Discrimination, Affect, and Cancer Risk Factors among African Americans

Abstract

Objectives

To examine whether stress or depressive symptoms mediated associations between perceived discrimination and multiple modifiable behavioral risk factors for cancer among 1363 African American adults.

Methods

Nonparametric bootstrapping procedures, adjusted for sociodemographics, were used to assess mediation.

Results

Stress and depressive symptoms each mediated associations between discrimination and current smoking, and discrimination and the total number of behavioral risk factors for cancer. Depressive symptoms also mediated the association between discrimination and overweight/obesity (p values < .05).

Conclusions

Discrimination may influence certain behavioral risk factors for cancer through heightened levels of stress and depressive symptoms. Interventions to reduce cancer risk may need to address experiences of discrimination, as well as the stress and depression they engender.

Stressful experiences with discrimination, defined as unfair treatment due to a person’s racial/ethnic characteristics, are unfortunately quite common among African Americans in the United States^1–3 and have been associated with poorer physical and psychological health.^4–8 Researchers have recently suggested that experiences of discrimination may at least partially contribute to racial/ethnic disparities in health,^7,9,10 including disparities in cancer incidence and survival that disproportionately affect African Americans [eg,^11,12]. Modifiable behavioral-based risk factors, such as overweight/obesity, inadequate fruit and vegetable intake, smoking, insufficient physical activity, and at-risk alcohol use have each been associated with increased cancer risk among diverse samples [eg,^13,14]. Therefore, one mechanism by which discriminatory events may affect cancer risk among African Americans is through associations with increased engagement in these modifiable risk factors. However, the relationships between discrimination and cancer risk behaviors, as well as the mechanisms underlying these potential associations, are unclear.

Findings regarding the associations between discrimination and behavioral risk factors for cancer in African Americans have been mixed. Perceived discrimination has been linked to higher rates of smoking,^15–22 problematic alcohol use,^16,23–26 and unhealthy eating habits,²³ but it has not been linked to insufficient/low physical activity rates^16,23,27 or weight status.^28,29 However, some research suggests that discrimination is associated with higher levels of physical activity, at least among some subgroups of African Americans [eg, young women ²³]. Given that insufficient/low physical activity and overweight/obesity are some of the most common modifiable risk factors among African Americans [cf.^30,31], research aimed toward understanding whether discrimination affects these behaviors in African Americans and how it operates is critical. The effects of discrimination on these and other behavioral risk factors for cancer among African Americans might be best understood through a more comprehensive biopsychosocial model that includes the potential for indirect effects via psychological factors, such as stress and depressive symptoms. The generation of new information on the mechanisms underlying relations between discrimination and health among African Americans, including the modifiable behavioral factors that influence health, is needed to better understand and address health inequities involving this population.³²

Biopsychosocial theory and the stress and coping framework suggest that discriminatory events are experienced as psychological stressors that can trigger unhealthy behaviors, presumably in efforts to alleviate distress.³³ Empirical research has linked perceived discrimination to elevated stress and depressive symptoms^32,33 as well as serious/severe psychological distress.^2,34 In turn, such psychological symptoms have been linked with a number of behavioral risk factors for cancer, including an unhealthy diet,^35–38 increased alcohol consumption,³⁸ insufficient/low physical activity rates,^35,36,39 being a smoker,^35,36,40,41 and being overweight or obese.^42,43 Consequently, stress and depressive symptoms may be the primary mechanisms through which discrimination affects modifiable cancer risk factors among African Americans. However, very few studies have formally examined this possibility by testing indirect effects in mediation analyses. We are aware of only 2 such studies, both of which focused on smoking. One study, conducted among young African American females, indicated that stress mediated the relationship between discrimination and current cigarette smoking.¹⁹ Another more recent study using a large population-based sample supported a similar pattern of results among adults.¹⁵ These initial results suggest the biopsychosocial model and the stress and coping framework may critically alter our understanding of how perceived discrimination affects engagement in other modifiable risk factors for cancer among African Americans. Research that elucidates biopsychosocial pathways linking discrimination with key health-related outcomes will be informative for the development of interventions to prevent the negative health consequences of discrimination.

