Abstract
Background
Baseline depressive symptoms are associated with subsequent adverse cardiovascular (CV) events in subjects with and without diabetes but the impact of persistent symptoms vs. improvement remains controversial.
Objective
Examine long-term changes in depressive symptoms in individuals with and without diabetes and the associated risk for adverse CV events.
Design
REGARDS is a prospective cohort study of CV risk factors in 30,000 participants aged 45 years and older.
Participants
N = 16,368 (16.5% with diabetes mellitus) who remained in the cohort an average of 11.1 years later and who had complete data.
Main Measures
Depressive symptoms were measured using the 4-item Centers for Epidemiologic Study of Depression (CES-D) questionnaire at baseline and again at a mean follow-up of 5.07 (SD = 1.66) years. Adjudicated incident stroke, coronary heart disease (CHD), CV mortality, and a composite outcome were assessed in a subsequent follow-up period of 6.1 (SD = 2.6) years.
Methods
The association of changes in depressive symptoms (CES-D scores) across 5 years with incident CV events was assessed using Cox proportional hazards modeling.
Key Results
Compared to participants with no depressive symptoms at either time point, participants without diabetes but with persistently elevated depressive symptoms at both baseline and follow-up demonstrated a significantly increased risk of incident stroke (HR (95% CI) = 1.84 (1.03, 3.30)), a pattern which was substantially more prevalent in blacks (HR (95% CI) = 2.64 (1.48, 4.72)) compared to whites (HR (95% CI) = 1.06 (0.50, 2.25)) and in those not taking anti-depressants (HR (95% CI) = 2.01 (1.21, 3.35)) in fully adjusted models.
Conclusions
The persistence of depressive symptoms across 5 years of follow-up in participants without diabetes identifies individuals at increased risk for incident stroke. This was particularly evident in black participants and among those not taking anti-depressants.
Supplementary Information
The online version contains supplementary material available at 10.1007/s11606-022-07449-w.
KEY WORDS: depressive symptoms, diabetes, stroke
Cardiovascular disease (CVD) is a leading cause of death, long-term disability, and cognitive impairment for older adults and occurs more commonly in individuals with diabetes mellitus1. Both interview-determined depression and a history of depressive symptoms are known risk factors for CVD including coronary heart disease2, incident stroke2–6, and stroke mortality7.
Several investigators have also explored whether changes in depressive symptoms across time influence stroke risk with conflicting results. In the initial study by Gilsanz et al.8, persistently elevated depressive symptoms as well as symptoms that were initially elevated but remitted over 2 years of follow-up were both associated with increased stroke risk. By contrast, in the Cardiovascular Health Study9, individuals with persistently elevated symptoms across two assessments 1 year apart were at increased risk of incident stroke while those with improved symptoms did not have significantly increased risk. In the nationally representative China Health and Retirement Longitudinal Study10, both persistent and remitted depressive symptoms were associated with an increased risk of incident CVD, with individuals with persistently elevated depressive symptoms having an almost fourfold increased risk of stroke over the subsequent 2-year follow-up period compared with those who never reported depressive symptoms.
The prevalence of major depressive symptoms in persons with diabetes mellitus (DM) is approximately twice as high as in those without DM, and these symptoms are associated with poorer adherence to treatment, worse glycemic control, and neuroendocrine changes 10–13. Similarly, poor metabolic control in persons with DM may worsen depressive symptoms 14. Our prior report showed that among participants with diabetes, depressive symptoms coupled with high levels of stress were associated with substantially increased risk of adverse cardiovascular outcomes.15 Therefore, it is critical to understand the extent to which change in depressive symptoms, including in individuals with DM, may influence subsequent cardiovascular disease, including stroke risk, and the relationship with CV risk factors3–7, 15, 16].
In summary, conflicting information is available on how changes in depressive symptoms over time impact risk for adverse CV/stroke outcomes in subjects with and without diabetes. Therefore, the purpose of the present study is to examine changes in depressive symptoms in a population-based sample of diverse subjects with and without diabetes at baseline and the associated risk for adverse cardiovascular events including stroke, nonfatal myocardial infarction (MI), and coronary heart disease (CHD) death.
METHODS
Study Population
The REasons for Geographic And Racial Differences in Stroke (REGARDS) study is a population-based, prospective cohort study of 30,239 non-Hispanic black and non-Hispanic white adults aged >45 years at baseline (2003–2007) that oversampled black participants and adults residing in the southeastern “Stroke Belt” region of the USA. A detailed description of the REGARDS study has been reported previously17. Briefly, demographics and medical history were obtained via a computer-assisted telephone interview, while anthropometrics, blood pressure, and blood and urine samples were obtained during a subsequent in-home examination. Race was self-reported. Participants were contacted every 6 months during follow-up.
Standard Protocol Approvals, Registrations, and Patient Consents
The Institutional Review Boards of participating institutions approved the study methods and written informed consent was obtained from all participants.
