|Home | About | Journals | Submit | Contact Us | Français|
Research that examines the associations of psychosocial factors with incident hypertension among African Americans (AA) is limited. Using Jackson Heart Study (JHS) data, we examined associations of negative affect and stress with incident hypertension and blood pressure (BP) progression among AA.
Our sample consisted of 1,656 normotensive participants at baseline (2000–2004) (mean age 47±12; 61% women). We investigated associations of negative affect (cynical distrust, anger-in, anger-out, and depressive symptoms) and stress (perceived stress, weekly stress inventory (WSI)-event, WSI-impact, and major life events) with BP progression (an increase by one BP stage as defined by JNC VII) and incident hypertension by examination 2 (2005–2008). Poisson regression analysis was utilized to examine the prevalence ratios (PRs; 95% confidence interval (CI)) of BP tracking and incident hypertension with psychosocial factors, adjusting for baseline age, sex, socioeconomic status (SES), and hypertension risk factors.
Fifty-six percentage of the sample (922 cases) had BP progression from 2005 to 2008. After adjustment for age, sex, and SES, a high anger-out score was associated with a 20% increased risk of BP progression compared to a low anger-out score (PR 1.20; 95% CI 1.05–1.36). High depressive symptoms score was associated with BP progression in the age, sex, and SES-adjusted model (PR 1.14; 95% CI 1.00–1.30). High WSI-event scores were associated with BP progression in the fully adjusted model (PR 1.21; 95% CI 1.04–1.40). We did not observe significant associations with any of the psychosocial measures and incident hypertension.
Psychosocial factors were associated with BP progression, with the strongest evidence for number of stressful events that occurred.
Hypertension is a risk factor for cardiovascular disease (CVD), affecting 77.9 million adults in the United States.1,2 While the prevalence among men and women is nearly equivalent (33.6% and 32.3%, respectively),3 previous studies report a higher prevalence of among African Americans (AA) than other groups.4,5 Although rates of control have improved among AA, disparities persist6,7 and continue to increase.3 Studies also report a higher incidence of hypertension in AA than Whites.8–10
While hypertension is attributed to known risk factors (i.e., age, diet, family history),5,11,12 psychosocial factors may also contribute to hypertension risk and related CVD risk factors.13–16 Yan et al. 17 found that greater hostility and time urgency–impatience were associated with a higher risk of hypertension among AA and White young adults. Cohen and colleagues18 discussed the impact of mental health factors (i.e., depression, anxiety, and stress) and the extent to which they are associated with CVD and related risk factors. Research has reported that greater levels of depressive symptoms were associated with incident hypertension,19,20 and that increased negative affect was associated with hypertension risk among AA and Whites.21
There is limited research on psychosocial factors and blood pressure (BP) risk in large AA samples. Previous studies have examined few psychosocial measures, which limits the ability to fully capture the extent to which psychological ill-being predicts BP outcomes. Since research has primarily examined psychosocial factors and incident hypertension as an outcome, it is important to examine the extent to which psychosocial factors are associated outcomes other than incident hypertension (such as BP progression and changes in systolic and diastolic BP). Using Jackson Heart Study (JHS) data, we hypothesized that higher levels of negative affect (cynical distrust, anger-in, anger-out, and depressive symptoms) and stress (global stress, weekly stress, and major life events (MLEs)) are associated with BP progression, incident hypertension, and changes in systolic and diastolic BP among AA between 2000–2004 and 2005–2008.
The JHS, an observational study of the etiology of CVD among AA, includes 5,301 men and women (21–95 years old) at the baseline examination (2000–2004) living in the tri-county area of Jackson, MS. Examination 2 (2005–2008) included 4,203 participants. Each examination included a physical examination, blood/urine analyses, and interview questions including psychosocial measures. Further details about data collection are described elsewhere.22,23 The study was approved by the Institutional Review Boards of the 3 participating universities and participants provided informed consent.
Standardized protocols were used for examinations 1 and 2. Systolic BP (SBP) and diastolic BP (DBP) were measured twice in the right arm of participants using the random-zero BP sphygmomanometer (Hawksley and Sons Limited, Sussex, England). The first BP measurement was taken after the participant rested for 5 minutes in a seated position; the second BP was taken after waiting an additional minute. The average of the 2 measurements comprised the BP measure for the examination.