The current study examined associations among perceived discrimination, stress and depressive symptoms, and multiple behavioral-based cancer risk factors (ie, overweight/obesity, inadequate fruit and vegetable intake, smoking, low physical activity, at-risk alcohol use, and the total number of risk factors) in a large sample of African American adults living in Houston, Texas. We hypothesized that stress and depressive symptoms would significantly mediate associations between perceived discrimination and behavioral risk factors for cancer.

METHODS

Participants and Procedures

Data were collected through a cohort study investigating associations among biopsychosocial factors and health behaviors in African American adults. Participants were recruited from a large church in Houston, Texas, through printed and televised media within the church and in-person solicitation (eg, during church services, at a church health fair). Individuals were eligible to participate if they were 18 years of age or older, resided in the Houston area, had a functional telephone number, and attended church (although church membership was not a requirement).

In total, 1501 African American adults enrolled in the study from December 2008 to July 2009. Procedures included the completion of surveys at the church where participants viewed questionnaire items on a computer screen and responded using the computer keyboard. Compensation for time and effort included a $30 Visa Debit Card following survey completion. Only participants with complete data on the measures described below (N = 1363) were included in the current study.

Measures

Sociodemographics

Sociodemographic variables included age, sex, partner status, total annual household income, educational level, and employment status. Sociodemographics were included as covariates in the analyses due to known associations with cancer risk behaviors.

Perceived discrimination

The Day-to-Day Unfair Treatment Scale was selected as the measure of perceived discrimination. This measure is a widely used, validated protocol with good psychometric properties.⁴⁴ The scale includes ten examples of discrimination experienced in different settings (eg, at work, getting medical care, getting services in a store or restaurant). Respondents were asked how often they experienced such forms of discrimination in their day-to-day lives based on their race/ethnicity or skin color, where the response categories were 1=“4 or more times,” 2=“2 or 3 times,” 3=“once,” or 4=“never.” Responses were reverse scored and summed so higher scores were indicative of greater perceived discrimination. The Cronbach’s alpha for the Day-to-Day Unfair Treatment Scale in this sample was 0.91.

Perceived stress

The Perceived Stress Scale-4 (PSS-4)⁴⁵ is a 4-item self-report scale that asks respondents to indicate how often they experience certain situations, such as “In the last month, how often have you felt that you were unable to control the important things in your life?” and “In the last month, how often have you felt confident about your ability to handle your personal problems?” (reverse scored). Responses were summed with a potential range of zero to 16, where higher scores indicate greater perceived stress. The Cronbach’s alpha for the PSS-4 in this sample was 0.73.

Depressive symptoms

The Center for Epidemiological Studies Depression 10-item scale (CESD-10) was used to assess the degree of depressive symptoms experienced over the past week.^46,47 Items include “I was bothered by things that usually don’t bother me” and “I felt hopeful about the future” (reversed scored). Responses were summed with a potential range of zero to 30, where higher scores indicate more depressive symptoms. The Cronbach’s alpha for the CESD-10 in this sample was 0.82.

Overweight/Obesity status

Overweight/ obesity status was determined based on staff-administered height and weight measurements, which were converted to body mass index (BMI; kg/m²). Participants with a BMI ≥ 25 were considered overweight/obese.

Fruit and vegetable intake

Fruit and vegetable intake was assessed with the NCI 5-A-Day fruit and vegetable questionnaire.⁴⁸ This questionnaire yielded a continuous variable of daily fruit and vegetable servings that was highly skewed. Because of this, we chose to focus on a binary outcome whereby participants were classified as meeting recommendations for daily intake (≥5 servings of fruits and vegetables a day) or not meeting recommendations for daily intake (<5 servings of fruits and vegetables a day).

