Associations of Cumulative Perceived Stress with Cardiovascular Risk Factors and Outcomes: Findings from The Dallas Heart Study

Study Population

The DHS is a multiethnic, population-based cohort study of residents in Dallas County. Self-Identified Black individuals were intentionally oversampled to encompass 50% of the cohort [14]. The original DHS participants were enrolled from 2000-2002. Participants and their spouses or significant others were invited to attend DHS phase 2 (DHS-2), a follow up examination performed between 2007-2009. DHS-2 included 3403 participants who underwent a comprehensive health examination including collection of extensive survey information, measurement of blood pressure, collection of blood and urine samples, cardiac and brain magnetic resonance imaging (MRI), accelerometry, and cardiac fitness testing [15].

Details of the DHS design and cohort have been reported previously [16]. For the present study, we included participants from DHS-2 who completed the questionnaire assessments outlined below and excluded participants with prevalent CVD or missing stress assessment data. The final study population included 2975 participants (see supplementary figure 1). All participants provided written informed consent and the study was approved by the University of Texas Southwestern Institutional Review Board

Generalized Perceived Stress (PSS-4)

Generalized perceived stress was assessed using the 4-item version of the Perceived Stress Score (PSS). The PSS is a reliable measure of an individual’s sense of control and confidence in handling circumstances over the past month and has been validated in low literacy populations[17, 18]. The four questions ask about the degree to which respondents feel in control of important aspects in their life, their confidence in ability to handle personal problems, how often they felt things were going their way and how often difficulties were overwhelming. Participants responded to each question on a five-point scale ranging from never (1) to very often (5). Of the four items, two of the questions were positively worded and thus were reverse coded. There is no established cut off for high stress on the PSS. Responses were summed to create a total score which ranged from 4-20.

Perceived Psychosocial Stress

The psychosocial variables: presence of depressive symptoms, lack of calm, and lack of energy were assessed through self-reported frequency of experiencing these feelings using 4- or 5-point Likert style responses. These questions were reproduced from the English version of the Medical Outcomes Study Short Form (SF-36), a 36-item questionnaire that measures eight dimensions of perceived health status and has been repeatedly validated for its internal reliability across multiple cohorts [19, 20]. The questions used are constituents of the mental component score, with a total score range from 3-15.

Perceived Financial Stress

Participants were asked to rate the stress they experienced in two areas: “Hard to afford basics” and “Hard to afford medical care”. Perceived stress associated with these measures of financial hardship was assessed using 4-point Likert scale for each question. Scores were summed to create a total score range of 2-8.

Perceived Neighborhood Stress

Participants were asked to respond using a 5-point Likert scale to 18 questions about perceptions of their neighborhood environment. A higher score on a 1-5 scale for each neighborhood perception question represented a less favorable perception of that neighborhood characteristic. Participants were also surveyed about the length of time they had lived in their neighborhood. Methods of the Neighborhood Perception Assessment have been described previously [11]. Briefly, principal components factor analyses was used to define factors from the neighborhood questionnaire data. Neighborhood questions with a loading score of 0.40 or higher were used to define the theme of each factor. Three factors were identified that were associated with neighborhood stress: (1) neighborhood violence, (2) physical environment, and (3) social cohesion. Numeric values assigned to Likert scale answers for a factor’s questions were summed to calculate a factor-related perception score. The total neighborhood perception score was the sum of factor-related perception scores and ranged from 10 to 49 with higher scores representing a more unfavorable perception of neighborhood environment.

Associations of Cumulative Stress Score with CVD Risk Factors, Health Behaviors and Intermediate Phenotypes

In linear regression models adjusting for age, gender, and race/ethnicity, higher CSS was associated with higher SBP, DBP, BMI, waist circumference, HbA1c, hs-CRP, and HOMA-IR (Table 2, Model 1). There was also a significant association between CSS and accelerometer measured sedentary time and an inverse association with MVPA. All of these associations persisted after additional adjustment for income and education. The prevalence of traditional risk factors, including diabetes mellitus, current smoking, and hypertension, was higher among those with higher cumulative stress after adjustment for age, sex, and race/ethnicity though only current smoking status and hypertension achieved statistical significance after additional adjustment for education and income (Table 3). No significant association was observed between CSS and intermediate CV phenotypes such as left ventricular mass, coronary artery calcium (CAC) score or cardiovascular biomarkers such as hs-TnT or NT-proBNP (Tables 2 and 3). Associations of individual components of the CSS with health behaviors and risk factors is shown in Supplementary Table 2. Testing for interaction of the CSS by race/ethnicity, gender and age demonstrated few interactions (Supplementary table 3). There were interactions between CSS and gender for BMI and waist circumference that were further investigated with subgroup analyses and showed a stronger association of cumulative stress with obesity measures among women than men.