For the primary analysis, we examined BP progression (increase) by at least one category according to the Joint National Committee (JNC VII) on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. We also performed the analysis using JNC VI criteria; however, since results varied little, we present only results for JNC VII criteria. We also examined incident hypertension and longitudinal changes in SBP and DBP. The categories utilized for BP progression included normal BP (<120/80mm Hg), prehypertension (120–139/80–89mm Hg), stage 1 (140–159/90–99mm Hg) and stage 2 (≥160/≥100) hypertension. Categories for BP progression were classified according to JNC VII guidelines which showed the level of progression and how much SBP and DBP progressed. Incident hypertension was defined as new cases of hypertension from examination 1 to 2 and was derived from a BP of 140/90mm Hg or higher, reported taking antihypertensive medications, pill bottle review, or being told by a provider that one has hypertension. Longitudinal changes in SBP and DBP were measured as continuous changes between examination 1 and 2.
Baseline psychosocial measures included cynical distrust, anger-in and -out, depressive symptoms, global stress, weekly stress inventory (WSI)-event, WSI-impact, and MLE. Cynical distrust was measured as a component of the Cook–Medley Hostility Scale,24 where participants were asked to answer “true or false” on such questions as “….It is safer to trust nobody.” We calculated a cynical distrust score where the total average of all “true” responses was divided by 13 (range 0–13). Cronbach’s alpha was 0.76.
Anger was measured using the Spielberger trait anger scale, a 16-item scale that assesses anger-in (8 items) and anger-out (8 items).25,26 Participants were asked to rate how often they reacted to: “I keep things in”; “I express my anger.” Each item is rated from never (0) to almost always (5). The average of both was derived and ranged from 0 to 23 (anger-in) and 0 to 22 (anger-out). In general, internal consistency coefficients for the scale were 0.86–0.95, and the range for test–retest reliability coefficients was 0.65–0.75.27 For this specific study, internal consistency was good for the scale: total anger (α = 0.81), anger-in (α = 0.77), and anger-out (α = 0.77).
Depressive symptoms were measured using the 20-item Centers for Epidemiologic Studies Depression Scale (CES-D) that measures frequency of depressive symptoms.28 Participants were asked about their mood over the past week, responding to each item (e.g., “I was bothered by things that usually don’t bother me”). Item ratings ranged from 0 (“rarely/none of the time”) to 3 (“most/all of the time”). This scale ranged from 0 to 60 with higher scores reflecting greater levels of depressive symptoms. Cronbach’s alpha was 0.82.
Global perceived stress was created for the JHS29 and adapted from standardized stress scales.30,31 Participants rated the extent of chronic stress from 8 domains (e.g., employment, relationships) over a 12-month period. Choices ranged from “not stressful” (=1) to “very stressful” (=3); summing items yields a total score ranging from 0 to 24. Cronbach’s alpha was 0.72.
The WSI, developed by Jones and Brantley,32,33 is an 87-item questionnaire that measured occurrence of minor irritants (e.g., work tasks, financial challenge) during the past week and the extent of stressfulness based on a 7-point Likert-type scale (1 = not stressful; 7 = extremely stressful). We measured WSI-event, the number of stressful events endorsed (range 0–87), and WSI-impact, the sum of the subjective ratings assigned to endorsed events (range 0–493).33 Cronbach’s alpha was 0.97.
MLEs were assessed using an 11-item life events inventory.34 The MLE, administered at annual follow-up, measured whether participants experienced events in the last 12 months by answering “yes” or “no” to select items (experiencing illness, being a victim of assault). The total score was calculated by summing the “yes” (=1) responses where the range was 1–11. Since the MLE is an index rather than a true scale, Cronbach’s alpha was not calculated.
We constructed cumulative scores for negative affect and stress measures.16 Cumulative negative affect was constructed by assigning a separate score of 1 (referent) to 3 to tertiles of the cynical distrust, anger-in and -out, and depression measures separately and summing across the 4 measures (range 4–12). Cumulative stress was constructed by assigning a separate score of 1 (referent) to 3 to tertiles of the global stress, WSI-event, WSI-impact, and MLE measures separately and summing across the 4 measures (range 4–12).
