| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Department of Health Policy & Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland.
2 Department of Population and Family Health Sciences, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland.
Address correspondence to Jennifer L. Wolff, Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, 624 N. Broadway, Rm. 692, Baltimore, MD 21205. E-mail: jwolff{at}jhsph.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionREFERENCES |
|---|
Methods. Generalized estimating equations were used to generate population-average logistic regression models of prevalent depression, using a sample of 420 disabled community-dwelling women aged 65 or older receiving informal care obtained from the Women's Health and Aging Study Caregiving Survey.
Results. Findings confirm a substantial prevalence of depression among older women with disabilities and support the hypothesis that perceived reciprocity and respect afforded by one's primary caregiver as well as adequacy of instrumental support all were associated with a lower likelihood of being categorized as depressed, even after controlling for sociodemographic, health, and psychosocial characteristics that are known to be related to depression.
Discussion. Perceived quality of informal care arrangements has a bearing on the psychological health of care recipients. Individuals in more reciprocal relationships and in relationships where they felt respected and valued were less likely to be depressed than their counterparts.
THE need for assistance with day-to-day activities is an important concern for many older individuals, their families and loved ones, and society at large. The majority of task assistance is provided informally by family and friends (e.g., Arno, Levine, & Memmott, 1999
); however, the effect of such arrangements on the quality of life of recipients is not well understood. Whereas recipients tend to be highly satisfied with the care that they receive (Kasper, Shore, & Penninx, 2000; McCann & Evans, 2002), conflict has also been found to be common (Clark & Stephens, 1996; McCann & Evans, 2002; National Alliance for Caregiving and American Association of Retired Persons, 1997). A nascent literature suggests that conflicted relationships (Burgener & Twigg, 2002; Newsom & Schulz, 1998), unbalanced transfers (Dunkle, 1985; Pruchno, Burant, & Peters, 1997), and unhelpful assistance (Clark & Stephens, 1996) are associated with adverse psychological consequences for recipients. These studies collectively suggest that informal care arrangements have a bearing on recipients' mental health; however, they have relied on subgroups of caregiving dyads or diseases, have not considered more than one aspect of the arrangement, and have largely not explored the pathway by which such arrangements might exert their influence.
This study draws upon social support literature and theory to examine how the quality of informal care arrangements might influence depression. In particular, we draw upon findings that negative exchanges profoundly influence psychological well-being (Rook, 1984; Schuster, Kessler, & Aseltine, 1990). Negative interactions have been defined as unpleasant encounters characterized by criticism, rejection, competition, violation of privacy, lack of reciprocity, as well as ineffective or excessive helping (e.g., Krause & Rook, 2003). Whereas negative exchanges might be relatively uncommon among more distal network members, where few resources are transferred, they have been found to be fairly common when extensive support is provided (Clark & Stephens, 1996; Newsom & Schulz, 1998), such as in a caregiving context. This study examines recipient perceptions of three aspects of the quality of informal care arrangements—reciprocity, respect, and adequacy of instrumental support—with regard to prevalent depression among a group of moderate to severely disabled, community-dwelling older women receiving informal care.
Equity theory posits that individuals seek to maintain symmetry in interpersonal relations; asymmetrical exchanges, such as receiving help with activities of daily living (ADLs), may cause distress (Fisher, Nadler, & DePaulo, 1983). Relationships with greater reciprocity have been found to be associated with higher levels of satisfaction (Krause, 1995) and psychological well-being (Dunkle, 1985; Pruchno, Burant, & Peters, 1997; Wright & Aquilino, 1998). This study hypothesizes that recipients whose advice and opinions are not solicited by a primary caregiver will be at greater risk for depression owing to their inability to reciprocate support received.
