- Email: [email protected]
Employment status, social ties, and caregivers’ mental health
ARTICLE IN PRESS
Social Science & Medicine 58 (2004) 1247–1256
Employment status, social ties, and caregivers’ mental health Carolyn C. Cannuscioa,*, Graham A. Colditzb,c, Eric B. Rimmb,d, Lisa F. Berkmane, Camara P. Jonesf, Ichiro Kawachie a
Merck Research Laboratories, Department of Epidemiology, Blue Bell, PA 19422, USA b Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA c Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women’s Hospital, Boston, MA, USA d Department of Nutrition, Harvard School of Public Health, Boston, MA, USA e Department of Health and Social Behavior, Harvard School of Public Health, Boston, MA, USA f Division of Adult and Community Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA
Abstract The purpose of this study of mid-life and older women was to assess the relation between informal care provision and depressive symptoms, taking into account concurrent demands on women’s time (including multiple caregiving roles and employment outside the home) as well as participants’ access to potentially supportive social ties. This cross-sectional study included women ages 46–71, free from major disease, who provided complete health and social information in the 1992 Nurses’ Health Study follow-up survey (n ¼ 61; 383). In logistic regression models predicting depressive symptoms, we examined the interaction between employment outside the home and informal care provision for a disabled or ill spouse or parent. We also investigated level of social ties, measured with the Berkman– Syme Social Network Index, as a potential modifier of the association between informal care provision and depressive symptoms. In all analyses, higher weekly time commitment to informal care for a spouse or parent was associated with increased risk of depressive symptoms. This relationship persisted whether women were not employed outside the home, were employed full-time, or were employed part-time. Higher weekly time commitment to informal care provision was associated with increased risk of depressive symptoms whether women were socially integrated or socially isolated. However, both informal care provision and social ties were potent independent correlates of depressive symptoms. Therefore, women who reported high spousal care time commitment and few social ties experienced a dramatic elevation in depressive symptoms, compared to women with no spousal care responsibilities and many social ties (OR for depressive symptoms=11.8; 95% CI 4.8, 28.9). We observed the same pattern among socially isolated women who cared for their parent(s) many hours per week, but the association was not as strong (OR for depressive symptoms=6.5; 95% CI 3.4, 12.7). In this cross-sectional study, employment status did not seem to confer additional mental health risk or benefit to informal caregivers, while access to extensive social ties was associated with more favorable caregiver health outcomes. r 2003 Elsevier Ltd. All rights reserved. Keywords: Caregivers; Mental health; Depressive symptoms; Social ties; Employment; USA
Introduction
*Corresponding author. Fax: +1-148-43442992. E-mail address: carolyn [email protected] (C.C. Cannuscio).
With the aging of the population, and with women’s increasing labor market participation, juggling of work and family responsibilities is increasingly common. More than 60% of all women are expected to provide
0277-9536/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0277-9536(03)00317-4
ARTICLE IN PRESS 1248
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
care to elderly relatives at some point in their lives (Abel, 1995), and approximately 30–40% of informal caregivers in the United States are also salaried workers (Stone, Cafferata, & Sangl, 1987; Brody & Schoonover, 1986). Researchers and the popular press have debated the potential health effects of women’s multiple social roles, with essentially two reigning perspectives. The first view is the ‘‘scarcity hypothesis,’’ which argues that multiple roles deplete women’s limited energy and resources, resulting in adverse health outcomes (Barnett & Baruch, 1987; Mui, 1992). The second concept is the ‘‘expansion hypothesis,’’ which asserts that benefits accrue to people who operate in multiple roles or domains. With multiple roles come opportunities for prestige, recognition, and financial reward, all of which can bolster women’s self-concept and wellbeing (Skaff & Pearlin, 1992). The scientific evidence regarding multiple roles and health is equivocal. In support of the scarcity hypothesis, Barnett and Baruch (1987) found that the number of roles a woman occupies predicts both role demand overload and role strain. Those authors also reported that motherhood is rarely associated with psychological wellbeing and frequently associated with distress, and that employed women sacrifice their own leisure activities in order to meet work and family demands. In support of the expansion hypothesis, Barnett, Marshall, and Singer (1992) reported that employed women generally exhibit better mental health than do non-employed women. Ozer (1995) also offered evidence that employed mothers are healthier than non-employed mothers, although this may be an artifact of selection bias or ‘‘the healthy worker effect.’’ There is some empirical evidence that employed elder caregivers experience lower levels of depressive symptoms than do non-employed caregivers (Rosenthal, Sulman, & Marshall, 1993). Barnett and Rivers (1996) suggest in She Works, He Works that families are thriving in an era of multiple work and family care demands. In contrast, Taylor, Ford, and Dunbar (1995) noted no significant difference between the health effects of caregiving among non-employed vs. employed women.
