Self-Regulatory Coping Behaviors and Stress Reactivity: Exploring the Environmental Affordance Model of Health Disparities

Abstract

Objectives

To explore the relationship between self-regulatory coping behaviors (SRCB) and hypothalamic pituitary adrenal (HPA) stress reactivity.

Methods

Data came from the Richmond Stress and Sugar Study (n=125, median age: 57 years, 46% non-Hispanic White, 48% African American). The relationships between 11 SRCB (“health-harming” [e.g., smoking] and “health-promoting” [e.g., exercising]) with HPA stress reactivity, indicated by salivary cortisol response to the Trier Social Stress Test, was assessed using multi-level modeling.

Results

Health-harming and health-promoting SRCB were positively correlated (+0.33, p<0.001). Several individual behaviors were related to HPA stress reactivity, for example, smoking and meditation were associated with shallower increases in cortisol (smoking: −13.0%, 95%CI: −20.9% to −4.3%; meditation: −14.0%, 95%CI: −22.0% to −5.1%). However, SRCB summary measures were unrelated to stress reactivity.

Discussion

Health-harming and health-promoting SRCB are inter-related. Specific behaviors, rather than groupings as health-harming versus -promoting, are related to HPA stress reactivity.

Keywords

stress coping disparities biopsychosocial HPA-axis

“We believe that… the interrelation between physical and mental health disparities is about the utilization of stress coping mechanisms over the life course...We argue that poor health behaviors actually buffer the cascade of the effect of the chronic activation of the HPA-axis on psychiatric health disorders… The weak hypothesis [of the Environmental Affordances Model] says that poor health behaviors mask the stress response cascade that indeed creates the development of mental disorders. The strong one says that poor health behaviors through their actions on the HPA-axis and other brain hormones actually interfere with the cascade of neural and hormonal events that ordinarily would lead over time to mental disorders.”

- James S. Jackson, Association for Psychological Science, 24^th Annual Convention, May 24-27, 2012.

Introduction

Social scientists have long hypothesized that differential exposure to, and opportunities to cope with, chronic stress are fundamental processes underlying the emergence and persistence of racial and economic disparities in morbidity and mortality (Thoits, 2010). Additional scholarship emphasized the salience of “neighborhood” factors (e.g., cohesion, trust, and safety; greenspace, community resources, and walkability; area-level economic indicators) in shaping exposure and responses to stress (Barrington et al., 2014; Schulz et al., 2005). This work helped generate major improvements in the measurement of stress and in identifying contextual factors that shape the stress process (Turner et al., 1995). However, this research was largely uninformed by the parallel growth in understanding of the neurobiological stress response (i.e., hypothalamic–pituitary–adrenal (HPA) axis, sympathetic nervous system) (Miller et al., 2009), and the emerging appreciation of the salience of stress for reinforced and learned behaviors (Koob, 2008; Koob & Volkow, 2016). Dr. James S. Jackson sought to integrate these two lines of scholarship through his workgroup the Social Neuroscience of Health Disparities (SNOHD). Many of the activities of SNOHD were focused on elaborating and testing hypotheses informed by the Environmental Affordances (EA) Model of Health Disparities that James put forth in his writings and academic talks (Jackson et al., 2010; Jackson & Knight, 2006; Mezuk et al., 2010, 2013, 2017). This paper continues this line of inquiry by exploring the “strong” hypothesis of the EA Model that James described in his quoted speech above.

One of the most difficult challenges to the scientific study of stress as a determinant of health is the need to simultaneously integrate data on social, psychological, and biological aspects of the stress process (Nusslock & Miller, 2016). Even today there are few datasets that have robust measures across all three aspects of this phenomenon, particularly with sufficiently large samples of racial/ethnic minorities or lower socioeconomic groups, to permit testing hypotheses about how stress contributes to health disparities. Numerous studies have explored diurnal cortisol or multi-system indicators of physiology (e.g., allostatic load, telomere length) as indicators of the biological consequences of stress exposure (Castro-Diehl et al., 2014; Meier et al., 2019); however, these are not measures of the neurobiological stress response. While chronic stress exposure and acute stress reactivity are conceptually distinct, they are thought to be linked through both learned responses and physiologic processes over time (Chida & Hamer, 2008); however, the relationship between self-reported stress exposure and the acute stress response remains unclear (e.g., self-report chronic stress has been associated with both elevated and blunted cortisol response) (Do et al., 2011; Meier et al., 2019; Wells et al., 2014). As such the notion that “stress” is a key determinant of health disparities as a hypothesis still requires further exploration of the specific mechanisms driving this relationship.

There is also a need to critically evaluate the ways in which stress intersects with health behaviors like smoking, diet, physical activity, and alcohol use. First, move beyond models that treat behavior solely at the individual level by acknowledging that neighborhood context shapes opportunities to engage in behaviors, whether those are health-promoting (e.g., exercise) or health-harming (e.g., smoking). Second, move beyond the traditional approach of treating behaviors simply as “confounders” of the stress–health relationship; the NIH Common Fund Science of Behavior Change (SOBC) Initiative (Keller-Allen & Li, 2014) provides one framework for doing so. The SOBC seeks to reframe efforts to promote “healthy lifestyle” behaviors (i.e., diet, physical activity, alcohol consumption, tobacco use) in a biologically informed way that reframes these as “goal-oriented” behaviors in a broad sense, embodying concepts like temporal discounting, short-term versus long-term goals, and positive and negative trade-offs to (in)action (Nielsen et al., 2018). SOBC draws on growing evidence, from both animal and human studies, that the stress process directly intersects with health behaviors (Koob, 2008). For example, experimental studies have shown that after being stressed, not only do rats prefer to consume foods high in sugar, but doing so reduces the amount of corticotropin releasing factor (a component of the HPA-axis) produced in response to the stressor (Foster et al., 2009). While data are still emerging, experimental studies in humans have shown that consuming sucrose-sweetened beverages alters HPA-axis stress reactivity compared to aspartame-sweetened beverages (Tryon et al., 2015). However, how the relationship between stress and self-regulatory behaviors relates to racial and socioeconomic disparities is not well understood.

The intersection of the stress process and self-regulatory behaviors sits at the center of the EA Model. Supplemental Figure 1 illustrates the model, with elaborations regarding the hypothesized mechanisms underlying each theorized path. Disparities in physical health are theorized to be shaped by differential exposure to chronic stress over the life course, in combination with greater engagement in “negative” health behaviors (i.e., tobacco use, poor diet, sedentary behavior). While differential exposure to chronic stress also shapes risk of poor mental health, the model argues for a more complex relationship for health behaviors and mental health outcomes. Specifically, the model hypothesizes that the same health behaviors that increase risk of medical morbidity long-term, when employed as self-regulatory, stress-coping behaviors, buffer the negative impact of chronic stress on mental health short-term. The “strong” hypothesis of this buffering effect is that it operates through neurobiological processes, predominantly the HPA-axis (Jackson & Knight, 2006). Analyses using independent samples have provided support for various paths of the EA Model (Boardman & Alexander, 2011; Jackson et al., 2010; Mezuk et al., 2010, 2017), although findings are not conclusive (Keyes et al., 2011) and the model is a subject of ongoing debate (Mezuk et al., 2021). In sum, while fundamentally situated within the social sciences, the EA Model provides an integrative, biologically informed, hypothesis-generating framework to guide interdisciplinary research on the ways that chronic stress and health behaviors contribute to mental and physical health, and health disparities, over the life course.

The goal of this paper is to examine one component of the EA Model by exploring the relationship between self-regulatory coping behaviors (SRCB) and the neurobiological stress response. This analysis uses data from the Richmond Stress and Sugar Study (RSASS), a racially and socioeconomically diverse cohort of adults at high risk of developing type 2 diabetes (T2D) that has robust measures of stress coping and the HPA-axis stress response. Together this makes the RSASS cohort an ideal setting for exploring questions related to the “strong” hypothesis of the EA Model articulated in the quote above: (1) Does the frequency of SRCB vary by neighborhood SES and race? and (2) Are SRCB associated with HPA-axis stress reactivity? While addressing these questions will not directly test the “strong” hypothesis in its entirety, this will be among the first efforts to examine the neurobiological stress response from the theoretical framework of the EA Model.

