Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Gerontologist. Author manuscript; available in PMC 2009 October 1.
Published in final edited form as:
Gerontologist. 2008 October; 48(5): 573–583.
PMCID: PMC2586339

Declining patient functioning and caregiver burden/health: The Minnesota Stroke Survey-Quality of Life after Stroke Study

Melissa M. Nelson, BS,1 Maureen A. Smith, MD, MPH, PhD,2 Brian C. Martinson, PhD,3 Amy Kind, MD,4,5 and Russell V. Luepker, MD, MS6



Caregivers of stroke patients may adapt to changes in patient functioning over time. If adaptation occurs, caregiver burden and health may be influenced more by worsening in patient functioning than static levels of functioning. This study examines the relationship between patients' baseline and changes in functioning and caregivers' subjective and objective burden as well as their health.

Design and Methods:

Only stroke patients who had caregivers were included in this analysis (N=356). Stroke patients (N=281) or their proxies (N=75) were interviewed within 4 months of hospital discharge and patients' medical records were abstracted. The primary caregiver also was interviewed at approximately the same time as the patient/proxy interview (N=356). In all but one of the 75 proxy cases, the proxy was the patient's caregiver. Binomial and ordinal logistic regression models were used.


Declining patient neurological functioning predicted greater objective burden and subjective burden relating to consequences for caregivers' personal lives, but did not predict caregiver health.


The impact a patient's stroke has on a caregiver's personal life and the number of hours spent caring for the patient appear to be a function of the changes of the patient's status over time rather than a function of a “snapshot” of their functioning at baseline. If these results are confirmed, interventions to protect caregivers may be indicated for stroke patients who continue to decline after hospital discharge.

Keywords: Declining functioning, objective burden, subjective burden, sense of competence, proxies, social exchange


There are currently over one million stroke survivors in the United States requiring assistance with daily tasks (Centers for Disease Control [CDC], 2001). In addition to the primary impact of stroke, stroke survivors often suffer from a number of physical and psychosocial complications (Davenport, Dennis, Wellwood, & Warlow, 1996; Turner-Stokes & Hassan, 2002). Because of the severity of stroke and the prevalence of post-stroke complications, as well as the sheer number of stroke survivors requiring care, the health impact on and burden experienced by caregivers of stroke patients1 may be substantial. These caregivers may have multiple demands from other aspects of their lives or they may have their own health problems (Anderson, Linto, & Stewart-Wynne, 1995; Chumbler, Pienta, & Dwyer, 2004). Increasingly, there is concern that the level of burden experienced by these caregivers may further affect their own well-being and health (Hughes, Giobbie-Hurder, Weaver, Kubal, & Henderson, 1999). Understanding the factors that influence caregiver health and burden will help health care providers identify and develop interventions to target caregivers at risk for physical or mental health problems. We aim to improve understanding of these factors by using social exchange theory, a theory regarding reciprocity between members of a relationship, to examine how caregiver outcomes are influenced by changes in patient functioning.

Burden can be measured both subjectively, evaluating the strain or dissatisfaction experienced by a caregiver while providing care, or objectively, with a count of how many hours per week a caregiver spends providing care. Prior research on the effect of patients' strokes on caregiver burden and health has found mixed results, although these studies usually have focused on static assessments of patient functioning without reference to changes over time. Some studies have found worse patient physical functioning to be associated with higher levels of subjective caregiver burden (Blake & Lincoln, 2000; Choi-Kwon, Kim, Kwon, & Kim, 2005; Heuvel, de Witte, Schure, Sanderman, & Meyboom-de Jong, 2001). Oftentimes, however, patient health status does not predict subjective caregiver burden (Anderson et al., 1995; McCullagh, Brigstocke, Donaldson, & Kalra, 2005), or the relationship is inconsistent across different time points (Bugge, Alexander, & Hagen, 1999). The relationship between patient physical health and caregiver health is even less well established (Clark et al., 2004; McCullagh et al., 2005). Also, despite the fact that many stroke patients have significant neurological deficits in addition to physical functioning deficits, the relationships between patient neurological functioning and caregiver outcomes have not been explored extensively (Anderson et al., 1995; Bugge et al., 1999).

We propose that understanding the role of patient physical and neurological functioning in caregiver outcomes requires examining change in patient functioning over time in addition to baseline functioning. However, measures of change in patient health are rare in the literature, and studies with these measures show inconsistent results (Beach, Schulz, Yee, & Jackson, 2000; Tooth, McKenna, Barnett, Prescott, & Murphy, 2005). One study found that changes in patients' functional independence and physical health were associated with time spent caregiving, social burdens, and caregiver strain, but they were not associated with emotional, physical, or developmental burdens. Furthermore, these relationships were inconsistent over time (Tooth et al., 2005). Little is known about the relationship between change over time in patient neurological functioning and caregiver outcomes.

