Eight-Year Depressive Symptom Trajectories and Incident Stroke: A 10-Year Follow-Up of the HRS (Health and Retirement Study)
Abstract
Background:
Evidence suggests a link between depressive symptoms and risk of subsequent stroke. However, most studies assess depressive symptoms at only one timepoint, with few examining this relationship using repeatedly measured depressive symptoms. This study aimed to examine the relationship between depressive symptom trajectories and risk of incident stroke.
Methods:
This prospective cohort included 12 520 US individuals aged ≥50 years enrolled in the Health and Retirement Study, free of stroke at study baseline (1998). We used the 8-item Center for Epidemiologic Studies Depression scale to assess depressive symptoms (high defined as ≥3 symptoms; low <3 symptoms) at 4 consecutive, biennial timepoints from 1998 to 2004. We assigned individuals to 5 predefined trajectories based on their scores at each timepoint (consistently low, decreasing, fluctuating, increasing, and consistently high). Using self-reported doctors’ diagnoses, we assessed incident stroke over a subsequent 10-year period from 2006 to 2016. Cox regression models estimated the association of depressive symptom trajectories with risk of incident stroke, adjusting for demographics, health behaviors, and health conditions.
Results:
During follow-up, 1434 incident strokes occurred. Compared with individuals with consistently low symptoms, individuals with consistently high depressive symptoms (adjusted hazard ratio, 1.18 [95% CI, 1.02–1.36]), increasing symptoms (adjusted hazard ratio, 1.31 [95% CI, 1.10–1.57]), and fluctuating symptoms (adjusted hazard ratio, 1.21 [95% CI, 1.01–1.46]) all had higher hazards of stroke onset. Individuals in the decreasing symptom trajectory group did not show increased stroke risk.
Conclusions:
Depressive symptom trajectories characterized by high symptoms at multiple timepoints were associated with increased stroke risk. However, a trajectory with depressive symptoms that started high but decreased over time was not associated with higher stroke risk. Given the remitting-relapsing nature of depressive symptoms, it is important to understand the relationship between depressive symptoms and stroke risk over time through repeated assessments.
Graphical Abstract
Stroke is one of the leading causes of long-term disability and mortality worldwide1 and the fifth leading cause of death in the United States.2 Stroke incidence has declined dramatically over recent decades, yet the rate of decline has slowed and the total economic and social impact of stroke is projected to rise with accelerated population aging and rising life expectancy.3 Identifying risk factors for stroke is, therefore, crucial in developing strategies to prevent stroke and its associated morbidity. While numerous classical risk factors such as cigarette smoking and diabetes have been identified, these do not fully account for observed risk.4 Several prospective studies have examined the depressive symptom-stroke association, yielding mixed results. Although some studies failed to find clear evidence that depressive symptoms are associated with increased risk of incident stroke,5,6 others reported a positive association between depressive symptoms and incident stroke.7,8 Several meta-analyses have shown that depression and depressive symptoms are prospectively associated with a 31% to 45% higher risk of stroke independently of smoking and comorbidities, highlighting the role of depression as a potential risk factor for stroke.9–11 However, the interpretation of the depressive symptom-stroke association is complicated by the vascular depression hypothesis, which posits that depressive symptoms may be a consequence of preexisting subclinical cerebrovascular diseases.12 Examining depressive symptom trajectories across repeated exposure assessments in relation to incident stroke may help address reverse causality (eg, depression of vascular origin) and improve causal inference. More conclusive evidence regarding the direction of these associations would have implications for whether treating depressive symptoms might be an effective strategy for stroke prevention.
Prior studies evaluating the depression-stroke association have used depressive symptoms at only a single timepoint,9–11 which fails to capture how depressive symptoms change over time and may not provide a full picture of the role of depression in stroke risk. Depressive symptoms are dynamic, that is, they demonstrate a pattern of remitting and relapsing.13 Some individuals may experience chronic depression, while others experience temporary increases in depressive symptoms (eg, in response to stressful life events), and still others may experience recurrent depression (ie, relapses after partial remission). Moreover, it remains unclear whether remitting symptoms might reduce risk of incident stroke. Investigating these changing temporal patterns of depressive symptoms and their relationship to stroke risk is crucial for assessing whether treating depression might be an effective stroke prevention strategy.
To date, only 2 studies have explored depressive symptom changes in relation to incident stroke risk, with mixed results.14,15 Furthermore, these studies evaluated depressive symptom changes only in relation to short-term stroke risk, within 2 years of the second assessment of depressive symptoms. Few reports have described the longer-term impact of depressive symptom trajectories on incident stroke. To address this research gap, we used data from the HRS (Health and Retirement Study) to assess associations of depressive symptom trajectories across a period of 8 years with incident stroke risk in a subsequent 10-year follow-up period. We hypothesized that individuals with trajectories patterned by high levels of depressive symptoms at multiple timepoints would be at higher risk of incident stroke.
Methods
Data used in this study is publicly available through the Institute for Social Research at the University of Michigan, Ann Arbor (https://hrsdata.isr.umich.edu/data-products/public-survey-data).
Study Population
The HRS is a nationally representative, longitudinal panel study of US adults aged ≥50 years and their spouses of any age, as previously described in detail.16,17 We used data from 1998 to 2016 for participants enrolled in 3 cohorts (1992, 1993, and 1998), which were merged in 1998 and followed biennially thereafter.18 The HRS is sponsored by the National Institute on Aging and conducted by the University of Michigan, and the Harvard T.H. Chan School of Public Health human subjects committee determined the current study to be exempt from institutional review board review.
