Depressive Symptoms after Ischemic Stroke: Population-based Comparisons of Patients and Caregivers with Matched Controls

Abstract

Background and Purpose:

Persistent depression after ischemic stroke is common in stroke survivors and may be even higher in family caregivers, but few studies have examined depressive symptom levels and their predictors in patient and caregiver groups simultaneously.

Methods:

Stroke survivors and their family caregivers (205 dyads) were enrolled from the national REasons for Geographic and Racial Differences in Stroke (REGARDS) study into the Caring for Adults Recovering from the Effects of Stroke (CARES) study approximately 9-months after a first-time ischemic stroke. Demographically-matched stroke free dyads (N = 205) were also enrolled. Participants were interviewed by telephone and depressive symptoms were assessed with the 20-item Center for Epidemiological Studies Depression scale.

Results:

Significant elevations in depressive symptoms (Ps < 0.03) were observed for stroke survivors (M = 8.38) and for their family caregivers (M = 6.42) relative to their matched controls (Ms = 5.18 and 4.62, respectively). Stroke survivors reported more symptoms of depression than their caregivers (P = 0.008). No race or sex differences were found, but differential prediction of depressive symptom levels was found across patients and caregivers. Younger age and having an older caregiver were associated with more depressive symptoms in stroke survivors, while being a spouse caregiver and reporting fewer positive aspects of caregiving were associated with more depressive symptoms in caregivers. The percentage of caregivers at risk for clinically significant depression was lower in this population-based sample (12%) than in previous studies of caregivers from convenience or clinical samples.

Conclusions:

High depressive symptom levels are common 9 months after first-time ischemic strokes for stroke survivors and family caregivers, but rates of depressive symptoms at risk for clinical depression were lower for caregivers than previously reported. Predictors of depression differ for patients and caregivers, and standards of care should incorporate family caregiving factors.

Among the most prevalent long-term effects of ischemic stroke are changes in the stroke survivor’s psychological and social functioning. This includes frequent problems with post-stroke depression up to 5 years after the stroke event,^{1, 2} with rates of persistent post-stroke depression estimated to range from 17% to 25%.^{2, 3} Stroke is often found to disproportionately affect blacks, who are more likely than whites to experience strokes^{4, 5} and to have poorer long-term functional outcomes after stroke.^6–8 However, black stroke survivors are typically not found to experience higher levels of post-stroke depression^{9, 10} and may be less likely in some clinical settings to receive a formal diagnosis of depression.^11–13

Stroke often affects the lives of both stroke survivors and their family caregivers, and elevated rates of symptoms of depression that persist for years after stroke have been found for stroke caregivers compared to matched non-caregiving controls.¹⁴ Interestingly, rates of clinically significant depression for family caregivers of stroke survivors have been reported to be as high as 30%¹⁵ to 45%.¹⁶ These rates for caregivers exceed reported rates of post-stroke depression for stroke survivors themselves. A recent systematic review and meta-analysis of 12 studies across 6 different countries reported a 40% rate of post-stroke depression among caregivers and found significantly more depressive symptoms among women, whites, and those caring for female stroke survivors.¹⁷ Subsequent to this meta-analysis, a more recent population-based study did not find any race differences in depressive symptom levels among stroke caregivers.¹⁸

There are relatively few studies that have examined levels of depressive symptoms, rates of clinically significant depression, and potential race and other demographic differences in depression among stroke survivors and their family caregivers simultaneously. We are aware of no previous study that has done so using a population-based sample. In this report, we present analyses of depressive symptoms for stroke survivors, family caregivers, and carefully matched control dyads from an ancillary study of the national REasons for Geographic and Racial Differences in Stroke (REGARDS) study. We focused specifically on evaluating whether there were significant racial differences in depressive symptoms between black and white stroke survivors and their family caregivers, and we examined the salient predictors of post-stroke depressive symptoms levels in both groups of participants.

Methods

In cooperation with the Institutional Review Board (IRB) of the University of Alabama at Birmingham (UAB), the REGARDS project facilitates data sharing through formal data use agreements. Investigators who wish to access the data should send their requests to regardsadmin@uab.edu.