Methods

Sample. Details of the study design and inclusion/exclusion criteria of the RSASS are described elsewhere (Mezuk et al., 2020). Briefly, targeted recruitment of adults aged 40–70 who were at elevated risk of T2D was facilitated using geocoded healthcare records. These records were mapped to a composite index of area SES (i.e., per-capita income, owner-occupied housing, home value, vacant housing, and household poverty) of the Richmond, VA metropolitan area, which was used to create the two-by-two sampling frame. The target sample was first divided into Census tracts in the upper and lower tertiles of area SES, and then within each tertile we sought to recruit an equal number of African Americans (AA) and non-Hispanic whites (NHW) to separate “race” from “place.” In total, 125 participants completed the baseline assessment: 40 higher-SES NHW, 33 higher-SES AA, 18 lower-SES NHW, 27 lower-SES AA, and 7 (3 from higher-SES) who identified as a race/ethnicity other than AA or NHW. This analysis only uses data from the baseline interview which was when HPA-axis stress reactivity was assessed. RSASS was approved by the Institutional Review Board at [BLINDED FOR REVIEW], and all participants provided written informed consent.

Exposure: Self-regulatory coping behaviors (SRCB). Interviewer-administered Computer-Assisted Personal Interviews (CAPI) were used to assess SRCBs using a modified version of a scale developed by Dr. Jackson initially fielded in the Health and Retirement Study (Mezuk et al., 2017). The questions addressed six “health-harming” (eat sweet foods, eat fried foods, eat snack foods, drink alcohol, smoke/chew tobacco products, and use distractions like TV/internet) and five “health-promoting” SRCB (exercise, prayer, seek advice, speak with a counselor, and meditate). Supplemental Table 1 provides the Spearman correlations among these 11 SRCBs.

Participants were oriented to the scale with the following text (emphasis in original): “I will first ask you how often you do certain behaviors to cope after having what you think is a stressful event or day. I will then ask you how much that behavior reduces your feelings of being stressed.” This analysis focuses on the frequency of engaging in coping behaviors, that is, the response to the “how often” portion of the question. Frequency of using each SRCB was recategorized as a three-level variable (Never=0, Hardly ever/Not too often=1, and Fairly often/Very often=2) due to small cells. Beyond these indicators of each behavior, they were also summed to create three overall measures: (1) engagement in the six “health-harming” behaviors (median [IQR]: 4 [2, 5]), (2) engagement in the five “health-promoting” behaviors (median [IQR]: 8 [6, 9]), and (3) engagement in any of the SRCBs (median [IQR]: 11 [9, 13]).

Exposure: Health behaviors and related characteristics. In addition to the SRCB, participants were asked about health behaviors and related characteristics, irrespective of stress and coping (i.e., whether they smoke tobacco, drink alcohol, exercise, have a large network of friends, etc.). These were also examined as exposures to assess to what degree the behavior itself, as opposed to its perceived coping utility, was associated with HPA stress reactivity. These behaviors and characteristics were: (1) smoking status; (2) alcohol use; (3) exercise per week; (4) sedentary time at home; (5) social network size of (5a) friends and (5b) family; and (6) indicators of adiposity, (6a) body mass index (BMI), and (6b) waist circumference. Supplemental Table 2 details the measurement of these behaviors and characteristics

Outcome: HPA-axis stress reactivity. Acute stress reactivity was assessed using the Trier Social Stress Test (TSST), a well-validated paradigm for evoking an HPA-axis response (Allen et al., 2017). Eight salivary cortisol samples were taken to characterize the HPA-axis response: two pre-stress taken 15 minutes apart, and six post-stress, beginning immediately after the task, and thereafter every 15 minutes. Details of how the TSST was implemented in RSASS have been described previously (Mezuk et al., 2020). Briefly, the TSST began at 5:00 pm for all participants after resting quietly for approximately 30 minutes. Participants were asked to give a 5-minute speech as to why they should be hired for a job, followed by 2-minutes of mental arithmetic, in front of a panel of three “judges” (research staff) who the participants believed were evaluating their performance. With a handful of exceptions, TSST panels were composed of three staff and included both sexes and both NHW and AA individuals. After the eighth (final) cortisol sample was taken (∼6:45 pm), the participant was offered a snack and debriefed on the procedure, including the true nature of the judge panel. Samples were processed in duplicate using ELISA (Salimetrics LLC) and optical density values were converted to ug/dL of cortisol using the manufacturer’s specifications.

Covariates. Demographic characteristics included age (in years), sex (male, female), race (NHW, AA, other), education (in years), marital status (married/cohabitating vs. other), and area (neighborhood) SES (higher vs. lower, as described above).

Analysis

We first examined the distribution of SRCB as a function of neighborhood SES and race using chi-squared tests. The correlations of SRCB with each other, and with individual cortisol values from the TSST, were assessed using the Spearman method. Poisson regression was used to estimate whether the number of SRCB (total, health-harming, health promoting) varied by race and neighborhood SES after accounting for covariates.

We conducted exploratory data analysis including fitting locally estimated scatterplot smoothing curves (LOESS) to examine the cortisol response during the TSST. From this analysis, we determined that one linear spline knot at 45 minutes (time 4) since the start of the TSST was sufficient to capture the non-linear cortisol response. We used piecewise 2-level linear mixed effects models with an autoregressive correlation structure to examine the associations between the SRCB and three features of the salivary cortisol response across the TSST (baseline, increasing slope, and decreasing slope). Models included participant-level random intercepts and random slopes for the time spline, the main predictors, and their interactions with the time spline and were adjusted for age (mean-centered), sex, race, education (mean-centered), marital status, and neighborhood SES. To improve reliability of model standard errors, we used restricted maximum likelihood estimation (REML) and robust standard errors with bias correction (McNeish, 2017).

Cortisol values were log-transformed in all models to increase normality. Therefore, to improve model interpretability, the estimated coefficients were exponentiated and are interpreted as either geometric means or percent differences. Visually, for the increase in cortisol during the TSST, positive effect estimates indicate a steeper slope; for the decrease in cortisol after the peak, positive effect estimates indicate a shallower slope. Analyses were conducted using R (4.0.3) and all p-values refer to two-tailed tests.

Results

Table 1 describes the baseline demographic characteristics of participants, overall and stratified by race/ethnicity and by neighborhood SES. While there was some variation by race, differences were more apparent by neighborhood SES. Participants who lived in lower SES neighborhoods had less education, were more likely to be married, had higher BMI, and were more likely to smoke, drink heavily, and spent more time sedentary at home. There were no significant differences in the summary indicators of SRCB by race or neighborhood SES, and Poisson regression modeling confirmed no significant differences in these summary indicators of SRCB by race or neighborhood SES even after accounting for covariates (data not shown).

Table 1.

Demographic characteristics, health behaviors, and self-regulatory coping behaviors by race and neighborhood SES.

		Race/Ethnicity*			Neighborhood Socioeconomic Status
	Overall (n=125)	Black (n=60)	White (n=58)	p-value	High (n=76)	Low (n=49)	p-value
	N(%) or Median [IQR]	N(%) or Median [IQR]	N(%) or Median [IQR]	p-value	N(%) or Median [IQR]	N(%) or Median [IQR]	p-value
Demographic Characteristics
Age	57 [52, 63]	56 [51, 62]	60 [53, 65]	0.04	59 [52, 64]	56 [52, 60]	0.13
Sex (% Female)	61 (48.8)	31 (51.7)	26 (44.8)	0.58	42 (55.3)	19 (38.8)	0.11
Years of education	16 [13, 17]	14 [12, 16]	16 [14, 17]	<0.01	17 [16, 17]	13 [12, 16]	<0.01
Marital status (% married)	64 (51.2)	39 (65.0)	24 (41.4)	0.02	28 (36.8)	36 (73.5)	<0.01
Health Behaviors
BMI (kg/m²)	28 [25, 33]	29 [25, 34]	28 [25, 32]	0.42	28 [25, 32]	30 [24, 36]	0.16
Tobacco use
Never	50 (40.3)	23 (38.3)	25 (43.9)	0.02	38 (50.7)	12 (24.5)	<0.01
Past	43 (34.7)	16 (26.7)	24 (42.1)		28 (37.3)	15 (30.6)
Current	31 (25.0)	21 (35.0)	8 (14.0)		9 (12.0)	22 (44.9)
Alcohol Use
Never/Past	43 (34.4)	24 (40.0)	15 (25.9)	0.15	22 (28.9)	21 (42.9)	0.16
Users: drinks/month	1 [1, 2]	2 [1, 2]	2 [1, 2]	0.76	1 [1, 2]	2 [1, 4]	<0.01
Sedentary Time Sitting at home (hours)	3 [2, 5]	4 [3, 6]	3 [2, 4]	<0.01	3 [2, 4]	4 [3, 7]	<0.01
Exercise
None	60 (48.0)	33 (55.0)	22 (37.9)	0.13	31 (40.8)	29 (59.2)	0.13
1–3 times/week	37 (29.6)	17 (28.3)	19 (32.8)		25 (32.9)	12 (24.5)
>3 times/week	28 (22.4)	10 (16.7)	17 (29.3)		20 (26.3)	8 (16.3)
Summed frequency of Self-regulatory Coping Behaviors
Total	11 [9, 13]	11 [9, 13]	11 [8, 13]	0.52	11 [8, 13]	12 [9, 14]	0.11
Health-promoting	8 [6, 9]	8 [6, 9]	8 [6, 9]	0.28	8 [5, 9]	8 [6, 9]	0.11
Health-harming	4 [2, 5]	4 [2, 5]	3 [2, 5]	0.73	3 [2, 5]	4 [1, 6]	0.48