We defined 1998 to 2004 as our exposure period (where T1 refers to 1998, T2 to 2000, T3 to 2002, and T4 to 2004) and 2006 to 2016 as our follow-up period. We restricted our sample to HRS participants aged ≥50 years in 1998, who did not report having a stroke before or during our exposure period. Of 18 256 eligible HRS respondents, we excluded individuals with >2 items missing on the depressive symptoms measure for 2 or more waves (n=3772, 20.7%), those who had a stroke during the exposure period (n=678, 3.7%), and those who died during the exposure period (n=1286, 7.0%). Our final analytic sample included 12 520 individuals (Figure).
Primary Exposures
We assessed depressive symptoms using the validated, modified 8-item version of the Center for Epidemiologic Studies-Depression (CES-D) scale.18 In each biennial questionnaire, participants were asked to respond (yes/no) whether they experienced each of the 8 symptoms in the past week. A summary score (range, 0–8) was created by summing the number of yes responses across the 8 items (with 2 positively framed items reverse-coded).18 We created depressive symptom trajectories using dichotomized CES-D summary scores for each wave (high versus low), based on a cutoff of ≥3 symptoms. This cutoff has been previously validated to have high sensitivity and specificity in identifying a major depressive episode, as defined by the Composite International Diagnostic Interview—Short Form.18,19 For individuals missing ≤2 items on the CES-D scale for any wave, the summary score for that wave was calculated using the available items, accounting for the number of items available. For individuals missing the full assessment on the CES-D summary score on one wave only, we imputed that wave’s summary score using multiple imputation by chained equations.20
Drawing on theory and prior empirical work, we constructed 5 depressive symptom trajectories (consistently low, decreasing, fluctuating, increasing, and consistently high) a priori, based on combinations of CES-D scores across the first 4 waves (T1–T4).21 Consistently low was defined as nonelevated depressive symptoms at a minimum of 3 timepoints and did not include high depressive symptoms at T1 or T4. Decreasing was defined as a trend of elevated depressive symptoms at (1) the first wave only, followed by continued nonelevated depressive symptoms across subsequent waves; or (2) the first 2 waves, followed by continued nonelevated depressive symptoms across subsequent waves. Increasing was defined as a trend of nonelevated depressive symptoms at (1) the first wave, followed by elevated depressive symptoms across subsequent waves; or (2) the first 2 waves, followed by continued elevated depressive symptoms across subsequent waves. Consistently high was defined as elevated depressive symptoms at a minimum of 3 timepoints across the assessment time period. Fluctuating encompassed the other depressive symptom trajectories that did not fit the above classifications. A visual depiction of the trajectories is provided in Table S1.
Stroke Outcomes
We defined incident stroke events as the first occurrence of stroke during the 10-year follow-up period (2006–2016), based on self- or proxy-reported doctor’s diagnosis (Has a doctor ever told you that you had a stroke?). For participants unavailable for a direct interview (eg, deceased), interviews were conducted with proxies (predominantly spouses). Reports of transient ischemic attacks were not systematically assessed and not coded as strokes, nor was stroke subtype information available.22 Prior work with HRS data has shown that associations between known risk factors and self-reported stroke incidence in the HRS corresponded well to associations in studies using clinically verified strokes.22 Furthermore, self-reported strokes in the HRS corresponded well with strokes coded according to the International Classification of Diseases in the Centers for Medicare and Medicaid Services records, with 74% sensitivity and 93% specificity.15
Covariates
We ascertained self-reported demographic and socioeconomic variables at baseline (T1): age (continuous), sex (male or female), race and ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic or other), highest level of education (less than high school, high school graduate/GED/some college, or 4-year college and above), household income (continuous), and marital status (married, separated or divorced, widowed, never married). We also obtained self-reported measures of health behaviors and health conditions at T1: physical activity (yes/no to engaging in vigorous activity ≥3 times/wk), alcohol consumption (none, moderate: <3 drinks/d or <18/wk, heavy: ≥3 drinks/d or ≥18/wk), body mass index (continuous, kg/m2 derived from self-reported height and weight), and cigarette smoking status (nonsmoker, former smoker, current smoker). Health conditions were assessed using responses (yes/no) to the question, Has a doctor ever told you that you had a (health condition)? for heart disease, diabetes, and hypertension. Self-reported health conditions in the HRS have been shown to have substantial agreement with medical records data, and prior studies using HRS have demonstrated that the self-reported health behavioral measures have good external validity.22–26
Statistical Analysis
Missingness overall was around 5%, the largest percentage of missing values were for the CES-D measure at T4 (4.79%). Health condition-related missing values were first replaced using information from the previous wave (if available). We imputed the remaining missing values using multiple imputation by chained equations (stroke missing values not imputed). The main analyses are applied to 10 imputed data sets.27
We examined the distribution of covariates in our overall sample and by trajectory group. After ascertaining that the proportional hazards assumption was not violated, Cox proportional hazards models estimated the hazard ratios (HR) and 95% CIs for the associations between depressive symptom trajectory groups and incident stroke, using the consistently low depressive symptom trajectory as the reference. We used clustered standard errors to account for the clustering of observations within marital dyads. For each analytic question, we evaluated a series of models. In model 1, we adjusted for age only. In model 2, we additionally adjusted for potential demographic confounders at T1 (sex, race and ethnicity, educational level, marital status, and household income). In model 3, we further adjusted for health behaviors and conditions at T1 (body mass index, smoking, alcohol consumption, physical activity, diabetes, and hypertension). We used interaction terms and stratified models to assess for effect modification by sex and race and ethnicity. All analyses were conducted using Stata (version 14, Stata Corp, College Station, TX). The study was conducted in accordance with the STROBE guideline for cohort studies (Supplemental Material).