Associations of Cumulative Stress Score with CVD Outcomes

Over a median of 12.4 years follow up, ASCVD occurred in 136, Global CV in 202 and All-Cause Mortality in 211 participants. In Cox proportional hazards models adjusting for age and gender, CSS was associated with higher incidence of ASCVD (HR 1.46 per standard deviation [SD] increment, 95% CI 1.25-1.71), global CVD (HR 1.39, 95% CI 1.22-1.58), and all-cause mortality (HR 1.42, 95% CI 1.25-1.61). After additional adjustment for race/ethnicity, income and education, associations were modestly attenuated but remained significant (Figure 1). In a fully adjusted model that added adjustment for traditional risk factors (Hypertension, Diabetes, Smoking, Hyperlipidemia), associations with ASCVD (HR 1.22, 95% CI 1.01-1.47) and Global CVD (HR 1.20, 95% CI 1.03-1.40) remained significant though associations with all-cause mortality were attenuated (HR 1.13, 95% CI 0.97-1.32(Figure 1). No interactions of the CSS with age, sex, race/ethnicity, income, or education were observed for any of the endpoints.

The individual stress components, PSS-4, psychosocial, financial and neighborhood stress, were each independently associated with ASCVD, Global CVD, and All-Cause Mortality in analyses adjusted for age and gender (supplementary figure 2, model 1). In the fully adjusted model, which added race/ethnicity, income, education, and traditional risk factors, neighborhood stress was independently associated with ASCVD and global CVD while psychosocial stress was associated with global CVD and all-cause mortality (supplementary figure 2, model 2).

Prior Literature

Existing literature has not directly explored associations of cumulative stress across multiple domains with incident CVD and mortality. Prior studies have demonstrated associations between increased occupational, marital, and psychosocial stress with cardiovascular disease [1–6, 24]. These studies have important limitations, however, including brief or single item assessments of stress and lack of generalizability to contemporary and diverse populations[25–27]. Moreover, studies of perceived stress and CVD outcomes have largely considered only single domain assessments. In the Jackson Heart Study population of African Americans, participants with moderate to high financial stress were found to have increased risk of incident coronary heart disease after controlling for traditional risk factors, sociodemographic factors, and access to care [9], findings concordant with those from our study. However, financial stress was not considered in the context of other stressors in that study.

Redmond et al investigated associations of perceived stress with CHD and all-cause-mortality in the REGARDS cohort[13] using the same PSS-4 included in our cumulative score. They found increased hazard for incident CHD among participants reporting the highest level of stress, with effect modification by income, such that stronger associations were seen in the low-income subgroup. While we did not find effect modification by income, we do see similar associations of higher cumulative stress score with lower income and education as was seen in REGARDS. A study by Albert et al investigated the relationship between a cumulative stress model and ideal cardiovascular health metrics as defined by the American heart Association’s 2020 strategic impact goals in the Women’s Health Study [28]. Similar to our findings, the investigators found that cumulative stress levels were higher in minoritized participants, with Hispanic, Black, and Asian participants having higher cumulative stress than White participants. Though the study did not report associations between their cumulative stress model and cardiovascular outcomes, it considered similar components of stress (financial, neighborhood, life trauma) as were included in our cumulative score. Thus, our study builds on these prior findings by demonstrating for the first-time associations of a cumulative stress assessment with important clinical CVD outcomes.

In our analyses, individual subcomponents of stress including the PSS-4 appeared to be less strongly correlated with CV outcomes than the cumulative stress score. Additionally, although in univariable analyses the individual subcomponents were all associated with incident ASCVD, Global CVD, and all-cause mortality, most of these relationships were attenuated after adjusting for sociodemographic factors and traditional risk factors. These findings suggest that assessments of individual domains of stress incompletely capture the cumulative impact of stress among diverse populations. In contrast to results from individual stress domains, CSS remained robustly associated with ASCVD and Global CVD after multivariable adjustments for demographics, income, education, and traditional risk factors. Beyond supporting the incremental value from the cumulative stress model over individual components, this finding also suggests that effects of cumulative stress on CVD may not be entirely mediated through traditional risk factors.

Mechanisms Linking Cumulative Stress with CVD Outcomes

It is possible that our study underestimates associations of perceived stress with risk factors, as well as the contribution of risk factor burden to increased CVD risk among individuals with increased stress, due to the single time point assessments of perceived stress, health behaviors and risk factors. However, it is also plausible that increased perceived stress enhances CVD risk through alternative pathways. Emerging data on the biology of stress and adversity has expanded understanding of the pathophysiology of chronic stress and disease risk. For example, a correlation between resting amygdala activity and aortic vascular inflammation has been demonstrated, that is linked through hematopoiesis stimulated by chronic stress. This relationship has been termed the “neural hematopoietic inflammatory axis”[29] [30]. We found significant associations of the CSS with hs-CRP in our study. Other inflammatory mediators of ASCVD, such as TNF alpha, Interleukin 6 and NFKB [33] have been associated separately with psychosocial stress and depression[34, 35]. The association between these markers and cumulative stress were not directly explored in the present study but are promising candidates for future studies.