Depressive Symptoms (Primary Exposure Variable) and Timing of CV Event Assessment
Depressive symptoms were assessed using the validated Centers for Epidemiologic Study of Depression (CES-D) 4-item self-reported measure,18 derived from the original CES-D 20-item measure19. The 4-item instrument asks how much of the time in the past week the participant has felt depressed, lonely, and sad, and had crying spells and uses the following responses: 0 points—“Rarely or none of the time (less than 1 day)”; 1 point—“Some or a little of the time (1–2 days)”; 2 points—“Occasionally or a moderate amount of the time (3–4 days)”; or 3 points—“Most or all of the time (5–7 days).” A summary score is then calculated with a range from 0 to 12. Participants with a CES-D-4 score ≥4 were classified as having elevated depressive symptoms18. This cut-point was chosen to be equivalent to the cut-point used in the original 20-item CES-D instrument, classifying more than 80% of the study sample similarly18. Depressive symptoms (CES-D) were initially assessed as part of the baseline REGARDS evaluation performed between 2003 and 2007. Additionally, depressive symptoms (CES-D) were assessed as part of a planned assessment of cognitive performance which was initiated in 2008, and subsequently performed at 2-year intervals. This introduced a small amount of variability in the time between the baseline and the second assessment of depressive symptoms (a.k.a., the first depressive assessment during follow-up). The average time between these two depressive symptom assessments was 5.07 ± 1.66 years. The present study utilizes these two measures of depressive symptoms to assess change in depressive symptoms across time. The “clock” for potential incident cardiovascular events then began running after the second depressive symptom assessment. The time to the potential incident cardiovascular event was measured in days from that point (reported in decimal years) and led to a time-to-event analysis described below.
The change in CES-D classification from baseline to follow-up assessment was examined and participants were categorized as follows: (1) no elevated depressive symptoms at either time point; (2) no elevated depressive symptoms at baseline but elevated depressive symptoms present at follow-up; (3) elevated depressive symptoms present at baseline but no elevated depressive symptoms present at follow-up; or (4) elevated depressive symptoms present at both baseline and follow-up assessments.
Incident CVD (Primary Outcome Variable)
The primary outcome was incident cardiovascular disease (CVD), defined as stroke, nonfatal myocardial infarction (MI), or coronary heart disease (CHD) death. Participants or their proxies were contacted every 6 months to assess hospitalizations and vital status. Medical records were retrieved for all potential stroke and CVD-related hospitalizations and deaths. All events were expert-adjudicated by medical review teams. Stroke events were defined based on the World Health Organization definition20 for those with focal neurologic symptoms lasting >24 h or concordant with the American Heart Association guidelines21 with supportive neuroimaging for those with symptoms lasting <24 h. MI was assessed following published guidelines22–25. Data from medical records, interviews with next-of-kin, death certificates, and the National Death Index were collected and reviewed by adjudicators to determine if the death was a CHD death.
Covariates
Age, race, sex, education (< high school education, high school graduate, some college, and ≥ college graduate), annual household income (<$20,000/year, $20,000–$34,999/year, $35,000–$74,999/year, $75,000/year or more, refused to answer), smoking (never smoker, former smoker, current smoker), and medication adherence15 were self-reported. An in-home visit was completed to obtain height, weight, blood pressure, an electrocardiogram, and blood and urine samples. Prevalent diabetes at baseline was defined as fasting glucose ≥126 mg/dL (or a non-fasting glucose ≥200 mg/dL for those failing to fast) or use of oral or injectable diabetes medications. Total cholesterol was measured using previously described methods15. Estimated glomerular filtration rate (eGFR) was estimated using the Chronic Kidney Disease Epidemiology Collaboration combined creatinine-cystatin C equation26. A careful medication history including statins, anti-hypertensives, and anti-depressants was obtained during the home visit and adherence estimated using a validated 4-item self-reported scale27. The perceived level of stress was also measured at baseline using a four-item version of the validated Cohen Perceived Stress Scale28, categorized as 0 (no stress), 1–8 (low to moderate), and >8 (high stress).
Statistical Analysis
The distribution of changes in depressive symptoms was compared between participants with and without DM using Pearson’s chi-square test. The initial analysis compared the demographic, socioeconomic, and CV risk parameters by elevated depressive symptoms’ status at baseline and stratified by diabetes status. Subsequently, the association between change in depressive symptoms and each of the three CV outcome measures of interest and a composite measure of all three was examined in a series of Cox proportional cause-specific hazards models. Additional Cox proportional hazards models then incrementally adjusted for the following groups of variables: (1) baseline demographic characteristics (age, race, sex, and region [rural vs. urban]); (2) baseline socioeconomic factors (income, education, marital status); (3) baseline medical conditions including BMI, systolic BP, total cholesterol, presence of atrial fibrillation, perceived stress, statin use, anti-hypertensive medication use, anti-depressant use, and estimated glomerular filtration rate (GFR); and (4) baseline health behaviors (current smoking, alcohol use, exercise, medication adherence). For models of either incident stroke or the composite CVD outcome, an age-race interaction term was additionally added to the demographic characteristics based on previous analyses29. Participants with no elevated depressive symptoms at either time point served as the referent group, against which the other three groups were compared. Additional analyses stratified by race and anti-depressant medication use were adjusted for diabetes status. All analyses were performed using SAS 9.4 (SAS, Cary, NC).
RESULTS
Participants
The present study included n = 2,700 (16.5%) participants with diabetes mellitus and n = 13,668 participants without diabetes at baseline for a total n = 16,368 participants who remained in the cohort an average of 11.1 years later and who had complete data available for analysis. The baseline characteristics of the study population are given in Table 1 by diabetes status and separated by presence or absence of depressive symptoms. In general, compared to those without depressive symptoms in both the diabetes and no diabetes subgroups respectively, participants with depressive symptoms were significantly younger, more likely to be black, less likely to have high school or higher education levels, less likely to have an income ≥ $35,000/year, and less likely to be married. Participants with baseline depressive symptoms were also more likely to report high levels of stress and current smoking, and to reside in the Stroke Belt. Notably, more than twice as many participants who reported depressive symptoms at baseline were taking an anti-depressant medication compared to those that did not report depressive symptoms (24% with depressive symptoms vs. 12% without depressive symptoms; see Table 1). This pattern was similar in those with and without diabetes mellitus.
Table 1.