*Excludes n=7 participants who reported race other than Black or non-Hispanic white. p-Value for age, BMI, education, sedentary time, drinks/month, and sums of self-regulatory coping behaviors derived from Wilcoxon rank-sum test comparing medians. Household income missing data on n=3.

Table 2 shows the distribution of individual SRCBs, and the reported effectiveness of those behaviors, as a function of race/ethnicity and neighborhood SES. While there were not substantial differences in SRCB by race, NHW were more likely to exercise (and reported it to be more effective), and less likely to smoke, pray, or talk to a counselor compared to AA. Similarly, while the differences were not substantial, those who lived in lower SES neighborhoods were significantly more likely to eat fried foods, eat snack foods, smoke (and reported it to be more effective), and marginally more likely to meditate and talk to a counselor, than those from higher SES neighborhoods. Participants who lived in higher SES neighborhoods were also marginally more likely to exercise and reported it to be more effective at reducing stress, than those who lived in lower SES neighborhoods.

Table 2.

Frequency and perceived effectiveness of SRCB by race and neighborhood SES.

		Race/Ethnicity*			Neighborhood SES
	Overall (n=125)	Black (n=60)	White (n=58)	p-value	Low (n=49)	High (n=76)	p-value
	N (%)	N (%)	N (%)	p-value	N (%)	N (%)	p-value
Sweet Food: Frequency
Never	28 (22.4)	18 (30.0)	8 (13.8)	0.10	11 (22.4)	17 (22.4)	0.46
Hardly Ever/Not often	61 (48.8)	27 (45.0)	33 (56.9)		21 (42.9)	40 (52.6)
Fairly/Very Often	36 (28.8)	15 (25.0)	17 (29.3)		17 (34.7)	19 (25.0)
Sweet Food: Perceived Effectiveness
Not at all	43 (34.4)	24 (40.0)	17 (29.3)	0.37	17 (34.7)	26 (34.2)	0.46
Not very much	48 (38.4)	23 (38.3)	23 (39.7)		16 (32.7)	32 (42.1)
Some/A great deal	34 (27.2)	13 (21.7)	18 (31.0)		16 (32.7)	18 (23.7)
Fried Food: Frequency
Never	49 (39.2)	23 (38.3)	23 (39.7)	0.91	18 (36.7)	31 (40.8)	0.02
Hardly Ever/Not often	57 (45.6)	27 (45.0)	27 (46.6)		18 (36.7)	39 (51.3)
Fairly/Very Often	19 (15.2)	10 (16.7)	8 (13.8)		13 (26.5)	6 (7.9)
Fried Food: Perceived Effectiveness
Not at all	80 (64.0)	40 (66.7)	36 (62.1)	0.28	30 (61.2)	50 (65.8)	0.87
Not very much	26 (20.8)	9 (15.0)	15 (25.9)		11 (22.4)	15 (19.7)
Some/A great deal	19 (15.2)	11 (18.3)	7 (12.1)		8 (16.3)	11 (14.5)
Snack Food: Frequency
Never	40 (32.0)	23 (38.3)	16 (27.6)	0.25	23 (46.9)	17 (22.4)	<0.01
Hardly Ever/Not often	55 (44.0)	21 (35.0)	29 (50.0)		13 (26.5)	42 (55.3)
Fairly/Very Often	30 (24.0)	16 (26.7)	13 (22.4)		13 (26.5)	17 (22.4)
Snack Food: Perceived Effectiveness
Not at all	68 (54.4)	36 (60.0)	28 (48.3)	0.43	29 (59.2)	39 (51.3)	0.67
Not very much	30 (24.0)	12 (20.0)	16 (27.6)		11 (22.4)	19 (25.0)
Some/A great deal	27 (21.6)	12 (20.0)	14 (24.1)		9 (18.4)	18 (23.7)
Alcohol: Frequency
Never	76 (60.8)	39 (65.0)	31 (53.4)	0.32	35 (71.4)	41 (53.9)	0.14
Hardly Ever/Not often	37 (29.6)	15 (25.0)	22 (37.9)		10 (20.4)	27 (35.5)
Fairly/Very Often	12 (9.6)	6 (10.0)	5 (8.6)		4 (8.2)	8 (10.5)
Alcohol Perceived Effectiveness
Not at all	84 (67.2)	44 (73.3)	34 (58.6)	0.23	36 (73.5)	48 (63.2)	0.27
Not very much	11 (8.8)	4 (6.7)	7 (12.1)		5 (10.2)	6 (7.9)
Some/A great deal	30 (24.0)	12 (20.0)	17 (29.3)		8 (16.3)	22 (28.9)
Smoking: Frequency
Never	97 (77.6)	41 (68.3)	51 (87.9)	0.04	31 (63.3)	66 (86.8)	<0.01
Hardly Ever/Not often	9 (7.2)	7 (11.7)	2 (3.4)		6 (12.2)	3 (3.9)
Fairly/Very Often	19 (15.2)	12 (20.0)	5 (8.6)		12 (24.5)	7 (9.2)
Smoking: Perceived Effectiveness
Not at all	101 (80.8)	44 (73.3)	51 (87.9)	0.11	34 (69.4)	67 (88.2)	0.02
Not very much	3 (2.4)	1 (1.7)	1 (1.7)		1 (2.0)	2 (2.6)
Some/A great deal	21 (16.8)	15 (25.0)	6 (10.3)		14 (28.6)	7 (9.2)
Exercise: Frequency
Never	29 (23.2)	21 (35.0)	8 (13.8)	0.02	16 (32.7)	13 (17.1)	0.09
Hardly Ever/Not often	47 (37.6)	20 (33.3)	22 (37.9)		14 (28.6)	33 (43.4)
Fairly/Very Often	49 (39.2)	19 (31.7)	28 (48.3)		19 (38.8)	30 (39.5)
Exercise: Perceived Effectiveness
Not at all	33 (26.4)	22 (36.7)	10 (17.2)	0.03	18 (36.7)	15 (19.7)	0.04
Not very much	9 (7.2)	2 (3.3)	6 (10.3)		1 (2.0)	8 (10.5)
Some/A great deal	83 (66.4)	36 (60.0)	42 (72.4)		30 (61.2)	53 (69.7)
Prayer: Frequency
Never	17 (13.6)	5 (8.3)	11 (19.0)	0.07	6 (12.2)	11 (14.5)	0.76
Hardly Ever/Not often	34 (27.2)	14 (23.3)	19 (32.8)		12 (24.5)	22 (28.9)
Fairly/Very Often	74 (59.2)	41 (68.3)	28 (48.3)		31 (63.3)	43 (56.6)
Prayer: Perceived Effectiveness
Not at all	22 (17.6)	8 (13.3)	13 (22.4)	0.43	6 (12.2)	16 (21.1)	0.10
Not very much	8 (6.4)	4 (6.7)	4 (6.9)		1 (2.0)	7 (9.2)
Some/A great deal	95 (76.0)	48 (80.0)	41 (70.7)		42 (85.7)	53 (69.7)
Social Support: Frequency
Never	10 (8.0)	7 (11.7)	3 (5.2)	0.33	5 (10.2)	5 (6.6)	0.73
Hardly Ever/Not often	57 (45.6)	24 (40.0)	29 (50.0)		21 (42.9)	36 (47.4)
Fairly/Very Often	58 (46.4)	29 (48.3)	26 (44.8)		23 (46.9)	35 (46.1)
Social Support: Perceived Effectiveness
Not at all	16 (12.8)	9 (15.0)	7 (12.1)	0.86	7 (14.3)	9 (11.8)	0.92
Not very much	15 (12.0)	6 (10.0)	7 (12.1)		6 (12.2)	9 (11.8)
Some/A great deal	94 (75.2)	45 (75.0)	44 (75.9)		36 (73.5)	58 (76.3)
Counselor: Frequency
Never	54 (43.2)	23 (38.3)	28 (48.3)	0.05	20 (40.8)	34 (44.7)	0.09
Hardly Ever/Not often	47 (37.6)	20 (33.3)	24 (41.4)		15 (30.6)	32 (42.1)
Fairly/Very Often	24 (19.2)	17 (28.3)	6 (10.3)		14 (28.6)	10 (13.2)
Counselor: Perceived Effectiveness
Not at all	55 (44.0)	24 (40.0)	28 (48.3)	0.54	21 (42.9)	34 (44.7)	0.16
Not very much	9 (7.2)	4 (6.7)	5 (8.6)		1 (2.0)	8 (10.5)
Some/A great deal	61 (48.8)	32 (53.3)	25 (43.1)		27 (55.1)	34 (44.7)
Distraction: Frequency
Never	12 (9.6)	7 (11.7)	5 (8.6)	0.77	3 (6.1)	9 (11.8)	0.23
Hardly Ever/Not often	27 (21.6)	11 (18.3)	13 (22.4)		8 (16.3)	19 (25.0)
Fairly/Very Often	86 (68.8)	42 (70.0)	40 (69.0)		38 (77.6)	48 (63.2)
Distraction: Perceived Effectiveness
Not at all	13 (10.4)	8 (13.3)	5 (8.6)	0.68	4 (8.2)	9 (11.8)	0.42
Not very much	15 (12.0)	7 (11.7)	6 (10.3)		4 (8.2)	11 (14.5)
Some/A great deal	97 (77.6)	45 (75.0)	47 (81.0)		41 (83.7)	56 (73.7)
Meditation: Frequency
Never	52 (41.6)	25 (41.7)	26 (44.8)	0.17	14 (28.6)	38 (50.0)	0.06
Hardly Ever/Not often	40 (32.0)	15 (25.0)	21 (36.2)		19 (38.8)	21 (27.6)
Fairly/Very Often	33 (26.4)	20 (33.3)	11 (19.0)		16 (32.7)	17 (22.4)
Meditation: Perceived Effectiveness
Not at all	55 (44.0)	26 (43.3)	28 (48.3)	0.74	17 (34.7)	38 (50.0)	0.17
Not very much	9 (7.2)	5 (8.3)	3 (5.2)		3 (6.1)	6 (7.9)
Some/A great deal	61 (48.8)	29 (48.3)	27 (46.6)		29 (59.2)	32 (42.1)