Smoking status

Smoking status was assessed with survey items resulting in classification as a current smoker (smoked ≥100 cigarettes in lifetime and currently smoke) or former smoker/never smoker (ie, smoked ≥100 cigarettes in lifetime but quit or smoked <100 cigarettes in lifetime).

Insufficient physical activity

Physical activity was assessed with the International Physical Activity Questionnaire - Short Format (IPAQ). The IPAQ is a self-report questionnaire used to measure the amount of time spent in moderate activity, vigorous activity, and walking during the past 7 days.⁴⁹ Time spent engaging in each type of activity was multiplied by the corresponding metabolic equivalent (MET) value, which is a metric used to quantify energy expenditure [ie, the ratio of energy expended during an activity to the energy expended during rest].⁵⁰ MET minutes were summed to arrive at the total weekly MET minutes spent in physical activity. Because resulting data were highly skewed, participants were classified as engaging in low or moderate/high rates of physical activity during the previous week based on total weekly MET minutes, the number of days per week engaged in physical activity, and the amount of time spent in each type of physical activity (see Guidelines for data processing and analysis of the IPAQ, 2005). The short version of the IPAQ has good test-retest reliability and acceptable criterion validity against the Computer Science Applications, Inc. accelerometer.⁴⁹ Participants were considered insufficiently active if they were categorized as having low activity during the previous week.

At-risk drinking

At-risk drinking was assessed using the Alcohol Quantity and Frequency Questionnaire, a self-report measure of the average alcohol consumption on each day of the week over the last 30 days. Males were classified as atrisk drinkers if they consumed an average of >14 drinks per week, and females were classified as atrisk drinkers if they consumed an average of >7 drinks per week.⁵¹

Total number of risk factors

The total number of behavioral cancer risk factors was determined by summing the number of risk factors for which the specified criteria were met (ie, overweight/obesity, not meeting fruit and vegetable intake recommendations, current smoking, low physical activity, and at-risk drinking). Scores could range from zero to 5 risk factors.

Data Analysis

This study was designed to examine the indirect effects of perceived discrimination on behavioral cancer risk factors through stress and depressive symptoms among a subset of participants with complete data from the parent project (Figure 1). First, any potential differences between included and excluded participants on the variables of interest were assessed, as data allowed. Next, a preliminary linear regression analysis was conducted among the included participants to examine relations between each sociodemographic variable and perceived discrimination while controlling for other sociodemographic variables. Finally, main analyses consisted of a series of single mediation models to assess the indirect effects of perceived discrimination on each behavioral cancer risk factor through stress and depressive symptoms. Each model was adjusted for age, sex, partner status, total annual household income, educational level, and employment status. Analyses involving overweight/ obesity status additionally controlled for fruit and vegetable intake and physical activity. These models were run using an INDIRECT macro⁵² in SPSS 19.0, which also provided information regarding the total effects of perceived discrimination on each behavioral cancer risk factor (ie, the direct associations between these variables). Indirect effects were specifically tested using a non-parametric, bias-corrected bootstrapping procedure. ⁵³ The bootstrapping procedure generated an empirical approximation of the sampling distribution of the product of the estimated coefficients in the indirect paths using 5000 resamples from the data set. The proportion of the effect that was mediated (PME) in each model was calculated by dividing the indirect effect by the total effect (PME = ab/c).⁵⁴

Figure 1.

Hypothesized Conceptual Model of the Total (c path) and Indirect Effect (ab paths) of Perceived Discrimination on Behavioral Risk Factors for Cancer Through Proposed Mediators

RESULTS

Participant Characteristics

The current study consisted of the 1363 participants with complete data (ie, no missing data). Participants (25.0% male) were 45.4 years old on average (SD=12.5). Three quarters of participants were employed, 49.4% reported at least a Bachelor’s Degree, and 35.2% reported an annual household income of at least $80,000. See Tables 1 and 2 for participant characteristics and their interrelations with other study variables.