Theoretical Model

To examine the effects of change in patient functioning on caregiver outcomes, we used principles from social exchange theory. Social exchange theory focuses on relationships between members of a dyad or a larger network. According to this theory, members of a dyad are dependent on one another and act in ways that increase positive outcomes for themselves. Additionally, these positive outcomes have diminished marginal utility over time; that is, if one member of a relationship receives the same positive outcome many times and in close succession, the relative value of this outcome to the individual decreases (Molm & Cook, 1995). Exchanges between members of a dyad can be either emotional or material in nature (Mutran & Reitzes, 1984).

While much of social exchange theory focuses on the structural aspects of relationships (such as who has more power), the subjective components of a relationship, such as satisfaction and loneliness, are also important (Rook, 1987). Most studies that have examined caregiving burden relating to dissatisfaction or patient demands within a social exchange framework focus on measures of patient functioning at one point in time rather than measures of changing functioning (Call, Finch, Huck, & Kane, 1999; Raschick & Ingersoll-Dayton, 2004; Wright & Aquilino, 1998). Because the needs of patients and thus the caregiving situation change with time, measures of change in patient functioning, rather than cross-sectional measures, may be more appropriate to use in studies of caregiver outcomes. When evaluating a relationship, members of a dyad look for changes in the status of the relationship (Molm 1991). Since a stroke is a sudden event, it provides a reference point from which change in a relationship might be evaluated. This may be especially relevant for a family member who goes from being in a reciprocal relationship to one that is more asymmetrical, where he/she begins providing care to the other member of the relationship.

The objective of this study is to examine the relationship between declining health of stroke patients and subsequent caregiver burden and health (see Figure 1). For this study, primary caregivers identified by patients will be studied. We expect that the magnitude of decline in the patient's neurological or physical status will be associated with both increased subjective and objective caregiver burden as well as subsequent declines in caregiver health.

Figure 1
Conceptual model of baseline patient functioning, changes in patient functioning, and caregiver burden and health. All relationships are mediated by sociodemographic characteristics. Dashed lines indicate possible relationships not examined in this study. ...

Previous studies suggest that static measures of patient functioning immediately after a stroke predict subjective burden at least some of the time (Blake & Lincoln, 2000; Choi-Kwon, Kim, Kwon, & Kim, 2005; Heuvel, de Witte, Schure, Sanderman, & Meyboom-de Jong, 2001). We believe the relationship between change in functioning and subjective caregiver burden will be stronger than the relationship between baseline functioning and burden. While a caregiver may be accustomed to a certain amount of asymmetry in reciprocity, a large decline in reciprocation from a patient due to his/her declining function might cause the caregiver to be more dissatisfied with the relationship.

Worsening patient functioning also might increase asymmetry in physical demands of providing care, leading to increased objective burden (time spent providing care). Objective burden could be impacted by both static and changing levels of functioning, because a patient will still require some set amount of care whether or not his/her functional ability is declining.

Additionally, interaction with a loved one suffering from declining physical or neurological functioning might lead to a greater emotional impact. The changing physical and emotional demands in a relationship might impact caregiver health. Similar to the relationship between patient functioning and subjective burden, we believe the relationship between change in functioning and caregiver health will be stronger than the relationship between static patient functioning and health. Caregivers of patients with larger declines in functioning may have a harder time adjusting to patients' increasing needs. Because prior research has shown more significant results with caregiver burden than caregiver health, we expect that this relationship will be weaker than the relationships for subjective and objective burden (Blake & Lincoln, 2000; Heuvel et al., 2001).

We will test the following three hypotheses:

  1. Larger changes in patient physical and neurological functioning after stroke will predict higher subjective caregiver burden.
  2. Larger baseline deficits and larger changes in patient physical and neurological functioning after stroke will predict higher objective caregiver burden.
  3. Larger changes in patient physical and neurological functioning after stroke will predict decreased caregiver health.

Stroke patients often cannot respond on their own behalf, particularly if the stroke is severe. In this case, enlisting a family member to provide a proxy response allows the challenges faced by these more severely ill or communication-impaired patients to be studied. However, studying caregiving in situations where the caregiver also acts as the proxy respondent raises issues as the caregiver's assessment of the patient may be related to his/her sense of burden or health status in ways that differ from caregivers who do not act as proxy respondents. We will examine the effects of proxy status within the context of each hypothesis.

To address the hypotheses, we used data from a longitudinal study of stroke patients, the Quality of Life After Stroke Study (QLASS), a component of the Minnesota Stroke Survey (MSS). This study was approved by the Institutional Review Boards of the University of Wisconsin and the University of Minnesota, and informed consent was obtained from all participants.

