for examining whether these domains remained separable or showed dimensional overlap.

RESEARCH IN CONTEXT

1. Systematic review: Prior work has consistently shown associations between depressive symptoms, cognitive impairment, and risk for Alzheimer's disease and related dementias. There is also abundant evidence linking exposure to adverse policing practices and poor mental health outcomes. Emerging research further situates these relationships within a life‐course exposomic framework, emphasizing the cumulative impact of chronic social and structural stressors on cognitive aging. However, few studies have examined whether exposure to structural adversity in the form of adverse policing, shapes the dimensional structure or measurement properties of depressive and cognitive assessments in later life.

2. Interpretation: Using longitudinal data from Black older adults in the Health and Retirement Study and applying a social exposomic framework, this study examined how chronic exposure to adverse policing relates to the conceptualization of depressive symptoms and cognition. Results indicate that depression and cognition largely remain distinguishable constructs; however, among individuals exposed to adverse policing, select depressive items related to effort and motivation and cognitive tasks tapping executive functioning exhibited consistent overlap. These findings suggest that chronic structural adversity may give rise to an intermediate dimension reflecting shared psychosocial and executive burden, complicating distinctions between affective and cognitive symptoms. Rather than solely increasing symptom burden, adverse policing exposures may shape how depression and cognition are expressed and measured in later life.

3. Future directions: Future work should more explicitly integrate structural adversity and criminal‐legal exposures into exposome‐based models of cognitive aging and dementia risk. Longitudinal investigations of measurement invariance, along with the development of assessment approaches that account for chronic structural stress, are needed to improve the validity of depression and cognitive measures in marginalized populations. Incorporating social exposome exposures into dementia research may enhance etiologic inference, refine risk characterization, and support more equitable approaches to Alzheimer's disease prevention and diagnosis.

2. METHODS

2.1. Sample

Participants were enrolled in HRS, a nationally representative longitudinal study of U.S. adults aged 51 and above. The HRS was approved by the University of Michigan's Institutional Review Board, and all participants provided informed consent prior to participation. For this analysis, we selected respondents who (1) self‐identified as being racialized as Black or African American, (2) were aged 65 or older in 2006 (the baseline year), and (3) had non‐missing data on depressive symptoms, cognitive test scores, and at least one adverse policing exposure variable. This yielded an analytic sample of N = 265 community‐dwelling older adults followed from 2006 to 2020. See Table 1 for additional sample characteristics. 

TABLE 1.

Participants racialized as Black (65 +) from the Health and Retirement Study (N = 265).

Characteristic	N	%	M (SD)
Born in U.S.	249	94
Live in U.S. South	145	56
Gendered as woman	185	72
Has HS diploma	112	44
Married	82	31
Exercise	67	25
Currently smoke	19	15
Individual‐level police exposure	36	14%
State‐level police exposure	123	46%
State‐level lynching exposure	257	97%
All adverse policing exposure	20	8%
Age			69 (3.88)
Depression at baseline (2002) with reverse coding			6.74 ()
Cognitive function at baseline (2002)			15.18 ()

Abbreviations: HS, high school; SD, standard deviation; U.S., United States.

2.2. Study variables

2.2.1. Exposure variable

We measured  adverse policing exposure in three ways: (1) historic or early life exposure to lynching through late‐life residence in a state with documentation of at least one lynching of a person racialized as Black between 1882 and 1968; (2) adulthood exposure to disproportionate policing through late‐life residence in a state with more than average police killings of people racialized as Black than the average of those racialized as White between 1985 and 1994; and (3) lifetime exposure to individual‐level negative police encounters.

The historic or early life exposure variable was calculated using the frequency of people racialized as Black who were lynched per U.S. state between 1882 and 1968 from the Tuskegee Archive. We coded a state as 1 if there was at least one lynching of a person racialized as Black. The adulthood exposure variable was calculated using data from the Global Disease Burden Dataset between 1985 and 1994. In this time frame, racially targeted strategies produced by the War on Drugs were already in motion, while the billionaire boost to prisons, police departments, and punitive measures as the primary approach to addressing crime had not yet been officially launched. We coded a state as 1 if the state has a mean of estimated police killings of people racialized as Non‐Latinx Black greater than the mean of estimated police killings of people racialized as Non‐Latinx White. The lifetime exposure variable was gathered from participants within HRS that self‐reported ever being unfairly stopped by a police officer or ever being in trouble with the police before the age of 18.