Baseline Characteristics of REGARDS Participants With and Without Diabetes, by Presence or Absence of Depressive Symptoms
| Characteristic | Diabetes | No diabetes | ||
|---|---|---|---|---|
| No baseline depressive symptoms | Baseline depressive symptoms | No baseline depressive symptoms | Baseline depressive symptoms | |
| N | 2351 | 349 | 12584 | 1084 |
| Age, mean years (SD) | 64.5 (8.3) | 62.6 (8.5) | 63.4 (8.9) | 61.4 (9.0) |
| Female, n (%) | 1324 (56.3) | 259 (74.2) | 7290 (57.9) | 819 (75.6) |
| Black, n (%) | 1334 (56.7) | 226 (64.8) | 4197 (33.4) | 467 (43.1) |
| Stroke Belt residence, n (%) | 1368 (58.2) | 232 (66.5) | 6866 (54.6) | 641 (59.1) |
| High school or more education, n (%) | 2054 (87.4) | 262 (75.1) | 11732 (93.2) | 927 (85.5) |
| Annual household income >$35,000, n (%) | 1016 (43.2) | 89 (25.5) | 7075 (56.2) | 383 (35.3) |
| Married, n (%) | 1351 (57.5) | 140 (40.1) | 8156 (64.8) | 476 (43.9) |
| Current smoker, n (%) | 260 (11.1) | 58 (16.7) | 1437 (11.5) | 219 (20.3) |
| Exercise, n (%) | 1512 (64.3) | 194 (55.6) | 8834 (70.2) | 622 (57.4) |
| Body mass index, mean (SD) | 32.6 (6.6) | 34.0 (7.0) | 28.5 (5.6) | 29.6 (4.5) |
| Alcohol use, n (%) | ||||
| None | 1686 (73.4) | 255 (75.4) | 6952 (56.1) | 635 (60.1) |
| Moderate | 562 (24.5) | 76 (22.5) | 4841 (39.1) | 366 (34.6) |
| Heavy | 48 (2.1) | 7 (2.1) | 600 (4.8) | 56 (5.3) |
| Use of hypertensive medication, n (%) | 1639 (69.7) | 252 (72.2) | 5069 (40.3) | 496 (45.8) |
| Perceived stress, n (%) | ||||
| No stress | 677 (28.8) | 17 (4.9) | 3396 (27.0) | 76 (7.0) |
| Low to moderate stress | 1618 (68.9) | 258 (73.9) | 8927 (70.9) | 752 (69.4) |
| High stress | 55 (2.3) | 74 (21.2) | 261 (2.1) | 256 (23.6) |
| Statin use, n (%) | 985 (41.9) | 149 (42.7) | 2768 (22.0) | 253 (23.3) |
| Anti-depressant use, n (%) | 293 (12.5) | 83 (23.8) | 1270 (10.1) | 257 (23.7) |
| Medication adherence, n (%) | ||||
| High | 1568 (66.7) | 208 (59.6) | 7926 (63.0) | 596 (55.0) |
| Good to moderate | 646 (27.5) | 106 (30.4) | 2939 (23.4) | 337 (31.1) |
| Low | 137 (5.8) | 35 (10.0) | 1719 (13.7) | 151 (13.9) |
| Systolic blood pressure, mmHg, mean (SD) | 129.9 (15.9) | 130.0 (15.8) | 125.1 (15.6) | 125.3 (16.3) |
| Total cholesterol, mean mg/dL (SD) | 182.6 (40.6) | 186.9 (43.7) | 197.8 (37.4) | 200.0 (37.5) |
| EGFR estimated, mean (SD) | 87.1 (22.0) | 89.3 (22.9) | 87.4 (17.0) | 90.9 (18.3) |
Change in Depressive Symptoms
Comparing the change in depressive symptoms across time, Table 2 shows the differential distribution of participants by the change in depressive symptoms category between those with and without diabetes (p<0.0001). Among those with diabetes, a similar percentage (7.6%) had depressive symptoms at one measurement (either at baseline or at the second time point) but not at the other assessment time point, suggesting that a similar percentage had remitted symptoms or new symptoms across the intervening 5-year period. A higher percentage (5.4% vs. 3.0%) of participants with diabetes had depressive symptoms at both assessment time points than was reported by participants without diabetes.
Table 2.
Change in Depressive Symptoms from Baseline to Follow-up
| Change in depressive symptom category | Diabetes (n= 2,700) | No diabetes (n= 13,668) |
|---|---|---|
| No depressive symptoms at baseline or at follow-up | 2145 (79.4%) | 11717 (85.7%) |
| Depressive symptoms at baseline but not at follow-up | 204 (7.6%) | 676 (5.0%) |
| No depressive symptoms at baseline but depressive symptoms at follow-up | 206 (7.6%) | 867 (6.3%) |
| Depressive symptoms at baseline and at follow-up (i.e., persistent symptoms) | 145 (5.4%) | 408 (3.0%) |
CV Events
The mean follow-up time for CVD outcome assessment after the second assessment of depressive symptoms was 6.1 (SD = 2.6) years. Crude incidence rates for each of the four CV outcomes by diabetes status are given in online Supplemental Table S1 and shows higher event rates among participants with diabetes. Tables 3 (participants with diabetes) and 4 (participants without diabetes) give the adjusted hazard ratios from the Cox proportional hazards modeling for adjudicated CV events during follow-up including individual CV outcomes (i.e., incident stroke, CHD, and CV mortality) as well as a composite outcome.
Table 3.