Table 1.

Participant Characteristics and Their Interrelations

Participant Characteristics	Mean (SD) / [Percentage]	N	1	2	3	4	5	6	7	8	9
1. Age	45.4 (12.5)	1363	1.000
2. Sex			.031	1.000
Female	[75.0]	1022
Male (REF)	[25.0]	341
3. Education			.048	.097**	1.000
<  Bachelor’s Degree (REF)	[50.6]	690
Bachelor’s Degree	[30.0]	409
≥ Master’s Degree	[19.4]	264
4. Annual household income			.055*	−.140**	.344**	1.000
< $40,000 (REF)	[25.0]	341
$40,000–79,999	[39.8]	542
≥ $80,000	[35.2]	480
5. Partner status			.179**	−.217**	.046	.421**	1.000
Partner	[43.6]	594
No Partner (REF)	[56.4]	769
6. Employment status			−.097**	−.045	.198**	.294**	.052	1.000
Employed	[75.2]	1025
Unemployed (REF)	[24.8]	338
7. Perceived discrimination	20.4 (7.3)	1363	−.017	−.085**	−.004	−.019	.011	−.009	1.000
8. Stress	4.6 (3.0)	1363	−.177**	.051	−.141**	−.165**	−.045	−.044	.148**	1.000
9. Depressive symptoms	5.7 (5.0)	1363	−.109**	.067*	−.174**	−.203**	−.102**	−.104**	.178**	.674**	1.000

^**p <. 01;

^*p < .05

Note.

Interrelations between participant characteristics were evaluated using Pearson correlations for continuous-continuous variable associations, Point-biserial correlations for continuous-binary associations, Spearman correlations for continuous-ordinal and ordinal-ordinal associations, Phi coefficients for binary-binary associations, and Rank biserial coefficient for ordinal-binary associations. Discrimination = Day-to-Day Unfair Treatment Scale; Stress = Perceived Stress Scale; Depressive symptoms = Center for Epidemiological Studies Depression 10-item scale. REF= Reference group.

Table 2.

Relations of Participant Characteristics and Behavioral Risk Factors for Cancer

	Overweight/ obese	Inadequate F&V intake	Current smoking	Insufficient physical activity	At-risk drinker	# of risk factors
Participant Characteristics	No (REF)/Yes N = 229/1134	No (REF)/Yes N = 1140/223	No (REF)/Yes N = 1247/116	No (REF)/Yes N = 1004/359	No (REF)/Yes N = 1296/67	Mean/Std 2.1 (±0.8)
1. Age	.223**	.007**	−.011	.004	−.056*	.056*
2. Sex	−.042	.040	−.139**	.157**	.006	.001
Female
Male (REF)
3. Education	−.080*	.002	−.297**	.054	−.136*	−.064*
< Bachelor’s Degree (REF)
Bachelor’s Degree
≥ Master’s Degree
4. Annual household income	.015	.001	−.214**	−.008	−.023	−.045
< $40,000 (REF)
$40,000–79,999
≥ $80,000
5. Partner status	.051	.015	−.024	−.008	−.029	−.004
Partner
No Partner (REF)
6. Employment status	.015	−.090**	−.068*	.043	−.034	.040
Employed
Unemployed (REF)
7. Perceived discrimination	.029	.010	.039	−.014	.048	.028
8. Stress	.013	−.022	.086**	.054*	.055*	.092**
9. Depressive symptoms	.045	−.026	.105**	.054*	.042	.112**

^**p < .01;

^*p < .05

Note.

Interrelations between participant characteristics and risk factors were assessed using Pearson correlations for continuous-continuous variable associations, Point-biserial correlations for continuous-binary associations, Spearman correlations for continuous-ordinal and ordinal-ordinal associations, Phi coefficients for binary-binary associations, and Rank biserial coefficient for ordinal-binary associations. Discrimination = Day-to-Day Unfair Treatment Scale; Stress = Perceived Stress Scale; Depressive symptoms = Center for Epide-miological Studies Depression 10-item scale. REF= Reference group. Std = standard deviation.