*Excludes n=7 participants who reported race/ethnicity other than NHW or AA.

Note. Perceived effectiveness imputed as “Not at all” for those who reported “Never” using a behavior.

As shown by Supplemental Table 1, SRCB generally had either null or positive correlations with each other, indicating that participants who engaged in “health-harming” behaviors were also likely to engage in “health-promoting” ones (i.e., r² =+0.24, p<0.001 for meditation and eating sweet foods). Summing across all behaviors, the correlation between health-harming and health promoting behaviors was +0.33, p<0.001. Supplemental Figure 2 shows the relationship between the SRCB and their proxies (i.e., eating snacks and BMI, using distractions and sedentary time, seeking social support, and network size). It shows that while the SRCB are modestly correlated with their proxies they are not interchangeable.

Turning to the outcome of HPA stress reactivity, Table 3 presents zero-order bivariate correlations between frequency of SRCB and salivary cortisol during the TSST. While the sum totals of the SRCB were not correlated with any aspect of stress reactivity, several individual behaviors were associated with different parts of the TSST response. For example, among the health-promoting behaviors, frequency of mediation was not associated with pre-TSST cortisol, but was significantly associated with a shallower increase, steeper decline, and less total cortisol produced in response to this acute stressor. Among the health-harming behaviors, frequency of eating sweet and snack foods were similarly associated with shallower increase, steeper decline, and less total cortisol. The only behavior consistently positively correlated with cortisol was frequency of alcohol consumption.

Table 3.

Spearman Correlations of Frequency and Effectiveness of SRCBs with Salivary Cortisol during the TSST.

Self-Regulatory Coping Behaviors	Log-transformed Salivary Cortisol Concentrations across 8 Collection Points								AUC
	Pre-TSST		Post-TSST						AUC
	−20 min	Immediately before	Immediately after	+15 min	+30 min	+45 min	+75 min	+90 min	Increase	Total
Sweet Food
Frequency	−0.13	−0.15	−0.19*	−0.16	−0.19*	−0.19*	−0.21*	−0.22*	−0.09	−0.19*
Effectiveness	−0.16	−0.14	−0.18	−0.07	−0.09	−0.12	−0.15	−0.17	0.03	−0.13
Fried Food
Frequency	−0.09	−0.04	−0.005	−0.05	−0.08	−0.08	−0.06	−0.05	0.001	−0.04
Effectiveness	−0.16	−0.05	−0.05	−0.0003	−0.05	−0.07	−0.07	−0.07	0.06	−0.04
Snack Food
Frequency	−0.17	−0.23*	−0.25**	−0.17	−0.21*	−0.20*	−0.19*	−0.16	0.01	−0.20*
Effectiveness	−0.16	−0.16	−0.22*	−0.10	−0.17	−0.17	−0.17	−0.18	−0.01	−0.16
Alcohol
Frequency	0.03	0.14	0.15	0.11	0.12	0.13	0.09	0.21*	0.08	0.14
Effectiveness	−0.01	0.06	0.08	−0.01	−0.003	0.01	−0.01	0.08	−0.02	0.02
Smoking
Frequency	0.09	0.01	0.003	−0.06	−0.12	−0.13	−0.07	−0.01	−0.22*	−0.06
Effectiveness	0.10	0.05	0.04	−0.03	−0.09	−0.10	−0.04	−0.01	−0.18	−0.03
Exercise
Frequency	−0.02	−0.04	−0.05	−0.09	−0.07	−0.12	−0.12	−0.17	−0.11	−0.09
Effectiveness	−0.07	−0.08	−0.08	−0.11	−0.09	−0.12	−0.10	−0.15	−0.10	−0.10
Prayer
Frequency	−0.07	−0.04	−0.08	0.02	0.02	0.01	0.08	0.04	0.04	0.03
Effectiveness	−0.06	−0.03	−0.04	0.03	0.04	0.02	0.09	0.08	−0.01	0.04
Social Support
Frequency	−0.06	0.02	0.08	0.005	0.02	0.02	−0.03	−0.05	0.15	0.02
Effectiveness	−0.15	−0.12	−0.06	−0.03	−0.03	−0.05	−0.08	−0.06	0.17	−0.05
Counselor
Frequency	0.13	0.11	0.08	0.02	0.0004	0.001	0.01	0.04	−0.03	0.04
Effectiveness	0.05	0.02	−0.01	−0.05	−0.05	−0.07	−0.05	0.01	−0.06	−0.03
Distractions
Frequency	0.05	0.04	−0.01	−0.06	−0.08	−0.09	−0.07	−0.12	−0.12	−0.06
Effectiveness	0.11	0.02	−0.002	−0.06	−0.03	−0.05	−0.04	−0.06	−0.16	−0.03
Meditation
Frequency	−0.07	−0.08	−0.16	−0.24**	−0.26**	−0.22*	−0.21*	−0.20*	−0.27**	−0.21*
Effectiveness	−0.05	−0.04	−0.12	−0.18*	−0.17	−0.15	−0.15	−0.13	−0.20*	−0.14
Summations
Total	−0.06	−0.05	−0.09	−0.14	−0.17	−0.18	−0.16	−0.13	−0.11	−0.13
Health-promoting	−0.01	−0.02	−0.03	−0.11	−0.11	−0.12	−0.10	−0.12	−0.12	−0.08
Health-harming	−0.09	−0.09	−0.09	−0.10	−0.15	−0.15	−0.14	−0.08	−0.06	−0.12

=p<0.05, **=p<0.01. AUC: Total Area Under the TSST Response Curve. AUC Increase: Area Under the Curve relative to the Increase.