We conducted 6 sensitivity analyses to assess the robustness of our study findings. First, we used a higher cutoff to determine high depressive symptoms (≥4 symptoms). Second, we kept the ≥3 symptoms cutoff, but required at least a 2-point change in depressive symptom levels between 2 consecutive assessments, to determine high depressive symptoms for that assessment. Third, we applied more stringent definitions for certain depressive symptom trajectories by excluding individuals in the increasing and decreasing trajectories who reported high or low depressive symptoms only at T4. We designed these analyses to assess whether our findings might differ based on how depressive symptom trajectory groups were defined. Fourth, we additionally adjusted for health conditions and behaviors at T4 to evaluate changes in effects due to potentially mediating health behavioral pathways. Fifth, we used inverse probability weighting to examine the impact of potential selection bias from excluding individuals based on our exposure period.28 Finally, we analyzed only participants with complete data on the exposure, outcome and covariates.
Results
Among 12 520 individuals, 1434 incident strokes occurred during the 10 years of follow-up (2006–2016). Tables 1 and 2 describe the demographic and health characteristics of the overall sample and by depressive symptom trajectory group. Among participants, mean age at T1 was 64.1 years and participants were predominantly female (61.7%), non-Hispanic White (77.9%), and married (68.6%). Individuals included in the analyses tended to be younger (64.1 versus 69.8 years), were more likely to self-identify as non-Hispanic White (77.9% versus 73.3%) and were more likely to be married (69.5% versus 63.0%; Table S2). Among participants, the majority (67.4%) had consistently low depressive symptoms over time (Table 1). Individuals with consistently high versus consistently low symptoms had lower average levels of education (less than high school: 47.7% versus 20.7%) and were less likely to be married (53.3% versus 73.3%).
| Overall sample | Depressive symptoms trajectory group | |||||
|---|---|---|---|---|---|---|
| Consistently low | Decreasing | Fluctuating | Increasing | Consistently high | ||
| Individuals, n (%) | 12 520 | 8439 (67.4) | 882 (7.0) | 778 (6.2) | 812 (6.5) | 1609 (12.9) |
| Sociodemographic characteristics | ||||||
| Age, y, mean (SD) | 64.1 (8.7) | 63.9 (8.5) | 63.9 (8.9) | 64.5 (9.2) | 65.7 (9.0) | 64.5 (9.2) |
| Males, n (%) | 4796 (38.3) | 3597 (42.6) | 296 (33.6) | 219 (28.2) | 276 (34.0) | 408 (25.4) |
| Race and ethnicity, n (%) | ||||||
| Non-Hispanic White | 9750 (77.9) | 6923 (82.0) | 601 (68.1) | 554 (71.2) | 641 (78.9) | 1031 (64.1) |
| Non-Hispanic Black | 1610 (12.9) | 917 (10.9) | 155 (17.6) | 136 (17.5) | 99 (12.2) | 303 (18.8) |
| Hispanic or other | 1159 (9.3) | 599 (7.1) | 126 (14.3) | 85 (11.3) | 72 (8.9) | 274 (17.0) |
| Highest degree in education, n (%) | ||||||
| Less than high school | 3307 (26.4) | 1743 (20.7) | 283 (32.1) | 263 (33.8) | 250 (30.8) | 768 (47.7) |
| High school | 6689 (53.4) | 4648 (55.1) | 458 (51.9) | 420 (54.0) | 432 (53.2) | 731 (45.4) |
| College and above | 2524 (20.2) | 2048 (24.3) | 141 (16.0) | 95 (12.2) | 130 (16.0) | 110 (6.8) |
| Income (in 1000s, USD), mean (SD) | 43.0 (65.8) | 47.0 (69.8) | 41.5 (57.7) | 36.2 (63.8) | 38.7 (51.5) | 28.1 (52.1) |
| Marital status, n (%) | ||||||
| Married | 8590 (68.6) | 6,189 (73.3) | 502 (56.9) | 476 (61.2) | 565 (69.6) | 858 (53.3) |
| Separated or divorced | 1481 (11.8) | 848 (10.1) | 134 (15.2) | 123 (15.8) | 83 (10.2) | 293 (18.2) |
| Widowed | 1908 (15.2) | 1056 (12.5) | 214 (24.3) | 148 (19.0) | 133 (16.4) | 357 (22.2) |
| Never married | 382 (3.1) | 249 (3.0) | 27 (3.1) | 22 (2.8) | 22 (2.7) | 62 (3.9) |
Includes individuals who have an imputed CES-D summary score. Number of missing values before imputation: race 0.01%; income 0.9%; marital status 1.3%. All others: no missing values. CES-D indicates Center for Epidemiologic Studies-Depression.