Whereas equity theory holds that individuals prefer balanced relationships, there are conceivably situations where individuals might maintain relationships that do not conform to this equation, such as among family members or when individuals are physically ill and there are no other available sources of assistance. The main effect theory for social support posits that interpersonal resource transfers directly influence physical and mental health on a daily basis (Cohen, Gottlieb, & Underwood, 2000). As one example, frequent phone calls and home visits to parents who are experiencing a new health problem might remind individuals that they are loved and that their health is important to others, thereby providing them with a sense of purpose and an incentive to sustain their physical and mental well-being. To this end, perceived respect afforded to an individual by the primary caregiver captures the overall balance of favorable and unfavorable interactions flowing from the caregiving relationship. This study hypothesizes that recipients whose primary caregivers are perceived not to take their wishes into consideration or to be unconcerned with their happiness and well-being will be at greater risk of depression.
Although assistance that is desired and helpful has been linked to positive affect, some efforts to provide support may be unwanted, ineffective, or excessive. In fact, perceived adequacy of care has been found to be a more important determinant of mental health than the amount of support received (George, Blazer, Hughes, & Fowler, 1989; Oxman, Berkman, Kasl, Freeman, & Barrett, 1992), with inadequate support being associated with depressive symptomatology (Clark & Stephens, 1996; George et al., 1989; Oxman et al., 1992; Oxman & Hull, 1997). This study hypothesizes that individuals whose needs for task assistance are not satisfied will be at greater risk for depression.
Last, it was hypothesized that lack of respect from a primary caregiver would be a particularly potent predictor of depression among individuals without adequate emotional support from their larger social network. Numerous studies have found emotional support to be a powerful predictor of health (e.g., Berkman, Leo-Summers, & Horwitz, 1992; George et al., 1989; Oxman & Hull, 2001). In this study, it was hypothesized that overall emotional support might buffer ill effects related to suboptimal relationships with a primary caregiver.
Figure 1 presents a framework that illustrates the relationship among the various constructs used in this study. Care recipient characteristics (1) include characteristics such as age, socioeconomic status, health, and social support that are related to structural features of the care arrangement (2) as well as resulting psychological states. Perceived quality of the care arrangement (3) mediates the extent to which care recipient (1) and care arrangement (2) characteristics lead to health outcomes by influencing health-promoting behaviors and neuroendocrine responses (e.g., stress and allostatic load). In this project, we examine the salience of relationship quality through measures of reciprocity, respect, and adequacy of instrumental support.
|
) and that changes in satisfaction with social support tend to precede changes in depression (Krause & Liang, 1989). Therefore, we believe that the dominant causal pathway is one in which recipients' perceptions of care arrangements influence subsequent mental health outcomes. Nevertheless, owing to the complexity of the relationships under study, inferences will be made with careful consideration of alternative explanations. In summary, the relationship between social support and health is well established; however, researchers have generally evaluated social support broadly as a function of an individual's global social network. This study draws upon social support literature and theory to examine the influence of informal care arrangements, net of these more global factors. Specifically, this study provides a unique opportunity to simultaneously evaluate the relationship of care arrangement characteristics and depression among a study population of disabled older women receiving informal care, controlling for individual characteristics known to be related to depression.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionREFERENCES |
|---|
; Kasper, Shapiro, Guralnik, Bandeen-Roche, & Fried, 1999; Simonsick, Maffeo, Rogers, Skinner, Davis, Guralnik, et al., 1997).
The WHAS-CGS is a longitudinal study of caregiving arrangements conducted among a subset of WHAS participants who were receiving informal (unpaid) assistance from family or friends. Participants and their primary caregivers were both interviewed annually on a variety of subjects, including the composition of the caregiving arrangement as well as attitudes and preferences on caregiving. WHAS participants without a caregiver (23.9%) were excluded from the study, as were women with paid caregivers only (14.9%). Care recipient and caregiver interviews were completed for 84% of eligible cases at baseline, representing a sample size of 420 women. Baseline findings from the study have been published elsewhere (Kasper et al., 2000).