Change in Depressive Scores from Baseline to Follow-up and Hazard Ratio for CV Events During Follow-up in Participants with Diabetes
| CESD change (baseline vs. first CESD follow-up assessment) | ||||
|---|---|---|---|---|
| Normal at both assessments | Improved at second assessment | Worsened at second assessment | Elevated at both assessments (persistent symptoms) | |
| Incident stroke outcome | ||||
| Model 1: CESD change | Ref | 1.00 (0.51, 1.98) | 1.24 (0.66, 2.31) | 1.85 (0.99, 3.45) |
| Model 2*: Model 1+demographic | Ref | 0.95 (0.48, 1.89) | 1.19 (0.64, 2.23) | 1.77 (0.94, 3.33) |
| Model 3*: Model 2+SES | Ref | 1.01 (0.50, 2.02) | 1.23 (0.66, 2.31) | 1.89 (1.00, 3.59) |
| Model 4*: Model 3+medical conditions/AFib | Ref | 0.81 (0.38, 1.71) | 1.06 (0.54, 2.08) | 1.38 (0.65, 2.94) |
| Model 5*: Model 4+health behaviors | Ref | 0.79 (0.36, 1.76) | 1.07 (0.54, 2.08) | 1.41 (0.65, 3.04) |
| Incident CHD outcome | ||||
| Model 1: CESD change | Ref | 0.82 (0.46, 1.48) | 1.36 (0.86, 2.16) | 1.08 (0.59, 1.99) |
| Model 2: Model 1+demographic | Ref | 0.93 (0.52, 1.68) | 1.48 (0.93, 2.36) | 1.24 (0.67, 2.29) |
| Model 3: Model 2+SES | Ref | 0.82 (0.45, 1.48) | 1.38 (0.87, 2.21) | 1.06 (0.57, 1.96) |
| Model 4: Model 3+medical conditions/AFib | Ref | 0.72 (0.39, 1.33) | 1.22 (0.74, 2.02) | 0.85 (0.43, 1.69) |
| Model 5: Model 4+health behaviors | Ref | 0.70 (0.37, 1.34) | 1.18 (0.70, 1.99) | 0.77 (0.37, 1.57) |
| CVD mortality outcome | ||||
| Model 1: CESD change | Ref | 1.03 (0.57, 1.85) | 1.72 (1.08, 2.75) | 0.76 (0.34, 1.73) |
| Model 2: Model 1+demographic | Ref | 1.17 (0.64, 2.13) | 1.96 (1.22, 3.15) | 0.97 (0.42, 2.20) |
| Model 3: Model 2+SES | Ref | 1.04 (0.57, 1.90) | 1.85 (1.15, 2.99) | 0.87 (0.38, 1.98) |
| Model 4: Model 3+medical conditions/AFib | Ref | 0.94 (0.51, 1.76) | 1.51 (0.89, 2.56) | 0.77 (0.33, 1.82) |
| Model 5: Model 4+health behaviors | Ref | 1.06 (0.56, 1.98) | 1.58 (0.93, 2.68) | 0.67 (0.26, 1.73) |
| Composite outcome | ||||
| Model 1: CESD change | Ref | 1.04 (0.71, 1.53) | 1.37 (0.98, 1.93) | 1.16 (0.74, 1.80) |
| Model 2: Model 1 + demographic | Ref | 1.14 (0.77, 1.69) | 1.46 (1.04, 2.06) | 1.28 (0.82, 2.00) |
| Model 3: Model 2 + SES | Ref | 1.06 (0.72, 1.58) | 1.40 (0.99, 1.98) | 1.18 (0.75, 1.85) |
| Model 4: Model 3 + medical conditions/AFib | Ref | 0.89 (0.59, 1.35) | 1.12 (0.76, 1.63) | 0.84 (0.51, 1.41) |
| Model 5: Model 4 + health behaviors | Ref | 0.92 (0.60, 1.42) | 1.10 (0.75, 1.61) | 0.76 (0.44, 1.30) |
*Model 1: CESD change; Model 2: Model 1 + age, race (age-race interaction for incident stroke and composite outcome), sex, belt; Model 3: Model 2 + education, income, marital status; Model 4: Model 3 + SBP, BMI, cholesterol, anti-hypertension medications, statin use, anti-depressant medication, eGFR, atrial fibrillation (AFib), perceived stress; Model 5: Model 4 + exercise, smoking, alcohol use, medication adherence
Stroke
When examining the participants with diabetes, relative to the subgroup with no depressive symptoms at baseline or at follow-up, the subgroup with elevated depressive symptoms at both baseline and follow-up had a marginally significant increased risk for incident stroke in Model 3 (1.89 (1.00, 3.59)) which adjusted for demographic characteristics and SES. This was attenuated leading to a non-significant 41% (HR (95% CI) = 1.41 (0.65, 3.04)) increased risk for incident stroke in the fully adjusted model (Table 3). However, participants without diabetes but with persistent depressive symptoms demonstrated a significant 84% (HR (95% CI) = 1.84 (1.03, 3.30)) increased risk for incident stroke in the fully adjusted model (Table 4). Among those with depressive symptoms at baseline but not present at follow-up and among those with no depressive symptoms at baseline but symptoms present at follow-up, in both those with and without diabetes, there was no significantly increased risk of incident stroke.
Table 4.