| Overall sample | Depressive symptoms trajectory group | |||||
|---|---|---|---|---|---|---|
| Consistently low | Decreasing | Fluctuating | Increasing | Consistently high | ||
| Individuals, n (%) | 12 520 | 8439 (67.4) | 882 (7.0) | 778 (6.2) | 812 (6.5) | 1609 (12.9) |
| Health behaviors | ||||||
| Physical activity (≥3× a week), n (%) | 6104 (48.8) | 4507 (53.4) | 407 (46.2) | 318 (40.9) | 360 (44.3) | 512 (31.8) |
| Drinking, n (%) | ||||||
| None | 5914 (47.2) | 3611 (42.8) | 441 (50.0) | 437 (56.2) | 428 (52.7) | 997 (62.0) |
| Moderate (<3 drinks per day or <18 per week) | 5627 (44.9) | 4167 (49.4) | 372 (42.2) | 272 (35.0) | 334 (41.1) | 482 (30.0) |
| Heavy (≥3 drinks per day or ≥18 per week) | 979 (7.8) | 661 (7.8) | 69 (7.8) | 69 (8.9) | 50 (6.2) | 130 (8.1) |
| BMI, kg/m2, mean (SD) | 27.9 (5.3) | 27.6 (4.9) | 28.4 (5.6) | 28.2 (6.2) | 28.5 (5.5) | 29.2 (6.5) |
| Smoking, n (%) | ||||||
| Nonsmoker | 5336 (42.6) | 3675 (43.6) | 367 (41.6) | 314 (40.4) | 333 (41.0) | 647 (40.2) |
| Former smoker | 5143 (41.1) | 3551 (42.1) | 348 (39.5) | 311 (40.0) | 338 (41.6) | 595 (37.0) |
| Current smoker | 1951 (15.6) | 1153 (13.7) | 161 (18.3) | 148 (19.0) | 132 (16.3) | 357 (22.2) |
| Health conditions | ||||||
| Heart condition (yes/no), n (%) | 1289 (10.3) | 729 (8.6) | 100 (11.3) | 112 (14.4) | 99 (12.2) | 249 (15.5) |
| Hypertension (yes/no), n (%) | 4228 (33.8) | 2546 (30.2) | 330 (37.4) | 327 (42.0) | 299 (36.8) | 726 (45.1) |
| Diabetes (yes/no), n (%) | 1122 (9.0) | 609 (7.2) | 95 (10.8) | 113 (14.5) | 84 (10.3) | 221 (13.7) |
Includes individuals who have an imputed CES-D summary score. Number of missing values before imputation: Physical activity 0.9%; BMI 2.3%; Smoking 0.7%; heart condition 0.8%; hypertension 0.8%; diabetes 0.8%. All others: no missing values. BMI indicates body mass index; and CES-D, Center for Epidemiologic Studies-Depression.
In our main analyses (Table 3), compared with individuals with consistently low depressive symptoms (reference), individuals with consistently high depressive symptoms had a higher risk for incident stroke using model 3 (adjusted HR, 1.18 [95% CI, 1.02–1.36]). Similarly, individuals with increasing (adjusted HR, 1.31 [95% CI, 1.10–1.57]) and fluctuating (adjusted HR, 1.21 [95% CI, 1.01–1.46]) depressive symptoms had higher risk for incident stroke, compared with the reference group. The effect estimates for incident stroke were not significantly different for individuals with decreasing depressive symptoms (adjusted HR, 1.02 [95% CI, 0.84–1.24]), when compared with the reference group. The tests for interactions showed no evidence of effect modification by sex, nor race and ethnicity, in the association between depressive symptom trajectories and incident stroke. Stratified results by sex and race and ethnicity are shown in Tables S3 and S4.
| Depressive symptoms trajectory group | No. of person-years | No. of cases | Total sample (N=12 520) | ||
|---|---|---|---|---|---|
| Model 1* HR (95% CI) | Model 2† HR (95% CI) | Model 3‡ HR (95% CI) | |||
| Consistently low (ref) | 82 110 | 871 | reference | reference | reference |
| Decreasing | 8533 | 100 | 1.09 (0.90–1.32) | 1.05 (0.87–1.28) | 1.02 (0.84–1.24) |
| Fluctuating | 7483 | 110 | 1.32 (1.10–1.59) | 1.27 (1.06–1.53) | 1.21(1.01–1.46) |
| Increasing | 7775 | 125 | 1.38 (1.16–1.65) | 1.36 (1.14–1.62) | 1.31(1.10–1.57) |
| Consistently high | 15 444 | 228 | 1.33 (1.16–1.52) | 1.25(1.09–1.44) | 1.18 (1.02–1.36) |
BMI indicates body mass index; HR‚ hazard ratio.
*
Model 1 adjusts for age only.
†
Model 2 additionally adjusts for sociodemographics (race and ethnicity, education, income, marital status, and sex).
‡
Model 3 additionally adjusts for health conditions and behaviors (physical activity, alcohol consumption, BMI, smoking status, heart conditions, diabetes, and hypertension).
The 6 sensitivity analyses showed that our findings were robust to different exposure classifications (Tables S5 through S7), were not largely attenuated by health behaviors and conditions (Table S8), and were not largely affected by loss to follow-up (Tables S9 and S10), although some attenuated associations for incident stroke with consistently high and fluctuating trajectories were evident.
Discussion
In this study of middle-aged and older US adults, individuals in consistently high, fluctuating and increasing depressive symptom trajectory groups had 18% to 31% higher hazard for developing incident stroke, relative to those with consistently low depressive symptoms. Our findings were robust to adjustment for several potential confounders and covariates that might lie on the pathway between elevated depressive symptoms and stroke risk. Our findings were also robust to changes in the depressive symptom trajectory definitions. Contrary to our hypothesis, we noted that the hazard for the decreasing depressive symptom trajectory group was not substantially different from those with consistently low depressive symptoms. Taken together, our results suggest that repeated occurrences of elevated depressive symptoms increase stroke risk, perhaps through accumulated cardiovascular damage over time. However, with sufficient time for recovery (ie, a period of repeated low depressive symptoms), even individuals who have experienced high depressive symptoms may not be at increased risk for stroke.