Enhanced amygdala activity and subclinical cardiovascular disease have been linked with chronic discrimination [31]. Upregulation of stress hormone secretion related to the sympathetic-adrenal-medullary or hypothalamic-pituitary-adrenal axes is another proposed mechanism for the observed association between chronic stress and CVD [32] [33]. Although we did not investigate a broad stress hormone panel, we found that HOMA-IR and HbA1c were associated with increased CSS, and it is plausible that chronic cumulative stress may impact insulin resistance and glycemic control, for example, through mechanisms such as the release of stress hormones like cortisol.

Strengths and Limitations

Our study has multiple strengths including enrollment of a large and diverse multiethnic population, use of validated instruments to measure perceived stress across multiple domains, assessment of multiple self-reported sociodemographic factors, deep phenotyping of participants, and rigorous adjudication of incident events during follow up. In addition to examining a cumulative stress instrument with synergistic components, we also investigated the association of individual stress components with cardiovascular outcomes.

Several important limitations merit comment. First, stress and CVD risk factors were only examined at a single time point, which limits causal inferences and consideration of the dynamic nature of stress over time. Second, although the study adjusted for CVD risk factors and health behaviors there remains a risk of residual confounding from unmeasured factors associated with adverse outcomes. Finally, even with a multidimensional model of cumulative stress we were unable to assess important factors such as discrimination based on gender, racial/ethnicity, or sexual orientation, traumatic life events, or occupational stress given lack of adequate survey data for these areas.

Clinical and Public Health Implications

Our findings suggest that screening for perceived stress and implementing efforts to reduce perceived stress levels and modify risk factors associated with stress could have favorable effects on CVD health. We found significant associations of cumulative stress with health behaviors such as smoking, sedentary time and moderate to vigorous physical activity, as well as CVD risk factors such as hypertension and obesity. Importantly, each of these behaviors and risk factors are modifiable. Thus, leveraging interventions aimed at health behavior and risk factor modification in individuals with high stress levels may be beneficial. Exercise, in particular, may have beneficial effects on both stress and CVD outcomes as it improves baroreflex dysregulation and blood pressure homeostasis observed with stress [37].

We found that associations of cumulative stress with CVD risk factors, health behaviors, and outcomes were largely similar across important subgroups defined by gender, race/ethnicity, age, income, and education. This suggests that pathophysiology of cumulative stress may be similar among diverse adults. On the other hand, the burden of chronic stress is born disproportionately by younger people, women, Black and Hispanic individuals, and those with lower education and income. For that reason, cumulative stress may be a particularly important risk marker or risk factor in minoritized individuals and those with lower socioeconomic status. In this regard, it may function like smoking, hypertension, obesity, and diabetes, which also demonstrate similar associations with outcomes but are more prevalent in individuals from minoritized groups and lower socioeconomic strata [38, 39]. It also stands to reason that high levels of stress may interfere with positive behavior change. Recognizing this differential stress burden is important when considering potential targeted efforts to mitigate stress-associated risk and to reduce health care disparities.

Targeted stress reduction or mitigation strategies such as transcendental meditation have been reported to decrease risk factors such as blood pressure ([40–42], smoking [43], and hypercholesterolemia [44], as well as reducing cardiovascular mortality in several small studies [45, 46] in diverse populations. Transcendental meditation practices, of which the goal is to achieve a wakeful hypometabolic state, aim to reduce the chronic emotional, physiological, and sympathetic arousal [40]. Community engaged programs are likely needed to co-design interventions with these stress reduction methods; these should focus on those most impacted by cumulative stressors in at risk populations and provide the social support and structure to sustain this change. Finally, social services interventions may help to in alleviate stress related to non-psychosocial stressors such as financial and neighborhood stress[47].

Future Directions

Coping mechanisms and resilience likely modify the effects of chronic stress on cardiovascular disease. Future studies should evaluate the relationships between resilience, chronic stress, and CVD outcomes and test whether individuals with high cumulative stress may have cardiometabolic benefits from behavioral interventions that focus on strengthening of coping and resilience mechanisms. Screening for stress as a potential comorbidity in primary care settings should be studied, and interventions aimed at mitigating perceived stress explored. Our findings suggest that strategies that consider general mitigation strategies targeted to a full spectrum of stress, rather than individual stressors, may have greater yield.