Design & Methods


The MSS is a long-standing study of acute cerebrovascular disease (Shahar et al., 1995). The Minnesota Stroke Survey Quality of Life After Stroke Study (MSS-QLASS) includes community-dwelling and institutionalized members of a large HMO in the St. Paul/Minneapolis area between the ages of 30 and 84. All of these patients who were discharged alive from an acute care facility between August 2001 and August 2004 with a diagnosis of acute cerebrovascular disease, specifically with International Classification of Diseases, 9th Revision (ICD-9) discharge codes of 431, 434, 436 in the first position, or 434, 436 in the second position (N=776), met criteria for inclusion in the study. All patients who were not known to have died during the initial hospitalization were sent an introductory letter and contacted by telephone. Sixty percent of patients (or the proxies identified for patients) that were approached for interviews agreed to participate in the study.

Patients, proxy respondents for patients, and primary caregivers were included in this analysis. Proxy respondents were used for patients who had difficulties communicating or who had severe impairment. In all but one case, the proxy was the patient's caregiver (so the caregiver answered questions for the patient and for him/herself). Seventy-six percent of patients/proxies in the original study identified a primary caregiver, and only patients/proxies with caregivers were included in this analysis (N=356). Consequently, the sample for our analysis included 281 patients and 75 proxies who identified an informal caregiver and were interviewed within four months of discharge, along with their 356 caregivers. While the focus of this analysis is caregiver outcomes, patient interview data were used in order to gain an understanding of how patient functioning could affect caregiver health and burden.

Data Collection

Patient/Proxy Interviews

Trained interviewers conducted in-person and telephone surveys of patients or their proxies within four months of and at approximately 12 months after hospital discharge. Only the interviews conducted within four months (median number of days = 55) of hospital discharge are included in this study.

Caregiver Interviews

Trained interviewers also conducted in-person and telephone surveys of caregivers identified by patients or serving as proxies for patients. All caregivers completed interviews.

Hospital Chart Abstraction

Trained registered nurses abstracted information from charts for index hospitalizations for all patients in this analysis. To ensure inter-rater reliability, each month a random sample of 5% of charts completed by one auditor in the previous month was re-abstracted by another auditor. Difference reports between original audits and re-audits were reviewed on a monthly basis. We conducted inter-rater reliability assessments for key data elements in the abstraction, identifying moderate to high kappa values (0.6 to 0.8) for most data elements.

Clinic Chart Abstraction

Trained registered nurses also abstracted information from stroke patients' outpatient clinic charts for all visits (median number = 2) occurring within one year of the stroke discharge. (Only visits occurring within four months of discharge were used for this analysis.) To ensure inter-rater reliability, a random sample of 7% of cases was re-audited during the outpatient audit process. Fair to moderate kappa values (0.4 to 0.6) were found for functional and neurological status. Moderate to high kappa values (0.6 to 0.8) were found for assessments of the presence or absence of specific symptoms.


The dependent variables were caregiver subjective and objective burden as well as caregiver health. The explanatory variables of interest were baseline measures of patient functioning as well as worsening in patient physical and neurological functioning. Baseline refers to functioning pre-hospitalization. We controlled for other patient, caregiver, and sociostructural characteristics.

Caregiver burden

Subjective caregiver burden was ascertained by the Sense of Competence Questionnaire (SCQ) in the caregiver interview (Vernooij-Dassen, Persoon, & Felling, 1996). This questionnaire was originally developed for caregivers of dementia patients but is reliable and valid for caregivers of stroke patients as well (Scholte op Reimer, de Haan, Pijnenborg, Limburg, & Bos, 1998). It includes 27 Likert scale questions and three subscales: subjective burden relating to consequences for caregivers' personal lives, dissatisfaction with patient, and dissatisfaction with own performance as a caregiver. Caregivers were asked to rate their agreement with sentences such as: “I feel my [patient] makes requests which I perceive to be over and above what he/she needs”, “I feel pleased about my interactions with my [patient]”, and “I feel that my [patient] seems to expect me to take care of him/her as if I were the only one he/she could depend on” (Scholte op Reimer, de Haan, Pijnenborg, et al., 1998). Intra-class correlation coefficients (ICCs) of the SCQ and its subscales are high in caregivers of stroke patients (dissatisfaction with patient as recipient of care: ICC= 0.84, dissatisfaction with self: 0.89, subjective burden relating to consequences for caregivers' personal lives: 0.92, entire scale: 0.93). Scores on the total SCQ have good clinical validity and are significantly associated with patients' functioning, especially with scores on the Mini-Mental State Examination (p=.03) and the Rankin Scale (p<.01) (Scholte op Reimer, de Haan, Pijnenborg, Limburg, and van den Bos, 1998). For this study, the scores on each subscale were standardized to a standard normal distribution; higher scores indicate higher levels of burden.

We measured objective caregiver burden by the number of hours per week caregivers reported providing care. The categories were coded so that higher numbers indicate higher levels of burden (3 = 40 hours or more per week, 2 = 20-39 hours per week, 1 = 10-19 hours per week, and 0 = less than 10 hours per week).

