Behavioral reactivity to acute stress among Black and White women with type 2 diabetes: The roles of income and racial discrimination

Abstract

This study investigated relationships of income and self-reported racial discrimination to diabetes health behaviors following an acute stressor. A total of 77 diabetic women (51% Black, 49% White) completed a laboratory public speaking stressor. That evening, participants reported same-day eating, alcohol consumption, and medication adherence; physical activity was measured with actigraphy, and the next morning participants reported sleep quality. Measures were repeated on a counterbalanced control day. There was no mean level difference in health behaviors between stressor and control days. On stressor day, lower income predicted lower physical activity, sleep quality, and medication adherence, and higher racial discrimination predicted more eating and alcohol consumed, even after accounting confounders including race and control day behaviors.

Keywords

chronic illness diversity health behavior lifestyle race

Introduction

Low socioeconomic status (SES) and exposure to racial discrimination are stressors that are adversely associated with health outcomes, particularly stress-related chronic diseases. Diabetes is a case in point. Low SES is linked to higher prevalence of diabetes (Connolly et al., 2000), worse diabetes health status (Gary-Webb et al., 2011) and increased diabetes-related mortality (Saydah and Lochner, 2010). Recent evidence suggests that racial discrimination may be associated with risk factors for diabetes such as obesity (Cozier et al., 2014) as well as worse metabolic control (Wagner et al., 2013b, 2015), autonomic functioning (Wagner et al., 2015) and cognitive decline (Crowe et al., 2010) among those with extant diabetes.

The deleterious effects of low SES and racial discrimination on long-term health outcomes may be explained, in part, by their effects on health behaviors. It has long been demonstrated that low-SES environments are adversely associated with mean levels of health behaviors (Barr et al., 2002; Colgan et al., 2004; Goldman and Smith, 2002; Relton et al., 2005; Virtanen et al., 2006). Yet, these associations cannot be simply attributed to financial constraints on a healthy lifestyle. Whereas it is true that low SES is associated with avoidance of healthy behaviors that incur costs (e.g. eating fruits and vegetables, adhering to medication), it is also true that low SES is associated with participation in unhealthy behaviors that incur costs (e.g. smoking; Nettle, 2010). Cost-neutral behaviors, such as physical activity, are also negatively affected by low SES. For example, in Canada where access to medical screening is universal, socioeconomic disparities in health screenings still exist (Vikky et al., 2006). Thus, characteristics of low-SES environments other than lack of financial resources per se must play a role in shaping health behaviors.

More recently, exposure to racial discrimination has also been associated with lower levels of healthy behaviors. Whereas findings are not entirely consistent, studies have linked higher racial discrimination to less healthy eating (Brodish et al., 2011), lower use of preventive screening procedures (Mouton et al., 2010) and poorer sleep quality and quantity (Grandner et al., 2012; Hicken et al., 2013). There is also some evidence that exposure to racial discrimination is associated with increased appetitive behaviors such as alcohol consumption (Yen et al., 1999), smoking (Guthrie et al., 2002) and other substance use (Brodish et al., 2011), possibly by reducing self-control (Gibbons et al., 2012).

Low-SES environments and racial discrimination may also make individuals more vulnerable to the deleterious effects of acute stressors, that is, these contextual factors may increase stress reactivity. Most studies examining stress reactivity have examined short-term physiological or emotional reactivity. For example, low SES has been associated with higher blood pressure reactivity (Wilson et al., 2000), higher amygdala activation (Phillips, 2011) and greater hostile responses to laboratory stressors (Krause et al., 2011). Similarly, discrimination has been associated with elevated blood pressure reactivity (Guyll et al., 2001; Thomas et al., 2006), endothelial reactivity (Wagner et al., 2015) and autonomic reactivity (Wagner et al., 2013, 2013a) in response to acute laboratory stressors. To the degree that an acute stressor can effect factors such as mood, attention and self-regulation (Jamieson et al., 2013), it could impact the performance of health behaviors. Yet, little is known about the effect of SES or racial discrimination on behavioral reactivity.

Health behaviors including healthy eating, physical activity, sleep quality, medication compliance and limiting alcohol consumption are key to successful diabetes self-management and affect glycemic control (Jones et al., 2003). The study reported here examined the effects of racial discrimination and SES on health behavior reactivity in Black and White women with type 2 diabetes. Our research question was “Do background stressors (i.e. low SES and racial discrimination) affect behavioral reactivity to an acute stressor?” We examined the relationships among discrimination and income on health behaviors under two conditions: on the day of a standardized mental stressor challenge and on a counterbalanced, control day without a stressor challenge. We hypothesized that higher discrimination and lower income would predict greater behavioral reactivity to the acute stressor. Specifically, we hypothesized that, compared to control day, on stressor day lower income and higher racist events would predict larger reductions in physical activity, sleep, and medication adherence and larger increases in eating and alcohol consumption.

Our outcomes included health behaviors known to be important for diabetes management. This study focused on health behaviors that occur, or may occur, daily for most people with diabetes. Thus, they are observable within a timeframe that renders them potentially vulnerable to disruption by a same-day acute stressor. Other, less frequent diabetes health behaviors, for example, A1c testing, occur less frequently and thus any given acute stressor would not be expected to effect it (unless a stressor just happened to occur on the day of a planned A1c test). Thus, we selected behaviors that are ideal for examining acute stress reactivity in vivo. Furthermore, each of these behaviors is clinically relevant. Each has a strong influence on long-term outcomes in persons with diabetes.