Covariates included baseline age (continuous), sex (men/women (referent)), education, cigarette smoking, physical activity, diet, body mass index, type 2 diabetes, total cholesterol, SBP, and DBP. Education was classified as less than high school (referent), high school graduate to some college, or college graduate and above. Cigarette smoking status was classified as current, former, and never (referent). Physical activity was measured as a continuous sum of the 4 index scores (Active Living, Work/Occupational, Home Life, and Sport) from the JHS physical activity instrument.35 Diet was assessed using a validated 158-item questionnaire.36 Measures related to the Dietary Approaches to Stop Hypertension trial37 (percent calories from fat, sodium (mg/1,000 kcal), potassium (mg/1,000 kcal), calcium (mg/kcal), and total fiber (g/1,000 kcal)) were examined. Body mass index was calculated as weight in kilograms divided by height in meters squared. Type 2 diabetes was defined according to American Diabetes Association (ADA) 2004 criteria as fasting glucose ≥ 126mg/dl, or confirmed medication usage from the medication inventory, or self-reported use of antidiabetic medications within the past 2 weeks of the examination, or self-reported diabetes diagnosis. Total cholesterol was assayed by the cholesterol oxidase method supplied by Boehringer Mannheim Diagnostics on a Roche COBAS Fara analyzer (Indianapolis, IN). Baseline hypertension was defined as a SBP ≥ 140mm Hg or DBP ≥ 90mm Hg and/or antihypertensive medication.
Descriptive statistics were presented, in that we characterized the sample by BP progression status. Differences were tested using chi-square or analysis of variance tests for categorical and continuous variables, respectively. Next, associations of psychosocial measures with each BP end point were estimated. Multivariable Poisson regression analysis estimated the associations of psychosocial factors with BP progression with robust SEs, where prevalence (risk) ratios (PRs; 95% confidence interval (CI)) estimated the prevalence of BP progression. Psychosocial factors were measured in tertiles in order to assess threshold effects and as continuous variables in SD units. P for trend, estimated for tertiles of each psychosocial measure, represented the linear trend across categories. A sensitivity test was conducted to examine associations of psychosocial factors with BP progression from normotensive status only to stage 1 and 2 hypertension, and associations with progression from prehypertensive status only to stage 1 and 2 hypertension. Poisson regression was used to estimate these models, and fully adjusted predicted values for the 2 outcomes were graphically displayed for select psychosocial variables.
The associations of psychosocial measures (SD units) with incident hypertension were estimated using multivariable Poisson regression. Poisson regression was used for this and the BP-progression analyses because prevalence of BP progression was 56%, and incident hypertension was 37%.38 The associations of psychosocial factors with changes in SBP and DBP between examination 1 and 2 were examined using multivariable linear regression to estimate mean differences (SEs) in changes in SBP and DBP. For participants taking antihypertensive medications, BP values were imputed by adding 10mm Hg to the SBP and 5mm Hg to the DBP in each of the analyses. We also performed imputations by adding 15mm Hg to the SBP and 10mm Hg to the DBP.39,40 Since results were similar, we reported those for the first imputation.
For each BP end point, models were estimated sequentially. Model 1 adjusted for age, sex, and psychosocial factors; model 2 adjusted for model 1 plus education; model 3 adjusted for model 2 plus smoking, sodium intake, calories from fat, total fiber, physical activity, body mass index, type 2 diabetes, and total cholesterol. In the BP progression and incident hypertension analyses, we added model 4 that adjusted for model 3 plus baseline SBP and DBP. In order to avoid multicolinearity, each psychosocial measure was analyzed in separate models. A test for interactions between sex and psychosocial measures was calculated in fully adjusted models to examine potential effect modification by sex. All reported P values correspond to 2-tailed tests and were significant at the 0.05 level. Due to performing multiple test scenarios that puts us at an increased risk for type I error, we have implemented a Bonferroni method of correction, for all comparisons and presented the adjusted P values for multiple testing in Tables 2–4. The following correction for multiple testing was used for the 3 negative affect measures: 0.05/3 = 0.0167 or 0.02. The following correction was used for the 3 stress measures: 0.05/3 = 0.0167 or 0.02. Both unadjusted and adjusted P values will be considered in our analysis. All analyses were performed using STATA 13.0 (STATA Corp, College Station, TX).