(EPESE) cohort, presence of a confidant, frequency of contact with family and friends, and church or other group membership all were related to lower levels of depressive symptoms. Oxman’s study also demonstrated that the following factors were associated with increased depressive symptoms: loss of a spouse or confidant; decline in the number of children seen weekly; and decline in emotional or tangible support available. Although these shifts in social ties may predict mental health status, social ties and support appear fairly consistent over time (Eurelings-Bontekoe, Diekstra, & Verschuur, 1995). There is some evidence that supportive social ties moderate the negative health effects (e.g., depressive symptoms and physical distress) of elder care (Rosenthal et al., 1993; Moritz, Kasl, & Berkman, 1995; Tausig, 1992; Lieberman & Fisher, 1995; Baumgarten et al., 1994; Gallagher, Wagenfeld, Baro, & Haepers, 1994). Cohen, Teresi, and Blum (1994) reported that 47% of a caregiver sample attended church or synagogue on a regular basis, and that religious attendance had the most potent direct and interactive effects on caregiver psychological wellbeing of any social network variable studied. Access to tangible support also may predict caregiver wellbeing. In Rosenthal’s et al. (1993) study of caregivers for long-stay patients, lower depressive symptom levels (measured by the Center for Epidemiologic Studies Depression Scale) were reported by caregivers who felt that someone else could assume their role if necessary. In the current study, we examined the association between informal care and depressive symptoms, exploring employment status and social ties as potential modifiers of that relationship. We hypothesized that employment outside the home would attenuate the association between informal care and depressive symptoms, in accordance with the expansion hypothesis. We also postulated that, compared to socially isolated caregivers, those with extensive social ties would be less likely to suffer depressive symptoms. Specific social ties may be particularly protective (e.g., large family size may represent access to tangible support and potential relief from informal care responsibilities).
Social ties An extensive literature documents the impact of social ties on health status. Social networks and support predict both short- and long-term mortality, especially from cardiovascular disease (see especially Berkman & Syme, 1979; House, Robbins, & Metzner, 1982). This finding is particularly strong and consistent among men (Berkman, Vaccarino, & Seeman, 1993). Research also points to a protective effect of supportive social ties on mental health status. In Oxman, Berkman, Kasl, Freeman, and Barrett’s (1992) research in the Established Population for the Epidemiologic Study of the Elderly
Methods Sample The current study is a cross-sectional analysis of women’s caregiving responsibilities and mental health, based on data from the 1992 Nurses’ Health Study (NHS) follow-up questionnaire. Participants enrolled in the NHS cohort were 30–55 years old at baseline in 1976, and 46–71 in 1992. Eligible women were registered nurses who were married and living in one of 11 states at
ARTICLE IN PRESS C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
the time of enrollment. A detailed description of the population is available elsewhere (Colditz, Manson, & Hankinson, 1997). The original NHS cohort included 121,700 nurses. After losses to death, follow-up, and non-response, 104,064 participants returned the 1992 survey on health and disease status. Of those women, 75,453 completed an additional section on social variables, which was included in the first mailing to the cohort but not in subsequent attempts to obtain updated information from study participants. Only those women who provided complete information on both health/disease status and social variables were included in the current analysis. We also excluded women who provided no information on caregiving responsibilities, who reported in 1992 (and in conflict with the cohort’s original inclusion criteria) that they had never been married, and who had been diagnosed previously with cancer, heart disease, or stroke. The final sample includes 61,383 women. Caregiving exposure measure Participants self-reported the number of hours per week they typically spent caring for various people. The question was framed as follows: ‘‘Outside of your employment, do you provide regular care for any of the following? (Mark one response on each line. For people to whom you do not provide regular care, mark ‘Zero hours’.)’’ The survey queries participants regarding care to a disabled or ill spouse, disabled or ill parent, disabled or ill other person, child, or grandchild. Since ‘‘care’’ is not defined specifically in the survey question, types of informal care assistance (e.g., with bathing, dressing, cooking, etc.) were not assessed. Mental health outcome We measured depressive symptoms using the five-item Mental Health Inventory (MHI-5), one of the eight health-related quality of life scales included in the Medical Outcomes Study 36-Item Short Form Health Survey (SF-36). The MHI-5 captures four major dimensions of mental health: anxiety, depression, loss of behavioral/emotional control, and psychological wellbeing (Ware, Snow, Kosinski, & Gandek, 1993; McHorney, Ware, & Raczek, 1993). The MHI-5 is scored from 0 to 100 and can be treated either as a continuous measure of mental health or as a dichotomous measure indicating the presence or absence of depressive/anxious symptoms. Higher-scoring participants enjoy better mental health, while those who score o52 are likely to satisfy the clinical diagnostic criteria for depression and related disorders. Using ROC analysis, Berwick et al. (1991) found the MHI to be a viable tool for the identification of clinical depression,