Potential biological mechanisms underlying the association between repeatedly elevated depressive symptoms and stroke include neuroendocrine dysregulation as characterized by sympathetic nervous system activation, dysregulation of the hypothalamic pituitary adrenocortical axis, platelet aggregation dysfunction, and immunologic/inflammatory factors.29 A prior meta-analysis examining depression treatment among depressed patients has shown that inflammation levels are reduced posttreatment, demonstrating how improved depressive symptoms may benefit stroke prevention.30 Depressive symptoms are also associated with unhealthy biobehavioral factors (eg, smoking, physical inactivity, poor diet, insomnia and poor sleep quality, lack of medication compliance, and obesity), which may increase vascular risk and are considered important stroke risk factors.6,8,29,31–34 A prior study on diabetic patients has shown that depression treatment can improve exercise rates and functional status, suggesting that improved depression may improve biobehavioral factors and ultimately stroke risk.35 Although we did not formally assess for mediating pathways, our sensitivity analysis additionally adjusting for health behaviors and conditions at T4 showed that risk estimates were modestly attenuated. This suggests that health behaviors may explain only a modest proportion of the observed association, or that our measures of biobehavioral factors were imprecise. In either case, future studies may want to explore other potential mechanisms that may mediate the association between depressive symptom trajectories and incident stroke.
In addition to the above mechanisms linking depressive symptoms to stroke more generally, there may also be trajectory-specific mechanisms. Increasing depressive symptoms, for example, may lead to more rapid onset of other conditions (eg, obesity, hypertension) that may increase stroke risk.36 Fluctuations in depressive symptoms may lead to greater inflammation or dysregulation in other biological processes linked to stroke risk, similar to findings on body weight fluctuations.37 Additional work will be needed to assess potential trajectory-specific mechanisms.
Individuals in different depressive symptom trajectory groups may also have different baseline characteristics, representing different subgroups in the overall population.38 For example, increasing depressive symptoms may indicate prodromal symptoms for other illnesses (ie, dementia) or represent a late onset of depressive symptoms in response to other physical illnesses, such as hypertension, high blood cholesterol, diabetes or undiagnosed silent strokes,21,39,40 that are also associated with higher risk of incident stroke.36 There may also be a shared genetic basis for depression and other physical illnesses that also influence stroke.41 For example, the fluctuating depressive symptom trajectory group may include a subgroup of individuals with bipolar or other cyclic disorders, with a different genetic basis than unipolar depression. Such disorders may be associated with stroke risk for varied reasons, including use of mood stabilizing medications for treating bipolar disorder (rather than effects of the disorder per se), or because of drug-induced weight gain and metabolic disturbances and poor access to health care for individuals with bipolar disorder.42,43
Previously, only 2 studies have examined depressive symptom changes on incident stroke, with inconclusive results.14,15 Further, both studies categorized depressive symptom changes across only 2 assessment periods 1 to 2 years apart. This study contributes substantively to the extant literature by examining different temporal patterns of depressive symptoms and providing evidence of its longer-term impact on incident stroke. Our findings suggest that even 2 timepoints may be insufficient for determining subgroups of high-risk individuals. To illustrate, one additional assessment of high or low depressive symptoms can determine different trajectory group assignments for an individual who reported high depressive symptoms in one wave, and low depressive symptoms in the following wave. An individual previously categorized as increasing or decreasing with only 2 assessments, may now be categorized in a completely different trajectory, with consequent change in stroke hazard as well.
Future research may evaluate how frequently depressive symptoms should be assessed, and how many assessments would be sufficient to chart the course of depressive symptoms to determine high-risk groups for stroke. Another area of research would be to elucidate further the potential underlying mechanisms linking depressive symptoms and stroke, with special attention to the possibility that one overall mechanism might not encompass all high-risk subpopulations indicated by the different depressive symptom trajectory groups. Other potential directions may be to examine by stroke subtype (which may have different causes) or assess the impact of antidepressants or other medication use, which were unavailable for this study. Antidepressant use has been identified as a pathway from high depressive symptoms to increased stroke risk or may also indicate that those who take medication have more severe depressive symptoms.44,45 Future studies may compare how treatment for high depressive symptoms (eg, antidepressant use or psychological therapies) may differentially affect stroke risk, compared with natural remission of depressive symptoms.
Strengths of our study include its prospective study design, repeated assessment of depressive symptoms, long-term follow-up, and wide range of covariates obtained using standardized protocols. With repeated assessments over a long exposure period, we were able to capture nuances in how depressive symptom patterns may differentially affect incident stroke risk and also minimize the likelihood of reverse causation. Several limitations are present in our study. Individuals with missing CES-D measures and those who experienced stroke during the exposure assessment period were excluded from the analyses, limiting generalizability of our results to individuals who provided CES-D measures and did not experience a stroke prior or concurrent to our exposure assessment period. Additionally, our study examined US adults ≥50 years, and as a result, our findings may not be generalizable to younger or non-US populations. However, our IPW results suggest that the effect estimates were only minimally attenuated. We also used predefined trajectory group definitions based on a dichotomous categorization of the CES-D scores, which is only one possible representation of the accumulative progression of depressive symptoms over time. However, we conducted a series of sensitivity analyses to test the effects of varying trajectory definitions, through which we demonstrated the robustness of our results. Additionally, differential adherence to medications (ie, for treatment of diabetes, hypertension or high cholesterol) might partially explain some of the observed relationships between depressive symptom trajectories and stroke risk. However, medication adherence can only be assessed through pharmacy records, which was not available for our analyses.