This study relies on a sample of 384 caregiving dyads where the recipient participated in both the WHAS and the WHAS-CGS; 36 individuals (8.6% of original WHAS-CGS sample of n = 420) were excluded from this study because they contributed fewer than two measurements of depression. Those who were excluded from these analyses tended to be older, less educated, in poor health status, and depressed compared with those that remained in the study, although differences were not statistically significant. Table 1 shows the composition of follow-up time for the study sample. Because of staggered enrollment into the WHAS, only a subset of 226 participants had the potential to participate in all six WHAS interviews; the minimum number of interviews for participants was two.
|
). This study adopted previously defined cut-points to categorize depression (Lyness et al., 1997; Norris, Gallagher, Wilson, & Winograd, 1987), with scores of 0–10 designating individuals as "not depressed" and scores of 11–30 designating individuals as "depressed." The outcome was missing for 196 of 1892 (10.4%) person–time observations.
Independent Variables
This study controls for a number of characteristics that are known to be related to depression. Time-invariant sociodemographic characteristics included baseline age in years, years of education, and race ("Black" versus "other").
Subjective health status was measured on a 5-point Likert scale ranging from "excellent" (1) to "poor" (5) and was modeled as a time-varying continuous covariate. Several ADLs (Katz, Ford, Moskowitz, Jackson, & Jaffe, 1963) and instrumental ADLs (IADLs) (Lawton & Brody, 1969) have been shown to measure the same underlying construct of functional disability (Fleishman, Spector, & Altman, 2002). In this study, these 10 ADL and IADL activities were combined to form a composite index of limitations in task functioning that was modeled as a time-varying continuous covariate. Baseline cognition was measured using the Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975). Women scoring below 18 (of 30) on the MMSE were excluded from participation in the WHAS. This study relied on previously defined cut-points (Anthony, LeResche, Niaz, Von Korff, & Folstein, 1982) in comparing study participants with MMSE scores of 18–23 with those scoring 24–30.
Time-varying psychosocial characteristics were "sense of personal mastery" and "perceived adequacy of emotional support." Women who both agreed with the statement "I can do just about anything I really set my mind to" and disagreed with the statement "I often feel helpless in dealing with the problems of life" were characterized as having a sense of personal mastery (Pearlin & Schooler, 1978). Perceived adequacy of emotional support was based on responses to "Could you have used more emotional support than you received in the last six months?"
Key Independent Variables
All care arrangement characteristics were treated as time-varying covariates. Duration refers to the length of time the primary caregiver occupied the caregiving role (<2 versus 2+ years). Relationship of primary caregiver to recipient integrates information about the caregiver's family ties to the respondent and also whether they live together ("spouse," "coresident child," "noncoresident child," or "others").
Reciprocity was captured by care recipient responses to "Does (caregiver) ask you for advice or talk over (his/her) problems with you?" (yes/no). Respect was based on care recipient's responses to two questions: "In helping you, does (caregiver) usually consider your wishes and opinions or just do what is easiest and quickest for (her/him)?" and "Would you say (caregiver) is concerned about your happiness and well-being?" ("very" versus "somewhat"/"not concerned"). Perceived adequacy of instrumental support was based on recipient's response to "Do you feel you get as much help day-to-day as you need?" ("most of the time" versus "some of the time"/"never").
Data Analysis
A data set was constructed with individual person–time observations that incorporated time-varying WHAS and WHAS-CGS variables. When time-varying independent covariates were missing, data were imputed by carrying previous values forward by no more than 1 year. Because the WHAS-CGS was administered annually and the WHAS was administered semiannually, data were carried forward up to two time periods for WHAS covariates and one time period for WHAS-CGS covariates. Item nonresponse among study covariates varied from 0.0% to 8.7% prior to imputation and decreased to <3.0% subsequent to imputation.
At the first stage of analysis, descriptive statistics were used to characterize the study population. Bivariate relationships between independent variables and depression were evaluated at baseline and over time using simple logistic regression. Because the effect of time on prevalent depression was relatively linear, time was modeled as a continuous covariate. Collinearity between covariates was examined with correlation matrices, the variance inflation factor, as well as close examination of change in covariate standard errors and regression coefficients in the model-fitting process.