1249
with high area under the curve (AUC=0.892). However, since the MHI-5 is not a clinical diagnostic instrument, women in this low-scoring range will be said to exhibit depressive or anxious symptoms (abbreviated ‘‘depressive symptoms’’), rather than depression per se. Employment status Participants reported whether they were currently retired, working as homemakers, working full-time (in nursing or another field), or working part-time (in nursing or another field). In all analyses, women were classified as not employed outside the home; employed full-time outside the home; or employed part-time outside the home. Measure of social ties Social ties were measured using the Berkman–Syme Social Network Index (SNI). The SNI measures marital status, number of close friends, relatives, and children and frequency of contact with these people, religious service attendance, and participation in social groups. Contact with a confidant also was reported. Using the method described by Kawachi et al. (1996), subjects were classified into four levels of social ties, ranging from low ties (Level I), to medium (Level II), to medium-high (Level III), to high (Level IV). Participants in Level I have few intimate contacts (not married; fewer than six close friends or relatives) and no religious service attendance or social group membership. Employment status and informal care provision Indicator variables were created for combinations of informal care type (to a disabled or ill spouse or parent) and hours per week (0; 1–8; 9–20; 21–35; 36+) and employment status (not employed outside the home; employed part-time; employed full-time). We created indicator variables, rather than multiplicative interaction terms, to maintain the integrity of original response categories and to avoid ‘‘force fitting’’ interaction models. Using logistic regression, odds ratios were estimated for each combination of employment and informal care responsibility. The outcome was presence (MHI-5 score o52) or absence (MHI-5 score >52) of depressive symptoms. Multivariate models controlled for additional caregiving responsibilities; age; marital status; subject’s and husband’s level of education; chronic disease (diabetes, rheumatoid arthritis); living arrangement; current cigarette smoking; and ethnicity. We examined the effect estimates (ORs) within strata of employment status (not working outside the home; working full-time outside the home; working part-time outside the home) in order to evaluate the presence/ absence of effect modification by employment status.
ARTICLE IN PRESS 1250
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
Social ties and care to a disabled/ill spouse or parent
Results
We used similar methods to evaluate the interaction between informal care and level of social ties. To investigate potential interaction between spousal care and social ties, we limited the set to married participants (n ¼ 49; 810), who were the only women ‘‘eligible’’ to provide spousal care (the exposure of interest). Therefore, for this analysis, the marital status indicator was removed from calculation of the SNI. In logistic regression models predicting the relative odds of depressive symptoms, we included indicator variables for every combination of spousal care hours per week (0; 1–8; 9–20; 21–35; 36+) and social ties (low; medium; medium–high; high). The reference group, including women presumed to have the lowest occurrence of depressive symptoms, was composed of married women with no spousal care reported and high levels of social ties. To analyze the interaction between parent caregiving and social ties, the analysis set included both married and unmarried women, but included only participants ‘‘eligible’’ to provide parent care—those who had at least one living parent when this variable was last assessed on the 1988 survey. Therefore, we used the latest information available regarding parents’ vital status. However, women who became parentless during the 1988–1992 follow-up period would be most likely to be classified as providing zero hours of parent care, thereby introducing potential bias. The reference group included women who had extensive social ties, at least one living parent, and no parent care responsibility. As in the case of spousal caregiving, we created indicator variables for every combination of parent care responsibility and social ties and used logistic regression to estimate the association between each combination and depressive symptoms. Multivariate models controlled for the following variables: additional caregiving responsibilities; age; subject’s and husband’s education; living arrangement; chronic illness (rheumatoid arthritis, diabetes); current cigarette smoking; ethnicity; employment status; and access to a confidant. The parent care model also controlled for marital status and number of parents alive in 1988. After exploring potential interactions between informal care and a composite measure of social ties, we also examined the individual components of the SNI, since previous reports indicate that certain social ties (e.g., religious participation) may be particularly beneficial. We used methods similar to those described above to examine the association with depressive symptoms for each element of the SNI, including religious service attendance, social group participation, number of close friends and relatives, and presence of a confidant.