Conclusions
Repeatedly elevated depressive symptoms may result in higher risk for incident stroke. However, risk for stroke may change alongside depressive symptom changes, as individuals with decreasing depressive symptoms over time showed no higher risk than those with consistently low symptoms. Increasing depressive symptoms or chronically high depressive symptoms may potentially indicate future cerebrovascular problems, suggesting a potential role for clinicians in stroke prevention (eg, routine screening for depressive symptoms or referrals to behavioral specialists for patients displaying increasing or repeatedly severe depressive symptoms). Further studies are needed to confirm these initial findings, to explore the underlying overall and trajectory-specific mechanisms of these associations, and to determine how frequently depressive symptoms should be assessed to best determine and treat high-risk groups for clinical care.
Article Information
Supplemental Material
STROBE Checklist
Tables S1–S10
Acknowledgments
This work contains results that are part of the doctoral thesis of Y.S.46
Footnote
Nonstandard Abbreviations and Acronyms
- CES-D
- Center for Epidemiologic Studies-Depression
- HR
- hazard ratio
- HRS
- Health and Retirement Study
- T1, T2, T3, T4
- time 1, time 2, time 3, time 4
Supplemental Material
File (str_stroke-2021-037768-t_supp1.pdf)
- Download
- 290.84 KB
References
1.
Johnson CO, Nguyen M, Roth GA, Nichols E, Alam T, Abate D, Abd-Allah F, Abdelalim A, Abraha HN, Abu-Rmeileh NM, et al. Global, regional, and national burden of stroke, 1990–2016: a systematic analysis for the global burden of disease study 2016. Lancet Neurol. 2019;18:439–458. doi: 10.1016/S1474-4422(19)30034-1
2.
Virani SS, Alonso A, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Chang AR, Cheng S, Delling FN, et al; American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics-2020 update: a report from the American heart association. Circulation. 2020;141:e139–e596. doi: 10.1161/CIR.0000000000000757
3.
Marsh JD, Keyrouz SG. Stroke prevention and treatment. J Am Coll Cardiol. 2010;56:683–691. doi: 10.1016/j.jacc.2009.12.072
4.
Bang OY, Ovbiagele B, Kim JS. Nontraditional risk factors for ischemic stroke: an update. Stroke. 2015;46:3571–3578. doi: 10.1161/STROKEAHA.115.010954
5.
Surtees PG, Wainwright NW, Luben RN, Wareham NJ, Bingham SA, Khaw KT. Psychological distress, major depressive disorder, and risk of stroke. Neurology. 2008;70:788–794. doi: 10.1212/01.wnl.0000304109.18563.81
6.
Salaycik KJ, Kelly-Hayes M, Beiser A, Nguyen AH, Brady SM, Kase CS, Wolf PA. Depressive symptoms and risk of stroke: the Framingham study. Stroke. 2007;38:16–21. doi: 10.1161/01.STR.0000251695.39877.ca
7.
Glymour MM, Maselko J, Gilman SE, Patton KK, Avendaño M. Depressive symptoms predict incident stroke independently of memory impairments. Neurology. 2010;75:2063–2070. doi: 10.1212/WNL.0b013e318200d70e
8.
Jonas BS, Mussolino ME. Symptoms of depression as a prospective risk factor for stroke. Psychosom Med. 2000;62:463–471. doi: 10.1097/00006842-200007000-00001
9.
Barlinn K, Kepplinger J, Puetz V, Illigens BM, Bodechtel U, Siepmann T. Exploring the risk-factor association between depression and incident stroke: a systematic review and meta-analysis. Neuropsychiatr Dis Treat. 2015;11:1–14. doi: 10.2147/NDT.S63904
10.
Dong JY, Zhang YH, Tong J, Qin LQ. Depression and risk of stroke: a meta-analysis of prospective studies. Stroke. 2012;43:32–37. doi: 10.1161/STROKEAHA.111.630871
11.
Li M, Zhang XW, Hou WS, Tang ZY. Impact of depression on incident stroke: a meta-analysis. Int J Cardiol. 2015;180:103–110. doi: 10.1016/j.ijcard.2014.11.198
12.
Alexopoulos GS, Meyers BS, Young RC, Campbell S, Silbersweig D, Charlson M. ‘Vascular depression’ hypothesis. Arch Gen Psychiatry. 1997;54:915–922. doi: 10.1001/archpsyc.1997.01830220033006
13.
McDougall GJ, Morgan S, Vaughan PW. Sixteen-month evaluation of depressive symptomatology in older adults. Arch Psychiatr Nurs. 2012;26:e13–e21. doi: 10.1016/j.apnu.2011.12.004
14.
Gilsanz P, Kubzansky LD, Tchetgen Tchetgen EJ, Wang Q, Kawachi I, Patton KK, Fitzpatrick AL, Kop WJ, Longstreth WT, Glymour MM. Changes in depressive symptoms and subsequent risk of stroke in the cardiovascular health study. Stroke. 2017;48:43–48. doi: 10.1161/STROKEAHA.116.013554
15.
Gilsanz P, Walter S, Tchetgen Tchetgen EJ, Patton KK, Moon JR, Capistrant BD, Marden JR, Kubzansky LD, Kawachi I, Glymour MM. Changes in depressive symptoms and incidence of first stroke among middle-aged and older US adults. J Am Heart Assoc. 2015;4:e001923. doi: 10.1161/JAHA.115.001923
16.
Juster FT, Suzman R. An overview of the health and retirement study. J Hum Resour. 1995;30:S7. doi: 10.2307/146277
17.
Sonnega A, Faul JD, Ofstedal MB, Langa KM, Phillips JW, Weir DR. Cohort profile: the health and retirement study (HRS). Int J Epidemiol. 2014;43:576–585. doi: 10.1093/ije/dyu067
18.