Change in Depressive Scores from Baseline to Follow-up and Hazard Ratio for CV Events During Follow-up in Participants without Diabetes
| CESD change (baseline vs. first CESD follow-up assessment) | ||||
|---|---|---|---|---|
| Normal at both assessments | Improved at second assessment | Worsened at second assessment | Elevated at both assessments (persistent symptoms) | |
| Incident stroke outcome | ||||
| Model 1: CESD change | Ref | 0.94 (0.58, 1.53) | 1.01 (0.66, 1.55) | 1.25 (0.72, 2.18) |
| Model 2*: Model 1+demographic | Ref | 1.10 (0.68, 1.80) | 1.08 (0.71, 1.65) | 1.54 (0.88, 2.69) |
| Model 3*: Model 2+SES | Ref | 1.05 (0.64, 1.72) | 1.05 (0.69, 1.61) | 1.44 (0.82, 2.53) |
| Model 4*: Model 3+medical conditions/AFib | Ref | 1.15 (0.69, 1.93) | 1.07 (0.69, 1.66) | 1.76 (0.98, 3.13) |
| Model 5*: Model 4+health behaviors | Ref | 1.14 (0.67, 1.94) | 1.11 (0.71, 1.72) | 1.84 (1.03, 3.30) |
| Incident CHD outcome | ||||
| Model 1: CESD change | Ref | 0.93 (0.61, 1.42) | 1.14 (0.81, 1.61) | 0.99 (0.58, 1.68) |
| Model 2: Model 1+demographic | Ref | 1.13 (0.74, 1.72) | 1.24 (0.88, 1.75) | 1.27 (0.75, 2.17) |
| Model 3: Model 2+SES | Ref | 1.06 (0.70, 1.63) | 1.21 (0.85, 1.71) | 1.18 (0.69, 2.02) |
| Model 4: Model 3+medical conditions/AFib | Ref | 0.97 (0.63, 1.52) | 1.15 (0.80, 1.64) | 1.08 (0.61, 1.89) |
| Model 5: Model 4+health behaviors | Ref | 1.01 (0.65, 1.57) | 1.13 (0.78, 1.62) | 1.04 (0.58, 1.87) |
| CVD mortality outcome | ||||
| Model 1: CESD change | Ref | 0.90 (0.55, 1.46) | 1.10 (0.74, 1.63) | 1.30 (0.76, 2.21) |
| Model 2: Model 1+demographic | Ref | 1.12 (0.69, 1.83) | 1.10 (0.74, 1.65) | 1.65 (0.96, 2.83) |
| Model 3: Model 2+SES | Ref | 1.01 (0.62, 1.64) | 1.07 (0.72, 1.60) | 1.46 (0.85, 2.52) |
| Model 4: Model 3+medical conditions/AFib | Ref | 0.89 (0.52, 1.52) | 0.93 (0.60, 1.44) | 1.45 (0.83, 2.53) |
| Model 5: Model 4+health behaviors | Ref | 0.94 (0.55, 1.61) | 0.89 (0.57, 1.40) | 1.41 (0.78, 2.52) |
| Composite outcome | ||||
| Model 1: CESD change | Ref | 0.99 (0.74, 1.32) | 1.15 (0.90, 1.47) | 1.15 (0.81, 1.64) |
| Model 2: Model 1+demographic | Ref | 1.19 (0.89, 1.59) | 1.23 (0.96, 1.57) | 1.46 (1.02, 2.08) |
| Model 3: Model 2+SES | Ref | 1.12 (0.83, 1.50) | 1.19 (0.93, 1.53) | 1.34 (0.94, 1.92) |
| Model 4: Model 3+medical conditions/AFib | Ref | 1.06 (0.78, 1.45) | 1.14 (0.88, 1.47) | 1.38 (0.95, 1.99) |
| Model 5: Model 4+health behaviors | Ref | 1.09 (0.79, 1.49) | 1.12 (0.87, 1.46) | 1.34 (0.91, 1.97) |
*Model 1: CESD change; Model 2: Model 1 + age, race (age-race interaction for incident stroke and composite outcome), sex, belt; Model 3: Model 2 + education, income, marital status; Model 4: Model 3 + SBP, BMI, cholesterol, anti-hypertension medications, statin use, anti-depressant medication, eGFR, atrial fibrillation (AFib), perceived stress; Model 5: Model 4 + exercise, smoking, alcohol use, medication adherence
Incident CHD, CVD Mortality, and the Composite Outcome
In the subgroup with diabetes who did not have depressive symptoms at baseline but did have these symptoms at the second assessment, the crude model (1.72 (1.08, 2.75)) and two partially adjusted models each demonstrated a significant increase in CV mortality; however, this was attenuated in the fully adjusted model. Among those without diabetes but who had depressive symptoms at both assessments, there was a significant 46% (1.46 (1.02, 2.08)) increase in risk for the composite outcome in Model 2 which adjusted only for demographic characteristics and change in CES-D; however, this increase in risk was attenuated in the fully adjusted model.
Stroke Analysis by Race and Anti-depressant Use
Of particular interest was the risk of incident stroke when examined in all participants both by race and subsequently by anti-depressant use in preplanned analyses (see Tables 5 and 6). Compared to black participants who were without depressive symptoms at both assessments, black participants with elevated depressive symptoms at both time points had a much higher risk of incident stroke in crude and fully adjusted models, suggesting that measured covariates did not attenuate the relationship (HR (95% CI) = 2.64 (1.48, 4.72)). This same comparison in whites revealed no increased risk of incident stroke (HR (95% CI) = 1.06 (0.50, 2.25)) in the fully adjusted model. Across both races, compared to participants who were without depressive symptoms at both assessments and were taking anti-depressants, participants with elevated depressive symptoms at both assessments but who were not taking anti-depressants had an increased risk of stroke in all adjusted models including a 2.0-fold higher risk of incident stroke in fully adjusted models (HR (95% CI) = 2.01 (1.21, 3.35)). This same comparison in those who were taking anti-depressants revealed no significantly increased risk of incident stroke (HR (95% CI) = 1.41 (0.52, 3.85)).
Table 5.