Characteristics of the study population Among women with no informal care responsibilities, 5.2% scored less than 52 on the MHI-5 (indicating presence of depressive symptoms), as did 11.0% of spousal caregivers and 6.7% of parent caregivers (Table 1). Spousal caregivers were older and parent caregivers were younger, on average, than women without informal care responsibilities. Both spousal and parent caregivers were more likely than noncaregivers to report high levels of social ties, suggesting that, in addition to the often discussed benefits of social ties, social integration also carried responsibilities or role demands. Employment status and informal care provision The association between informal care provision and depressive symptoms is similar among women who are not employed outside the home, who are employed parttime, and who are employed full-time (Table 2). Independent of employment status, there is a dose– response relation between higher time commitment to care provision and greater risk of depressive symptoms. The magnitude of the association is similar in all employment categories. For example, among women who do not work outside the home, high-level spousal caregivers (36+ h/week) are 3.5 times more likely than non-caregivers to experience depressive symptoms (multivariate OR=3.5; 95% CI 2.7, 4.5). Among women who are employed full-time, the association between high-level spousal care and depressive symptoms is equally strong (multivariate OR=3.4, 95% CI 2.2, 5.1). While high time commitment to spousal caregiving is associated with an approximate three-fold increase in depressive symptoms (controlling for employment status), the association between parent care and depressive symptoms is not as strong. For example, among women who do not work outside the home, those who provide 36+ h/week of parent care are 1.5 times more likely than non-caregivers to report depressive symptoms (multivariate OR=1.5; 95% CI 1.0, 2.1). We observed similarly moderate associations between parent care and depressive symptoms among women employed part-time (multivariate OR=1.5; 95% CI 0.9, 2.3) and full-time (multivariate OR=1.8; 95% CI 1.3, 2.6). The similarity of the association between caregiving and depressive symptoms across strata indicates limited effect modification by employment status. For women without informal care responsibilities, employment outside the home is associated with somewhat lower depressive symptoms. Among women who are not informal caregivers, those who are employed part-time are approximately 20% less likely than
ARTICLE IN PRESS C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
1251
Table 1 Characteristics of the Nurses’ Health Study participants in 1992a Characteristic
Women with no informal elder care roles
Women who care for a disabled/ill spouse
Women who care for a disabled/ill parent
Women who care for a disabled/ill other person
n Depressive symptoms (%) Age (%) o50 50–54 55–59 60–64 65–69 70+
46,141 5.2
3123 11.0
8617 6.7
5450 6.4
13.4 21.0 21.3 19.0 19.7 5.5
7.0 13.0 19.0 21.5 29.4 10.3
14.4 27.1 26.2 19.6 10.9 1.8
10.4 18.4 21.0 20.4 23.3 6.5
Marital status (%) Married Divorced Separated Widowed
81.3 7.5 1.1 10.0
96.3 1.2 0.6 1.7
84.6 5.7 0.7 8.7
79.0 8.4 0.8 11.5
Social ties (%) Low Medium Medium–high High Confidant (yes) (%)
5.6 27.3 19.6 47.5 93.7
4.2 20.2 21.7 53.9 92.6
3.8 21.4 19.3 55.6 94.7
3.0 22.3 14.3 60.5 95.5
a
Age-adjusted. Participants could report multiple caregiving roles.