Steffick, DE. Documentation of Affective Functioning Measures in the Health and Retirement Study [Internet]. Institute for Social Research, University of Michigan; 2000 [cited Jul 8, 2018]. http://hrsonline.isr.umich.edu/sitedocs/userg/dr-005.pdf.
19.
Glymour MM, Yen JJ, Kosheleva A, Moon JR, Capistrant BD, Patton KK. Elevated depressive symptoms and incident stroke in Hispanic, African-American, and white older Americans. J Behav Med. 2012;35:211–220. doi: 10.1007/s10865-011-9356-2
20.
Azur MJ, Stuart EA, Frangakis C, Leaf PJ. Multiple imputation by chained equations: what is it and how does it work? Int J Methods Psychiatr Res. 2011;20:40–49. doi: 10.1002/mpr.329
21.
Mirza SS, Wolters FJ, Swanson SA, Koudstaal PJ, Hofman A, Tiemeier H, Ikram MA. 10-year trajectories of depressive symptoms and risk of dementia: a population-based study. Lancet Psychiatry. 2016;3:628–635. doi: 10.1016/S2215-0366(16)00097-3
22.
Glymour MM, Avendano M. Can self-reported strokes be used to study stroke incidence and risk factors? Evidence from the health and retirement study. Stroke. 2009;40:873–879. doi: 10.1161/STROKEAHA.108.529479
23.
Fisher, GG, Faul, JD, Weir, DR, Wallace, RB. Documentation of Chronic Disease Measures in the Health and Retirement Study [Internet]. Institute for Social Research, University of Michigan; 2005 [cited Jul 26, 2018]. https://hrs.isr.umich.edu/sites/default/files/biblio/dr-009.pdf.
24.
Okura Y, Urban LH, Mahoney DW, Jacobsen SJ, Rodeheffer RJ. Agreement between self-report questionnaires and medical record data was substantial for diabetes, hypertension, myocardial infarction and stroke but not for heart failure. J Clin Epidemiol. 2004;57:1096–1103. doi: 10.1016/j.jclinepi.2004.04.005
25.
Wallace RB, Herzog AR. Overview of the health measures in the health and retirement study. J Hum Resour. 1995;30:S84–S107. doi: 10.2307/146279
26.
Jenkins, KR, Ofstedal, MB, Weir, D. Documentation of Health Behaviors and Risk Factors Measured in the Health and Retirement Study (HRS/AHEAD) [Internet]. Institute for Social Research, University of Michigan; 2008 [cited Jun 24, 2019]. http://hrsonline.isr.umich.edu/sitedocs/userg/dr-010.pdf.
27.
von Hippel PT. How many imputations do you need? A two-stage calculation using a quadratic rule. Sociol Methods Res. 2020;49:699–718. doi: 10.1177/0049124117747303
28.
Hernán MA, Hernández-Díaz S, Robins JM. A structural approach to selection bias. Epidemiology. 2004;15:615–625. doi: 10.1097/01.ede.0000135174.63482.43
29.
Pan A, Sun Q, Okereke OI, Rexrode KM, Hu FB. Depression and risk of stroke morbidity and mortality: a meta-analysis and systematic review. JAMA. 2011;306:1241–1249. doi: 10.1001/jama.2011.1282
30.
Strawbridge R, Arnone D, Danese A, Papadopoulos A, Herane Vives A, Cleare AJ. Inflammation and clinical response to treatment in depression: a meta-analysis. Eur Neuropsychopharmacol. 2015;25:1532–1543. doi: 10.1016/j.euroneuro.2015.06.007
31.
Sotelo JL, Nemeroff CB. Depression as a systemic disease. Pers Med Psychiatry. 2017;1:11–25. doi: 10.1016/j.pmip.2016.11.002
32.
Strine TW, Mokdad AH, Dube SR, Balluz LS, Gonzalez O, Berry JT, Manderscheid R, Kroenke K. The association of depression and anxiety with obesity and unhealthy behaviors among community-dwelling US adults. Gen Hosp Psychiatry. 2008;30:127–137. doi: 10.1016/j.genhosppsych.2007.12.008
33.
Joynt KE, Whellan DJ, O’Connor CM. Depression and cardiovascular disease: mechanisms of interaction. Biol Psychiatry. 2003;54:248–261. doi: 10.1016/s0006-3223(03)00568-7
34.
Gangwisch JE, Malaspina D, Posner K, Babiss LA, Heymsfield SB, Turner JB, Zammit GK, Pickering TG. Insomnia and sleep duration as mediators of the relationship between depression and hypertension incidence. Am J Hypertens. 2010;23:62–69. doi: 10.1038/ajh.2009.202
35.
Williams JW, Katon W, Lin EH, Nöel PH, Worchel J, Cornell J, Harpole L, Fultz BA, Hunkeler E, Mika VS, et al; IMPACT Investigators. The effectiveness of depression care management on diabetes-related outcomes in older patients. Ann Intern Med. 2004;140:1015–1024. doi: 10.7326/0003-4819-140-12-200406150-00012
36.
Musselman DL, Evans DL, Nemeroff CB. The relationship of depression to cardiovascular disease: epidemiology, biology, and treatment. Arch Gen Psychiatry. 1998;55:580–592. doi: 10.1001/archpsyc.55.7.580
37.
Zou H, Yin P, Liu L, Liu W, Zhang Z, Yang Y, Li W, Zong Q, Yu X. Body-weight fluctuation was associated with increased risk for cardiovascular disease, all-cause and cardiovascular mortality: a systematic review and meta-analysis. Front Endocrinol (Lausanne). 2019;10:728. doi: 10.3389/fendo.2019.00728
38.