Caregiver health

Caregivers were asked to rate their present health on a scale of 1 to 5 (1=excellent, 5=poor). Self-rated health has been found to predict mortality (Area under the receiver operated curve (AUC) / c-statistic: 0.74), hospitalization (AUC: 0.63), and outpatient service use (AUC: 0.61) as well as longer measures of health status such the SF-36 and Seattle Index of Comorbidity (DeSalvo, Fan, McDonell, & Fihn, 2005).

Patient physical functioning

We measured baseline physical functioning in the time period prior to stroke with the reverse-scored Barthel Index (range 0-100, with higher values indicating worse functioning). The Barthel Index (BI) measures activities of daily living (ADLs), such as the abilities to dress and bathe oneself (Mahoney & Barthel, 1965). It predicts patient functioning in the future and has high test-retest and inter-rater reliability (Kappa values = 0.87-0.99 and 0.75-0.99 respectively) (Cohen & Marino, 2000; Wolfe, Taub, Woodrow, & Burney, 1991). Patients reported their retrospective assessment of ADL difficulties prior to stroke and their assessment of current ADL difficulties at the time of the interview. Worsening in patient physical functioning was measured by subtracting the BI score from the time period prior to stroke from the BI score at the time of the interview.

Covinksy et al. (2000) found that retrospective reports of physical functioning are valid measures of patient health status prior to a hospitalization for an acute illness. Patients who were interviewed during a hospital admission and reported being independent in an ADL prior to admission were more likely to be independent in that ADL after recovery from the acute illness than those who were dependent prior to admission (p < .001 for each ADL). This indicates that retrospective measures of physical functioning provide a suitable baseline from which to measure change in functioning due to stroke. Also, medical records have been found to be unreliable sources of information about functional status (Bogardus et al., 2001), further justifying the use of patient recall to obtain this information.

We used PROC MI in SAS to impute Barthel scores for a small percentage of individuals (2.5% of Barthel Index scores). Because the Barthel scores at the time of hospital discharge were highly correlated with the Barthel scores at the time of the interview, we were unable to distinguish change in physical function from prior to stroke to hospital discharge and from hospital discharge to the time of the interview. As a result, we included a single variable in our analyses measuring change in physical function from prior to stroke to the time of the interview.

Patient neurological functioning

Baseline neurological functioning was measured by patient report of the occurrence of a previous stroke. We recognize that history of stroke is not an ideal measure of prior neurological functioning and that other conditions affect neurological status. However, studies suggest that 37%-62% of patients have long-term residual neurological deficits after stroke (see Duncan, 1994). Although we recognize the limitations, these studies suggest that inclusion of prior stroke as a proxy measure for baseline neurological deficit is supported.

Worsening in patient neurological functioning as a result of the most recent stroke was measured with both the Canadian Neurological Scale (CNS) and with a variable indicating a notation in the patient's medical record of worsening neurological functioning. The CNS uses data abstracted from hospital records and rates functioning in the following areas: consciousness (awake, drowsy, or coma), orientation, expression and receipt of speech, face weakness, and limb weakness and motor responses (Goldstein & Chilukuri, 1997). The score on the CNS ranges from 0 to 11, with higher values indicating worse functioning. In patients who have initially survived stroke, the CNS has been found to have high inter-rater reliability (r=.91) (Goldstein & Chilukuri, 1997) and to be predictive of death within six months (Cote et al., 1989; Goldstein & Chilukuri, 1997).

We constructed the second measure of worsening in neurological functioning through abstraction of clinic records. If worsening in neurological functioning was noted in clinic records at any point during the time period from hospital discharge to the last physician visit, this variable was coded as “1”, for worsening neurological functioning. If worsening was not noted in the clinic records, a value of 0 was assigned to this variable.

Other patient characteristics

The gender, age, race, marital status, and education level of patients were gathered from patient and proxy interviews.

The Charlson comorbidity index (range 0-9, with higher values indicating more comorbid conditions) was used to assess the number of comorbid conditions. It is a weighted sum of diagnoses (the diagnoses of cerebrovascular disease and hemiplegia were excluded). It has been found to have good test-retest reliability (intraclass correlation coefficient = 0.92) (Katz, Chang, Sangha, Fossel, & Bates, 1996). Also, it has high predictive validity for poor outcomes at discharge and mortality for stroke patients (De Groot, Beckerman, Lankhorst, & Bouter, 2003), with a one point increase in the scale corresponding with “a 15% increase in the odds of a poor outcome at discharge (p<0.005) and a 29% increase in the odds of death by 1 year (p<0.001)” (Goldstein, Samsa, Matchar, & Horner, 2004, p. 1943).

Caregiver characteristics

The gender, age, and educational level of caregivers were collected during the caregiver interviews.

Sociostructural characteristics

Sociostructural characteristics were recorded during interviews of patients, proxies, and caregivers. Patients were asked if they used any paid services at home (homemakers, nurses, or other services) during the time period beginning immediately after leaving the hospital. Caregivers also were asked if they lived with the patient at the time of hospital discharge.