This study uses generalized estimating equations (GEEs) to model the relationship between caregiving characteristics and prevalent depression. GEE is an ideal method for this type of analysis as it accounts for correlated error terms (Diggle, Liang, & Zeger, 1994; Stokes, Davis, & Koch, 2000). GEE generates population-based estimates by applying a working covariance structure across subjects, treating correlation as a nuisance parameter and assuming relative independence between subjects. An exchangeable correlation structure was selected for this study, although GEE is robust to misspecification of the correlation structure.
Because GEE assumes that missing outcomes are missing completely at random—a strong assumption—attrition analyses were conducted to understand the characteristics of the study population with censored data. Baseline characteristics of censored study participants were compared with those of individuals with complete data. A total of 282 of the 384 study participants (73%) completed all possible self-reported interviews. The 102 individuals with missing or proxy interviews were significantly more likely to be older, to be less well educated, to have task limitations, and to have lower MMSE scores but were no different in terms of depression or self-reported health status at baseline. Sensitivity analyses were conducted to evaluate results from regression models using all available person–time observations versus person–time observations for only those with complete self-reported data. Results for complete data analyses were of the same direction but generally slightly smaller in effect than analyses presented herein.
Bivariate regression models tested the relationship of individual covariates to depression. Two multivariate regression models are described that evaluate the influence of care arrangement characteristics on prevalent depression after controlling for key individual sociodemographic, health, and psychosocial factors that are known to be associated with depression. Multivariate models include a main effects model as well as a second model that includes key covariates along with several interaction terms. As score statistics provide conservative estimates of the strength of relationships of interest, score statistics rather than Wald tests were used in reporting individual covariate results from multiple regression analyses.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionREFERENCES |
|---|
2 years. Few care recipients reported receiving inadequate task assistance (16.7%), that their caregiver was not concerned with their happiness (12.8%), or that their caregiver did not consider their wishes and opinions (12.0%).
|
|
Of the variables related to perceived quality of support, lack of reciprocity was consistently related to a greater likelihood of depression; women whose caregiver did not ask them for their advice or opinions were 33–36% more likely to be depressed than those whose advice was solicited (p <.05). The effect of inadequate instrumental support on depression, which was statistically significant in the bivariate model, was much weaker and no longer statistically significant after controlling for individual-level variables. This suggests that the relationship between adequacy of instrumental support and depression might be confounded in part by other care recipient characteristics and factors related to the structure of the care arrangement.
Respect was also a significant predictor of care recipient well-being. Care recipients reporting their primary caregiver was not concerned with their happiness and well-being were approximately 70% more likely to be depressed than their counterparts (p <.05). Care recipients without adequate emotional support from their social network and who reported that their primary caregiver did not consider their wishes and opinions had a greater likelihood of being depressed than women who reported their caregiver did consider their wishes and opinions, although the effect diminished over time (p <.05).
Key covariate interactions of interest were examined. Most notably, the main effect of the variable "caregiver fails to consider wishes and opinions" changed from a modest but statistically insignificant greater likelihood of depression without a time interaction to more than double the likelihood of being depressed (p <.05) after taking time into account, but with its effect diminishing over time (p <.05). In addition to single covariate interactions with time, the hypotheses were tested that care recipients without adequate emotional support would be particularly at risk for depression if they perceived their primary caregiver not to be concerned with their happiness and well-being or not to consider their wishes and opinions. Whereas the interaction effect between emotional support and primary caregivers' being unconcerned with care recipient happiness was related to an increased risk of depression, it did not reach statistical significance or alter the direction of the main effects. However, results for the interaction between emotional support and primary caregiver's failure to consider care recipient's wishes and opinions yielded a significant interaction term (odds ratio = 1.75, p <.05), suggesting that the relationship between the care recipient and the primary caregiver may be important to emotional well-being, even among individuals with adequate emotional support.