2.2.2. Depressive symptoms

We used HRS participant responses from in‐person or online interviews. Depressive symptoms were measured using a modified version of the Center for Epidemiologic Studies Depression (CES‐D) scale consisting of eight items with total scores ranging from 0 to 8. ⁷³ Participants were asked to reflect upon the past week and rate “No” or “Yes” if they experienced specific symptoms during that time. Items (i.e., emotions or feelings) included feeling lonely, depressed, sad, happy, enjoyed life, lacked motivation, activities took effort, and experienced restless sleep. We reverse‐coded necessary items to ensure that greater depressive symptoms were scored as 1. We averaged responses per individual, whereby higher scores indicated greater depressive symptoms. Depressive symptoms were assessed in participants every 2 years between 2006 and 2020.

2.2.3. Cognitive performance

For global cognitive test performance, we used test scores from the modified Telephone Interview for Cognitive Status (TICS), a telephone‐administered cognitive assessment commonly used in population‐based aging studies to assess global cognitive performance and screen for possible cognitive impairment. ⁷⁴ The test captures immediate and delayed 10‐word recall, serial 7s subtraction, and a backwards counting test starting from 20. For the immediate word recall task, the administrator read a list of 10 nouns and asked participants to recall as many words as possible, in any order. The delayed word recall task was administered approximately 5 minutes later wherein participants were asked to recall the previously presented nouns. Immediate recall and delayed recall scores ranged from 0 to 10 each. For the serial 7s subtraction task, participants were asked to subtract 7 a total of five times beginning with the number 100 (score range 0 to 5). Backwards counting required participants to count backwards from 20 correctly for 10 consecutive numbers, and they were given two trials to complete the task successfully. Responses were coded as 0 (incorrect/unable to complete), 1 (incorrect on first trial, correct on second trial), or 2 (correct on first trial). TICS summary scores ranged from 0 to 27, with higher scores indicating better cognitive performance. Although modified TICS summary scores have been used to identify possible cognitive impairment in prior HRS research, the present analyses examined dimensional patterns across individual cognitive performance indicators rather than assigning clinical impairment diagnoses. We used the imputations provided by HRS for missing data.

2.3. Data analysis

We linked the independent and dependent variables by U.S. Federal Information Processing Standard (FIPS) code. We averaged depressive symptom and global cognitive test scores per year across four participant groups: (1) participants with historical lynching exposure (N = 257); (2) participants with disproportionate state‐level police killings exposure (N = 123); (3) participants with self‐reported lifetime negative law enforcement encounters (N = 36); and (4) participants exposed to all three adverse policing indicators (N = 20). Exposures are not mutually exclusive, as participants may experience more than one simultaneously. These participant groups reflect adverse policing indicators at different levels of influence and timing of exposures. This approach is consistent with exposome and social determinants frameworks that emphasize exposures across levels of influence and across the life‐course. ⁸ , ⁴⁹ , ⁷⁵

We used descriptive statistics to examine trajectories of average depressive symptoms and global cognitive test performance among participant groups between 2006 and 2020. We then performed an exploratory factor analysis (EFA) to investigate the dimensionality of the CES‐D and TICS by participant group. To determine factorability, we used the Kaiser–Meyer–Olkin Measure of Sampling Adequacy test (KMO > 0.5) and Bartlett's test of sphericity (p < 0.05). Eigenvalues of 1 or greater were used as the criterion for determining a factor. We expect an item structure guided by the conceptual distinction between depressive symptoms and cognitive performance. Therefore, the expected structure may be a bi‐factor one where CES‐D items load on a depressive symptom dimension because they reflect affective, motivational, interpersonal, and somatic depressive symptoms. Separately, TICS items load on a cognitive performance dimension because they assess episodic memory and executive functioning. This conceptually informed expected bi‐factor structure was used as the baseline model for testing whether depressive symptoms and cognitive performance remained separable or showed overlap across adverse policing exposure groups. Although we are aware of the challenges to this approach, we found these criteria to be appropriate for the exploratory nature of the research question.

Factor loadings were interpreted using sample‐size‐informed thresholds, with higher loading thresholds applied to smaller participant groups to reduce overinterpretation of unstable factor patterns. For the participant group historic or early life exposure to lynchings, we expect items to load onto their respective factors at a magnitude of 0.35 or above. For the participants grouped as adulthood exposure to disproportionate policing killings, we expect loadings of 0.50 or above. For all other participant groups (n ≤ 50), we expect 0.75 and above. Because depressive symptoms and global cognitive performance items can have different directionality, we also performed a set of exploratory analyses where depressive symptoms were coded in the same direction as cognitive items. Analyses were computed in statistical software SPSS (version 31) and R (version 4.4.1).