Linear regression was used to examine the relationship between change in patient functioning and each of the standardized subscales of the SCQ (caregiver subjective burden). Standardized beta coefficients were calculated. We used ordinal logistic regression to examine the relationships between change in patient functioning and caregiver health and between change in patient functioning and caregiver objective burden. Odds ratios and 95% confidence intervals were calculated. All models included the following groups of variables: patient physical functioning, patient neurological functioning, other patient characteristics, caregiver characteristics, sociostructural characteristics, and proxy status. For adjustment purposes, we included the total number of physician visits and the total number of days between discharge and interview as continuous variables. Stata version 9 was used to perform the analyses.


Descriptive Statistics

Most (70%) caregivers were women, and their mean age was 59 (Table 1). The patients' mean age was 68 years. Also, 51% of the patients were male, and 68% were married. Most of the sample had acquired at least a high school education (89% of the caregivers and 85% of the patients). Prior to stroke, most patients had good physical functioning: the mean value of the Barthel Index was 8.4. Scores could range from 0 to 100, with higher levels indicating more difficulties with ADLs. Approximately a third of the patients (28%) had suffered a previous stroke. 156 patients had worsening physical functioning as a result of the stroke, 194 maintained their functioning, and 6 patients showed improvement after the stroke. No patients showed improved neurological functioning as a result of the stroke.

Table 1
Descriptive statistics for independent variables

Proxy status is moderately correlated with patient health. Pairwise correlations between proxy status and functioning variables are as follows: ADL functioning prior to stroke: 0.27, change in ADL functioning prior to stroke to 2 months post stroke: 0.58, occurrence of a previous stroke: 0.08, Canadian Neurological Scale: 0.40, and worsening neurological functioning measured by a physician: 0.17.

Caregiver reports of their sense of competence, objective burden and self-rated health are presented in Table 2. The SCQ scores were standardized to have a mean of zero and a standard deviation of one in the regression analyses. Twenty-seven percent of caregivers reported high levels of objective caregiver burden (more than 20 hours per week spent caring for the patient). Average caregiver self-rated health was between very good and good at the interview (mean score=2.4), although 51 caregivers rated their health as being fair or poor.

Table 2
Descriptive statistics for caregiver burden and health

Predictors of subjective caregiver burden

After adjustment, subjective burden was affected by worsening in patient neurological, but not physical, functioning (Table 3). Greater stroke severity, as measured by the CNS from hospital records, increased subjective burden relating to consequences for caregivers' personal lives (standardized β =0.149, p<.05). Caregivers who were proxy respondents for patients had greater subjective burden relating to consequences for their personal lives (standardized β = 0.183, p<.05). Neither of the other two types of subjective burden (burden relating to dissatisfaction with the patient as a recipient of care and burden concerning the caregiver's dissatisfaction with his or her performance as a caregiver) was associated with static measures of patient functioning or with measures of change in patient functioning.

Table 3
Regression results for subjective caregiver burden

Some of the other patient and caregiver characteristics influenced caregiver burden. Subjective burden relating to dissatisfaction with the patient as a recipient of care and dissatisfaction with self as caregiver increased with caregiver age. Patients with less than a high school education also were related to increased caregiver burden concerning dissatisfaction with the patient as a recipient of care. Sociostructural characteristics did not influence any of the subjective caregiver burden subscales.

Predictors of objective burden and caregiver health

After adjustment, objective caregiver burden increased with worsened patient neurological, but not physical, functioning. Greater stroke severity, as measured by the CNS from hospital records, increased objective burden (OR=1.14, 95% CI = [1.05, 1.30]). As was the case with subjective burden relating to consequences for caregivers' personal lives, objective burden was greater for caregivers who were proxies (OR = 2.92, 95% CI = [1.41, 6.04]). Neither baseline measures nor worsening patient physical and neurological functioning predicted changes in caregiver health.

Objective burden was lower when the patient was male, but it was higher when a patient had received paid assistance after being discharged from the hospital or made more visits to see a physician. Increased caregiver age and lower patient education were associated with decreased caregiver health. Caregiver health was better if a patient had received paid assistance after being discharged from the hospital.

Descriptive differences between proxies and non-proxies

Patients with proxy respondents were older and had more physical comorbidities than those who did not have proxies. We found that the impact of the stroke was much worse for patients with proxies: the mean increase in ADL difficulties from the time period prior to stroke to the interview was only 5.7 points for patients without proxies but 38.3 points for patients with proxies. Also, on the CNS, the mean score for patients without proxies was 4.9, and it was 7.1 for patients with proxies. Twenty-nine percent of patients without proxies and 48% of patients with proxies had an indication of decline in neurological functioning noted in their clinic records during the time between hospital discharge and the time of the interview.

