Previous studies of the association between symptoms of anxiety or depression and coronary artery calcium (CAC) have produced heterogeneous results. Our aim was to investigate whether psychopathological symptoms were associated with CAC in a cross-sectional analysis of the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) baseline. We analyzed data from 4,279 ELSA-Brasil subjects (aged 35 to 74 years) from the São Paulo site without previous cardiovascular disease who underwent CAC score assessment at baseline. Prevalent CAC was defined as a CAC score >0. Anxiety and depressive symptoms were assessed using the Clinical Interview Schedule–Revised (CIS-R). We built binary logistic regression models to determine whether CIS-R scores, anxiety, or depression were associated with prevalent CAC. Prevalent CAC was found in 1,211 subjects (28.3%). After adjustment for age and gender, a direct association between CIS-R scores and prevalent CAC was revealed (odds ratio for 1-SD increase: 1.12; 95% confidence interval [CI] 1.04 to 1.22). This association persisted after multivariate adjustment (odds ratio for 1-SD increase 1.11; 95% CI 1.02 to 1.20). No independent associations were found for specific diagnoses of anxiety or depression and prevalent CAC. In post hoc models, a significant interaction term (p = 0.019) suggested a stronger association in older subjects. In conclusion, psychopathological symptoms were directly associated with coronary atherosclerosis in the ELSA-Brasil baseline in adjusted models, and this association seems to be stronger in older subjects.
Coronary heart disease, depressive and anxiety disorders are responsible for a large burden of disease. Although some co-morbidity between these prevalent conditions may be expected by chance, most of the evidence suggests a specific link between depressive and anxiety disorders and higher cardiovascular risk, although findings from other studies are equivocal. Previous studies on the association between anxiety or depression symptoms and coronary artery calcium (CAC) scores have reported heterogeneous results as well. In the largest study to date, Diez Roux et al studied 6,789 subjects from the Multiethnic Study of Atherosclerosis. No association was found between CAC scores and symptoms of depression or anxiety using the Center for Epidemiology Studies–Depression (CES-D) and Spielberger Trait Anxiety Inventory (STAI) scales. This is consistent with the findings of other authors. In contrast, Stewart et al analyzed data from 2,171 subjects from the CARDIA cohort and found that a CES-D score ≥16 was predictive of incident CAC after 5 years of follow-up. In this study, we studied a subsample of 4,279 Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) subjects from the São Paulo site who underwent CAC score assessment at baseline to verify whether the Clinical Interview Schedule–Revised (CIS-R) scores, and the diagnoses of anxiety or depression disorders were associated with CAC scores.
Methods
The ELSA-Brasil design, objectives, and cohort profile have been published in detail elsewhere. Briefly, it is a multicenter cohort study of 15,105 civil servants from 6 institutions in different Brazilian cities (5,061 in São Paulo site). Active or retired employees of the 6 institutions aged 35 to 74 years were eligible for the study. The baseline assessment consisted of a 7-hour examination, including questionnaires, medical measurements, and laboratory examinations. The baseline assessment took place from August 2008 to December 2010 and included the application of the CIS-R to assess nonpsychotic psychiatric morbidity. In addition, the subjects at the ELSA-Brasil site in São Paulo were invited to perform a computed tomographic (CT) examination to quantify CAC. Approvals were granted by the institutional review boards of all the centers, and all the subjects signed an informed consent form.
Trained interviewers using an adapted Brazilian-Portuguese version of the CIS-R assessed the mental symptoms and diagnoses in the ELSA-Brasil baseline. The CIS-R is a short, straightforward, structured interview validated in Brazilian Portuguese for measuring and diagnosing community nonpsychotic psychiatric morbidity and can be applied by lay interviewers with reliability. CIS-R scores range from 0 to 72. Subjects with a score of ≥12 were classified as having common mental disorder. Diagnoses of generalized anxiety disorder, obsessive-compulsive disorder, and major depressive disorder using the CIS-R questionnaire are based on the International Classification of Diseases, Tenth Revision ( ICD-10 ).
The CAC examination was performed with a 64-detector computed tomography scanner (Brilliance 64; Philips Healthcare, Best, The Netherlands). An ECG-gated prospective calcium score examination with a tube potential of 120 kV and a tube current adjusted to body habitus was performed. Images were reconstructed in 2.5-mm slice thickness using standard filtered back projection. CT images were evaluated in blinded form, by an experienced cardiologist, using semiautomatic software (Calcium Scoring; Philips Workstation). The CAC scores were expressed in Agatston units.
There are 4,317 subjects from the ELSA-Brasil site in São Paulo who underwent a CT scan for CAC evaluation and had no previous myocardial infarction, angina, stroke, or coronary revascularization. We excluded 38 subjects (0.8%) with a body mass index <18.5 kg/m 2 . Therefore, the study sample consisted of 4,279 subjects (1,948 men and 2,331 women).