Analyses failed to support any statistically significant differences between the included participants (N = 1363) and the additional participants from the parent project with incomplete data (p values ≥ .44).

Preliminary Analysis

See Table 3 for results of a preliminary analysis assessing differences in perceived discrimination as a function of sociodemographic variables. Controlling for the other sociodemographic variables, results indicated that men reported higher levels of perceived discrimination than women (β=−1.54, t= −3.27, p = .001).

Table 3.

Adjusted Relations of Sociodemographics and Perceived Discrimination

Variables	Unstandardized Coefficients β	Std. Error	t	p
Age	−.010	.016	−.620	.533
Sex
Female	−1.540	.471	−3.27	.001
Male (REF)
Education
< Bachelor’s Degree (REF)
Bachelor’s Degree	−.543	.473	−1.15	.251
≥ Master’s Degree	1.042	.566	1.84	.066
Annual household oncome
< $40,000 (REF)
$40,000–79,999	−.186	.536	−.350	.728
≥$80,000	−.368	.612	−.600	.548
Partner status
Partner	−.002	.449	−.010	.996
No Partner (REF)
Employment status
Employed	−.158	.481	−.330	.742
Unemployed (REF)

Note.

Results represent relations between each sociodemographic variable and perceived discrimination controlling for the other sociodemographic variables, as assessed using linear regression.

Main Analyses

None of the direct associations between perceived discrimination and behavioral risk factors were significant (p values ≥ .06); however, there were a few significant indirect effects. Specifically, perceived discrimination had an indirect effect on smoking status (CIs_.95= .0006, .0096; .0018, .0115) and the total number of risk factors (CIs_.95= .0006, .0028; .0012, .0036) through both stress and depressive symptoms. Specifically, greater discrimination was associated with greater stress and depressive symptoms, which were associated with a higher likelihood of current smoking and a greater number of behavioral risk factors for cancer. Additionally, the indirect effect of discrimination on overweight/obesity status through depressive symptoms was significant (CIs_.95= .0013, .0108). Specifically, greater discrimination was associated with greater depressive symptoms, which were associated with higher odds of overweight/obesity. Finally, both stress and depressive symptoms were suppressors of the effect of perceived discrimination on insufficient physical activity (CIs_.95= .0001, .0060; .0005, .0069). Suppression occurs when the association between perceived discrimination and insufficient physical activity increases (rather than decreases) when the mediator is included in the model.⁵⁵ This inconsistent mediation (suppressor effect) occurred because the directional relation between discrimination and insufficient physical activity was negative, whereas the directional relations between stress/depressive symptoms and insufficient physical activity were positive in this sample. No indirect effects emerged for fruit and vegetable intake or at-risk drinking. See Table 4 for a summary of the mediation results.

Table 4.

Results of Adjusted Models Examining Relations of Discrimination and Behavioral Risk Factors Through Proposed Mediators

Mediator = Stress
	Coefficient Estimates					BCa
Behavioral Risk Factor	a path	b path	c path	c’ path	ab	95%	CI	PME (%)
Overweight/obese	0.063	0.0488	0.0121	0.0089	0.0031	−0.0003	0.0069	-
Inadequate F&V intake	0.063	−0.0150	0.0054	0.0064	−0.0010	−0.0045	0.0023	-
Current smoking	0.063	0.0755	0.0113	0.0066	0.0047	0.0006	0.0096	41.59
Insufficient physical activity	0.063	0.0427	−0.0002	−0.0030	0.0027	0.0001	0.0060	<0
At-risk drinking	0.063	0.0532	0.0311	0.0278	0.0033	−0.0022	0.0092	-
Total number of risk factors	0.063	0.0246	0.0034	0.0019	0.0015	0.0006	0.0028	44.12
Mediator = Depressive symptoms
	Coefficient Estimates					BCa
Behavioral Risk Factor	a path	b path	c path	c’ path	ab	95%	CI	PME (%)
Overweight/obese	0.124	0.0456	0.0121	0.0069	0.0056	0.0013	0.0108	46.28
Inadequate F&V intake	0.124	−0.0170	0.0054	0.0076	−0.0020	−0.0065	0.0016	-
Current smoking	0.124	0.0498	0.0113	0.0043	0.0062	0.0018	0.0115	54.87
Insufficient physical activity	0.124	0.0271	−0.0002	−0.0040	0.0034	0.0005	0.0069	<0
At-risk drinking	0.124	0.0193	0.0311	0.0287	0.0024	−0.0038	0.0082	-
Total number of risk factors	0.124	0.0180	0.0034	0.0012	0.0022	0.0012	0.0036	64.71