Caregivers who acted as proxy respondents had more subjective and objective burden than those who did not act as proxies, but caregiver health did not differ between the two groups. Twenty-one percent of caregivers of patients without proxies and 49% of caregivers of patients with proxies reported high levels of objective caregiver burden (more than 20 hours per week spent caring for the patient).


As would be expected if caregivers are impacted by changes in reciprocity between themselves and patients, worsening in patient functioning from before to after a stroke significantly predicted subjective burden relating to consequences for caregivers' personal lives and objective burden, but baseline patient status did not. Caregiver health and subjective burden relating to dissatisfaction with the self as caregiver or with the patient, however, were not impacted by changes in patient functioning. Caregivers who were proxy respondents experienced greater subjective burden relating to consequences for caregivers' personal lives and greater objective burden than caregivers who were not proxy respondents.

Similar to other studies (Choi-Kwon et al., 2005; Scholte op Reimer, de Haan, Rijnders, Limburg, & Bos, 1998; Tooth et al., 2005), we found subjective burden resulting from consequences on the caregiver's personal life to be more salient than other dimensions of burden in caregivers of stroke patients. These other studies focused either on change in patient functional status (Tooth et al., 2005) or on baseline measures of patient disability (Choi-Kwon et al., 2005, Scholte op Reimer, de Haan, Rijnders, et al., 1998) but not both. Our study adds significantly to this nascent literature by including measures of both baseline and change in health and by incorporating neurological functioning in addition to physical functioning. We were able to show that change, rather than baseline functioning, affects subjective caregiver burden the most and that change in neurological functioning might be more relevant than changes in physical functioning.

Worsening in patient functioning, rather than baseline functioning, also predicts caregiver objective burden. In contrast to Tooth et al. (2005), we did not find that worsening in patient physical functioning predicted objective burden, but we did find that worsening neurological functioning predicted objective burden. These results were contrary to our expectations that objective burden would be predicted mainly by the patient's baseline health status. We believed that objective burden would be more stable than subjective burden in the face of worsening patient health status as caregivers might have an established amount of care to provide that corresponded with the consequences of the stroke, regardless of whether or not the caregiver was satisfied with the amount of asymmetry in the relationship. Instead, our results support the conclusion that caregivers who have not had a chance to adjust to increasing neurological deficits in the patient spend more time providing care. Our study is one of very few that have examined hours spent caring as an outcome measure.

Similar to other studies, we did not find a relationship between worsening patient functioning and caregiver health (Beach et al., 2000; Tooth et al., 2005). We did, however, find that worse caregiver health was predicted for caregivers who were proxies. Other studies of caregiver health that include baseline measures of patient functioning have either not found a relationship between patient baseline health and caregiver health (Clark et al., 2004), or they have found the relationship to occur inconsistently over time (McCullagh et al., 2005). To our knowledge, no other studies have simultaneously examined the effect of changes in patient physical and neurological functioning on caregiver health. Other authors have suggested that caregiver health is predicted more by caregiver burden than by changes in patient functional status (Hughes et al., 1999; McCullagh et al., 2005). Future studies will be needed to address the possible link between changes in patient functioning, caregiver burden, and caregiver health.

As with all studies, we have several limitations. First, we only analyzed data from interviews of caregivers and patients within four months of hospital discharge. We might find significant relationships between patient changes in functioning and caregiver health and burden if we study a longer time period or had a greater sample size. Secondly, our sample is not ethnically or racially diverse. It is, however, representative of the Minneapolis-St. Paul metro area. Data were obtained from a large HMO that enrolls approximately 10% of the area population. Also, for this analysis, we combined variables representing changes in physical functioning from “prior to stroke to discharge” and from “discharge to the interview” into one measure as these variables were highly correlated. This did not allow us to distinguish change between these two time periods, but the high correlation suggests that little change occurred. Finally, our R-squared values for overall model fit were low (see Tables Tables33 and and4),4), but this is often the case even with extremely good models (Schemper, 2003).

Table 4
Regression results for objective burdenb and caregiver health

Because we used data from a study that was initially designed to examine other outcomes, we lacked some information that would have been useful for our analyses. Specifically, we did not have explicit measures of reciprocity in the caregiver-patient relationship, and the burden measures reflect objective and subjective changes in the relationship from only the caregiver's point of view. Understanding the patient's emotional experience might improve our understanding of caregiver burden. We also did not have information on unpaid assistance received by caregivers. Unpaid assistance from other family members or friends may lessen the impact of changes in patient status on caregiver burden.

The magnitude of change, not patient's baseline status, influences a stroke's degree of impact on a caregiver's personal life and the number of hours spent caring for the patient. Our results suggest that a social exchange framework is useful for examining the effects of stroke patient health status on caregiver subjective and objective burden. Caregivers may not be adversely affected by patient characteristics and needs to which they have adapted, but unexpected or large changes in patient neurological functioning cause changes in the symmetry of a relationship and might lead to increased caregiver subjective and objective burden and perhaps even poorer health.