Fifty-three (53) participants with missing data on hypertension and 3,301 hypertensives at baseline were excluded. Two hundred ninety-one (291) participants had missing data on education, smoking, physical activity, and diet variables. This left 1,656 participants free of hypertension at baseline in the final sample (mean age 47±12; 61% women). The median follow-up period was 5.0±0.8 years. Fifty-six percentage (56%) of the sample progressed at least one BP stage from 2000–2004 to 2005–2008 (Table 1). Incident hypertension occurred among 37.4% of normotensives at baseline, and mean increase in SBP and DBP were 6.5 and 3.7mm Hg, respectively.
Table 2 presents associations of each psychosocial factor with BP progression in sex-pooled models. There were moderate correlations between cynical distrust and anger-in (Pearson’s coefficient: 0.24; P < 0.001), depressive symptoms and cynical distrust (0.32; P < 0.001), and depressive symptoms and anger-in (0.45; P < 0.001). There were also weak-to-moderate correlations between the stress measures: MLE and WSI-event (0.08; P < 0.01), global stress and WSI-impact (0.39; P < 0.001).
A high anger-out score was associated with a 20% increased risk of BP progression compared to a low anger-out score (PR 1.20; 95% CI 1.05–1.36) (model 1). The association remained significant after adjustment for socioeconomic status (SES). There was a graded association of anger-out with BP progression (P trend = 0.02). The high anger-out values in models 1 and 2 were also significant after adjustment for multiple testing. The standardized regression analysis in model 2 confirmed this association; the risk of BP progression increased by 6% for each 1-SD increase in anger-out (PR 1.06; 95% CI 1.01–1.12). The association attenuated after adjustment for risk factors in model 3. Participants with high depressive symptoms scores had a 14% increased risk of BP progression compared to those with low scores (PR 1.14; 95% CI 1.00–1.30) after adjustment for age, sex, and SES. This association weakened after adjustment for risk factors. After adjustment for multiple testing, high depressive symptoms in model 2 was not significant; however, the continuous depressive symptoms value remained significant (P = 0.02). WSI-event analyses revealed that a greater number of minor stressors was associated with higher risk of BP progression in the 3 models (model 3, P trend = 0.04). WSI-impact was inversely associated with BP progression in each model, high (vs. low) MLE scores were associated with a 13% increased prevalence of BP progression, but estimates attenuated in the fully adjusted model. Participants with high cumulative stress scores had a 24% increased risk of BP progression than those with low scores (PR 1.24; 95% CI 1.04–1.48) after full adjustment. Associations were graded, and standardized regression coefficients confirmed this association. Associations did not change when full models were adjusted for baseline SBP and DBP. Findings remained significant when considering adjusted P values for multiple testing. There was no evidence of effect modification by sex in fully adjusted models, as indicated by the nonsignificant P values for interaction (data not shown).
Select psychosocial factors were marginally associated with BP progression from normotensive status only to stage 1 and stage 2 hypertension (P < 0.1). In fully adjusted models, anger-out, depressive symptoms, WSI-event, and cumulative stress were positively associated with risk of BP progression from prehypertensive status to stage 1 and 2 hypertension (P < 0.05) (Supplementary Figure).
We did not observe a significant association with any of the psychosocial factors and incident hypertension (Table 3). There was attenuation in the PRs from models 1 to 3 for most measures, but no significant associations.
While changes in SBP and DBP were examined, psychosocial factors were only associated with changes in DBP (Table 4). In the fully adjusted model, anger-out was associated with changes in DBP (mean difference in DBP: 2.73, P < 0.05). After full adjustment, high (vs. low) cumulative stress was associated with changes in DBP (mean difference in DBP: 3.67, P < 0.05). This association was graded from low to high cumulative stress (P for trend = 0.08) and was not significant according to the P value (0.02) for multiple testing.
This study investigated associations of psychosocial factors with BP end points among AA. To our knowledge, this is the first study to examine multiple psychosocial factors and BP end points in a large sample of AA. We found that high anger-out, depressive symptoms, WSI-event, MLE, and cumulative stress scores were associated with risk of BP progression. Associations of psychosocial factors with BP outcomes did not vary by sex. Psychosocial factors were not associated with incident hypertension. We did not observe a significant association of any of the psychosocial factors with incident hypertension. Anger-out and cumulative stress were significantly associated with changes in DBP.