For logistic regression modeling, CIS-R scores were standardized to mean 0 and SD 1 within the study sample. For main analyses, CAC scores were dichotomized as 0 or >0 Agatston units. Prevalent CAC was defined as a CAC score >0. Race was self-defined as Black, Brown, White, Asian, or Native. Because of the small number of subjects self-defined as Asian or Native, we classified subjects in these 2 groups as other races. Educational level was self-reported and categorized as lower than high school, high school, and college or above. Family monthly income was also self-reported and categorized as ≤US$1,244, US$1,245 to US$3,319, and ≥US$3,320. Body mass index was defined as weight divided by squared height and expressed in kilogram per square meter. Hypertension was defined as the reported use of medications to treat hypertension, systolic blood pressure ≥140 mm Hg, or diastolic blood pressure ≥90 mm Hg at ELSA-Brasil baseline assessment. Diabetes was defined as a medical history of diabetes mellitus, the reported use of medications to treat diabetes mellitus, fasting serum glucose ≥7.0 mmol/L (126 mg/dl), HbA1c levels ≥48 mmol/L (6.5%), or 2-hour oral glucose tolerance test ≥11.1 mmol/L (200 mg/dl). Dyslipidemia was defined as the reported use of lipid-lowering treatment or low-density lipoprotein cholesterol level ≥3.36 mmol/L (130 mg/dl). Smoking status was self-reported as never, past, or current.
Continuous variables are presented as mean (SD). Categorical variables are presented as absolute numbers and proportions. The chi-square tests, t tests, and Kruskal-Wallis tests were used where applicable. We built binary logistic regression models to determine whether CIS-R scores (as a continuous variable), common mental disorder, generalized anxiety disorder, obsessive-compulsive disorder, or major depressive disorder diagnoses were associated with prevalent CAC. These models are presented (a) crude, (b) adjusted for age and gender, and (c) fully adjusted, that is, adjusted for variables in model (b) plus race, educational level, family monthly income, BMI, hypertension, systolic blood pressure, diabetes, dyslipidemia diagnoses, and smoking status. We also ran sensitivity analyses setting the cutoff for the CAC score at 400 Agatston units.
After observing a positive association between CIS-R scores and prevalent CAC, we ran post hoc interaction models to determine whether this association was homogeneous in men and women and across the age range. We built models including all variables in fully adjusted models and included interaction terms for (a) CIS-R scores and gender and (b) CIS-R scores and age, as a continuous variable. As we found a significant interaction term for CIS-R scores and age, we present adjusted models stratified by age, using a cutoff at the median age for the sample (50 years). Statistical analyses were performed using R software (R Core Team, Vienna, Austria), version 3.1.2. The significance level was set at 0.05.
Results
Table 1 lists the characteristics of the study sample at ELSA-Brasil baseline. Prevalent CAC was found in 275 of 2,103 subjects (13%) aged <50 years and 936 of 2,176 subjects (43%) aged ≥50 years. From the 1,211 subjects with prevalent CAC, 806 (67%) had a CAC score between 0.1 and 99.9, 271 (22%) had a CAC score between 100 and 399.9, and 134 (11%) had a CAC score ≥400 Agatston units.