Our results also can be used to identify caregivers at risk for increased burden and poor health outcomes. In particular, caregivers of patients who are more severely ill, have communication difficulties, or have large declines in neurological functioning after stroke might be at risk for decreased health. If these results are confirmed, interventions to protect caregivers may be indicated for severely ill or communication-impaired stroke patients who continue to decline after hospital discharge. Currently, interventions to address the needs of caregivers have been disappointing and criticized as methodologically weak (Knapp, Young, House, & Forster, 2000). More complex interventions to assist carers have been proposed but await randomized controlled trials to evaluate effectiveness (Robinson et al., 2005). Specifically, assessment and targeting of these interventions to caregivers of certain stroke patients may be indicated. Assuming caregivers accompany patients to physician visits, physicians of stroke patients who have experienced significant declines in functioning also may be engaged and can refer caregivers to specific interventions, support groups, or other community resources.

Table 5
Differences between patients who did and did not have proxies
Table 6
Differences between caregivers who were and were not proxies


The authors gratefully acknowledge the contributions of Alexandra Wright and Jinn-ing Liou with their help in data preparation.


This work was supported by NIH/NINDS R01-NS39028-01A1. Melissa Nelson was supported by a grant (T32 HS-000036) from the Agency for Healthcare Research and Quality.


1While the term “care recipient” more accurately refers to people who are recovering from a stroke post-hospital discharge, we use “patient” to improve readability.