We studied women because, compared to men, women’s diabetes self-care is more compromised by socioeconomic and psychosocial factors (de Melo et al., 2013), there is some evidence that the relationship between stress and diabetes is stronger for women (Agardh et al., 2003) and because some studies have shown that women have stronger behavioral reactivity to stressors and negative affect (Saladin et al., 2012). Physiological measures were also collected and are reported elsewhere (Wagner et al., 2015, 2013b, 2015).

Methods

Participants

A total of 77 women (39 Black, 38 White) with type 2 diabetes mellitus (T2DM) participated as part of a study on racial discrimination and cardiovascular risk in diabetes. Inclusion criteria for Black women were having two parents of African descent, being born and raised in the United States and identifying as Black or African American. For Whites, inclusion criteria were two parents of European descent, being born and raised in the United States, and identifying as White, Caucasian, or European American. Women were excluded if they self-identified as Hispanic or if they had known or suspected (e.g. angina) coronary artery disease, acute medical or psychiatric problems, drug or alcohol use disorder, or lack of reliable transportation to the study site. We recruited participants from print and radio advertisements and state employee paycheck inserts.

Procedures

Procedures were approved by the University of Connecticut Health Center Institutional Review Board. Consented individuals provided demographic information and completed a medical history, brief physical exam and self-report questionnaires with a research nurse at the University of Connecticut Health Center Clinical Research Center (CRC). Participants were given a glucometer and a supply of strips and were instructed in their use.

On a subsequent day, participants abstained from tobacco, exercise, caffeine, medications, food and beverages (except water) for 8 hours before a morning laboratory session. After a fasting blood draw, participants ate a standardized breakfast that consisted of a nutritional shake with 45 g of carbohydrate. Participants self-administered their morning diabetes medications per their usual regimen. Participants rested for approximately 30 minutes. Then, participants performed the public speaking stressor. Upon conclusion of the speech stressor, participants reported subjective units of distress (SUDS) and racial attributions regarding the stressor.

The public speaking stressor was adapted from Guyll et al. (2001). Participants defended themselves against a false accusation of shoplifting, a laboratory challenge that has been validated in subsequent research (Lepore et al., 2006). A research assistant, trained by the investigator (J.W.), described the fictional situation and explained the speech task to each participant. The fictional person making the accusation was identified as a security guard, but was not described in terms of race, age, gender, or other characteristics. Speeches were videotaped and were observed live by three confederates who appeared to be taking notes on the speech. To increase the stressfulness of the task, participants were told that the videotape would be reviewed at a later time to assess the speaker’s intelligence and believability. Participants were then instrumented with an accelerometer and instructed in its use. They were also given a paper-and-pencil diary to be completed at home for recording sleep quality, medication adherence, and eating behaviors. Participants were instructed when and how to complete the diary and were socialized to the meaning of each question. Participants were compensated US$50 for this visit.

On a separate day that served as a control condition, instead of performing the public speaking stressor, participants watched a video on gardening. All measures were collected on the same schedule and the 1-day diary was repeated. Participants were compensated US$50 for this visit. Control and stressor days were counterbalanced within race.

Measures

Predictors

Self-reported exposure to racial discrimination

Frequency of self-reported racial discrimination was assessed with the 18-item Schedule of Racist Events (SRE) (Landrine and Klonoff, 1996). Participants completed the SRE at baseline on their first study visit. The SRE measures exposure to discriminatory situations in which the respondent is the direct and individual target of racial discrimination. The scale yields a “frequency” subscale and a “stressfulness” subscale. We used the “frequency” subscale in all analyses and used a 2-month timeframe to capture proximal associations. Because we were investigating discrimination among both White and Black women, we modified the scale’s original wording “… because you are Black” to read “… because of your race or ethnicity.” We recently reported that our race-neutral version shows acceptable psychometric qualities and differential item functioning in a sample of 302 Black and White participants (Feinn et al., 2014). The distribution of scores was positively skewed and kurtotic (skewness = 1.7, standard error (SE) of skewness = 0.3; kurtosis = 2.0, SE of kurtosis = 0.6), but a log transformation nearly normalized the distribution (skewness = 1.2, SE of skewness = 0.3; kurtosis = 0.8, SE of kurtosis = 0.5). For ease of interpretation, we report the untransformed mean and standard deviation (SD) in Table 1.

Table 1.

Descriptive statistics and unadjusted correlations.

		M	SD	Correlations
		M	SD	1	2	3	4	5	6	7	8	9	10	11
1	Race^a	0.5	0.5	1
2	Income^b	2.7	2.0	.24*	1
3	SRE	23.6	7.1	−.31**	−.19^†	1
4	CES-D	11.8	9.4	.05	−.39**	.10	1
5	Stressor SUDS	5.8	2.8	.11	.10	.01	.21	1
6	Stressor racial attribution	4.2	4.1	−.83**	−.22^†	.27*	−.07	−.06	1
7	Physical activity	7.3	2.5	.03	.40**	−.02	−.20	−.08	−.15	1
8	Sleep quality^c	2.3	1.4	−.08	.34**	.09	−.15	−.24^†	.10	.15	1
9	Out-of-control eating^d	0.9	1.2	.00	−.14	.32**	.12	.09	−.19	.11	−.13	1
10	Medication adherence^d	3.7	0.8	.15	.23^†	.11	−.40**	.08	−.20	−.06	.24^†	.05	1
11	Alcohol consumed	3.7	0.9	.19	−.07	.23*	−.16	.07	−.05	−.07	.13	.06	.08	1

SD: standard deviation; SRE: Schedule of Racist Events; CES-D: Center for Epidemiological Studies Depression scale; SUDS: subjective units of distress.