| Variable | Men | Women | Total (N = 4279) | ||||
|---|---|---|---|---|---|---|---|
| CAC score = 0 (N = 1174) | CAC score > 0 (N = 774) | All men (N = 1948) | CAC score = 0 (N = 1894) | CAC score > 0 (N = 437) | All women (N = 2331) | ||
| Age (years) mean (SD) | 47.6 (7.5) | 55.7 (9.0) | 50.8 (9.1) | 49.0 (7.7) | 58.3 (7.7) | 50.7 (8.5) | 50.8 (8.8) |
| White | 626 (54%) | 470 (62%) | 1096 (57%) | 1117 (59%) | 273 (64%) | 1390 (60%) | 2486 (59%) |
| Brown | 314 (27%) | 166 (22%) | 480 (25%) | 378 (20%) | 65 (15%) | 443 (19%) | 923 (22%) |
| Black | 178 (15%) | 75 (10%) | 253 (13%) | 282 (15%) | 52 (12%) | 334 (14%) | 587 (14%) |
| Other (Asian / Native) | 37 (3%) | 49 (6%) | 86 (4%) | 103 (5%) | 37 (9%) | 140 (6%) | 226 (5%) |
| Educational level | |||||||
| Lower than high school | 227 (19%) | 156 (20%) | 383 (20%) | 173 (9%) | 68 (16%) | 241 (10%) | 624 (15%) |
| High school | 551 (47%) | 243 (31%) | 794 (41%) | 810 (43%) | 154 (35%) | 964 (41%) | 1758 (41%) |
| College or above | 396 (34%) | 375 (48%) | 771 (40%) | 911 (48%) | 215 (49%) | 1126 (48%) | 1897 (44%) |
| Family monthly income | |||||||
| ≤ US$1244 | 472 (40%) | 219 (28%) | 691 (36%) | 550 (29%) | 96 (22%) | 646 (28%) | 1337 (31%) |
| US$1245 – 3319 | 483 (41%) | 262 (34%) | 745 (38%) | 915 (48%) | 198 (46%) | 1113 (48%) | 1858 (44%) |
| ≥ US$ 3320 | 214 (18%) | 288 (37%) | 502 (26%) | 422 (22%) | 141 (32%) | 563 (24%) | 1065 (25%) |
| Body-mass index (kg/m 2 ) mean (SD) | 27.1 (4.4) | 27.6 (4.2) | 27.3 (4.3) | 27.4 (5.2) | 28.1 (5.1) | 27.5 (5.1) | 27.4 (4.8) |
| Hypertension | 320 (27%) | 365 (47%) | 685 (35%) | 413 (22%) | 189 (43%) | 602 (26%) | 1287 (30%) |
| Diabetes mellitus | 205 (17%) | 249 (32%) | 454 (23%) | 259 (14%) | 128 (29%) | 387 (17%) | 841 (20%) |
| Dyslipidemia | 636 (54%) | 504 (65%) | 1140 (59%) | 969 (51%) | 316 (72%) | 1285 (55%) | 2425 (57%) |
| Smoker | |||||||
| Never | 604 (51%) | 330 (43%) | 934 (48%) | 1144 (60%) | 211 (48%) | 1355 (58%) | 2289 (53%) |
| Past | 392 (33%) | 303 (39%) | 695 (36%) | 486 (26%) | 136 (31%) | 622 (27%) | 1317 (31%) |
| Current | 178 (15%) | 141 (18%) | 319 (16%) | 264 (14%) | 90 (21%) | 354 (15%) | 673 (16%) |
| CIS-R score mean (SD) | 6.3 (6.8) | 5.9 (6.9) | 6.1 (6.9) | 9.7 (8.5) | 8.9 (8.5) | 9.5 (8.5) | 8.0 (8.0) |
| Common Mental Disorder | 219 (19%) | 118 (15%) | 337 (17%) | 627 (33%) | 125 (29%) | 752 (32%) | 1089 (25%) |
| Major Depression Disorder | 31 (3%) | 11 (1%) | 42 (2%) | 111 (6%) | 23 (5%) | 134 (6%) | 176 (4%) |
| Generalized Anxiety Disorder | 102 (9%) | 62 (8%) | 164 (9%) | 311 (17%) | 63 (15%) | 374 (16%) | 538 (13%) |
| Obsessive-Compulsive Disorder | 18 (2%) | 11 (1%) | 29 (1%) | 48 (3%) | 9 (2%) | 57 (2%) | 86 (2%) |
In bivariate analyses, subjects with prevalent CAC were older, had higher income, higher frequencies of hypertension, diabetes, and dyslipidemia (p <0.001 for all comparisons, both in men and women), current smoking (p = 0.001 for men and p <0.001 for women), and higher BMI (p = 0.005 for men and p = 0.006 for women). Women had significantly higher CIS-R scores than men (p <0.001). We also found significantly higher frequencies of common mental disorder, generalized anxiety disorder, obsessive-compulsive disorder, and major depressive disorder in women (p <0.001 for all comparisons).
Table 2 lists the odds ratios in crude and adjusted models for the association between CIS-R scores and/or psychiatric conditions and prevalent CAC. Crude models showed an inverse association among major depression disorder, generalized anxiety disorder, common mental disorder and CIS-R scores, and prevalent CAC. However, these inverse associations seem to be driven by age and gender distributions in the sample. In fact, after adjustment for age and gender, these inverse associations disappeared, and a positive association between prevalent CAC and continuous CIS-R scores was revealed. This association persisted after adjustment for educational level, family monthly income, BMI, hypertension, systolic blood pressure, diabetes, dyslipidemia diagnoses, and smoking status. In our sensitivity analyses, using a cutoff at 400 Agatston units, 1 SD in CIS-R scores were significantly associated with higher CAC scores in the age- and gender-adjusted models (odds ratio [OR] 1.24; 95% confidence interval [CI] 1.002 to 1.51; p = 0.048). However, after multivariate adjustment, this association lost statistical significance (p = 0.30), because of, at least in part, the small number of subjects with a CAC score of 400 Agatston units or higher in our sample.