Figure 2 shows the odds of depression for each group in the interaction between satisfaction with emotional support and primary caregiver's consideration of care recipient wishes and opinions. Women who were satisfied with emotional support from their overall social network but who reported their primary caregiver failed to take into consideration their wishes and opinions were 76% more likely to be depressed than their counterparts without adequate emotional support but who perceived their primary caregivers to consider their wishes and opinions (p <.05).
|
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionREFERENCES |
|---|
Three specific findings from this study merit particular attention. First, care recipients in less reciprocal arrangements were at a 33–36% greater risk of depression, even when other risk factors for depression were controlled. This finding is consistent with equity theory in that more balanced relationships may be associated with psychological well-being because they support feelings of self-worth and self-esteem. This study confirms prior reports of the importance of reciprocity among informal caregiving dyads (Douglass, 1997; Dunkle, 1985; Pruchno, Burant, & Peters, 1997) and warrants attention by social scientists interested in unraveling the mechanism by which social support influences health.
Second, care recipients who perceived that they were afforded lower levels of respect were at greater risk of depression. Care recipients who perceived their primary caregivers not to be concerned with their happiness and well-being were approximately 70% more likely to be depressed than their counterparts, a finding that held in both simple and multivariate models. The relationship with the second respect variable, "caregiver considers your wishes and opinions," was more complex. A statistically nonsignificant effect was observed at the bivariate level, but the introduction of two individual interaction terms ("time" and "adequacy of emotional support") substantially attenuated the main effect and yielded statistically significant interaction terms in both simple and multivariate models.
Women who perceived that their primary caregiver failed to consider their wishes and opinions were more likely to be depressed than their counterparts, and this relationship was stronger and statistically significant among women who were satisfied with the emotional support provided by their social network, as shown in Figure 2. Care recipients interact with a primary caregiver intensively, about very personal issues, and it may be that the level of respect afforded by one's primary caregiver is more influential for mental health than the adequacy of emotional support from one's overall social network. This interpretation is in keeping with research suggesting conflict among intimate social ties to be more influential for psychological well-being than more distant network members (Abbey, Abramis, & Caplan, 1985).
Third, although inadequate instrumental support was significantly related to depression in bivariate analyses, it was not an important predictor of depression after controlling for care recipient sociodemographic, health, and psychosocial characteristics. These results suggest that the relationship between perceived adequacy of instrumental support and depression may be confounded by need for assistance or other care recipient factors. Whereas this finding is at odds with what was anticipated, it is not entirely inconsistent with other studies that have found emotional support to be more strongly related to depression than instrumental support (George et al., 1989; Oxman et al., 1992). Because this sample of older disabled women all received some informal assistance, it may be that subjective measures of instrumental assistance are not as useful a measure as in more heterogeneous populations. Nonetheless, the finding that adequacy of instrumental assistance was not predictive of depression in the multivariate model is important as it suggests that the mere provision of assistance may be less important to care recipient mental health than the circumstances surrounding its provision.
There are several limitations of this study that must be mentioned. Of particular note is that whereas the outcome of interest, depression, is most certainly influenced by care arrangements, it is not a direct consequence of them. Moreover, although care arrangements exert a powerful influence over care recipient lives, it is impossible to fully separate the caregiving network from individuals' larger social networks. It is therefore not possible to definitively state that the relationships observed in this study are due primarily to care arrangement characteristics independent of other social circumstances. Moreover, as shown in Figure 1, perceptions of adequacy or reciprocity of arrangements are not necessarily isolated from psychological well-being. As discussed at the outset, we cannot make inferences regarding causality, however, we believe the dominant causal pathway is one in which recipients' perceptions of arrangements influence subsequent mental health outcomes.