Race and Anti-depressant Use Comparisons. Persistent Depressive Symptoms (CESD Change Baseline vs. First CESD Follow-up Assessment) and Hazard Ratio for Incident Stroke During Follow-up, by Race
| Normal at both assessments—white | Elevated at both assessments—white | Normal at both assessments—black | Elevated at both assessments—black | |
|---|---|---|---|---|
| Model 1: CESD change, diabetes status | Ref | 0.92 (0.46, 1.86) | Ref | 2.17 (1.29, 3.64) |
| Model 2*: Model 1+demographic | Ref | 1.11 (0.55, 2.25) | Ref | 2.31 (1.37, 3.91) |
| Model 3*: Model 2+SES | Ref | 1.03 (0.51, 2.09) | Ref | 2.46 (1.44, 4.18) |
| Model 4*: Model 3+medical conditions | Ref | 1.04 (0.50, 2.19) | Ref | 2.58 (1.45, 4.58) |
| Model 5*: Model 4+health behaviors | Ref | 1.06 (0.50, 2.25) | Ref | 2.64 (1.48, 4.72) |
Table 6.
Race and Anti-depressant Use Comparisons. Persistent Depressive Symptoms (CESD Change Baseline vs. First CESD Follow-up Assessment) and Hazard Ratio for Incident Stroke During Follow-up, by Anti-depressant Use
| Normal at both assessments—no anti-depressant use | Elevated at both assessments—no anti-depressant use | Normal at both assessments—on anti-depressant use | Elevated at both assessments—on anti-depressant | |
|---|---|---|---|---|
| Incident stroke outcome | ||||
| Model 1: CESD change, diabetes status | Ref | 1.50 (0.94, 2.41) | Ref | 1.46 (0.62, 3.46) |
| Model 2*: Model 1+demographic | Ref | 1.68 (1.04, 2.71) | Ref | 1.75 (0.72, 4.24) |
| Model 3*: Model 2+SES | Ref | 1.65 (1.02, 2.66) | Ref | 1.68 (0.69, 4.10) |
| Model 4*: Model 3+medical conditions | Ref | 1.93 (1.16, 3.21) | Ref | 1.28 (0.49, 3.34) |
| Model 5*: Model 4+health behaviors | Ref | 2.01 (1.21, 3.35) | Ref | 1.41 (0.52, 3.85) |
DISCUSSION
Cohort studies have suggested a relationship between baseline depressive symptoms and the subsequent occurrence of stroke3–5. Recent data from the REGARDS cohort study by Ford et al.6 add to these findings, demonstrating a 54% increased risk of incident stroke over 9 years in individuals reporting depressive symptoms at baseline. The present study examined the association between the change in depressive symptoms across 5 years and adjudicated cardiovascular outcomes in African American and white subjects with and without diabetes mellitus, followed for up to 6 additional years. Persistent depressive symptoms were not significantly associated with incident CHD or CV death in fully adjusted models but were significantly associated with stroke risk in participants without diabetes.
The present findings demonstrate that the chronic persistence of elevated depressive symptoms over an average of 5 years in individuals without diabetes who reported depressive symptoms at baseline was associated with an increased risk for incident stroke (84% in fully adjusted models) compared to individuals who never reported depressive symptoms. While a similar trend was evident in participants with diabetes, the increase in risk for incident stroke was not statistically significant, likely because of the smaller sample size with fewer events. Of particular interest from a health equity perspective is the finding that the risk for incident stroke among black participants with persistent depressive symptoms at both time points was dramatically increased compared to those with no symptoms at either time point even in fully adjusted models, while a similar risk was not evident in white participants. These data add to the race-specific findings reported by Ford et al.6 using only baseline depressive symptoms in which there were limited differences by race. This finding raises continuing questions regarding equitable access to health care and the potential long-term impact of adverse social determinants of health that may be associated with persistently elevated depressive symptoms. Similarly, our findings point to the potential impact of not taking anti-depressant medications in the setting of persistently increased depressive symptoms, as this was also associated with dramatically increased stroke risk compared to those without symptoms at either time point and not on anti-depressants.
These findings have important implications for screening and evaluation of depressive symptoms. In particular, identification of individuals who have persistent depressive symptoms appears to identify a subgroup at substantially increased risk for incident stroke during follow-up. This suggests that depressive disorders are chronic diseases, and that lack of anti-depressant treatment may be associated with incident stroke. Further, this suggests that additional efforts should be undertaken to monitor depressive symptoms during follow-up and to consider if anti-depressant treatment is warranted.
The mechanisms that underlie this increased stroke risk in individuals with persistent depressive symptoms are not well understood. As described by Sun et al.30, depressive symptoms have been associated with cardiac arrhythmia, enhanced insulin resistance, more white matter lesions, and increased levels of platelet activation, fibrinogen, and C-reactive protein, all of which might influence stroke risk. Further, persistent depressive symptoms may also lead to reduced physical activity, increased smoking and/or alcohol consumption, increased obesity, medication non-adherence, and inadequate blood pressure control. In the present study, those with depressive symptoms at baseline had less physical activity, were more likely to be current smokers, had higher BMI and cholesterol values, and had modestly greater anti-hypertensive use than did those without depressive symptoms. Although controlled for in the analysis as covariates, the potential persistence of this pattern may suggest the longitudinal impact of these and other unmeasured risk factors on cerebrovascular integrity and function.