Frequency: higher value denotes more frequent use of behavior: Coded as 0=Never, 1=Hardly ever/Not often, 2=Fairly often/Very often. Perceived Effectiveness: higher value denotes higher effectiveness: Coded as 0=Not at all, 1=Not very much, 2=Some/A great deal.

We then examined the behaviors individually. The panels of Supplemental Figure 3 show salivary cortisol response to the TSST as a function of the 11 individual SRCB and for 7 proxy health behaviors and characteristics (e.g., smoking, alcohol use, exercise, sedentary activities, social network size, adiposity indicators) and Table 4 provides the estimates for each panel presented as percent change. In some cases, the self-regulatory and health behaviors had similar relationships with HPA stress reactivity. For example, frequently smoking to cope with stress was associated with a 13% shallower increase in HPA reactivity compared to those who never smoke to cope (−13.0%, 95%CI: −20.9% to −4.3%, p<0.006). Similarly, current smokers had a shallower increase in reactivity (−14.5%, 95%CI: −22.8% to −5.3%, p<0.003) and a shallower rate of decrease compared to never smokers (5.6%, 95%CI: 0.6 to 10.8, p<0.03). For alcohol, while the association was not statistically significant for the SRCB, the pattern was similar as for drinking behavior more generally: heavy alcohol use was associated with higher baseline cortisol (65.2%, 95%CI: −0.2%–173.3%, p=0.05) and steeper rate of decrease (−6.6%, −12.9%–0.03%, p=0.05) compared to never/former drinkers. Frequent use of distractions as an SRCB was associated with 71% higher baseline cortisol (70.6%, 95%CI: 3.1%–182.3%, p<0.04) as compared to infrequent use; similarly, participants who reported the most sedentary time at home at had 51% higher baseline cortisol (51.2%, 95%CI: 5.2–117.2%, p<0.03) compared to the least sedentary. Exercise, whether as an SRCB or health behavior, was not associated with HPA stress reactivity. None of the eating-related SRCB, whether snacks, fried or sweet foods, nor their proxies of BMI and waist circumference, were significantly associated with HPA reactivity in the adjusted models.

Table 4.

Model estimates for the panels of Supplemental Figure 3 of the relationships between individual SRCBs and proxy behaviors/characteristics with HPA-axis stress reactivity.

	Level of behavior*
	Fairly/Very Often	Hardly Ever/Not Often % difference (95% CI)	p -value	Never % difference (95% CI)	p -value
Smoking as an SRCB (Panel A)
Baseline	[ref.]	−21.4 (−60.0, 54.5)	0.46	−12.7 (−43.8, 35.7)	0.53
Rate of Increase	[ref.]	5.4 (−12.5, 26.9)	0.56	14.9 (4.5, 26.4 )	0.01
Rate of Decrease	[ref.]	−4.1 (−11.3, 3.6)	0.26	−4.0 (−9.0, 1.1)	0.12
Smoking status (Panel B)	Never	Past		Current
Baseline	[ref.]	−0.1 (−24.7, 32.4)	0.99	13.1 (−23.8, 68.0)	0.53
Rate of Increase	[ref.]	−4.0 (−13.6, 6.7)	0.45	−14.5 (−22.8, −5.3)	0.003
Rate of Decrease	[ref.]	0.7 (−3.7, 5.3)	0.76	5.6 (0.6, 10.8)	0.03
Alcohol Use as an SRCB (Panel C)
Baseline	[ref.]	−46.9 (−68.4, −10.9)	0.02	−39.6 (−63.1, −1.1)	0.05
Rate of Increase	[ref.]	5.9 (−12.1, 27.5)	0.53	−2.0 (−16.3, 14.8)	0.80
Rate of Decrease	[ref.]	−2.2 (−11.1, 7.7)	0.64	−1.7 (−10.4, 7.8)	0.69
Alcohol Use (Panel D)	Never/Past	Moderate/Occasional		Heavy/Frequent
Baseline	[ref.]	0.4 (−23.3, 31.4)	0.98	65.2 (−0.2, 173.3)	0.05
Rate of Increase	[ref.]	2.4 (−6.4, 12.1)	0.60	6.5 (−7.2, 22.3)	0.35
Rate of Decrease	[ref.]	−1.1 (−5.2, 3.1)	0.60	−6.6 (−12.9, 0.03)	0.05
Exercise as an SRCB (Panel E)
Baseline	[ref.]	−20.8 (−40.0, 4.5)	0.10	7.6 (−25.5, 55.4)	0.69
Rate of Increase	[ref.]	6.0 (−4.7, 18.0)	0.28	3.0 (−6.7, 13.8)	0.55
Rate of Decrease	[ref.]	2.7 (−1.8, 7.4)	0.24	4.7 (−0.2, 9.8)	0.06
Exercise as a behavior (Panel F)	None	1-3 times/week		>3 times/week
Baseline	[ref.]	−15. (−29.1, 34.9)	0.92	13.5 (−17.7, 56.6)	0.43
Rate of Increase	[ref.]	8.1 (−3.5, 21.1)	0.18	−2.1 (−11.4, 8.2)	0.68
Rate of Decrease	[ref.]	−1.4 (−6.2, 3.7)	0.59	−1.2 (−5.2, 3.0)	0.57
Sedentary activities as an SRCB (Panel G)
Baseline	[ref.]	11.9 (−18.5, 53.7)	0.48	−41.4 (−64.6, −3.0)	0.04
Rate of Increase	[ref.]	8.5 (−4.7, 23.5)	0.21	6.9 (−10.7, 28.0)	0.44
Rate of Decrease	[ref.]	0.1 (−4.7, 0.02)	0.98	4.32 (−2.4, 11.6)	0.20
Sedentary time at home (Panel H)	Tertile 1	Tertile 2		Tertile 3
Baseline	[ref.]	25.5 (−5.4, 66.6)	0.11	51.2 (5.2, 117.2)	0.03
Rate of Increase	[ref.]	−12.9 (−22.3, −2.3)	0.02	−20.5 (−29.4, −10.6)	<0.001
Rate of Decrease	[ref.]	2.5 (−2.2, 7.3)	0.30	7.1 (1.7, 12.7)	0.01
Social Support as an SRCB (Panel I)
Baseline	[ref.]	−4.0 (−25.6, 23.8)	0.76	47.9 (−23.2, 204.8)	0.26
Rate of Increase	[ref.]	−5.2 (−13.4, 3.7)	0.24	−10.0 (−26.8, 10.7)	0.29
Rate of Decrease	[ref.]	1.5 (−2.5, 5.6)	0.47	6.2 (−2.4, 15.6)	0.15
Social Network—Family (Panel J)	0–3	4–5		6+
Baseline	[ref.]	−40.9 (−56.8, −18.9)	0.002	−0.1 (−25.3, 33.6)	0.99
Rate of Increase	[ref.]	8.3 (−2.9, 20.8)	0.15	6.5 (−3.6, 17.6)	0.21
Rate of Decrease	[ref.]	−0.69 (−6.0, 4.9)	0.80	−0.19 (−4.4, 4.2)	0.93
Social Network—Friends (Panel K)	0–3	4–5		6+
Baseline	[ref.]	6.3 (−25.9, 52.6)	0.74	4.6 (−23.9, 43.8)	0.78
Rate of Increase	[ref.]	6.4 (−3.2, 16.9)	0.20	12.7 (1.5, 25.1)	0.03
Rate of Decrease	[ref.]	−5.0 (−9.2, −0.6)	0.03	−3.9 (−8.4, 0.7)	0.09
Sweet Food as an SRCB (Panel L)
Baseline	[ref.]	−7.0 (−31.1, 25.6)	0.63	19.9 (−14.3, 67.7)	0.28
Rate of Increase	[ref.]	8.7 (−1,3, 19.6)	0.09	2.7 (−8.2, 14.8)	0.64
Rate of Decrease	[ref.]	0.5 (−4.1, 5.4)	0.83	1.8 (−3.6, 7.6)	0.51
Fried Food as an SRCB (Panel M)
Baseline	[ref.]	−7.0 (−40.1, 44.2)	0.73	12.5 (−26.5, 72.3)	0.58
Rate of Increase	[ref.]	7.5 (−5.8, 22.7)	0.27	3.6 (−8.8, 17.6)	0.58
Rate of Decrease	[ref.]	−2.4 (−9.1, 4.7)	0.48	−2.2 (−8.7, 4.7)	0.51
Snack Food as an SRCB (Panel N)
Baseline	[ref.]	18.6 (−11.7, 59.2)	0.25	30.2 (−6.5, 81.2)	0.12
Rate of Increase	[ref.]	6.4 (−3.9, 17.9)	0.23	3.7 (−6.2, 14.6)	0.47
Rate of Decrease	[ref.]	−5.4 (−10.0, −0.4)	0.03	−3.6 (−8.0, 1.0)	0.13
BMI (Panel O)	Tertile 1	Tertile 2		Tertile 3
Baseline	[ref.]	−9.7 (−35.4, 26.3)	0.55	−17.9 (−46.2, 25.4)	0.36
Rate of Increase	[ref.]	7.5 (−2.7, 18.7)	0.15	0.5 (−10.9, 13.3)	0.94
Rate of Decrease	[ref.]	1.3 (−3.6, 6.4)	0.60	3.2 (−2.9, 9.8)	0.30
Waist Circumference (Panel P)	Tertile 1	Tertile 2		Tertile 3
Baseline	[ref.]	−15.6 (−39.6, 18.0)	0.32	−22.1 (−46.1, 12.6)	0.18
Rate of Increase	[ref.]	9.4 (−0.6, 20.3)	0.06	8.0 (−7.4, 2.6)	0.20
Rate of Decrease	[ref.]	−2.5 (−7.4, 2.6)	0.33	−2.4 (−8.1, 3.7)	0.42
Prayer as an SRCB (Panel Q)
Baseline	[ref.]	11.3 (−19.2, 53.3)	0.51	20.3 (−19.6, 80.1)	0.36
Rate of Increase	[ref.]	−6.0 (−15.3, 4.4)	0.25	−1.7 (−12.2, 10.0)	0.75
Rate of Decrease	[ref.]	−0.4 (−4.5, 4.0)	0.86	−0.4 (−6.1, 5.7)	0.90
Counselor as an SRCB (Panel R)
Baseline	[ref.]	−18.1 (−41.8, 15.2)	0.25	−17.9 (−40.2, 12.8)	0.22
Rate of Increase	[ref.]	4.6 (−6.2, 16.7)	0.41	1.9 (−4.2, 17.8)	0.24
Rate of Decrease	[ref.]	1.9 (−3.6, 7.7)	0.50	−0.2 (−4.9, 4.8)	0.95
Meditation as an SRCB (Panel S)
Baseline	[ref.]	18.5 (−17.4, 70.0)	0.35	12.2 (−20.2, 57.6)	0.50
Rate of Increase	[ref.]	4.0 (−5.9, 14.9)	0.44	16.3 (5.5, 28.3)	0.003
Rate of Decrease	[ref.]	−0.2 (−5.3, 5.2)	0.94	−4.1 (−8.2, 0.1)	0.05