Health behavior outcomes reported here are for stressor day.

0 = Black, 1 = White, mean values represent proportions.

1 = 0–US$10,000, 2 = US$11,000–US$20,000, 3 = US$21,000–US$40,000, 4 = US$41,000–US$60,000, 5 = US$61,000–US$80,000, 6 = US$81,000–US$100,000, 7 = more than US$100,000.

4-point Likert scale from 0 = “not at all well” to 3 = “very well.”

Responses on a 4-point Likert scale from 0 = “not at all” to 3 = “definitely.”

p < .01, *p < .05, ^†p < .10.

Income

Participants selected from the following options their household annual income before taxes: 1 = 0–US$10,000, 2 = US$10,001–US$20,000, 3 = US$20,001–US$40,000, 4 = US$40,001–US$$60,000, 5 = US$60,001–US$80,000, 6 = US$80,001–US$100,000, 7 = more than US$100,000. Participants reported income at baseline on their first study visit. Income was treated as numerical in all analyses.

Distress

Upon conclusion of the speech, participants reported SUDS (Wolpe, 1958) from 0 to 10 with 0 = “not at all stressed” and 10 = “extremely stressed.” They also reported racial attribution by indicating the degree to which they believed that the wrongful accusation was based on their race by answering the question “To what degree do you think your race was a reason for being accused of shoplifting?” on a scale from 1 = “not at all” to 10 = “completely.”

Daily health behavior outcomes

We assessed associations between our brief, daily measures of health behaviors on the control day and global, retrospective measures of diabetes self-care measures, including the Summary of Diabetes Self-Care Activities (SDSCA) (Toobert et al., 2000), the diabetes Self-Care Inventory (SCI) (Weigner et al., 2005) and the Alcohol Use Disorder Identification Test (AUDIT) (Babor et al., 2001 [1992]). Participants completed the SDSCA and the SCI at baseline on their first study visit. For these correlations, we chose outcome on the control day because our hypothesis was precisely that the speech task would disrupt routine self-care activities on the stressor day.

Physical activity

For the 24 hours following both stressor and control conditions, participants were instrumented with Actigraph GT1M physical activity monitors. It is a small device worn on the forearm that looks like a watch and reliably records step count and activity (Welk et al., 2004). Calibration is automatic with software, and data are downloaded with a USB port and analyzed offline. Our actigraphy step count was positively associated with the Exercise subscale of the SDSCA, r = .44, p = .000.

Sleep quality

On the morning following both stressor and control conditions, participants self-reported sleep quality. Participants were asked to respond to a question regarding sleep quality, “How well did you sleep last night?” Response options were on a 4-point Likert scale from 0 = “not at all well” to 3 = “very well.”

Medication adherence

On the evening of both stressor and control conditions, adherence to medications for diabetes was assessed with one question that asked, “Since you left the laboratory today, did you take the correct dose of your diabetes medication?” Respondents answered on a 4-point Likert scale from 0 = “not at all” to 3 = “definitely” such that higher scores indicated higher adherence.

Eating

On the evening of both stressor and control conditions, participants self-reported eating behavior. Based on qualitative work regarding racial discrimination and eating behaviors (Wagner et al., 2011), we developed one question for this study that asked, “Since you left the laboratory today, have you felt out of control of your eating?” Respondents answered on a 4-point Likert scale from 0 = “not at all” to 3 = “definitely.”

Alcohol consumption

On the evening of both stressor and control conditions, participants self-reported alcohol consumption, with a question adapted from a validated daily assessment of drinking (Kranzler et al., 2004). We asked, “Since you left the laboratory today, how many servings of alcohol did you drink? Remember, one serving of alcohol is a 12-ounce bottle of beer, an 8-ounce glass of wine, or a shot of liquor.”

Covariates

Race

Race was assessed at baseline on the first study visit. It was assessed per self-report and dummy coded where participants were asked to choose from “White, Caucasian, European American” or “Black, African American.”

Depressive symptoms

Depressive symptoms are consistently associated with low SES (Lorant et al., 2003), reports of racial discrimination (Chou et al., 2012) and lower diabetes self-management behaviors (Gonzales et al., 2008). Negative affect also biases reports of past experiences (Blaney, 1986; Bower, 1981). Therefore, we sought to determine whether any effects of low SES or racial discrimination on health behaviors were better accounted for by depressive symptoms. Depressive symptoms over the past week were measured at baseline during the first study visit with the Center for Epidemiological Studies Depression scale (CES-D) (Radloff, 1977). The 20-item scale has high reliability; coefficient α in this sample was .81.