3. RESULTS

Among the full analytical sample (N = 265), the average age at baseline was 70 years (SD = 3.87). Approximately 72% of the participants were women, 44% had at least a high school education, and most participants resided in the U.S. South (56%). Fewer participants (15%) smoked at baseline and exercised at least once a week (25%). The highest mean levels of chronic health conditions and alcohol use were observed among participants who experienced all three adverse policing exposures (see Table 1).

3.1. Participants with historical lynching exposure

At baseline, participants (n = 257) reported an average of 1.26 (SD = 1.69; range = 0–8) depressive symptoms, and an average cognitive performance score of 13.95 (SD = 3.9; range = 0–27). The EFA yielded a bi‐factor solution where Factor 1 includes all depressive symptoms. Factor 2 includes all cognitive performance items in addition to depressive symptoms, felt activities were efforts (0.533), felt unmotivated (0.406), and feeling depressed (0.353). Felt activities were efforts loaded higher on Factor 2 than Factor 1 (0.429) with a confirmatory factor analysis (CFA) based on the EFA bi‐factor model suggesting comparable factor loadings for both factors. This model yielded poor fit using root mean square error of approximation (RMSEA) and Tucker–Lewis Index (TLI), marginal fit with Comparative Fit Index (CFI), and good fit with standardized root mean square residual (SRMR). The CFAs based on the expected bi‐factor model and a unidimensional model yielded poor fit (see model fit indices in Table 2, CFA factor loadings in Table 3, and EFA factor loadings in Table S1).

TABLE 2.

Confirmatory factor analyses for reverse coded depressive symptoms and cognitive function items among participants racialized as Black (65 +) from the Health and Retirement Study.

		Individual‐level police exposure (N = 36)					State‐level police exposure (N = 123)
		Expected model		Unidimensional model	EFA model		Expected model		Unidimensional model	EFA model
Construct	Items	Factor 1	Factor 2	Factor 1	Factor 1	Factor 2	Factor 1	Factor 2	Factor 1	Factor 1	Factor 2	Factor 3
Depressive symptoms	Felt sad	0.957		0.961	0.977		0.885		0.878	0.889
	Feeling depressed	0.899		0.896	0.895		0.863		0.869	0.861
	Felt lonely	0.811		0.803	0.795		0.811		0.812	0.810
	Happy	0.803		0.803	0.790		0.719		0.708	0.723
	Sleep was restless	0.615		0.615			0.577		0.577	0.572
	Enjoyed life	0.669		0.668			0.561		0.552	0.564
	Felt unmotivated	0.720		0.726			0.553		0.561	0.543
	Felt activities were efforts	0.486		0.483		0.491	0.294		0.313			0.564
Cognitive function	Immediate word recall		1.175	0.130		1.059		0.900	0.238		0.917
	Delayed word recall		0.756	0.047		0.841		0.911	0.204		0.897
	Series 7s subtraction		0.285	0.412				0.463	0.139		−0.040	0.767
	Backward counting task		−0.113	0.568				0.366	0.155			0.610

		State‐level lynching exposure (N = 257)					All adverse policing exposure (N = 20)
		Expected model		Unidimensional model	EFA model		Expected model		Unidimensional model	EFA model
Construct	Items	Factor 1	Factor 2	Factor 1	Factor 1	Factor 2	Factor 1	Factor 2	Factor 1	Factor 1	Factor 2	Factor 3
Depressive symptoms	Felt sad	0.883		0.877	0.891		0.942		0.953	0.913
	Feeling depressed	0.883		0.888	0.855	0.136	0.890		0.886	0.948
	Felt lonely	0.801		0.800	0.799		0.853		0.840	0.899
	Happy	0.716		0.706	0.730		0.827		0.830	−0.823		1.694
	Sleep was restless	0.647		0.647	0.640		0.716		0.697	0.726
	Enjoyed life	0.538		0.529	0.550		0.761		0.756
	Felt unmotivated	0.598		0.606	0.551	0.214	0.657		0.668			0.651
	Felt activities were efforts	0.435		0.454	0.360	0.336	0.350		0.338		0.266
Cognitive function	Immediate word recall		0.953	0.227		0.936		0.708	0.026		1.325
	Delayed word recall		0.853	0.192		0.861		1.234	−0.011		0.663
	Series 7s subtraction		0.437	0.247		0.456		−0.076	0.301
	Backward counting task		0.322	0.178		0.333		−0.236	0.685			0.667

Note: Individual‐level police exposure are participants that reported unfairly being stopped by a police officer (at any point in their life) or being in trouble with the police at any point in time (2006 and 2008 interviews) and being in trouble with the police before the age of 18 (2010 interview); state‐level police exposure are participants living in a state where the mean of estimated deaths of people racialized as Non‐Latine Black due to police violence is greater than the mean of estimated deaths of people racialized as Non‐Latine White; state‐level lynching exposure are participants living in a state where at least one lynching of a person racialized as Black occurred.