Byers AL, Vittinghoff E, Lui LY, Hoang T, Blazer DG, Covinsky KE, Ensrud KE, Cauley JA, Hillier TA, Fredman L, Yaffe K. Twenty-year depressive trajectories among older women. Arch Gen Psychiatry. 2012;69:1073–1079. doi: 10.1001/archgenpsychiatry.2012.43
39.
Fiske A, Wetherell JL, Gatz M. Depression in older adults. Annu Rev Clin Psychol. 2009;5:363–389. doi: 10.1146/annurev.clinpsy.032408.153621
40.
Mirza SS, de Bruijn RF, Direk N, Hofman A, Koudstaal PJ, Ikram MA, Tiemeier H. Depressive symptoms predict incident dementia during short- but not long-term follow-up period. Alzheimers Dement. 2014;10(5 Suppl):S323–S329.e1. doi: 10.1016/j.jalz.2013.10.006
41.
Wassertheil-Smoller S, Qi Q, Dave T, Mitchell BD, Jackson RD, Liu S, Park K, Salinas J, Dunn EC, Leira EC, et al. Polygenic risk for depression increases risk of ischemic stroke: from the stroke genetics network study. Stroke. 2018;49:543–548. doi: 10.1161/STROKEAHA.117.018857
42.
Prieto ML, Cuéllar-Barboza AB, Bobo WV, Roger VL, Bellivier F, Leboyer M, West CP, Frye MA. Risk of myocardial infarction and stroke in bipolar disorder: a systematic review and exploratory meta-analysis. Acta Psychiatr Scand. 2014;130:342–353. doi: 10.1111/acps.12293
43.
Chen PH, Tsai SY, Pan CH, Chang CK, Su SS, Chen CC, Kuo CJ. Mood stabilisers and risk of stroke in bipolar disorder. Br J Psychiatry. 2019;215:409–414. doi: 10.1192/bjp.2018.203
44.
Chen Y, Guo JJ, Li H, Wulsin L, Patel NC. Risk of cerebrovascular events associated with antidepressant use in patients with depression: a population-based, nested case-control study. Ann Pharmacother. 2008;42:177–184. doi: 10.1345/aph.1K369
45.
Glassman AH, Roose SP, Bigger JT. The safety of tricyclic antidepressants in cardiac patients. Risk-benefit reconsidered. JAMA. 1993;269:2673–2675.
46.
Soh, Y. Psychosocial Factors and Risk of Incident Cardiovascular Disease Among Middle-Aged and Older Adults. [Doctoral Dissertation]. Harvard T.H. Chan School of Public Health; 2020.
Information & Authors
Information
Published In
Copyright
© 2022 American Heart Association, Inc.
Versions
You are viewing the most recent version of this article.
History
Received: 20 October 2021
Revision received: 30 March 2022
Accepted: 13 April 2022
Published online: 23 May 2022
Published in print: August 2022
Keywords
Subjects
Authors
Disclosures
Disclosures Dr Berkman reports compensation from Fidelity Investments for consultant services. The other authors report no conflicts.
Sources of Funding
The HRS (Health and Retirement Study) is sponsored by the National Institute on Aging (NIA U01AG009740) and is conducted by the University of Michigan.
Metrics & Citations
Metrics
Citations
Download Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Select your manager software from the list below and click Download.
- Association between depression symptom severity and total cost of care: Findings from a large, 2-year, claims-based, retrospective population health study, Journal of Affective Disorders, 368, (41-47), (2025).https://doi.org/10.1016/j.jad.2024.09.056
- A Retrospective Analysis of the Underlying Health Status of Patients Treated for Stroke in the Emergency Department of a Community Hospital Situated in a Health Professional Shortage Area, Cureus, (2024).https://doi.org/10.7759/cureus.68150
- Association of both depressive symptoms scores and specific depressive symptoms with all-cause and cardiovascular disease mortality, Annals of General Psychiatry, 23, 1, (2024).https://doi.org/10.1186/s12991-024-00509-x
- Adult child educational attainment and older parents’ psychosocial outcomes during the COVID-19 pandemic, BMC Public Health, 24, 1, (2024).https://doi.org/10.1186/s12889-024-19425-6
- Changes in muscle strength and risk of cardiovascular disease among middle-aged and older adults in China: Evidence from a prospective cohort study, Chinese Medical Journal, 137, 11, (1343-1350), (2024).https://doi.org/10.1097/CM9.0000000000002968
- Trajectories of pain and their associations with long-term cognitive decline in older adults: evidence from two longitudinal cohorts, Age and Ageing, 53, 8, (2024).https://doi.org/10.1093/ageing/afae183
- Eight-year total, cognitive-affective, and somatic depressive symptoms trajectories and risks of cardiac events, Translational Psychiatry, 14, 1, (2024).https://doi.org/10.1038/s41398-024-03063-y
- Evaluation of Social Isolation Trajectories and Incident Cardiovascular Disease Among Middle-Aged and Older Adults in China: National Cohort Study, JMIR Public Health and Surveillance, 9, (e45677), (2023).https://doi.org/10.2196/45677
- Adverse childhood and adulthood experiences and risk of new-onset cardiovascular disease with consideration of social support: a prospective cohort study, BMC Medicine, 21, 1, (2023).https://doi.org/10.1186/s12916-023-03015-1
- Depressive Symptom Trajectories and Cognition Among Older American Couples: A Dyadic Perspective, Journal of Aging and Health, 35, 3-4, (282-293), (2022).https://doi.org/10.1177/08982643221125838
Loading...
View Options
Get Access
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginPurchase Options
Purchase this article to access the full text.
eLetters(0)
eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.
Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.