Past-month diabetes self-care

We accounted for adherence to diabetes health behaviors during the past month in order to isolate influences on self-care behaviors on the day of the stressor specifically. The diabetes SCI (Weigner et al., 2005) measures self-reported diabetes management during the past 1–2 months. The 15 items measure adherence to medication, exercise, eating, blood glucose monitoring, wearing a medic alert, clinic attendance and preparedness for hypoglycemia. It was administered at baseline during the first study visit.

Data analysis

To describe the sample, means, SDs, and frequencies were calculated and zero-order correlations were calculated among variables. Where either predictor (either SES or discrimination) showed an association with an outcome on stressor day that predictor was regressed on that respective behavioral outcome (physical activity, eating, sleep quality, or medication adherence). Covariates included race, depressive symptoms, past-month diabetes self-care and the value for the outcome on the control day. This approach, in which a baseline value is utilized as a covariate in the model (rather than as a level of time factor or as part of the dependent variable within the analysis of group differences), is generally preferred (Rausch et al., 2003) because assumptions of the repeated measures analysis with the pretest as a level of the time factor are usually not met. For each outcome, we first tested a model including main effects of discrimination and income. Where a main effect was significant, it was tested in a final model that included covariates. We also tested for an interaction between discrimination and income and none were near significant and so only the main effects are reported in the results. To adjust for multiple comparisons, we applied the Benjamini and Hochberg (1995) false discovery rate to those behavior outcomes that were tested in a final, adjusted model. Standardized regression coefficients are presented.

Results

Participants

Table 1 presents descriptive statistics and bivariate correlations among main study variables. All participants were adult women aged M = 55.8 (SD = 11.7) years, and T2DM duration of M = 8.7 (SD = 8.1) years and adequate glycemic control, A1c M = 6.9 (SD = 1.5). Of these, 13 percent controlled their diabetes with diet only, 56 percent with oral agents, and 31 percent used insulin. Modal annual income (25%) was US$21,001–US$40,000 per year and about one-third (31.2%) had high school education or less. Compared to Whites, Blacks had significantly lower income, χ²(1, N = 76) = 4.16, p = .04. They also had significantly higher A1c (M = 7.3, SD = 1.6 vs M = 6.4, SD = 1.4; p = .01) and reported more discrimination in the prior 2 months (M = 25.8, SD = 8.4 vs M = 21.5, SD = 4.9; p = .01).

Laboratory speech stressor

In a paired t-test, SUDS increased from baseline (M = 1.2, SD = 1.7) to speech stressor (M = 5.8, SD = 2.8), p < .05. According to SUDS scores, Black and White participants did not significantly differ in how stressful they found the speech stressor. However, Black women reported significantly higher racial attribution for the accusation of shoplifting (M = 7.6, SD = 2.7) than did White women (M = 0.9, SD = 1.8), p < .05.

Racial attribution was not related to physical activity, sleep, out-of-control eating, alcohol consumption, or medication adherence on stressor day nor on control day; correlations all p > .10. In paired t-tests comparing health behaviors on stressor and control days, there were no mean differences for any of the health behaviors, all p > .05.

Physical activity

As can be seen in Table 1, lower income, but not racial discrimination, was associated with lower step count on stressor day. In adjusted analyses, lower income predicted lower step count on stressor day even after controlling for step count on control day, past-month diabetes self-care behaviors, race and depressive symptoms (β = .21, t = 2.34, p = .038). Physical activity on control day was a significant predictor of physical activity on stressor day (β = .70, t = 8.81, p < .001).

Sleep

As can be seen in Table 1, lower income, but not racial discrimination, was associated with poorer sleep quality on stressor day. In adjusted analyses, lower income predicted poorer sleep quality on stressor day even after controlling for sleep quality on control day, past-month diabetes self-care behaviors, race and depressive symptoms (β = .33, t = 2.64, p = .028). Sleep quality on control day was a significant predictor of sleep quality on stressor day (β = .37, t = 3.09, p < .01).

Out-of-control eating

As can be seen in Table 1, frequency of racial discrimination, but not income, was associated with more out-of-control eating on stressor day. In adjusted analyses, the association between higher log-transformed racist events and more out-of-control eating on stressor day was significant after controlling for out-of-control eating on control day, past-month diabetes self-care behaviors, race and depressive symptoms (β = .40, t = 3.36, p = .010).

Alcohol consumption

As can be seen in Table 1, frequency of racial discrimination, but not income, was associated with higher alcohol consumption on stressor day. In adjusted analyses, the association between higher log-transformed discrimination and more drinks consumed was significant, after controlling for drinks consumed on control day, past-month diabetes self-care behaviors, race and depressive symptoms (β = .25, t = 2.14, p = .046).

Medication adherence

As can be seen in Table 1, lower income, but not racial discrimination, was marginally associated with lower medication adherence on stressor day. In adjusted analyses controlling for medication adherence on control day, past-month diabetes self-care behaviors, race and depressive symptoms, the association between lower income and lower medication adherence became nonsignificant (β = .030, t = .06, p = .87). Higher depressive symptoms were a significant predictor of lower medication adherence on stressor day (β = −.42, t = −2.94, p < .01).