High WSI-event and cumulative stress scores significantly predicted the risk of BP progression, which suggests that the number and accumulation of stressors are important for understanding BP progression. One study found that anger was associated with progression from prehypertension to hypertension over a period of 3–6 years among AA.41 Similarly, we found an association of anger-out with progression from prehypertension to hypertension after full adjustment, which suggests that the relationship is not fully explained by risk factors.
We did not observe significant interactions by sex between psychosocial factors and BP end points, although research has found that anger and depression significantly predicted incident hypertension and changes in SBP among women.20 Räikkönen et al. 20 also reported that an increase in psychological distress over time can affect individuals physiologically, which may be indicated in our pooled analysis. Previous research has also found that stress is associated with CVD risk primarily in men,42 which was not found in our study.
Previous studies have reported associations of psychosocial measures with incident hypertension.20–22 One study of normotensive, middle-aged AA and White women (n = 541) found high levels of anger and anxiety were associated with hypertension risk.20 Conversely, we did not observe any significant associations of any of the psychosocial factors with incident hypertension, which is consistent with one study that found no associations of depressive symptoms with incident hypertension among AA.43 One potential reason we did not observe any significant associations may have been due to not accounting for time to diagnosis of events as previous studies have used.
Our study also examined longitudinal changes in SBP and DBP, which has been examined by a few studies. One multiethnic study found associations of depressive symptoms and slight changes in SBP (+2.4mm Hg) and DBP (+0.8mm Hg) over a 2-year period.43 We found associations of anger-out and cumulative stress and changes in DBP, which supports the association of stress on BP.
One novel feature of our study included the examination of multiple psychosocial measures. Recent studies of psychosocial factors and BP have examined few negative affect measures.41,43 We, however, examined 4 dimensions of negative affect and stress among AA. The examination of more than one dimension is important for representing the broad array of psychological ill-being which is important for predicting BP disparities. The cumulative exposure of these factors (vs. one at a time) may be a truer depiction of how multiple stressors predict chronic diseases among AA, a highly stressed population. Cumulative negative affect did not predict BP outcomes, which is similar to another study that found no associations between cumulative negativity and subclinical disease in a multiethnic sample.16
The findings of this study should be evaluated in the context of several limitations. The study was conducted in a single metropolitan area in the Southeastern United States, possibly limiting its generalizability to other AA populations. The study sample was restricted to individuals that did not have hypertension at baseline. We recognize this may result in a survivor bias; however, the focus of the study was to identify normotensive individuals that developed hypertension over time by way of hypertension incidence and BP progression. We were unable to estimate incident hypertension using time to diagnosis. Participants that developed hypertension between examination 1 and 2 were likely placed on antihypertensive medications upon diagnosis. Although this was accounted for in our imputations that considered those taking antihypertensive medication, this may have affected our longitudinal analysis. The strengths of this study include utilizing JHS, the largest study of CVD in AA, to examine multiple dimensions of negative affect and stress and BP progression. The use of multiple BP end points is another strength. The use of BP progression longitudinally enabled us to draw causal inferences about psychosocial factors and BP progression unlike cross-sectional studies.
In summary, psychosocial factors were associated with BP progression, with the strongest evidence for number of stressful events and accumulation of stressors that occurred, possibly contributing to the clustering of stressors among this AA population. Considering the high levels of stress and affective states reported by AA may be important for preventing the development of hypertension, and the extent to which coping strategies (e.g., family networks) might mitigate the association of psychosocial factors with BP end points among AA should be considered.
Supplementary materials are available at American Journal of Hypertension (http://ajh.oxfordjournals.org).
The authors declared no conflict of interest.
We thank the Jackson Heart Study team (University of Mississippi Medical Center, Jackson State University, and Tougaloo College) and participants for their long-term commitment that continues to improve our understanding of the epidemiology of cardiovascular and other chronic diseases. The Jackson Heart Study is supported by contracts HHSN268201300046C, HHSN268201300047C, HHSN268201300048C, HHSN268201300049C, and HHSN268201300050C from the National Heart, Lung, and Blood Institute (NHLBI) and the National Institute on Minority Health and Health Disparities (NIMHD). Dr Sims is supported by the grants P60MD002249 and U54MD008176 from the NIMHD. C.D.F. and M.S. should be regarded as joint first authors.