Participants

The present analyses are based on 205 survivors of an adjudicated ischemic stroke event and their 205 primary family caregivers. All participants were enrolled in the Caring for Adults Recovering from the Effects of Stroke (CARES) study,^{8, 14, 19} which is an ancillary project to the national REGARDS study. Stroke dyads were individually matched with 205 control dyads that consisted of participants with no history of stroke (stroke free controls) and non-caregiving family members of those stroke free controls (family member controls).

The REGARDS study is an ongoing epidemiological investigation of stroke incidence that enrolled 30,239 participants 45 years of age or older from 2003 to 2007. Residents of the stroke belt of the United States and blacks/African Americans were oversampled by design.²⁰ Only individuals who identified themselves as either white/Caucasian or black/African American were enrolled into the REGARDS study because of the specific interest in that racial comparison. Individuals who reported being Hispanic were excluded. The design, recruitment, and enrollment procedures for the REGARDS project are described in more detail elsewhere.^{8, 20}

The 205 stroke survivors and 205 stroke-free controls included in the present analyses were participants of the REGARDS project and the CARES study, whereas the 205 family caregivers and 205 family member controls were participants of the CARES study only. The stroke survivors and stroke free controls reported no previous history of stroke during their initial REGARDS baseline interview (see below).

Each participant received $20 per interview for their participation in the CARES project. As approved by the IRB of UAB, all participants gave verbal informed consent at the beginning of the telephone interviews, and written informed consent was later obtained from those participants who also completed the in-home visits.

Procedures

The data used in the present analyses were collected from multiple REGARDS and CARES data collection sessions. A summary of those sessions and the measures obtained are provided below:

REGARDS baseline interview.

Stroke survivors and stroke free controls completed a baseline computer-assisted telephone interview (CATI) at the time of their enrollment into the REGARDS study. During that interview, participants were asked if a physician had ever told them that they had a stroke. Participants who answered “yes” were excluded from the present analyses, and those without a self-reported history of stroke were included. In addition, all REGARDS participants were asked if they had a family member who would be available to help them if they experienced a serious illness or became disabled at some point in the future. If they answered “yes” (more than 80% did so²¹), they were further asked who that person was, their relationship with that person, and whether they lived with that person. These questions served to identify a potential family caregivers if such a need would arise in the future and were used to identify the family member control participants in the control dyads.

All REGARDS participants were also administered a 4-item version of the Center for Epidemiological Studies Depression Scale²² to screen for depressive symptoms at enrollment into the REGARDS study. Scores ranged from 0 to 12, with higher scores indicating more depressive symptoms.

REGARDS follow-up interviews and stroke event adjudication.

Follow-up telephone interviews were conducted every six months by the REGARDS team after enrollment that included questions about hospitalizations, emergency department visits, and symptoms that might indicate possible stroke events. Medical records were subsequently obtained for possible stroke events and examined by at least two trained adjudicators. Variables extracted from records that confirmed stroke events were the date of the stroke, type of stroke (ischemic vs. hemorrhagic), length of the acute hospital stay (if any), and the discharge status of that hospitalization. Only those participants with adjudicated ischemic strokes were included in the present analyses as stroke survivors.

CARES enrollment.

Once stroke survivors were identified by the REGARDS team, those participants were referred for further screening and possible enrollment into the CARES study. Stroke survivors were eligible to participate in the CARES study if 1) they were community-dwelling 9 months after the stroke event (i.e., residents of nursing homes and other institutional residential settings were excluded); 2) they had a family member or close friend who was either serving as an informal caregiver or had served in that capacity at some point after the stroke; and 3) that caregiver was also willing to participate in the CARES project. Potential caregivers were not excluded based on any criteria relating to the type, duration, or amount of care provided. Formal caregivers (i.e., paid professional healthcare workers) were excluded from the caregiving group. Among eligible stroke survivor-caregiver dyads, approximately 80% agreed to participate.⁸ For these dyads, initial baseline interviews were conducted by telephone approximately 9 months after the stroke event (Median = 9.5, Min = 7.1, Max = 12.8, Interquartile range = 2.8).