Discussion

This study demonstrated that SES and racial discrimination are unique predictors of diabetes health behaviors in response to acute stress. Behavioral reactivity to an acute stressor was exacerbated in those with lower income or greater exposure to racial discrimination. Our data also show that these contextual factors may not affect all health behaviors similarly. Specifically, lower SES was associated with lower physical activity and worse sleep quality following acute stress, whereas racial discrimination was associated with more out-of-control eating and higher alcohol consumption following acute stress. Contextual factors such as low SES and exposure to racial discrimination may prime individuals to respond in a maladaptive way to acute stressors. Past research has shown similar priming effects for physiological and emotional reactivity, and this study adds to that literature by examining behavioral reactivity.

Lower income was associated with a greater decrease in step count on stressor day compared to control day. Or, put another way, people from low-SES backgrounds exhibited more stress reactivity, in terms of physical activity, than did their counterparts from higher SES backgrounds. These findings complement existing studies; lower SES has consistently been associated with lower levels of physical activity (Estabrooks et al., 2003; Giles-Corti and Donovan, 2002). Individuals in low-SES environments may live in high-crime neighborhoods where outdoor activities such as walking are unsafe. They may live in settings where running and cycling are made difficult by the built environment. They may also lack affordable access to recreational activities such as swimming and fitness classes. This combination of factors may create norms that favor sedentary behavior and low social support for physical activity (Okun et al., 2003). Acute stressors can also affect physical activity. Prospective studies show that acute distress may decrease subsequent physical activity (Lutz et al., 2007). Our findings expand these literatures by showing that acute stressors may further decrease levels of routine physical activity among those from low-SES backgrounds. Based on the work by Lutz et al. (2010) showing that acute stressors can increase or decrease exercise depending upon one’s routine exercise habits, we hypothesize that low-SES individuals, who face structural barriers to physical activity to begin with may view exercise as “one more burden” when experiencing distress, whereas higher SES individuals may view it as a stress reliever.

Lower income was associated with lower sleep quality on stressor day compared to control day. Or, put another way, people from low-SES backgrounds exhibited more stress reactivity, in terms of sleep quality, than did their counterparts from higher SES backgrounds. These findings are in line with the existing literature. Lower SES has consistently been associated with poorer sleep (Grandner et al., 2010). Several characteristics of low-SES environments are associated with impaired sleep including unemployment, being unmarried, family conflict, and depression (El-Sheikh et al., 2012) and we submit that poor housing conditions such as noise and safety concerns may also play a role. Studies clearly demonstrate that distress interferes with sleep and some show that the adaptive sleep patterns that help with recovery from stress, especially rapid eye movement sleep, are absent in persons with high anxiety (Suchecki et al., 2012). The current study extends this literature by showing that lower SES is associated with worsened sleep quality in response to an acute stressor.

Frequency of racial discrimination was associated with a greater increase in out-of-control eating and more alcohol consumption on stressor day compared to control day. Or, put another way, people with higher racial discrimination exhibited more stress reactivity, in terms of consumption, than did their counterparts with lower discrimination. These findings are consistent with previous studies. The literature has shown that racial stressors decrease self-control (Gibbons et al., 2012) and increase appetitive behaviors (Guthrie et al., 2002). Our own qualitative work suggests that these effects may be strengthened when individuals are faced with an acute stressful event (Wagner et al., 2011). This study supports these findings by demonstrating an association between higher levels of racial discrimination and higher levels of eating and drinking in response to an acute stressor. We hypothesize that coping with discrimination may deplete the psychological resources necessary for self-regulation in the face of an acute stressor, leaving those individuals with more discrimination vulnerable to appetitive responses.

Two points regarding measurement are important to note in reference to the association between discrimination and out-of-control eating. First, we used the “frequency” subscale of the SRE in order to approximate exposure to discrimination rather than responses to it (as is intended to be captured in the “stressfulness” subscale). Yet, whereas exposure to a given stressor (e.g. discrimination) and the distress associated with such an exposure are different, in the context of a subjective, self-report measure, they are inextricably linked. This is a measurement problem that plagues the stress and coping literature and it pertains to primary appraisals according to the framework of Folkman and collegues (1986). That is, in order for an exposure to rise to the level of awareness and subsequent endorsement on a self-report measure, it is usually associated with distress. An exposure that does not cause distress will likely not be reported as an exposure. Similar to other life event checklists, the SRE captures only those exposures that are perceived, recalled, and endorsed and are thus subject to several sources of bias and may reflect a more general responsivity to stressors.

Second, we used a single-item measure for eating behavior, which may not fully capture the behaviors of interest in a comprehensive way. Yet, prior research has established the validity and reliability of single-item measures of perceived psychosocial stress (Littman et al., 2006). Indeed, diary measures—even brief ones—actually overcome several common sources of measurement reactivity associated with retrospective, global reports of behavior which are usually much longer. Most notably, diary methods limit the recall error generated by the schema-based processing typically created when research participants are required to recall experiences and behaviors that occurred days, weeks, or even months ago, or to report how they “usually” behave (Littman et al., 2006).

Finally, we point out that exposure to discrete incidents of racial discrimination, and racism, are not synonymous. Racism is not only acts of prejudice and discrimination, but their occurrence in the broader context of unequal access to power and privilege. Thus, whereas White participants report experiences of prejudice and discrimination, we do not suggest here that they are victims of systematic and institutional racism per se. Notwithstanding, having experienced racial discrimination does indeed appear to have a deleterious influence on behavioral reactivity to acute stress, regardless of race.