*The levels of all SRCBs are Fairly often/Very often (reference) versus Hardly ever/Not often versus Never. The levels of each related health behavior/characteristics are specified with that behavior and are operationalized in Supplemental Table 2.

To aid in interpretation, values in the table are shown as percent difference, and values in Supplemental Figure 2 are shown as log-transformed salivary cortisol. All estimates are adjusted for age, sex, educational attainment, marital status, and neighborhood SES.

Of those behaviors that had proxies, social support was the only instance where the SRCB was unrelated to stress reactivity, but the proxy was. Compared to those with a small family network, participants with the largest networks had 40% lower cortisol at baseline (−39.6%, 95%CI: −56.5% to −16.3%, p<0.003). Friend network size was unrelated to baseline cortisol but was associated with the TSST response: those with the largest friend networks had a higher rate of increase (14.9%, 95%CI: 3.3%–27.8%, p<0.01) and steeper decline (−3.9%, 95% CI: −8.5%–0.9%, p=0.11) than those with the smallest networks. For the three SRCB that did not have equivalents/proxies (e.g., Supplemental Figure 3 Panels Q to S: prayer, counselor/professional help, and meditation), only meditation was associated with HPA reactivity. Compared to those who never use meditation, participants who most frequently use meditation had a 14% shallower increase (−14.0%, 95%CI: −22.0% to −5.1%, p<0.003); estimates were similar but more modest for occasionally using meditation.

Discussion

This study explored the relationship between a range of SRCB and HPA stress reactivity in a racially and economically diverse sample of adults at elevated risk of developing T2D. The primary findings are four-fold: First, while not traditionally conceptualized as a means of self-regulating emotional distress, health behaviors related to eating, smoking, alcohol, and exercise are commonly reported as efforts to cope after a stressful experience. The frequency of engaging in SRCB does not differ substantially by race, with modest variation by neighborhood SES. Second, “health-harming” and “health-promoting” SRCB were positively correlated, indicating that individuals engage in a diverse set of behaviors. Third, while the summation indicators of SRCB were not associated with HPA stress reactivity, several individual SRCB, both “health-promoting” and “health-harming,” for example, both smoking and meditating, were associated with shallower increases in cortisol (i.e., weaker response) to the TSST. Finally, the analysis of proxy health behaviors and characteristics, assessed outside the context of stress coping, shows that the associations between the SRCB and HPA reactivity are broadly consistent, suggesting that it is the behavior itself, rather than the self-identified use of that behavior to cope with stress, that underlies the relationship between these indicators of self-regulation and HPA stress reactivity.

Consistent with prior work, engagement in “health-promoting” behaviors was positively correlated with engagement in “health-harming” ones. Such behaviors can be mutually reinforcing, as individuals who engage in more health-promoting behaviors may be doing so to compensate for past health-harming behaviors (Knäuper et al., 2004), and vice-versa as individuals who engage in more health-promoting behaviors may feel more comfortable engaging in health-harming behaviors (Mullen & Monin, 2016). Moreover, the null findings between the aggregate SRCB indicators and HPA reactivity are consistent with a recent longitudinal study that assessed the day-to-day relationships among promoting/harming behaviors with cortisol and found that positive behaviors, such as healthy diet and exercise, were either only weakly or unrelated to cortisol responses (Strahler et al., 2021). This suggests that the nature of the association between health-promoting behaviors (i.e., healthy diet, exercise, meditation) and HPA-axis functioning is better understood as one of a threshold, rather than dose-response, relationship (Pascoe et al., 2017). As a result, it may be more difficult to identify significant effects of such behaviors on neurobiology as they may only emerge after continuous and sustained engagement.

The EA Model situates social disparities within the Life Course Framework, noting that the relationships between stress, behavior, and health vary over time (Ben-Shlomo & Kuh, 2002). Timescale is also fundamental to the “strong” hypothesis of the EA Model regarding the consequences of stress and SRCB for mental versus physical health outcomes. The Model hypotheses that SRCB serve as a means to preserve short-term mental health in situations of chronic stress and few coping resources; while the exact time scale is still unclear, experimental evidence of stress reactivity and “health-harming” behaviors (i.e., consuming high fat/sugar foods, using nicotine or other drugs) suggests it may be on the order of hours, days, or weeks (Koob, 2008; Margittai et al., 2016). In contrast, the long-term physical health consequences of engaging in SRCB, whether those are “health-promoting” (e.g., exercise and meditation) or “health-harming” (e.g., tobacco and alcohol), are on the timescale of decades. Over those decades, not only do “health-harming” SRCB cause morbidity (i.e., diabetes, cardiovascular disease, cancer), but those morbidities in turn also negatively impact mental health. Drawing on the addiction perspective, it is likely that the relative effectiveness of SRCB to preserve mental health declines over time, much like the development of tolerance for psychoactive substances (Koob & Volkow, 2016). As such, over the life course, for social groups with higher exposure to chronic stress and ready access to health-harming behaviors, the stress-coping relationship changes from a zero-sum game concerning mental versus physical health to a lose-lose one.