While this study has many strengths including the population-level nature of the cohort, a longer follow-up (5 years) for assessing change in depressive symptoms, and the adjudication of CV outcomes, there are limitations. The study only included black and white individuals over the age of 45 years at baseline and generalization to other population groups is limited. Depressive symptoms were measured using the CES-D 4-item screener rather than a formal interviewer-determined clinical diagnosis of major depressive disorder. Depressive symptoms were measured at two time points an average of 5 years apart and this may not reflect the changing nature of depressive symptoms. Stroke risk factors such as smoking, blood pressure control, obesity, and health behaviors were only measured at baseline and were not repeated at the 6-month follow-up telephone contacts and glycemic control across time in those with diabetes was not assessed. Likewise, the duration of CV risk factors such as hypertension and diabetes was not collected in REGARDS. Participants with CV events that occurred after the baseline visit but before the second CES-D assessment were excluded in order to measure the subsequent impact of change in depressive symptoms; this subgroup may have similar or even higher risks for subsequent stroke and/or other CV events.
In summary, the present study demonstrates that the persistence of depressive symptoms across 5 years of follow-up in those without diabetes identifies individuals who are at increased risk for incident stroke. Further, we demonstrate that the risk is particularly evident among black participants and among those not taking anti-depressants. This suggests the need to actively screen for depressive symptoms, and to pursue treatment that will successfully reduce elevated depressive symptoms.
Supplementary Information
(DOCX 26 kb)
Acknowledgements
This research project is supported by cooperative agreement U01 NS041588 co-funded by the National Institute of Neurological Disorders and Stroke (NINDS) and the National Institute on Aging (NIA), National Institutes of Health, Department of Health and Human Service. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NINDS or the NIA. Representatives of the NINDS were involved in the review of the manuscript but were not directly involved in the collection, management, analysis or interpretation of the data. The authors thank the other investigators, the staff, and the participants of the REGARDS study for their valuable contributions. A full list of participating REGARDS investigators and institutions can be found at: https://www.uab.edu/soph/regardsstudy/.
Funding
NINDS and NIA; U01 NS041588;
Declarations
Conflict of Interest
The authors declare that they do not have a conflict of interest.
Footnotes
No prior presentation of this data
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Benjamin EJ, Muntner P, Alonso A, et al. Heart Disease and Stroke Statistics-2019 Update: A Report from the American Heart Association. Circulation. 2019;139:e56–e528. doi: 10.1161/CIR.0000000000000659. [DOI] [PubMed] [Google Scholar]
- 2.Harshfield EL, Pennells L, Schwartz JE, Willeit P, Kaptoge S, Bell S, et al. Association between Depressive Symptoms and Incident Cardiovascular Diseases. JAMA. 2020;324(23):2396–2405. doi: 10.1001/jama.2020.23068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Larson SL, Owens PL, Ford D, Eaton W. Depressive Disorder, Dysthymia, and Risk of Stroke: Thirteen-Year Follow-up from the Baltimore Epidemiologic Catchment Area Study. Stroke. 2001;32:1979–1983. doi: 10.1161/hs0901.094623. [DOI] [PubMed] [Google Scholar]
- 4.Li C, Bai Y, Tu P, Lee Y, Huang Y, Chen T, Chang W, Su T. Major Depressive Disorder and Stroke Risks: A 9-year Follow-up Population-based, Matched Cohort Study. PLoS ONE. 2012;7(10):e46818. doi: 10.1371/journal.pone.0046818. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Dong J, Zhang Y, Tong J, Qin L. Depression and Risk of Stroke: A Meta-analysis of Prospective Studies. Stroke. 2012;43:32–37. doi: 10.1161/STROKEAHA.111.630871. [DOI] [PubMed] [Google Scholar]
- 6.Ford CD, Gray MS, Crowther MR, Wadley VG, Austin AL, Crowe MG, Pulley L, Unverzagt F, Kleindorfer DO, Kissela BM, Howard VJ. Depressive Symptoms and Risk of Stroke in a National Cohort of Black and White Participants From REGARDS. Neurol Clin Pract. 2021;11(4):e454–e461. doi: 10.1212/CPJ.0000000000000983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Everson SA, Roberts RE, Goldberg DE, Kaplan GA. Depressive Symptoms and Increased Risk of Stroke Mortality Over a 29-year Period. Arch Int Med. 1998;158:1133–38. doi: 10.1001/archinte.158.10.1133. [DOI] [PubMed] [Google Scholar]
- 8.Gilsanz P, Walter S, Tchetgen EJT, Patton KK, Moon JR, Capistrant BD, et al: Change in Depressive Symptoms and Incidence of First Stroke among Middle-Aged and Older US Adults. J Am Heart Assoc. 2015; 10.1161/JAHA.115.001923 [DOI] [PMC free article] [PubMed]
- 9.Gilsanz P, Kubzansky LD, Tchetgen EJT, Wang Q, Kawachi I, Patton KK, et al. Change in Depressive Symptoms and Subsequent Risk of Stroke in the Cardiovascular Health Study. Stroke. 2017;48:43–48. doi: 10.1161/STROKEAHA.116.013554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Li H, Qian F, Hou C, Li X, Gao Q, Luo Y, et al: Longitudinal Changes in Depressive Symptoms and Risks of Cardiovascular Disease and All-Cause Mortality: A Nationwide Population-Based Cohort Study. J Gerontol A Biol Sci Med Sci 2019; doi:10.1093/gerona/glz228 [DOI] [PMC free article] [PubMed]