We found that depressive symptoms, which are common in persons with diabetes (Anderson et al., 2001), and are related to worse diabetes outcomes (de Groot et al., 2001), predicted lower medication adherence in response to acute stress. It is well documented that depressive symptoms interfere with medication adherence in diabetes (Gonzales et al., 2007). Our findings add to this literature by showing that persons experiencing depressive symptoms may be particularly vulnerable to non-adherence during times of acute stress. Interventions aimed at improving depression and adherence in diabetes (Safren et al., 2014) should attend to the acute effects of mental stress.

Racial discrimination was not related to medication adherence in response to the stressor. The overall high levels of medication adherence may have created ceiling effects. Yet, these results are consistent with the findings of Peek et al. (2011) from much larger samples with greater variance, which demonstrated that racial discrimination was not associated with specific diabetes self-management behaviors (e.g. self-monitoring of blood glucose, foot exams, attending diabetes classes). On the other hand, our nonsignificant findings for an association between SES and medication adherence is not consistent with the literature. Previous research has, in fact, demonstrated associations between lower income and poorer adherence to medication regimens (Adams et al., 2003; Piette et al., 2004). It is likely that these relationships were attenuated in this study sample; due to the fact that most participants had full insurance coverage, cost of medication was not problematic even among low-SES individuals. The relationship between lower income and poorer medication adherence may be more evident in a sample with greater heterogeneity in insurance coverage.

Whereas both low SES and discrimination influenced reactivity, they did so differentially. Future studies should examine the mechanisms through which different stressors may have unique effects on various behaviors using designs that emphasize ecological validity by assessing the effects of naturally occurring stressors over a longer period of time.

Research designs using laboratory stressors are limited in that they often employ a single stressor that is contrived and decontextualized. Furthermore, the laboratory setting restricts the range of available coping responses and often does not permit extended periods of observation. Whereas these shortcomings are somewhat true of our stressor (i.e. public speaking), our measurement scheme was a significant improvement on many laboratory designs. Specifically, we examined behavioral reactivity and collected ambulatory measures, in vivo, over 24 hours post-stressor.

Limitations

This study has some limitations. First, this sample had overall good glycemic control and with high rates of insurance coverage, access to medical care, no serious medical comorbidities, reliable transportation to the study site, and with a low, but not very low, modal level of income. Second, other than an objective measure of physical activity, outcomes were per self-report which are vulnerable to demand characteristics. Although depressive symptoms (CES-D) were covaried, we likely did not capture all relevant individual difference variance that could contribute to a congruent bias reporting across variables. For example, recent evidence also suggests that individual differences in personality and genotype can affect appetitive responses to a stressor (Capello and Markus, 2014). Sleep was assessed with a single item that did not allow us to examine factors such as sleep latency, efficiency, or total sleep time. This may be why a recent study (Slopen and Williams, 2014) reported associations between discrimination and sleep that we did not. Future research is encouraged to examine ranked income, which recent studies suggest may be superior to absolute levels in explaining some outcomes (Boyce et al., 2010; Wood et al., 2012).

Finally, this study was limited to African American and White participants and so may not generalize to other racial/ethnic groups. These limitations are generally outweighed by strengths including tight control of stressor exposure, repeated measures on a control day, a laboratory stressor linked to in vivo health behaviors, and a well-characterized clinical sample for whom the respective health behaviors are particularly important.

Footnotes

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

References

Adams

Mah

Soumerai

. (2003) Barriers to monitoring of blood glucose among adults with diabetes in an HMO: A cross sectional study. BMC Health Services Research 3: 6.

Agardh

Ahlbom

Andersson

. (2003) Work stress and low sense of coherence is associated with type 2 diabetes in middle-aged Swedish women. Diabetes Care 26(3): 719–724.

Anderson

Freedland

Clouse

. (2001) The prevalence of comorbid depression in adults with diabetes: A meta-analysis. Diabetes Care 24(6): 1069–1078.

Babor

Higgins-Biddle

Saunders

. (2001 [1992]) AUDIT—The Alcohol Use Disorders Identification Test. Guidelines for Use in Primary Health Care. Geneva: WHO.

Barr

Somers

Speizer

. (2002) Patient factors and medication guideline adherence among older women with asthma. Archives of Internal Medicine 162(15): 1761–1768.

Benjamini

Hochberg

(1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B 57: 289–300.

Blaney

(1986) Affect and memory: A review. Psychological Bulletin 99(2): 229–246.

Bower

(1981) Mood and memory. American Psychologist 36(2): 129–148.

Boyce

Brown

Moore

(2010) Money and happiness: Rank of income, not income, affects life satisfaction. Psychological Science 21(4): 471–475.

10.

Brodish

Cogburn

Fuller-Rowell

. (2011) Perceived racial discrimination as a predictor of health behaviors: The moderating role of gender. Race and Social Problems 3(3): 160–169.

11.

Capello

Markus

(2014) Differential influence of the 5-HTTLPR genotype, neuroticism, and real-life acute stress exposure on appetite and energy intake. Appetite 77(1): 85–95.

12.

Chou

Asnaani

Hofmann

(2012) Perception of racial discrimination and psychopathology across three U.S. ethnic minority groups. Cultural Diversity & Ethnic Minority Psychology 18(1): 74–81.

13.

Colgan

Gospel

Petrie

. (2004) Does rear seatbelt use vary according to socioeconomic status? Journal of Epidemiology and Community Health 58(11): 929–930.