Note.

Results were calculated using the SPSS INDIRECT macro syntax for estimating models with continuous (or binary) outcomes and continuous mediators. Ordinary least squares regression is used for continuous outcomes. The a path is the relation between perceived discrimination and the mediator. The b path is the relation between the mediator and the behavioral risk factor, controlling for perceived discrimination. The c path is the total effect of perceived discrimination on the behavioral risk factor. The c’ path is the relation of perceived discrimination and the behavioral risk factor in the mediated model. ab is the mean of the indirect effect estimates calculated across 5000 bootstrap samples. BCa 95% CI= Bias corrected and accelerated 95% confidence interval. PME = proportion of the mediated effect. Discrimination = Day-to-Day Unfair Treatment Scale; Stress = Perceived Stress Scale; Depressive symptoms = Center for Epidemiological Studies Depression 10-item scale. Each model was adjusted for age, sex, partner status, total annual household income, educational level, and employment status. Analyses involving overweight/obesity status additionally controlled for fruit and vegetable intake and physical activity.

DISCUSSION

As suggested by biopsychosocial theory and the stress and coping framework, results of the current study support that perceived discrimination is a psychosocial stressor that may heighten stress and depressive symptoms, which in turn may engender unhealthy behavioral patterns, perhaps in attempt to cope with psychological symptomatology. Specifically, findings suggest that perceived discrimination may indirectly influence smoking via its effects on stress and depressive symptoms, and overweight/obesity status via its effects on depressive symptoms. The results linking perceived discrimination and current smoking via stress complement prior findings citing similar indirect effect among young African Americans females¹⁹ and a population-based sample of adults.¹⁵ The current study, however, was the first to examine and highlight the indirect role of depressive symptoms in relations between perceived discrimination and smoking and overweight/obesity, respectively. Results also support indirect effects of perceived discrimination on the total number of behavioral risk factors for cancer. Greater perceived discrimination was associated with higher stress and depressive symptoms, which were associated with a greater number of behavioral risk factors for cancer. This result is noteworthy given that, although each modifiable behavior examined in this study is thought to increase cancer risk independently, the combination of multiple risk factors may work synergistically to increase risk for chronic disease and related mortality to a greater degree.^56–59 As a result, church-based interventions to reduce cancer risk among African Americans, at least among African Americans from the South, may need to address experiences of discrimination as well as the stress and depression that these may engender.