- 11.Lustman PJ, Clouse RE. Depression in diabetic patients: the relationship between mood and glycemic control. J Diabetes Complications. 2005;19(2):113–22. doi: 10.1016/j.jdiacomp.2004.01.002. [DOI] [PubMed] [Google Scholar]
- 12.Golden SH. A review of evidence for a neuroendocrine link between stress, depression and diabetes mellitus. Curr Diabetes Review. 2007;3:252–259. doi: 10.2174/157339907782330021. [DOI] [PubMed] [Google Scholar]
- 13.Katon WJ, Young BA, Russo J, Lin EHB, Ciechanowski P, Ludman EJ, Von Korff MJ. Association of depression with increased risk of severe hypoglycemic episodes in patients with diabetes. Ann Fam Med. 2013;11:245–250. doi: 10.1370/afm.1501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Pan A, Lucas M, Sun Q, van Dam RM, Franco OH, Manson JE, Willett WC, Ascherio A, Hu FB. Bidirectional association between depression and type 2 diabetes mellitus in women. Archives Internal Med. 2010;170(21):1884–91. doi: 10.1001/archinternmed.2010.356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cummings DM, Kirian K, Howard G, Howard VJ, Yuan Y, Muntner P, Kissela B, Redmond N, Judd SE, Safford MM. Consequences of Co-Morbidity of Elevated Stress and/or Depressive Symptoms and Incident Cardiovascular Outcomes in Diabetes: Results from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study. Diabetes Care. 2016;39:101–109. doi: 10.2337/dc15-1174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cummings DM, Lutes LD, Littlewood K, Solar C, Carraway M, Kirian K, Patil S, Adams A, Ciszewski S, Edwards S, Gatlin P, Hambidge B. Randomized Trial of a Tailored Cognitive Behavioral Intervention in Type 2 Diabetes with Comorbid Depressive and/or Regimen-related Distress Symptoms—12-Month Outcomes from COMRADE. Diabetes Care. 2019;42:841–848. doi: 10.2337/dc18-1841. [DOI] [PubMed] [Google Scholar]
- 17.Howard VJ, Cushman M, Pulley L, Gomez CR, Go RC, Prineas RJ, Graham A, Moy CS, Howard G. The Reasons for Geographic and Racial Differences in Stroke Study: Objectives and Design. Neuroepidemiology. 2005;25(3):135–143. doi: 10.1159/000086678. [DOI] [PubMed] [Google Scholar]
- 18.Melchior LA, Huba GJ, Brown VB, Reback CJ. A short depression index for women. Educ Psychol Meas. 1993;53:1117–1125. doi: 10.1177/0013164493053004024. [DOI] [Google Scholar]
- 19.Radloff LS. The CES-D Scale: A self-report depression scale for research in the general population. Appl Psychol Meas. 1977;3:385–401. doi: 10.1177/014662167700100306. [DOI] [Google Scholar]
- 20.Stroke-1989: Recommendations on stroke prevention, diagnosis, and therapy. Report of the WHO Task Force on Stroke and other Cerebrovascular Disorders. Stroke 1989;20:1407-31 [DOI] [PubMed]
- 21.Sacco RL, Kasner SE, Broderick JP, Caplan LR, Connors JJ, Culebras A, et al. An updated definition of stroke for the 21st century: a statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2013;44:2064–2089. doi: 10.1161/STR.0b013e318296aeca. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Thygesen K, Alpert JS, White HD, Jaffe AS, Apple FS, Galvani M, et al. Universal definition of myocardial infarction. Circulation. 2007;116:2634–2653. doi: 10.1161/CIRCULATIONAHA.107.187397. [DOI] [PubMed] [Google Scholar]
- 23.Luepker RV, Apple FS, Christenson RH, Crow RS, Fortmann SP, Goff D, et al. Case definitions for acute coronary heart disease in epidemiology and clinical research studies: a statement from the AHA Council on Epidemiology and Prevention; AHA Statistics Committee; World Heart Federation Council on Epidemiology and Prevention; the European Society of Cardiology Working Group on Epidemiology and Prevention; Centers for Disease Control and Prevention; and the National Heart, Lung, and Blood Institute. Circulation. 2003;108:2543–2549. doi: 10.1161/01.CIR.0000100560.46946.EA. [DOI] [PubMed] [Google Scholar]
- 24.Prineas R, Crow R, Blackburn H. The Minnesota code manual of electrocardiographic findings: Standards and procedures for measurement and classification. Boston, MA: Wright-OSG; 1982. [Google Scholar]
- 25.Prineas R, Crow R, Zhang ZM. Minnesota code manual of Electrocardiographic findings. 2. London, UK: Springer-Verlag; 2010. [Google Scholar]
- 26.Walther CP, Gutiérrez OM, Cushman M, Judd SE, Lang J, McClellan W, Muntner P, Sarnak MJ, Shlipak MG, Warnock DG, Katz R, Ix JH. Serum albumin concentration and risk of end-stage renal disease: the REGARDS study. Nephrol Dial Transplant. 2018 Oct 1;33(10):1770-1777. 10.1093/ndt/gfx331. [DOI] [PMC free article] [PubMed]
- 27.Morisky DE, Green LW, Levine DM. Concurrent and predictive validity of a self-reported measure of medication adherence. Med. Care. 1986;24:67–74. doi: 10.1097/00005650-198601000-00007. [DOI] [PubMed] [Google Scholar]
- 28.Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav. 1983;24:385–396. doi: 10.2307/2136404. [DOI] [PubMed] [Google Scholar]
- 29.Howard VJ, Kleindorfer DO, Judd SE, McClure LA, Safford MM, Rhodes JD, Cushman M, Moy CS, Soliman EZ, Kissela BM, Howard G. Disparities in Stroke Incidence Contributing to Disparities in Stroke Mortality. Annals Neurol. 2011;69:619–627. doi: 10.1002/ana.22385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Sun J, Ma H, Yu C, Lv J, Guo Y, Bian Z, et al. Association of major depressive episodes with stroke risk in a prospective study of 0.5 million Chinese adults. Stroke. 2016;47(9):2203–2208. doi: 10.1161/STROKEAHA.116.013512. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
(DOCX 26 kb)