14.

Connolly

Unwin

Sherri

. (2000) Diabetes prevalence and socioeconomic status: A population based study showing increased prevalence of type 2 diabetes mellitus in deprived areas. Journal of Epidemiology & Community Health 54(3): 173–177.

15.

Cozier

Coogan

. (2014) Racism, segregation, and risk of obesity in the Black Women’s Health Study. American Journal of Epidemiology 179(7): 875–883.

16.

Crowe

Sartori

Clay

. (2010) Diabetes and cognitive decline: Investigating the potential influence of factors related to health disparities. Journal of Aging and Health 22(3): 292–306.

17.

de Groot

Anderson

Freedland

. (2001) Association of depression and diabetes complications: A meta-analysis. Psychosomatic Medicine 63(4): 619–630.

18.

de Melo

de Sa

Gucciardi

(2013) Exploring differences in Canadian adult men and women with diabetes management: Results from the Canadian community health survey. BMC Public Health 13(1): 1089.

19.

El-Sheikh

Kelly

Bagley

. (2012) Parental depression and children’s sleep: The role of family conflict. Journal of Child Psychology and Psychiatry 53(7): 806–814.

20.

Estabrooks

Lee

Gyurcsik

(2003) Resources for physical activity participation: Does availability and accessibility differ by neighborhood socioeconomic status? Annals of Behavioral Medicine 25(2): 100–104.

21.

Feinn

Schumann

Tennen

. (2014) Differential item functioning on the Schedule of Racist Events: Exploring a modified scale for samples with black and white participants. Ethnicity & Disease 24(4): 406-412.

22.

Gary-Webb

Baptiste-Roberts

Pham

. (2011) Neighborhood socioeconomic status, depression, and health status in the Look AHEAD (Action for Health in Diabetes) study. BMC Public Health 11: 349.

23.

Giles-Corti

Donovan

(2002) Socioeconomic status differences in recreational physical activity levels and real and perceived access to a supportive physical environment. Preventive Medicine 35(6): 610–611.

24.

Goldman

Smith

(2002) Can patient self-management help explain the SES health gradient? PNAS 99(16): 10929–10934.

25.

Gonzales

Peyrot

McCarl

. (2008) Depression and diabetes treatment non adherence: A meta-analysis. Diabetes Care 31(12): 2398–2403.

26.

Gonzales

Safren

Cagliero

. (2007) Depression, self-care, and medication adherence in type 2 diabetes: Relationships across the full range of symptoms severity. Diabetes Care 30(9): 2222–2227.

27.

Grandner

Hale

Jackson

. (2012) Perceived racial discrimination as an independent predictor of sleep disturbance and daytime fatigue. Behavioral Sleep Medicine 10(4): 235–249.

28.

Grandner

Patel

Gehrman

. (2010) Who gets the best sleep? Ethnic socioeconomic factors related to sleep complaints. Sleep Medicine 11(5): 470–478.

29.

Guthrie

Young

Williams

. (2002) African American girls’ smoking habits and day-to-day experiences with racial discrimination. Nursing Research 51(3): 183–190.

30.

Guyll

Matthews

Bromberger

(2001) Discrimination and unfair treatment: Relationship to cardiovascular reactivity among Black and European American women. Health Psychology 20(5): 315–325.

31.

Hicken

Lee

Ailshire

. (2013) “Every shut eye, ain’t sleep”: The role of racism-related vigilance in racial/ethnic disparities in sleep difficulty. Race and Social Problems 5(2): 110–112.

32.

Jamieson

Koslov

Nock

. (2013) Experiencing discrimination increases risk taking. Psychological Science 24(2): 131–139.

33.

Jones

Edwards

Vallis

. (2003) Changes in diabetes self-care behaviors make a difference in glycemic control: The Diabetes Stages of Change (DiSC) study. Diabetes Care 26(3): 732–737.

34.

Kranzler

Armeli

Feinn

. (2004) Targeted naltrexone treatment moderates the relations between mood and drinking behavior among problem drinkers. Journal of Consulting and Clinical Psychology 72(2): 317–327.

35.

Krause

Horbert

Goetz

. (2011) Social class rank, threat, vigilance, and hostile reactivity. Personality and Social Psychology Bulletin 37(10): 1376–1388.

36.

Landrine

Klonoff

(1996) The Schedule of Racist Events: A measure of racial discrimination and a study of its negative physical and mental health consequences. Journal of Black Psychology 22(2): 144–168.

37.

Lepore

Revenson

Weinberger

. (2006) Effects of social stressors on cardiovascular reactivity in black and white women. Annals of Behavioral Medicine 31(2): 120–127.

38.

Littman

White

Satia

. (2006) Reliability and validity of 2 single-item measures of psychosocial stress. Epidemiology 17(4): 398–403.

39.

Lorant

Deliege

Eaton

. (2003) Socioeconomic inequalities in depression: A meta-analysis. American Journal of Epidemiology 157(2): 98–112.

40.

Lutz

Lochbaum

Lanning

. (2007) Cross-lagged relationships among leisure-time exercise and perceived stress in blue-collar workers. Journal of Sport & Exercise Psychology 29(6): 687–705.

41.