As the name indicates, the EA Model provides a framework for generating hypotheses of how the connections between stress and self-regulatory behaviors, over the lifespan, contribute to disparities in mental and physical health as a function of factors like race, SES, and place. While not the focus of this analysis, it is important to be explicit as to the fundamental origin of these disparities: there is nothing biologically intrinsic about race, ethnicity, or social positioning as a cause of health disparities. Rather, because of systemic factors like racism, and discriminatory policies in political participation, housing, education, wealth distribution, criminal justice, etc., minority groups have been historically marginalized and disenfranchised (Williams et al., 2019). The legacy of these policies is evident today in neighborhood and school segregation, employment and wage gaps, mass incarceration, and access to quality healthcare. Leaders in the field have called for more structural, rather than individualistic, approaches to understanding race and racism as determinants of health (Neblett Jr., 2019). Given this fundamental origin of racial health inequities, what is the role of the neurobiological stress response in furthering our scientific understanding? As we have argued previously: “We believe that race, rather than a category of thingness, is actually a category of experiences... it is the accumulation of differential exposures and experiences associated with these categorizations that produces racial disparities in health over the life course” (Mezuk et al., 2013). It is from this foundation that the EA Model seeks to understand how adversities tethered to racial group membership and SES serve as conduits for life experiences, including those that involve the stress response.

Limitations and strengths. While the RSASS sampling frame was designed to decompose “race” and “place,” as a result it is non-representative (the Richmond, VA metropolitan area is approximately 57% NHW, 29% AA) (US Census Bureau, 2019) and lacks adequate representation of all racial/ethnic minority groups. While we sought to balance race groups on area-level SES, because of residential segregation this does not mean the contextual environments within stratum of neighborhood SES are equivalent for NHW and AA (Massey & Rugh, 2014). This analysis was limited to adults in middle-age, and future research on younger populations when health behaviors onset is needed to understand how these relationships emerge and change over the life course. This is particularly important given racial and SES differences in development of obesity in childhood and subsequent onset of T2D. This analysis tested multiple predictors of HPA reactivity, and the robustness of these findings needs to be confirmed in future research. Strengths of this study include direct measurement of the HPA-axis response to an acute stressor, a mechanism heretofore only hypothesized as central to the EA Model. Another strength is the comparison of many SRCBs to their equivalent health behaviors, of which only the latter have been widely used in tests of the EA Model to date.

These findings have implications for promoting health behavior change for adults at high risk of T2D like the participants in RSASS. Consistent with prior research (Nandi et al., 2014; Petrovic et al., 2018), participants from lower SES neighborhoods had higher BMI, were more likely to smoke, drink alcohol more frequently, spent more time sedentary, and were less likely to exercise, even if these differences were not all statistically significant. In contrast, we found that there was less variation in SRCB as a function of race or neighborhood SES. Rather, all groups, NHW and AA, and lower and higher SES, reported engaging in aggregate “health-harming” and “health-promoting” SRCB to similar degrees, even if they differed in specific behaviors. However, as posited by social science theory more broadly, norms and contextual factors influence behaviors, and thus these findings do not exclude culturally normative and nuanced reasons for the observed differences between specific behaviors (e.g., the salience of prayer as a coping mechanism among AA (Glover et al., 2019)).

When considering these findings in relation to individual health behavior change, it is important to consider the systemic and contextual factors that influence the options (affordances) within and across social groups. These findings, when viewed within the context of the SOBC Initiative, call for behavioral interventions that address how race, place, and SES combine to shape self-regulatory behaviors. Proposed interventions should incorporate indicators of stress response systems and seek to address the ways in which self-regulatory behaviors may be deterred or enhanced by cultural, environmental, social, and interpersonal factors.

Footnotes

Acknowledgments

This work would not have been possible without the mentorship of Dr. James S. Jackson. We are grateful to have been able to learn from him, and he is deeply missed. The authors would like to acknowledge the intellectual contributions of the SNOHD working group members who helped elaborate the EA Model over the years, particularly Jamie Abelson.

Declaration of conflicting interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Richmond Stress and Sugar Study is supported by the American Diabetes Association (1-16-ICTS-082, to Mezuk). Additional support for this manuscript provided by the National Institute of Health through the Michigan Integrative Well-Being and Inequality Training program (R25-AT010664, to Mezuk) and the Michigan Center for Urban African American Aging Research (P30AG015281-21).

Author’s Note

For submission to the Journal of Aging and Health, Special issue honoring the work of Dr. James S. Jackson.

ORCID iD

Briana Mezuk

References

Allen

A. P.

Kennedy

P. J.

Dockray

Cryan

J. F.

Dinan

T. G.

Clarke

(2017). The trier social stress test: Principles and practice. Neurobiology of Stress, 6(February), 113–126. https://doi.org/10.1016/j.ynstr.2016.11.001

Barrington

W. E.

Stafford

Hamer

Beresford

S. A. A.

Koepsell

Steptoe

(2014). Neighborhood socioeconomic deprivation, perceived neighborhood factors, and cortisol responses to induced stress among healthy adults. Health & Place, 27(May), 120–126. https://doi.org/10.1016/j.healthplace.2014.02.001

Ben-Shlomo

Kuh

(2002). A life course approach to chronic disease epidemiology: Conceptual models, empirical challenges and interdisciplinary perspectives. International Journal of Epidemiology, 31(2), 285–293. https://doi.org/10.1093/ije/31.2.285

Boardman

J. D.

Alexander

K. B.

(2011). Stress trajectories, health behaviors, and the mental health of black and white young adults. Social Science & Medicine, 72(10), 1659–1666. https://doi.org/10.1016/j.socscimed.2011.03.024

Castro-Diehl

Diez Roux

A. V.

Seeman

Shea

Shrager

Tadros

(2014). Associations of socioeconomic and psychosocial factors with urinary measures of cortisol and catecholamines in the Multi-Ethnic Study of Atherosclerosis (MESA). Psychoneuroendocrinology, 41(March), 132–141. https://doi.org/10.1016/j.psyneuen.2013.12.013

Chida

Hamer

(2008). Chronic psychosocial factors and acute physiological responses to laboratory-induced stress in healthy populations: A quantitative review of 30 years of investigations. Psychological Bulletin, 134(6), 829–885. https://doi.org/10.1037/a0013342

D. P.

Diez Roux

A. V.

Hajat

Auchincloss

A. H.

Merkin

S. S.

Ranjit

Shea

Seeman

(2011). Circadian rhythm of cortisol and neighborhood characteristics in a population-based sample: The multi-ethnic study of atherosclerosis. Health & Place, 17(2), 625–632. https://doi.org/10.1016/j.healthplace.2010.12.019

Foster

M. T.

Warne

J. P.

Ginsberg

A. B.

Horneman

H. F.

Pecoraro

N. C.

Akana

S. F.

Dallman

M. F.

(2009). Palatable foods, stress, and energy stores sculpt corticotropin-releasing factor, adrenocorticotropin, and corticosterone concentrations after restraint. Endocrinology, 150(5), 2325–2333. https://doi.org/10.1210/en.2008-1426

Glover

L. M.

Bertoni

A. G.

Golden

S. H.

Baltrus

Min

Y.-I.

Carnethon

M. R.

Taylor

Sims

(2019). Sex differences in the association of psychosocial resources with prevalent type 2 diabetes among African Americans: The Jackson heart study. Journal of Diabetes and Its Complications, 33(2), 113–117. https://doi.org/10.1016/j.jdiacomp.2018.11.005

10.

Jackson

J. S.

Knight

K. M.

(2006). Race and self-regulatory health behaviors: The role of the stress response and the HPA axis in physical and mental health disparities. In Schaie

K. W.

Cartensen

(Eds.), Social structures, aging and self-regulation in the elderly (pp. 189–207). Springer.

11.

Jackson

J. S.

Knight

K. M.

Rafferty

J. A.

(2010). Race and unhealthy behaviors: Chronic stress, the HPA axis, and physical and mental health disparities over the life course. American Journal of Public Health, 100(5), 933–939. https://doi.org/10.2105/AJPH.2008.143446

12.

Keller-Allen

(2014). Science of behavior change Common Fund program fourth annual meeting of investigators (p. 70). https://commonfund.nih.gov/sites/default/files/FINAL_SOBC_June2014_Annual_mtg_report_REV_10-15-14_508.pdf

13.

Keyes

K. M.

Barnes

D. M.

Bates

L. M.

(2011). Stress, coping, and depression: Testing a new hypothesis in a prospectively studied general population sample of U.S.-born Whites and Blacks. Social Science & Medicine, 72(5), 650–659. https://doi.org/10.1016/j.socscimed.2010.12.005

14.