For each consented and enrolled stroke dyad, a matched, stroke-free control dyad was also recruited and enrolled into the CARES study. The control dyad consisted of a participant from the REGARDS study who had not experienced any strokes and their likely or available family caregiver if that person was to become disabled in the future. After a stroke dyad was enrolled, a list of up 15 stroke-free REGARDS participants was generated who matched the enrolled stroke survivor on age (± 5 years), sex, race, relationship with their potential caregiver, and whether they lived with that potential caregiver. For example, if an enrolled black female stroke survivor was being cared for by a daughter who lived in a separate residence, then a list of up to 15 stroke-free black female participants from the REGARDS database was generated, all of whom reported in their REGARDS baseline interviews that they would be cared for by a daughter who did not live with them if they were to experience a serious illness or disability in the future. Potential control dyads were randomly selected from the list of eligible matches and contacted until one eligible, matched, stroke-free control dyad agreed to participate in the CARES study for that enrolled stroke dyad. Potential control dyads were excluded if the REGARDS participant reported any history of stroke or if their potential family caregiver reported a history of providing extended caregiving assistance to that REGARDS participant or to any another person.

CARES baseline interviews.

After explaining the study, screening for eligibility, and obtaining verbal informed consent, trained research interviewers obtained data on demographic information, general physical health, and psychosocial functioning from all four groups of participants (stroke survivors, family caregivers, stroke-free controls, family member controls). Age was determined from the participant’s reported date of birth. Education was coded into four categories (less than high school, high school graduate, some college, and college graduate and beyond) based on self-report during the interview. Sex, race (black or white), and marital status (married or unmarried) were dichotomous variables based on self-report. Descriptive information on the nature of the relationship between dyad members (e.g., co-residing, spouse vs. non-spouse relationship) was also obtained.

As part of the CARES interview, all participants completed the full, 20-item Center for Epidemiological Studies-Depression Scale (CES-D).²³ Participants indicated the frequency of each symptom over the past week on a 0 (rarely) to 3 (most of the time) scale. Scores potentially range from 0 to 60, with higher scores indicating more depressive symptoms and scores of 16 or more considered to indicate risk of significant clinical depression.

In addition, caregivers completed additional instruments on the nature of care provided and their reactions to that care. Two measures were included in the present analyses based on our previous work on the number of problems of the stroke survivor as observed by the caregiver and the number of benefits the caregivers reported from their caregiving experiences.¹⁹ For the number of problems measure, the caregiver indicated if the stroke survivor displayed any of 28 possible problems with activities of daily living (ADLs), instrumental ADLs (IADLs), memory or cognitive problems, communication difficulties, and emotional or other behavioral problems. This measure has been previously shown to have good psychometric properties including high internal consistency and a strong correlation of −0.62 with the Barthel index of ADLs collected during the in-home visit.¹⁹ The measure of benefits or positive aspects of caregiving consisted of 11 yes/no questions about such possible benefits. This measure has also been previously validated.^{19, 24}

For stroke dyads, an additional in-home visit was conducted approximately 1 year (Median = 11.9 months) after the stroke event to collect additional information on long-term functional outcomes after stroke.⁸ The in-home sessions were conducted for 178 (87%) of the stroke survivors in the present analyses and in 126 of these cases, the sessions were also attended by the family caregiver. The stroke survivors and the family caregivers (if present) were both asked to display their current prescription medications, and these medications were recorded and later categorized off-line by trained research staff to determine if participants were taking antidepressant medications at the time of the in-home visit.

Statistical analyses

Initial multiple regression analyses were used to test differences between stroke and control dyads on the CES-D measure of depressive symptoms obtained approximately 9-months after the stroke event for the stroke dyads. Separate analyses were used to compare 1) stroke survivors with their matched stroke-free controls, and 2) family caregivers with the matched family member controls. Covariates included race, sex, age, and education category. Standardized effect sizes were calculated by taking the difference in covariate-adjusted means and dividing by the overall standard deviation. Additional multiple regression analyses were conducted to examine the predictors of depressive symptom level for stroke survivors and family caregivers separately. In addition to including the demographic variables (race, sex, age, and education category) as predictors, we also examined predictive effects for aspects of the caregiving relationship (co-residence, spouse vs. non-spouse relationship), for the number of functional problems of the stroke survivor as reported by the caregiver, and for the caregiver’s score on the positive aspects of caregiving measure.