Table 2 Combinations of informal care and employment outside the home: relative odds of exhibiting depressive symptoms (MHI-5o52), with 95% confidence intervalsa Caregiving responsibilities
Among women who do not work outside the home
Among women who work part-time outside the home
Among women who work full-time outside the home
n No care to a disabled spouse
22,963 21,449 (ref.) 607 2.0 (1.5, 253 2.4 (1.6, 132 1.7 (0.8, 522 3.5 (2.7,
14,891 14,223 0.8 (0.7, 342 1.0 (0.6, 117 2.1 (1.1, 57 1.6 (0.6, 152 3.1 (2.0,
22,804 21,858 0.8 (0.7, 506 1.4 (1.0, 172 1.6 (0.9, 101 2.6 (1.5, 167 3.4 (2.2,
Disabled spouse 1–8 h/week Disabled spouse 9–20 h/week Disabled spouse 21–35 h/week Disabled spouse 36+ h/week
No care to a disabled parent Disabled parent 1–8 h/week Disabled parent 9–20 h/week Disabled parent 21–35 h/week Disabled parent 36+ h/week a
20,368 (ref.) 1527 0.9 (0.7, 502 1.3 (0.9, 166 1.2 (0.6, 400 1.5 (1.0,
2.7) 3.7) 3.3) 4.5)
1.1) 1.8) 2.1) 2.1)
12,541 0.8 (0.7, 1619 0.9 (0.7, 388 1.1 (0.8, 134 1.3 (0.7, 209 1.5 (0.9,
0.9) 1.6) 3.7) 4.2) 5.0)
0.9) 1.1) 1.6) 2.4) 2.3)
19,236 0.8 (0.7, 2530 0.9 (0.7, 580 1.0 (0.7, 177 1.3 (0.8, 281 1.8 (1.3,
0.9) 1.9) 2.7) 4.6) 5.1)
0.8) 1.0) 1.4) 2.2) 2.6)
Models control for all other caregiving responsibilities, age, marital status, subject’s and husband’s level of education, living arrangement, chronic conditions (diabetes and rheumatoid arthritis), current cigarette smoking, and ethnicity. Within each stratum of employment status, the trend of increasing risk with increasing time commitment was statistically significant.
ARTICLE IN PRESS 1252
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
homemakers and retired women to experience depressive symptoms (multivariate OR=0.8; 95% CI 0.7, 0.9). A similar decrease in depressive symptoms was observed among non-caregivers when women employed full-time were compared to women not employed outside the home (multivariate OR=0.8; 95% CI 0.7, 0.9). Social ties and informal care provision We found a dose–response relation within each category of social ties, so that a higher number of weekly hours of spousal care provision is associated with higher risk of depressive symptoms (Table 3). However, at any given level of time commitment to caregiving, women with few social ties are more likely to exhibit depressive symptoms than are women with many social ties. Because of the strong independent associations with depressive symptoms for both informal care provision and social ties, women with high time commitment to informal care and few social ties represent an extremely high-risk group. For example, women with high spousal caregiving responsibility (36+ h/week) and few social ties experience a dramatic elevation in depressive symptoms, compared to women with no spousal care responsibilities and many social ties (multivariate
OR=11.8; 95% CI 4.8, 28.9). The same pattern holds in the case of parent care, although the magnitude of the association is not as strong as it is for spousal care. Again, the greatest elevation in depressive symptoms is observed among high-level parent caregivers with few social ties (multivariate OR=6.5; 95% CI 3.4, 12.7). Despite the marked elevation in depressive symptoms observed in these subgroups of caregivers, there is limited evidence of statistical interaction between spousal care provision and social ties. In other words, the incremental increase in depressive symptoms with increasing hours of care provision is equally strong among women with low, medium, medium–high, or high levels of social ties. Within each category of social ties, women who report 36+ h/week of spousal care are approximately three to four times more likely than women with no care responsibilities to experience depressive symptoms. In the case of parent care, high-level caregivers are approximately two times more likely than non-caregivers to report depressive symptoms, at most levels of social ties. The only exception appears to be among women with the most social ties, who do not experience a care-related increase in depressive symptoms. In this socially integrated group, even women who report
Table 3 Combinations of spousal/parent care and social ties in the Nurses Health Study: relative odds of exhibiting depressive symptoms (MHI-5 o52)a Caregiving responsibilities
Among women with high levels of social ties
Among women with medium–high levels of social ties
Among women with medium levels of social ties
Among women with low levels of social ties
n (married women) No care to a disabled spouse
27,901 26,312 (ref.) 756 1.1 (0.8, 1.7) 428 2.6 (1.8, 3.8) 405 4.1 (3.0, 5.7)
10,812 10,176 1.6 (1.4, 289 2.9 (1.9, 175 4.0 (2.5, 172 4.5 (2.8,
8343 7803 2.6 (2.4, 239 3.8 (2.5, 148 3.8 (2.3, 153 8.3 (5.5,
1389 1280 3.7 (3.1, 4.4) 53 6.2 (3.1, 12.6) 30 8.5 (3.5, 20.5) 26 11.8 (4.8, 28.9)
18,248 14,005 (ref.) 2870 1.0 (0.9, 1.3) 1004 1.3 (0.9, 1.7) 369 1.1 (0.7, 1.9)
7323 5784 1.5 (1.3, 1039 1.9 (1.5, 340 2.5 (1.7, 160 3.4 (2.1,
Disabled spouse 1–8 h/week Disabled spouse 9–35 h/week Disabled spouse 36+ h/week
n (women with X1 parent alive) No care to a disabled parent Disabled parent 1–8 h/week Disabled parent 9–35 h/week Disabled parent 36+ h/week a
1.8) 4.4) 6.4) 7.2)
1.8) 2.5) 3.6) 5.6)
9096 7425 2.4 (2.1, 1089 3.0 (2.4, 380 3.3 (2.4, 202 5.1 (3.4,
2.9) 5.6) 6.3) 12.5)
2.7) 3.8) 4.6) 7.5)
1845 1540 3.8 (3.2, 178 3.9 (2.5, 67 3.8 (1.9, 60 6.5 (3.4,
4.6) 6.1) 7.6) 12.7)
To assess interaction between spousal care and social ties, set is limited to married women; for parent care the set is limited to women who have at least one living parent. In both cases, models control for all other caregiving responsibilities, age, subject’s and husband’s education, living arrangement, chronic illness (rheumatoid arthritis, diabetes), current cigarette smoking, ethnicity, employment status, and access to a confidant. Parent care model also controls for marital status and number of parents alive in 1988. Within each stratum of social ties, the trend of increasing risk with increasing caregiving time commitment was statistically significant.