Abbreviation: EFA, exploratory factor analysis.

TABLE 3.

Model fit for confirmatory factor analyses for reverse coded depressive symptoms and cognitive function items among participants racialized as Black (65 +) from the Health and Retirement Study.

	Individual‐level police exposure (N = 36)			State‐level police exposure (N = 123)			State‐level lynching exposure (N = 257)			All adverse policing exposure (N = 20)
Fit statistics	Expected model	Unidimensional model	EFA model	Expected model	Unidimensional model	EFA model	Expected model	Unidimensional model	EFA model	Expected model	Unidimensional model	EFA model
Chi‐squared	125.188	168.436	33.983	172.589	342.671	131.651	284.553	609.802	233.876	123.424	141.903	50.847
RMSEA	0.195	0.243	0.212	0.135	0.208	0.115	0.13	0.2	0.12	0.258	0.285	0.179
SRMR	0.202	0.159	0.152	0.114	0.165	0.092	0.105	0.149	0.079	0.225	0.167	0.145
CFI	0.767	0.63	0.896	0.832	0.596	0.886	0.851	0.641	0.881	0.645	0.557	0.874
TLI	0.71	0.548	0.831	0.791	0.506	0.849	0.814	0.561	0.843	0.558	0.458	0.817

Abbreviations: CFI, Comparative Fit Index; EFA, exploratory factor analysis; RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual; TLI, Tucker–Lewis Index.

Table S1. Exploratory factor analyses for reverse coded depressive symptoms and cognitive function items among participants racialized as Black (65+) from the Health and Retirement Study.

3.2. Participants with disproportionate state‐level police killings exposure

At baseline, participants (n = 123) reported depressive symptoms averaged 1.36 (SD = 1.68; range = 0–8), while cognitive performance items averaged 13.99 (SD = 3.84; range = 0–27). The EFA produced a 3‐factor solution where all depressive symptom items loaded onto Factor 1 except felt activities were efforts, which loaded highly (0.584) onto Factor 3 with serial 7s (0.714) and backwards counting (0.630). Factor 2 included all recall items. A CFA based on the EFA three‐factor model yielded poor fit using RMSEA and TLI and marginal fit for SRMR and CFI. The CFAs based on the expected bi‐factor model and a unidimensional model yielded poor fit across indices.

3.3. Participants with self‐reported lifetime negative law enforcement encounters

At baseline, participants (n = 36) reported depressive symptoms averaged 1.5 (SD = 1.99; range = 0–8), and cognitive performance items averaged 14.36 (SD = 4.39; range = 0–27). The EFA yielded a 3‐factor solution where Factor 1 includes most depressive symptoms (felt sad, feeling depressed, felt lonely, happy) and Factor 2 included cognitive performance items (immediate and delayed recall). Factor 3 included loadings greater than 0.20 for depressive and cognitive performance items. Most notably, serial 7s subtraction loaded on Factor 3 (0.745) with felt unmotivated (0.660), feeling depressed (0.592), felt sad (0.590), and felt activities were efforts (0.575). With these items loading higher in Factor 1, the model was run as a bi‐factor model. A CFA based on the EFA bi‐factor model yielded poor fit using RMSEA, SRMR, and TLI, but marginal fit with CFI. The CFAs based on the expected bi‐factor model and a unidimensional model both yielded poor fit across all indices.