Results

Participant Characteristics

Descriptive data for the stroke and control dyads are presented in Table 1. These data confirm the success of the case-by-case matching of the stroke survivors and stroke-free controls on age, sex, and race. In addition, stroke and control dyads were closely matched on relationship and on whether the two dyad members resided in the same residence. There were no differences between caregivers and family member controls on their age (P = 0.64) or education levels (P = 0.50). The 4-item depression screener from the original REGARDS interview indicated the exact same level of depressive symptoms at baseline between participants who later had first-time ischemic strokes and their matched controls who remained stroke-free (P = 1.00).

Table 1.

Descriptive Information for Ischemic Stroke Dyads and Matched Control Dyads

	Stroke Dyads		Control Dyads
	Stroke Survivors	Family Caregivers	Stroke-Free Controls	Family Member Controls
N	205	205	205	205
Age, Mean (SD)	75.8 (7.4)	62.3 (13.7)	74.6 (7.7)	61.7 (12.9)
Sex, N (%) female	102 (49.8)	158 (77.1)	102 (49.8)	158 (77.1)
Race, N (%) African American	77 (37.6)	76 (37.1)	77 (37.6)	78 (38.1)
Education, N (%)  < High School  High School Grad  Some College  College Grad	28 (13.7) 53 (25.9) 55 (26.8) 69 (33.7)	11 (5.4) 54 (26.3) 66 (32.2) 73 (35.6)	22 (10.7) 64 (31.2) 44 (21.5) 75 (36.6)	15 (7.3) 45 (22.0) 63 (30.7) 82 (40.0)
Marital Status, N (%) Married	121 (59.0)	148 (79.6)	124 (60.5)	153 (82.3)
Dyad Relationship, N (%)  Spouse/Partner  Adult Child  Other Family  Non-Family	100 (48.8) 79 (38.5) 19 (9.3) 7 (3.4)		100 (48.8) 81 (39.5) 17 (8.3) 7 (3.4)
Co-Residing, N (%) Yes	126 (61.5)		126 (61.5)
Pre-Stroke 4-item CES-D, M (SD)	0.97 (1.78)	---	0.97 (1.85)	---
9-months Post-Stroke Full CES-D, M (SD)	8.38 (8.90)	6.42 (8.88)	5.18 (6.94)	4.62 (7.04)
CESD ≥ 16, N (%)	39 (19.0)	24 (11.7)	15 (7.3)	12 (6.3)
Antidepressant Use 9-months Post-Stroke, N (%) Yes	(N = 178) 25 (14.0)	(N = 126) 20 (15.9)	---	---

Among those for whom antidepressant medication data were available, 25 of 178 (14%) stroke survivors and 20 of 126 (16%) caregivers were taking antidepressant medications approximately one year after the stroke event. There were no race or sex differences in antidepressant use rate among stroke survivors, but white caregivers were more likely than black caregivers to be taking antidepressants (22% vs. 5%, respectively, P = 0.013).

Stroke vs. Control Differences in Depressive Symptoms

The initial multiple regression analyses indicated that, after adjusting for race, sex, age, and education category, stroke survivors had higher covariate-adjusted CES-D scores than stroke-free controls (M’s = 8.91 and 5.73, respectively, P < 0.0001) and family caregivers had higher covariate-adjusted CES-D scores than family member controls ( M’s = 6.75 and 4.95, respectively, P = .02). These covariate-adjusted means are displayed in Figure 1. Standardized effect sizes indicated that depressive symptoms were 0.39 standard deviation units (SDUs) higher for stroke survivors than for their matched stroke-free controls and 0.22 SDUs higher for family caregivers compared to the family member controls.

Figure 1.

Adjusted means and standard errors of the mean for depressive symptoms for stroke dyads (survivors and family caregivers) and control dyads (stroke-free controls and family member controls).