  • Anderson CS, Linto J, Stewart-Wynne EG. A population-based assessment of the impact and burden of caregiving for long-term stroke survivors. Stroke. 1995;26:843–849. [PubMed]
  • Beach SR, Schulz R, Yee JL, Jackson S. Negative and positive health effects of caring for a disabled spouse: Longitudinal findings from the Caregiver Health Effects Study. Psychology and Aging. 2000;15:259–271. [PubMed]
  • Blake H, Lincoln NB. Factors associated with strain in co-resident spouses of patients following stroke. Clinical Rehabilitation. 2000;14:307–314. [PubMed]
  • Bogardus S, Towle V, Williams C, Desai M, Inouye S. What does the medical record reveal about functional status? A comparison of medical record and interview data. Journal of General Internal Medicine. 2001;16:728–736. [PMC free article] [PubMed]
  • Bugge C, Alexander H, Hagen S. Stroke patients' informal caregivers: Patient, caregiver, and service factors that affect caregiver strain. Stroke. 1999;30:1517–1523. [PubMed]
  • Call K, Finch M, Huck S, Kane R. Caregiver burden from a social exchange perspective: Caring for older people after hospital discharge. Journal of Marriage and the Family. 1999;61:688–699.
  • Centers for Disease Control Prevalence of disabilities and associated health conditions among adults--United States, 1999. Centers for Disease Control Surveillance Summaries: Morbidity and Mortality Weekly Report. 2001;50:120–125. [PubMed]
  • Choi-Kwon S, Kim HS, Kwon SU, Kim JS. Factors affecting the burden on caregivers of stroke survivors in South Korea. Archives of Physical Medicine and Rehabilitation. 2005;86:1043–1048. [PubMed]
  • Chumbler NR, Pienta AM, Dwyer JW. The depressive symptomatology of parent care among the near elderly. Research on Aging. 2004;26:330–351.
  • Clark PC, Dunbar SB, Shields CG, Viswanathan B, Aycock DM, Wolf SL. Influence of stroke survivor characteristics and family conflict surrounding recovery on caregivers' mental and physical health. Nursing Research. 2004;53:406–413. [PubMed]
  • Cohen ME, Marino RJ. The tools of disability outcomes research functional status measures. Archives of Physical Medicine and Rehabilitation. 2000;81(Suppl 2):S21–S29. [PubMed]
  • Cote R, Battista RN, Wolfson C, Boucher J, Adam J, Hachinski V. The Canadian Neurological Scale: Validation and reliability assessment. Neurology. 1989;39:638–643. [PubMed]
  • Covinsky K, Palmer R, Counsell S, Pine Z, Walter L, Chren M. Functional status before hospitalization in acutely ill older adults: Validity and clinical importance of retrospective reports. Journal of the American Geriatrics Society. 2000;48:164–169. [PubMed]
  • Davenport RJ, Dennis MS, Wellwood I, Warlow CP. Complications after acute stroke. Stroke. 1996;27:415–420. [PubMed]
  • De Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity: A critical review of available methods. Journal of Clinical Epidemiology. 2003;56:221–229. [PubMed]
  • DeSalvo KB, Fan VS, McDonell MB, Fihn SD. Predicting mortality and healthcare utilization with a single question. Health Services Research. 2005;40:1234–1246. [PMC free article] [PubMed]
  • Duncan PW. Stroke disability. Physical Therapy. 1994;74:399–407. [PubMed]
  • Duncan PW, Lai SM, Tyler D, Perera S, Reker DM, Studenski S. Evaluation of proxy responses to the Stroke Impact Scale. Stroke. 2002;33:2593–2599. [PubMed]
  • Goldstein LB, Chilukuri V. Retrospective assessment of initial stroke severity with the Canadian Neurological Scale. Stroke. 1997;28:1181–1184. [PubMed]
  • Goldstein LB, Samsa GP, Matchar DB, Horner RD. Charlson Index comorbidity adjustment for ischemic stroke outcome studies. Stroke. 2004;35:1941–1945. [PubMed]
  • Heuvel, van den ET, de Witte LP, Schure LM, Sanderman R, Meyboom-de Jong B. Risk factors for burn-out in caregivers of stroke patients, and possibilities for intervention. Clinical Rehabilitation. 2001;15:669–677. [PubMed]
  • Hughes SL, Giobbie-Hurder A, Weaver FM, Kubal JD, Henderson W. Relationship between caregiver burden and health-related quality of life. Gerontologist. 1999;39:534–545. [PubMed]
  • Katz J, Chang L, Sangha O, Fossel A, Bates D. Can comorbidity be measured by questionnaire rather than medical record review? Medical Care. 1996;34:73–84. [PubMed]
  • Knapp P, Young J, House A, Forster A. Non-drug strategies to resolve psychosocial difficulties after stroke. Age and Ageing. 2000;29:23–30. [PubMed]
  • Mahoney FI, Barthel DW. Functional evaluation: The Barthel Index. Maryland State Medical Journal. 1965;14:61–65. [PubMed]
  • McCullagh E, Brigstocke G, Donaldson N, Kalra L. Determinants of caregiving burden and quality of life in caregivers of stroke patients. Stroke. 2005;36:2181–2186. [PubMed]
  • Molm LD, Cook KS. Social exchange and exchange networks. In: Cook KS, Fine GA, House JS, editors. Sociological perspectives on social psychology. Allyn and Bacon; Needham Heights, MA: 1995. pp. 209–235.
  • Mutran E, Reitzes DC. Intergenerational support activities and well-being among the elderly: A convergence of exchange and symbolic interaction perspectives. American Sociological Review. 1984;49:117–130.
  • Neumann PJ, Araki SS, Gutterman EM. The use of proxy respondents in studies of older adults: Lessons, challenges and opportunities. Journal of the American Geriatrics Society. 2000;48:1646–1654. [PubMed]
  • Raschick M, Ingersoll-Dayton B. The costs and rewards of caregiving among aging spouses and adult children. Family Relations. 2004;53:317–325.
  • Robinson L, Francis J, James P, Tindle N, Greenwell K, Rodgers H. Caring for carers of people with stroke: Developing a complex intervention following the Medical Research Council framework. Clinical Rehabilitation. 2005;19:560–571. [PubMed]
  • Rook K. Reciprocity of social exchange and social satisfaction among older women. Journal of Personality and Social Psychology. 1987;52:145–154.
  • Schemper M. Predictive accuracy and explained variation. Statistics in Medicine. 2003;22:2299–2308. [PubMed]
  • Scholte op Reimer W, de Haan R, Pijnenborg J, Limburg M, van den Bos, G. Assessment of burden in partners of stroke patients with the Sense of Competence Questionnaire. Stroke. 1998;29:373–379. [PubMed]
  • Scholte op Reimer W, de Haan R, Rijnders P, Limburg M, van den Bos, G. The burden of caregiving in partners of long-term stroke survivors. Stroke. 1998;29:1605–1611. [PubMed]
  • Shahar E, McGovern PG, Sprafka JM, Pankow JS, Doliszny KM, Luepker RV, et al. Improved survival of stroke patients during the 1980s: The Minnesota Stroke Survey. Stroke. 1995;26:1–6. [PubMed]
  • Tooth L, McKenna K, Barnett A, Prescott C, Murphy S. Caregiver burden, time spent caring and health status in the first 12 months following stroke. Brain Injury. 2005;19:963–974. [PubMed]
  • Turner-Stokes L, Hassan N. Depression after stroke: A review of the evidence base to inform the development of an integrated care pathway. Clinical Rehabilitation. 2002;16(Part 1 Diagnosis frequency and impact):231–247. [PubMed]
  • Vernooij-Dassen MJ, Persoon JM, Felling AJ. Predictors of sense of competence in caregivers of demented persons. Social Science & Medicine. 1996;43:41–49. [PubMed]
  • Wolfe CD, Taub NA, Woodrow EJ, Burney PG. Assessment of scales of disability and handicap for stroke patients. Stroke. 1991;22:1242–1244. [PubMed]
  • Wright D, Aquilino W. Influence of emotional support exchange in marriage on caregiving wives' burden and marital satisfaction. Family Relations. 1998;47:195–204.