3.4. Participants exposed to all three adverse policing indicators

At baseline, participants (n = 20) reported depressive symptoms averaged 2.15 (SD = 2.11; range = 0–8), and cognitive performance items averaged 13.40 (SD = 3.95; range = 0–27). The EFA produced a 3‐factor solution where all depressive symptom items loaded moderate‐to‐high onto Factor 1 except felt activities were efforts (0.308). Backwards counting also loaded moderately high on Factor 1 (0.619). Factor 2 included highly loading recall items, and moderately loading felt activities were efforts (0.570), though this item shared variance with Factor 1 (0.308) and Factor 2 (0.350). Three depressive symptom items loaded moderately onto Factor 3: felt sad (0.584), happy (0.589), and felt unmotivated (0.542), with one item, enjoyed life, loading higher than on Factor 1 (0.823). These items loaded with serial 7s (0.568) and backwards counting (0.662). A CFA based on the EFA 3‐factor model revealed high factor loadings for Factor 1 comprised of felt sad, feeling depressed, felt lonely, and sleep was restless; Factor 2 with recall items; and Factor 3 with happy, felt unhappy, and backwards counting. This model yielded poor fit using RMSEA, SRMR, and TLI, but marginal fit with CFI. Again, CFAs based on the expected bi‐factor and unidimensional models yielded poor fit.

4. DISCUSSION

This study examined the dimensional overlap between depressive symptoms and cognitive performance among older Black adults exposed to structural racism, particularly through adverse policing experiences, using an exposome‐informed perspective. Because this was a within‐group study of older Black adults, these findings should not be interpreted as direct evidence of between‐group dementia disparities. Rather, they identify potential disparity‐relevant mechanisms by which adverse policing may shape depressive symptom expression, cognitive performance, and their measurement in AD/ADRD‐related research.

Our findings suggest that chronic exposure to structural racism, operationalized here through adverse policing, shapes not only individual symptoms but also may shape the fundamental measurement structure of depressive‐symptom and cognitive performance items relevant to AD/ADRD. Among participants with historical lynching exposure, depressive symptom and cognitive performance items generally separated into two dimensions, although effort‐related depressive symptoms also loaded with cognitive performance items. Specifically, Factor 1 included all depressive symptoms, while Factor 2 included all cognitive performance items in addition to “felt activities were efforts,” “felt unmotivated,” and “feeling depressed.”

Among participants exposed to disproportionate state‐level police killings, a three‐factor solution emerged, with most depressive symptom items loading onto one factor, recall items loading onto a second factor, and “felt activities were efforts” loading with serial 7s and backwards counting on a third factor. Among participants with self‐reported lifetime negative law enforcement encounters, a three‐factor solution also emerged, with most depressive symptoms loading onto one factor, recall items loading onto a second factor, and serial 7s subtraction loading with motivational and affective depressive symptom items on a third factor. Finally, among participants exposed to all three adverse policing indicators, model fit was poorest and factor patterns were least stable, with depressive symptoms, recall items, effort‐related symptoms, serial 7s, and backwards counting loading across multiple factors. Furthermore, participants with multiple adverse policing exposures demonstrated the poorest model fit, lowest cognitive performance, and highest depressive symptoms over time, suggesting both greater symptom severity and a mismatch between existing measurement tools and lived experiences. These findings suggest that cumulative exposure may be associated with greater measurement complexity; however, given sample size limitations, these findings should be interpreted cautiously.

Across exposure groups, depressive symptoms and cognitive performance generally loaded onto separate latent dimensions, with core affective items such as “felt sad,” “feeling depressed,” and “felt lonely” consistently loading on the depression factor and both recall items loaded on cognition. However, all other items infrequently load highly, load negatively, or occasionally load onto a third factor. Among all groups, items like serial 7s subtraction and backwards counting did not consistently load onto the cognitive factor. In some cases, backwards counting even loaded negatively. These somatic and executive control items may reflect a shared underlying construction (e.g., executive burden) that becomes more pronounced under conditions of chronic stress or institutional threat, suggesting that this latent overlap is relevant to both measurement approaches and lived experiences. These overlapping items were only observed when depressive symptoms were coded in the same direction as cognitive performance (i.e., fewer depressive symptoms were associated with better cognition). When coded in opposite directions (more depressive symptoms with better cognitive performance), the items were pulled distinctly into separate factors. These findings are consistent with literature linking motivational deficits and cognitive strain as overlapping domains in depression and neurocognitive disorders, particularly in groups exposed to sustained adversity. ²⁰ , ³³ , ³⁸ , ³⁹ , ⁴⁰

Additionally, unidimensional models consistently showed poor fit, with cognitive items loading weakest, especially in the individual‐level and cumulative exposure groups, where recall items fell below 0.15. The EFAs performed relatively better (KMO > 0.5, Bartlett's p < 0.05), though model fit remained poor by RMSEA and CFI standards, with only marginal fit using SRMR. The repeated emergence of a third factor in these analyses, coupled with the poor fit of standard bi‐factor and unidimensional models, suggests that depressive symptoms and cognitive performance were not fully captured by either a single shared dimension or completely separate dimensions. Instead, certain somatic, motivational, and cognitive effort‐related items appear to form an intermediate dimension that may reflect cumulative psychosocial burden, particularly for those exposed to chronic surveillance, community violence, or other manifestations of institutionalized racism.