Predictors of Depressive Symptoms

The analyses that examined the predictors of depressive symptoms within stroke survivor and family caregiver groups separately are summarized in Table 2. The R² statistics from these two analyses were similar; 0.31 for stroke survivors and 0.34 for caregivers. Race, stroke survivor sex, and family caregiver sex were not found to be significant predictors in either analysis after controlling for other covariates. Among the stroke survivors, younger age, older caregiver age, less than a high school education, and the total number of functional problems reported by the caregiver in the CARES interview (9 months after stroke) were all significant predictors. The total number of functional problems was also predictive of caregiver depression, but a different pattern of findings emerged for many of the other predictors when caregiver depressive symptoms were examined. Stroke survivor education and caregiver education were significant predictors of caregiver depression, with more symptoms of depression reported by caregivers if their stroke survivor had a college education, but fewer symptoms reported if the caregiver had achieved that level of education. In addition, spouse caregivers had significantly higher levels of depressive symptoms than non-spouse caregivers after controlling for other covariates, and caregivers who reported more positive aspects of caregiving also reported less depressive symptoms.

Table 2.

Predictors of Stroke Survivor and Caregiver Depressive Symptoms

	Stroke Survivor Analysis			Family Caregiver Analysis
Predictor	Estimate	Standard Error	P	Estimate	Standard Error	P
Intercept	7.231	2.934	0.0146	5.087	2.854	0.0763
Stroke Survivor Age (centered at 75)	−0.300	0.095	0.0018	0.177	0.092	0.0567
Caregiver Age (centered at 60)	0.148	0.066	0.0264	−0.071	0.064	0.2725
Stroke Survivor Sex (Female vs. Male)	1.698	1.771	0.3388	2.316	1.724	0.1805
Family Caregiver Sex (Female vs. Male)	0.227	1.558	0.8844	1.949	1.516	0.2001
Race (African American vs. White)	−1.615	1.299	0.2153	−0.722	1.264	0.5687
Stroke Survivor Education (less than high school vs. college graduate)	4.757	2.050	0.0214	−0.074	1.994	0.9705
Stroke Survivor Education (high school graduate vs. college graduate	−1.283	1.609	0.4261	−4.700	1.565	0.0030
Stroke Survivor Education (some college vs. college graduate)	−0.206	1.494	0.8903	−4.600	1.453	0.0018
Caregiver Education (less than high school vs. college graduate)	0.255	2.570	0.9212	4.065	2.500	0.1057
Caregiver Education (high school graduate vs. college graduate	−0.889	1.583	0.5750	3.356	1.540	0.0305
Caregiver Education (some college vs. college graduate)	1.816	1.455	0.2135	4.651	1.415	0.0012
Living situation (Co-residing vs. Separate residences)	−3.390	1.765	0.0563	−2.065	1.717	0.2308
Relationship (Spouse vs. Non-Spouse caregiver)	−2.248	2.669	0.4008	6.235	2.600	0.0173
Total Number of ADL, IADL, Memory, or Behavior Problems	0.603	0.100	<0.0001	0.620	0.097	<0.0001
Positive Aspects of Caregiving	−0.107	0.174	0.5412	−0.821	0.170	<0.0001

Additional descriptive analyses indicated that 39 (19.0%) of stroke survivors and 24 (11.7%) of family caregivers exceeded the CES-D threshold of being at risk for significant clinical depression (see Table 1). Stroke survivors also reported significantly more depressive symptoms than their family caregivers (P = 0.008). As we reported previously,²⁵ there was a moderate correlation between stroke survivor and family caregiver CES-D scores (r = 0.32, P < .0001). This association was remarkably consistent across dyad types as defined by co-residence status and spouse vs. non-spouse caregiving relationship (all rs ranged from 0.28 to 0.41).

Discussion

This study confirms the presence of significantly elevated symptoms of depression 9–10 months after first-time ischemic strokes for stroke survivors and for their family caregivers. Data from the REGARDS baseline CATI showed equivalence in pre-stroke depression levels between participants who subsequently experienced ischemic strokes and their individually-matched stroke-free controls. This reduces the possibility that the elevated depressive symptoms observed after first-time ischemic stroke events reflected longstanding or previous problems with depressive symptomatology. Although pre-stroke comparisons were not available for family caregivers and their family member controls, post-stroke comparisons indicated that stroke caregivers reported significantly more problems with depressive symptoms than their demographically-matched family member controls.