Despite the presence of indirect effects for some behavioral cancer risk factors, perceived discrimination did not affect all of the risk factors examined via stress or depressive symptoms. In the case of at-risk alcohol use, this may be attributable to the low prevalence of at-risk drinking (4.9%) in the current church-going sample, which is consistent with research suggesting that religious attendance is associated with lower alcohol use in African Americans.⁶⁰ Thus, null findings in this regard may not be generalizable to more population-based samples of African Americans. Similarly, the current results failed to support indirect associations of discrimination and the failure to meet daily recommendations for fruit and vegetable intake via stress or depressive symptoms. Future studies might explore whether discrimination affects other aspects of dietary consumption via relations with stress and depressive symptoms, such as an increased consumption of foods with high concentrations of fats and sugars (ie, “comfort foods”). A number of studies suggest that stress, and the experience of negative affect more generally, decreases the quality of dietary intake in this regard, particularly among women [cf. ^38,61–65]. Finally, results indicated that stress and depression were suppressors of the relation between perceived discrimination and insufficient physical activity, rather than mediators of that relation. These results were driven by the negative directional relation between discrimination and insufficient physical activity countered with the positive directional relation between stress/depressive symptoms and insufficient physical activity. Although this relation was not statistically significant in this sample, at least one previous study has supported significant links between perceived discrimination and higher rates of physical activity, but only among young African American women.²³ The literature on the relation between discrimination and physical activity, however, is mixed overall with most studies citing null effects as was the case in this sample.^16,23,27 Additional research, perhaps using a more complex model than the one tested in the current study, is needed to better understand how discrimination might affect physical activity and through what mechanisms. The potentially moderating role of sex might also be of interest in this regard.

In the current study, there were no significant direct relations between discrimination and modifiable behavioral risk factors for cancer, including overweight/obesity, inadequate fruit and vegetable intake, smoking, insufficient/low physical activity, or at-risk alcohol use. These results differ from some previous studies that have supported significant associations between discrimination and current smoking,^15–22 problematic alcohol use,^16,23–26 and less healthy eating.²³ However, results are consistent with other studies citing a lack of association between discrimination and insufficient physical activity rates^16,23,27 and body mass index. ^28,29 Discrepancies between the current findings and those of other studies may reflect differences in sample characteristics or measurement tools. Future studies in this area should seek to include a more diverse population of African American adults.

The present study was cross-sectional in nature, precluding assumptions of temporal associations or causality. In addition, the potential for reverse causality in mediational analyses using crosssectional data cannot be ruled out. Future studies might examine mediational pathways between discrimination and behavioral cancer risk factors using longitudinal data that can elucidate temporal associations. An additional limitation concerns the convenience sample of church-based African American adults from a large metropolitan city in the South. Participants were largely female and generally well-educated, and results may not generalize to other populations. Future research should seek to include larger, population-based samples to examine indirect relations of discrimination and behavioral risk factors through stress and depressive symptoms. Also, the current study exclusively focused on everyday discrimination. Relations of other conceptualizations of discrimination [eg, discrimination associated with major life events⁶⁶] and behavioral risk factors for cancer were not examined in this study but could be the focus of future research. In addition, the extent to which recall bias affected reports of perceived discrimination is unknown, as is information on the recency or chronicity of discrimination experienced. Finally, the current study examined 2 potential psychosocial mediators of relations between discrimination and risk factors: stress and depressive symptoms. Future research might address other potential mediators of these relations, such as serious/severe psychological distress,^2,34 which were not measured in the current study.

Despite limitations, the current study is strengthened by a large sample of African American adults, assessment of multiple behaviors related to cancer risk, statistical control of important sociodemographic covariates, and examination of psychosocial mechanisms underlying associations between discrimination and cancer risk. Results suggest that experiences with everyday discrimination may engender engagement in certain behavioral risk factors for cancer through their associations with greater stress and depressive symptoms. In light of these results, future research should strive to identify protective factors that might improve coping with discrimination and/or prevent discrimination- associated stress from triggering cancer risk behaviors. Although research does not clearly support the use of any one protective strategy in this regard, social support, racial/ethnic identity, and religious coping each show some promise for preventing mental and physical health effects of discrimination.^8,67–70 It is our hope that this ongoing area of research will lead to the development and dissemination of interventions that pay close attention to psychological responses to discrimination (ie, stress and depressive symptoms) in order to reduce cancer risk, improve quality of life, and eliminate health disparities for African Americans. The current work adds to the understanding of how discrimination might potentially affect the health of African American adults, but more preventative work is needed on a societal level to combat racism by eliminating the structural and social determinants of current discriminatory practices across the nation.