Knäuper

Rabiau

Cohen

Patriciu

(2004). Compensatory health beliefs: Scale development and psychometric properties. Psychology & Health, 19(5), 607–624. https://doi.org/10.1080/0887044042000196737

15.

Koob

G. F.

(2008). A role for brain stress systems in addiction. Neuron, 59(1), 11–34. https://doi.org/10.1016/j.neuron.2008.06.012

16.

Koob

G. F.

Volkow

N. D.

(2016). Neurobiology of addiction: A neurocircuitry analysis. Lancet Psychiatry, 3(8), 760–773. https://doi.org/10.1016/S2215-0366(16)00104-8

17.

Margittai

Nave

Strombach

van Wingerden

Schwabe

Kalenscher

(2016). Exogenous cortisol causes a shift from deliberative to intuitive thinking. Psychoneuroendocrinology, 64(February), 131–135. https://doi.org/10.1016/j.psyneuen.2015.11.018

18.

Massey

D. S.

Rugh

J. S.

(2014). Segregation in post-civil rights America: Stalled integration or end of the segregated century? Du Bois Review : Social Science Research on Race, 11(2), 205–232. https://doi.org/10.1017/S1742058X13000180

19.

McNeish

(2017). Small sample methods for multilevel modeling: A colloquial elucidation of REML and the Kenward-Roger correction. Multivariate Behavioral Research, 52(5), 661–670. https://doi.org/10.1080/00273171.2017.1344538

20.

Meier

H. C. S.

Hussein

Needham

Barber

Lin

Seeman

Diez Roux

(2019). Cellular response to chronic psychosocial stress: Ten-year longitudinal changes in telomere length in the multi-ethnic study of atherosclerosis. Psychoneuroendocrinology, 107(September), 70–81. https://doi.org/10.1016/j.psyneuen.2019.04.018

21.

Mezuk

Abdou

C. M.

Hudson

Kershaw

K. N.

Rafferty

J. A.

Lee

Jackson

J. S.

(2013). “White box” epidemiology and the social neuroscience of health behaviors: The environmental affordances model. Society and Mental Health, 3(2), 79–95. https://doi.org/10.1177/2156869313480892

22.

Mezuk

Del Toro

Gough Courtney

Kirk

K. F.

Zhang

Spears

E. C.

Green

Lee

Hudson

(2021). Theory and empiricism: A comment on “Interrogating the environmental affordances model” by Pamplin and colleagues. Social Science & Medicine, 285(September), 114281. https://doi.org/10.1016/j.socscimed.2021.114281

23.

Mezuk

Lexima

Kalesnikava

V. A.

Fleming

Montgomery

Tuktur

Winston

Perrin

P. B.

Green

Wheeler

D. C.

(2020). Stress reactivity as a contributor to racial and socioeconomic disparities: Rationale and baseline results from the richmond stress and sugar study. Psychosomatic Medicine, 82(7), 658–668. https://doi.org/10.1097/PSY.0000000000000830

24.

Mezuk

Rafferty

J. A.

Kershaw

K. N.

Hudson

Abdou

C. M.

Lee

Eaton

W. W.

Jackson

J. S.

(2010). Reconsidering the role of social disadvantage in physical and mental health: Stressful life events, health behaviors, race, and depression. American Journal of Epidemiology, 172(11), 1238–1249. https://doi.org/10.1093/aje/kwq283

25.

Mezuk

Ratliff

Concha

J. B.

Abdou

C. M.

Rafferty

Lee

Jackson

J. S.

(2017). Stress, self-regulation, and context: Evidence from the Health and Retirement Survey. SSM - Population Health, 3(December), 455–463. https://doi.org/10.1016/j.ssmph.2017.05.004

26.

Miller

Chen

Cole

S. W.

(2009). Health psychology: Developing biologically plausible models linking the social world and physical health. Annual Review of Psychology, 60, 501–524. https://doi.org/10.1146/annurev.psych.60.110707.163551

27.

Mullen

Monin

(2016). Consistency versus licensing effects of past moral behavior. Annual Review of Psychology, 67, 363–385. https://doi.org/10.1146/annurev-psych-010213-115120

28.

Nandi

Glymour

M. M.

Subramanian

S. V.

(2014). Association among socioeconomic status, health behaviors, and all-cause mortality in the United States. Epidemiology, 25(2), 170–177. https://doi.org/10.1097/EDE.0000000000000038

29.

Neblett

E. W.

Jr. (2019). Racism and health: Challenges and future directions in behavioral and psychological research. Cultural Diversity and Ethnic Minority Psychology, 25(1), 12–20. https://doi.org/10.1037/cdp0000253

30.

Nielsen

Riddle

King

J. W.

NIH Science of Behavior Change Implementation Team Aklin

W. M.

Chen

Clark

Collier

Czajkowski

Esposito

Ferrer

Green

Hunter

Kehl

King

Onken

Simmons

J. M.

Stoeckel

Stoney

Weber

(2018). The NIH science of behavior change program: Transforming the science through a focus on mechanisms of change. Behaviour Research and Therapy, 101(February), 3–11. https://doi.org/10.1016/j.brat.2017.07.002

31.

Nusslock

Miller

G. E.

(2016). Early-life adversity and physical and emotional health across the lifespan: A neuroimmune network hypothesis. Biological Psychiatry, 80(1), 23–32. https://doi.org/10.1016/j.biopsych.2015.05.017

32.

Pascoe

M. C.

Thompson

D. R.

Jenkins

Z. M.

Ski

C. F.

(2017). Mindfulness mediates the physiological markers of stress: Systematic review and meta-analysis. Journal of Psychiatric Research, 95(December), 156–178. https://doi.org/10.1016/j.jpsychires.2017.08.004

33.

Petrovic

de Mestral

Bochud

Bartley

Kivimäki

Vineis

Mackenbach

Stringhini

(2018). The contribution of health behaviors to socioeconomic inequalities in health: A systematic review. Preventive Medicine, 113(August), 15–31. https://doi.org/10.1016/j.ypmed.2018.05.003

34.

Schulz

A. J.

Kannan

Dvonch

J. T.

Israel

B. A.

Allen

3rd James

S. A.

House

J. S.

Lepkowski

(2005). Social and physical environments and disparities in risk for cardiovascular disease: The healthy environments partnership conceptual model. Environmental Health Perspectives, 113(12), 1817–1825. https://doi.org/10.1289/ehp.7913

35.

Strahler

Wurst

Fuchs

Wunsch

(2021). Joint associations of regular exercise and healthy diet with psychobiological stress reactivity in a healthy male sample. Stress, 24(6), 1–14. https://doi.org/10.1080/10253890.2021.1878496

36.

Thoits

P. A.

(2010). Stress and health: Major findings and policy implications. Journal of Health and Social Behavior, 51(Suppl), S41–S53. https://doi.org/10.1177/0022146510383499

37.

Tryon

M. S.

Stanhope

K. L.

Epel

E. S.

Mason

A. E.

Brown

Medici

Havel

P. J.

Laugero

K. D.

(2015). Excessive sugar consumption may be a difficult habit to break: A view from the brain and body. The Journal of Clinical Endocrinology and Metabolism, 100(6), 2239–2247. https://doi.org/10.1210/jc.2014-4353

38.

Turner

R. J.

Wheaton

Lloyd

D. A.

(1995). The epidemiology of social stress. American Sociological Review, 60(1), 104. https://doi.org/10.2307/2096348

39.

US Census Bureau . (2019). Census profile: Richmond, VA metro area. Census Reporter. http://censusreporter.org/profiles/31000US40060-richmond-va-metro-area/

40.

Wells

Tremblay

P. F.

Flynn

Russell

Kennedy

Rehm

Van Uum

Koren

Graham

(2014). Associations of hair cortisol concentration with self-reported measures of stress and mental health-related factors in a pooled database of diverse community samples. Stress (Amsterdam, Netherlands), 17(4), 334–342. https://doi.org/10.3109/10253890.2014.930432

41.

Williams

D. R.

Lawrence

J. A.

Davis

B. A.

(2019). Racism and Health: Evidence and needed research. Annual Review of Public Health, 40, 105–125. https://doi.org/10.1146/annurev-publhealth-040218-043750