No race or sex differences were found in post-stroke depression for either stroke survivors or for family caregivers. Our findings on race are consistent with several previous studies and further document that blacks do not report more problems with post-stroke depression than whites.^{9, 10, 13} This is interesting in light of multiple studies that have found worse outcomes for blacks than whites on many other functional outcomes after stroke.^6–8 The finding of no race difference for caregivers on depressive symptoms is consistent with the work of Skolarus and colleagues,¹⁸ who found that black caregivers for stroke survivors enrolled in the National Health and Aging Trends Study reported more hours of care provision, but not more depressive symptoms, than white caregivers. Racial differences and disparities have complex sociocultural, socioeconomic, and environmental causes,²⁶ and these factors deserve greater attention in future research.

We found that 19% of stroke survivors and 12% of caregivers had CES-D scores suggestive of risk for clinically significant depression. This rate for the stroke survivors is generally consistent with the 17%³ to 25%² rates of long-term, significant depression estimated from previous studies. The 12% rate of significant depression risk for caregivers, however, is markedly below the 40% rate of significant depressive symptoms found for stroke caregivers in a meta-analysis¹⁷ and the 45% rate reported for a clinical convenience sample of stroke caregivers in Canada.¹⁶ Many of the studies included in the meta-analysis assessed caregiver depression closer in time to the stroke event, and it is often not clear what cut point the investigators used when classifying significant depressive symptomology across the multiple instruments used in the literature. However, the 45% prevalence rate reported in the Canadian study used the same CES-D measure that we employed and examined caregiver depressive symptoms more than one year, on average, after the stroke event. We have observed previously in the CARES study that the strokes sampled in our national population-based approach tend to be milder and less impairing, on average, than what is often reported from more clinical or convenience samples.^{19, 27} Population-based samples tend to find lower rates of depressive symptoms in caregivers than what is reported from clinical or convenience samples,²⁸ and may be less biased by possible selection factors. We believe, therefore, that the rate of significant, long-term depressive symptoms among the larger population of stroke caregivers is lower than what has been previously reported from registries and clinical samples, although the level of depressive symptoms in our caregiving sample remains elevated compared to matched non-caregiving controls.

Consistent with previous findings,^{15, 29} we found post-stroke levels of depressive symptoms to be moderately linked on a long-term basis between stroke survivors and their family caregivers. However, differences in the present analyses emerged when we examined predictors of post-stroke depression. Although the caregiver’s report of the number of functional problems of the stroke survivor was associated with caregiver and with stroke survivor depressive symptoms levels, caregiver depression was further predicted by the caregiving relationship (spouse caregivers with higher depression) and by the education levels of the members of the dyad.

Our study is unique in providing an examination of post-stroke depressive symptoms in stroke survivors and their family caregivers simultaneously from a population based sample. Limitations include possible selection biases from some of the exclusion criteria. All stroke survivors reported no history of previous stroke, were community-dwelling, and had a family caregiver who was also willing to participate in the project. Our findings, therefore, are only generalizable to first-time ischemic stroke survivors living in the community, and different findings would likely emerge if survivors of recurrent strokes or hemorrhagic strokes were included. The results may also not generalize to other stroke populations, such as stroke survivors without available family caregivers and those residing in nursing homes or other institutional settings.

Conclusions

The treatment and prevention of post-stroke depression is increasingly prioritized in standards of care for stroke,³⁰ but significant depression continues to be a common problem that may persist for months or even years. Persistent problems with depression are present not only for stroke survivors, but also for their family caregivers in many cases. However, rates suggestive of risk for clinically significant depression were lower in this population-based sample of caregivers compared to previously reported rates from convenience samples. Antidepressant medications represent one type of intervention, but our data suggest that these medications are not commonly used 1-year after first-time ischemic strokes by either stroke survivors or their family caregivers. Family-based interventions and caregiver supports might provide additional relief for depressive symptoms and other psychological problems after stroke, not only for stroke survivors, but also for the family members who typically care for them.