From an exposomic perspective, these results illustrate how sustained social adversities impact affective and cognitive outcomes potentially through chronic stress and allostatic load pathways. Prior research highlights how chronic psychosocial stress may drive accelerated biological aging, inflammation, and neuroendocrine dysregulation, all of which increase AD/ADRD risk. ⁵⁹ , ⁶⁰ , ⁶¹ , ⁶² Our findings expand upon this literature by demonstrating that the exposome of structural racism does not simply elevate symptom severity; it may influence the measurement properties and psychometric validity of standard assessments. For example, chronic hypervigilance in response to adverse police presence may resemble depressive symptoms on standardized scales, emerging from a distinct sociocultural process. ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁶⁹ , ⁷⁰ The consistently poor fit among individuals with all three exposures underscores the limitations of traditional measurement frameworks when applied to groups enduring lifelong structural trauma. As cognitive health disparities persist, future work must interrogate how institutionalized mistreatment and structural inequity shape how depressive symptoms and cognitive performance are conceptualized and measured across diverse groups.

4.1. Limitations

Several limitations should be considered when interpreting these findings. It is unclear why state‐level exposure to historic or early life lynchings performed differently from other exposures. Perhaps the historic nature of the variable makes it so that only two factors were extracted. The historic lynching data are distinctly geographically patterned in the United States, which is reflected in the sample as most participants racialized as Black live in the U.S. South, where most lynchings occurred (56%). Individual‐level exposure to adverse policing yielded a different number of factors and loading, and this exposure included encounters prior to age 18 that may have also included lynchings. It was also not possible to explore how these constructs are perceived among participants without any exposure to adverse policing as only a few were not exposed in this study.

Sample sizes for specific participant groups, like the cumulative exposure (n = 20), may have limited the stability and precision of factor solutions. Although conservative loading thresholds were applied and factorability was confirmed, replication in larger samples and the use of appropriate comparison strategies are necessary to verify the robustness of the observed patterns. Additionally, anxiety, hypervigilance, avoidance, and other trauma‐related symptoms were not assessed. This is an important limitation given that adverse policing experiences may be associated with trauma‐related symptoms in addition to depressive symptoms. As such, we cannot determine whether overlapping or cross‐loading item patterns also reflected depressive symptoms solely, broader psychological distress, or trauma‐related responses to adverse policing. Future research should incorporate anxiety and trauma symptom measures to better distinguish affective‐cognitive pathways linking adverse policing to dementia‐related brain health.

Finally, these findings should be interpreted as exploratory and hypothesis‐generating. Although exploratory models showed improved relative fit compared with the expected model, they did not consistently demonstrate good absolute fit. Thus, findings should be interpreted as suggestive evidence of possible dimensional overlap, rather than confirmation of a stable alternative factor structure. Poor fit may reflect sample size limitations, exposure‐group heterogeneity, or measurement constraints of the CES‐D and TICS items. Longitudinal modeling and tests of measurement invariance across exposure groups may provide more substantial evidence of structural convergence or divergence. Future research should incorporate culturally validated measures, qualitative insights, and intersectional frameworks to more accurately capture the neurocognitive consequences of structural racism across the lifespan.

4.2. Conclusion

Our study highlights the urgent need for interdisciplinary exposome research that integrates social epidemiology, neuropsychology, gerontology, and public health. Understanding dementia disparities requires approaches that move beyond traditional biobehavioral models to acknowledge and measure the complex interplay of social, political, economic, and environmental exposures across the life‐course. Exposure to structural racism via adverse policing, may shape how depressive symptoms and cognitive performance are expressed, conceptualized, and measured. Among participants exposed to state‐level adverse policing or individual‐level negative law enforcement encounters, findings suggested dimensional overlap between some depressive‐symptom and cognitive performance items. By explicitly incorporating structural racism and measurement science into exposome research, scientists and practitioners can more accurately characterize the etiological pathways through which social adversity impacts cognitive health and dementia risk.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts to report. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

All participants provided informed consent prior to participation.

Supporting information

Supporting Information: alz71790‐sup‐0001‐tableS1.docx

Supporting Information: alz71790‐sup‐0002‐SuppMat.pdf

DATA AVAILABILITY STATEMENT

These data are publicly available at the Health and Retirement Study databases.