Lutz

Stults-Kolehmainen

Bartholomew

(2010) Exercise caution when stressed: Stages of change and the stress-exercise participation relationship. Psychology of Sport and Exercise 11(6): 560–567.

42.

Mouton

Carter-Nolan

Makambi

. (2010) Impact of perceived racial discrimination on health screening in black women. Journal of Health Care for the Poor and Underserved 21(1): 287–300.

43.

Nettle

(2010) Why are there social gradients in preventative health behavior? A prospective from behavioral ecology. PLoS ONE 5(10): e13371.

44.

Okun

Ruehlman

Karoly

. (2003) Social support and social norms: Do both contribute to predicting leisure-time exercise? American Journal of Health Behaviors 27(5): 493–507.

45.

Peek

Wagner

Tang

. (2011) Self-reported racial/ethnic discrimination in healthcare and diabetes outcomes. Medical Care 49(7): 618–625.

46.

Phillips

(2011) Socioeconomic status, amygdala reactivity, and selective attention to threat. Doctoral Dissertation, University of Pittsburgh, Pittsburgh, PA.

47.

Piette

Heisler

Wagner

(2004) Problems paying out-of-pocket medication costs among older adults with diabetes. Diabetes Care 27(2): 384–391.

48.

Radloff

(1977) The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement 1(3): 385–401.

49.

Rausch

Maxwell

Kelley

(2003) Analytic methods for questions pertaining to a randomized pretest, posttest, follow-up design. Journal of Clinical & Adolescent Psychology 32(3): 467–486.

50.

Relton

Hammal

Rankin

. (2005) Folic acid supplementation and social deprivation. Public Health Nutrition 8(3): 338–340.

51.

Safren

Gonzales

Wexler

. (2014) A randomized controlled trial of cognitive behavioral therapy for adherence and depression (CBT-AD) in patients with uncontrolled type 2 diabetes. Diabetes Care 37(3): 625-633.

52.

Saladin

Gray

Carpenter

. (2012) Gender differences in craving and cue reactivity to smoking and negative affect/stress cues. American Journal of Addictions 21(3): 210–220.

53.

Saydah

Lochner

(2010) Socioeconomic status and risk of diabetes-related mortality in the U.S. Public Health Report 125(3): 377–388.

54.

Slopen

Williams

(2014) Discrimination, other psychosocial stressors, and self-reported sleep duration and difficulties. Sleep 37(1): 147–156.

55.

Suchecki

Tiba

Machado

(2012) REM sleep rebound as an adaptive response to stressful situations. Frontiers in Neurology 3(41): 1–12.

56.

Thomas

Nelesen

Malcarne

. (2006) Ethnicity, perceived discrimination, and vascular reactivity to phenylephrine. Psychosomatic Medicine 68(5): 692–697.

57.

Toobert

Hampson

Glasgow

(2000) The summary of diabetes self-care activities measure: Results from 7 studies and a revised scale. Diabetes Care 23(7): 943–950.

58.

Vikky

Phillips

Hopman

(2006) Determinants of a healthy lifestyle and use of preventive screening in Canada. BMC Public Health 6: 275. DOI: 10.1186/1471-2458-6-275.

59.

Virtanen

Kivimaki

Vahtera

. (2006) Employment status and differences in one-year coverage of physician visits: Different needs or unequal access to services? BMC Health Services Research 6: 123.

60.

Wagner

Lampert

Tennen

. (2015) Exposure to discrimination and heart rate variability reactivity to acute mental stress in women with diabetes. Stress and Health 31(3): 255-262.

61.

Wagner

Osborne

Mendenhall

. (2011) Beliefs about racism and health among African American women with diabetes: A qualitative study. Journal of the National Medical Association 103(3): 224–232.

62.

Wagner

Tennen

Feinn

. (2013b) Racial discrimination and metabolic control in women with type 2 diabetes. Ethnicity & Disease 23(4): 421–427.

63.

Wagner

Tennen

Feinn

. (2015) Self-reported discrimination, diabetes distress, and continuous blood glucose in women with type 2 diabetes. Journal of Immigrant and Minority Health 17(2):566-573.

64.

Wagner

Tennen

Finan

. (2013) Self-reported racial discrimination and endothelial reactivity to acute stress in women. Stress and Health 29(3): 214-221.

65.

Weigner

Butler

Welch

. (2005) Measuring diabetes self-care: A psychometric analysis of the Self-Care Inventory Revised with adults. Diabetes Care 28(6): 1346–1352.

66.

Welk

Schaben

Morrow

(2004) Reliability of accelerometry-based activity monitors: A generalizability study. Medicine & Science in Sports & Exercise 36(9): 1637–1645.

67.

Wilson

Kliewer

Plybon

. (2000) Socioeconomic status and blood pressure reactivity in healthy black adolescents. Hypertension 35: 496–500.

68.

Wolpe

(1958) Psychotherapy by Reciprocal Inhibition. Stanford, CA: Stanford University Press.

69.

Wood

Boyce

Moore

. (2012) An evolutionary based social rank explanation of why low income predicts mental distress: A 17 year cohort study of 30,000 people. Journal of Affective Disorders 136(3): 882–888.

70.

Yen

Ragland

Greiner

. (1999) Racial discrimination and alcohol-related behavior in urban transit operators: Findings from the San Francisco Muni Health and Safety Study. Public Health Reports 114(5): 448–458.