A second major limitation of this study is that many of the measures have not been previously employed and therefore have not been subject to extensive tests of their psychometric properties. The single-item nature of most variables is not ideal in that they are not as reliable as could be expected with a well-designed multi-item scale. Although less than ideal, there were no other measures of the attributes of interest among a data set that is otherwise very rich. Moreover, examination of item change over time indicated that the covariates of interest remained stable or changed in a systematic way, which suggests general temporal stability. Should future researchers be interested in continuing this line of inquiry, additional studies to assess the psychometric properties of these and other relevant measures would strengthen the evidence regarding the relationship of caregiving arrangements for recipients of care.
Given the study sample used, it is important to be cognizant that these findings cannot be generalized to all populations that receive task assistance, such as men, dementia patients, or individuals who receive paid care only; these populations and their relationships with caregivers are likely to vary from our study sample in numerous ways. Although this is a limitation of our study, we believe the higher prevalence of depression among women and the fact that women disproportionately outlive their spouses (Federal Interagency Forum on Aging Related Statistics, 2000) make this sample of particular interest. Moreover, evidence from the social support literature suggests that women are commonly susceptible to stronger adverse psychological effects from social interactions, particularly with regard to nondiscretionary relationships such as with family (Kawachi & Berkman, 2001; Leffler, Krannich, & Gillespie, 1986; Schuster et al., 1990; Seeman, 2000) who typically are the providers of informal care (National Alliance for Caregiving and American Association of Retired Persons, 1997).
As in any longitudinal study, missing data constitute an important issue that can undermine the validity of findings. It is important to emphasize that complete self-reported data were available for nearly two thirds of study participants. Moreover, the effects of missing data were assessed by replicating all analyses for the subsample of 282 study participants with 1,381 self-reported person–time interviews. Although generally slightly smaller in effect, results for this subsample were in the same direction as for the full sample, lending further credibility to these findings.
There are a number of strengths associated with this study. First, the relatively large and homogeneous sample enhances the internal validity of these analyses. Whereas many studies evaluating the influence of social support have focused on extremes, the homogeneity of this study in terms of age, gender, geographic region of residence, need, and access to informal task assistance enhances the inferences that can be drawn related to the care arrangement characteristics. Moreover, the availability of a wide range of information for study participants allowed us to account for a variety of factors that are related to depression and thereby to minimize the potential for an omitted variable bias. Lastly, the availability of time-varying covariates minimizes misclassification biases that can arise in longitudinal studies that rely exclusively on time-invariant characteristics.
Results from this study suggest that informal care arrangements have a bearing on the psychological well-being of care recipients. Individuals in more reciprocal relationships and in relationships where they feel respected and valued are less likely to be depressed than their counterparts. This study has implications for researchers interested in understanding the influence of social ties for health as well as for long-term care policy makers seeking to improve home and community-based programs and support for informal caregivers. Better understanding of the relationship between informal care and mental health is particularly relevant given the high prevalence of depression and need for task assistance among the disabled community-dwelling older population and the adverse health effects and quality-of-life implications of depression.
|
Acknowledgments |
|---|
The authors thank Michelle Carlson, Kevin Frick, Thomas Glass, and Judith Kasper for their comments and advice.
|
Footnotes |
|---|
Received for publication June 16, 2003. Accepted for publication October 29, 2003.
|
REFERENCES |
|---|
|
TOPAbstractMethodsResultsDiscussionREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  E. Cahill, L. M. Lewis, F. K. Barg, and H. R. Bogner "You Don't Want to Burden Them": Older Adults' Views on Family Involvement in Care Journal of Family Nursing, August 1, 2009; 15(3): 295 - 317. [Abstract] [PDF]
|
|
|
|
 |
|
 J. R. Yee, S. A. Cavigelli, B. Delgado, and M. K. McClintock Reciprocal Affiliation Among Adolescent Rats During a Mild Group Stressor Predicts Mammary Tumors and Lifespan Psychosom Med, November 1, 2008; 70(9): 1050 - 1059. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
