Understanding the health effects of family caregiving is clearly an important area for research. Sparks and colleagues' paper addresses predictors of the mental and physical health in wives, husbands, and daughters caring for relatives with dementia. In our response to their study, we have highlighted some of the findings we found most interesting and have raised issues with respect to current research in caregiving.
Despite relatively large differences in the mean ages of the three groups included in the Sparks study-65 years for wives, 71 years for husbands, and 51 years for daughters-the three groups are remarkably similar in physical and mental health. Likewise, the three groups are similar in social position, years of caregiving, and the dementia severity of the care receiver. These similarities in health and other variables coexist with relatively large differences in how the three groups appraise caregiving: daughters (M = 17.4) and wives (M - 14.6) reported higher burden than did husbands (M = 6.6), and wives reported greater worry (M = 30.2) than did husbands (M= 22.3) and daughters (M = 22.7). These statistically significant differences correspond to effect sizes in the .60 to .80 range.
What do the three groups' similarities in health mean in the presence of moderate to large differences in cognitive appraisal? Is there a sample selection factor operating that affects these similarities and differences, or are they accurately portraying wives, husbands and daughters who are caregivers? We do not know the answer. Results of numerous studies of health and aging conclude that higher socioeconomic status, or SES (social position in the Sparks study) is a powerful predictor of health (Maddox, 1987). Thus the relatively high SES of the sample in the Sparks study may contribute to the similarities in health across groups. This restriction of range in SES may contribute to the lack of a significant association between social position and health in the correlation and regression analyses.
Although not stated as a purpose of the study, a question that seems to underlie it is: Does caregiving cause health problems in caregivers? Although quite a bit of research in the past has focused on this topic, results are conflicting. In their meta-analysis, Schulz, O'Brien, Bookwala, and Fleissner (1995) concluded that dementia caregivers had elevated levels of depressive symptoms compared to noncaregivers but that the evidence for physical health effects was less clear. Contrary to the conclusions by Schulz and associates, Sparks and colleagues found that mean scores of caregiving wives, husbands and daughters on the Brief Symptom Inventory were not significantly higher than normative data for elderly nonpatient adults. In another study of caregiver health that also used the Brief Symptom Inventory, Neundorfer (1991) found results similar to those of Sparks and associates; although depression and anxiety scores of dementia caregivers were slightly higher than age norms, they were not significantly different. If caregiving has deleterious effects on mental health of caregivers, it may be that the Brief Symptom Inventory is not sensitive enough to detect them.
One indirect approach to examining health effects of caregiving is to examine the correlation between years of caregiving and health. In the Sparks study, although there is no clear link between years spent in caregiving and mental health based on the simple correlations in Table 2 (r= -.12, .11, and .00 for wives, husbands, and daughters, respectively), it appears that years spent in caregiving may be associated with poorer physical health, as shown in Table 3. Using a one-tailed alpha for r, there is some suggestion that more years of caregiving is related to poorer physical health for daughters (r = .44, p < .01), and less so for wives (r = .24, p = .05). In the multiple regression analysis, the ability of years caregiving to predict the physical health of daughters remains, even after controlling for five other predictors including age. In the metaanalysis by Schulz and colleagues (1995), duration of caregiving was examined as a correlate of caregiver depression in three studies and was shown to be associated with more depression in one of three studies, but in the other two studies no relationship was found. In none of the studies of physical health in the meta-analysis was duration of caregiving studied. Sparks's finding of an association between years caregiving and physical health in daughters and wives may be an important one, especially taken together with the considerably higher burden experienced by daughters and wives compared to husbands.
Another of the intriguing findings in Sparks's study was that worry and burden were correlated with poorer mental and physical health in husbands (correlations ranged from .32 to .52) and daughters (correlations ranged from .32 to .58), but not in wives (correlations ranged from -.02 to .22). Moderate correlations between caregiver burden and health are not uncommon, so the lack of significant correlations for wives between the predictors of worry and burden and the outcomes of mental and physical health is baffling. A remaining question concerns inferences about causation, which cannot be made based on these cross-sectional results. Do increased worry and burden lead to poor health, or is it poor health that causes worry and burden? As Sparks and associates suggest, these findings argue for longitudinal studies of predictors of caregiver health.
Predictors included in this study explained only modest percents of variance in mental health of wives (23%) and husbands (20%). For daughters, the percent of explained variance in mental health (38%) was more robust. Percents of explained variance in physical health were 13% for wives, 24% for husbands, and 29% for daughters. What other variables might predict physical and mental health in spouses? If we were going to study predictors of caregiver health, one set of variables we would include is health behaviors. Does the caregiver get regular exercise, good nutrition, adequate rest, and relaxation? Does the caregiver avoid substances that lead to illness, such as cigarettes and too much alcohol? We would include preexisting health status. We would also look at the quality of the environment within which the caregiver lives and provides care. Based on our work on the concept of mutuality (Archbold, Stewart, Greenlick, & Harvath, 1990) and the work of others on relationship quality and health (Burman & Margolin, 1992; Coyne & Smith, 1991), we would hypothesize that a positive relationship between the caregiver and care receiver would be associated with better health of the caregiver. Schulz and colleagues (1995) suggest that the variables of caregiver self-esteem and mastery, contagion of affective symptoms between care receiver and caregiver, and anticipatory bereavement, may be important in explaining caregiver health outcomes. Sparks and colleagues, like most researchers studying dementia family caregiving, use a stress and coping framework based on the work of Lazarus and Folkman (1984). This framework was initially developed to study situation-specific coping, that is, coping with individual everyday problems in the life context within which they arise (e.g., coping with the flat tire you get on the way to work). The application of this framework to the more global and unbounded situation of caregiving over time to a family member with dementia may require more refinement than has yet occurred.
Sparks and associates have contributed to the growing body of literature on caregiving and its effects on health. Like most of the studies of caregiving and its health effects, this one is limited by its use of a nonrepresentative sample and its cross-sectional design. Clearly, a direction for research in this area is to examine samples that are more representative in terms of SES, race and culture. A full understanding of the complex interrelationships among caregiving and health will not be understood without longitudinal studies.
In conclusion, from a nursing perspective, we think it is now important to identify predictors of negative health effects of caregiving that are amenable to intervention (like Sparks's worry), to develop screening tools to identify caregivers who are at high risk for negative health effects, and to evaluate interventions to improve or maintain caregiver health that are targeted to predictors of health decline.
Archbold, P. G., Stewart. B. J., Greenlick, M. R., & Harvath, T. (1990). Mutuality and preparedness as predictors of caregiver role strain. Research in Nursing & Health, 73, 375-384.
Burman, B., & Margolin, G. (1992). Analysis of the association between marital relationships and health problems: An interactional perspective.p.Psychological Bulletin, 772(1), 39-63.
Coyne, J. C., & Smith, D. A. F. (1991). Couples coping with a myocardial infarction: A contextual perspective on wives' distress. Journal of Personality and Social Psychology, 67,404-412.
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer Publishing Co.
Maddox, G. L. (1987). Aging differently. The Gerontologist, 27, 557-564.
Neundorfer, M. M. (1991). Coping and health outcomes in spouse caregivers of persons with dementia. Nursing Research, 40, 260-265.
Schulz, R., O'Brien, A. T., Bookwala, J., & Fleissner, K. (1995). Psychiatric and physical morbidity effects of dementia caregiving: Prevalence, correlates, and causes. The Gerontologist, 35, 771-791.
Barbara J. Stewart, PhD
Patricia G. Archbold, RN, DNSc, FAAN
School of Nursing, Oregon Health Sciences University…