ARTICLE IN PRESS C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
36+ h/week of parent care are no more likely than noncaregivers to report depressive symptoms (multivariate OR=1.1; 95% CI 0.7, 1.9).
Elements of the SNI Specific SNI components were assessed and are summarized in Table 4. As we found in analysis of the composite index, we again observed that less socially connected women with high care responsibilities consistently experienced the most extreme elevation in depressive symptoms. Several types of social ties were found to alter the association between spousal caregiving and depressive symptoms. For instance, women who provide 36+ h/week of care to a disabled spouse and rarely/never attend religious services experience a dramatic elevation in depressive symptoms (multivariate OR=5.7; 95% CI 3.9, 8.3), compared to women who provide no spousal care and regularly attend religious services. The increase in depressive symptoms was less pronounced among women with high spousal care responsibilities and frequent religious service attendance (multivariate OR=3.7; 95% CI 2.9, 4.6). We observed particular elevations in depressive symptoms among women with high caregiving time commitment and: no regular participation in social group activities; five or fewer close family ties; five or fewer close friends; and no confidant. Again, we observed similar patterns, but weaker associations, among women who provided care to disabled or ill parents (data not shown).
1253
Discussion This study reports several key findings. First, there is a dose–response relation between higher informal care time commitment and increased risk of depressive symptoms, and the association is especially strong among spousal caregivers. Second, the association between informal care and depressive symptoms is not modified by employment status. The likelihood of exhibiting depressive symptoms is similar among caregivers who are not employed outside the home, who are employed part-time, and who are employed full-time. Third, there are strong independent associations with depressive symptoms for both informal care and level of social ties. Therefore, women with high time commitment to caregiving (especially for a spouse) and few social ties experience a marked elevation in risk of depressive symptoms. Strengths and limitations The primary limitation of this study is its crosssectional design, which thwarts our ability to infer causality in the associations between informal care, employment status, social ties, and depressive symptoms. The study is also limited by potential response bias: healthier women may have been more likely to return the 1992 follow-up questionnaire, including the social variables section analyzed in this paper. If this type of bias is operative, the observed associations between care provision and depressive symptoms may underestimate the true relationship between these factors.
Table 4 Spousal caregiving among women with and without specific social ties: relative odds of depressive symptoms (MHI-5o52)a
Religious service attendance Regular Infrequent/never Group social activities Regular None Number of close family ties Six or more Five or fewer Number of close friends Six or more Five or fewer Confidant Yes No a
Spousal care zero (h/week)
Spousal care 1–8 h/week
Spousal care 9–35 h/week
Spousal care 36+ h/week
(ref.) 1.4 (1.3, 1.6)
1.4 (1.1, 1.8) 3.4 (2.4, 4.8)
2.1 (1.6, 2.8) 4.4 (3.0, 6.5)
3.7 (2.9, 4.6) 5.7 (3.9, 8.3)
(ref.) 1.8 (1.7, 2.0)
1.6 (1.2, 2.2) 2.9 (2.2, 3.9)
2.7 (2.0, 3.7) 3.8 (2.7, 5.3)
4.6 (3.5, 6.0) 5.3 (4.0, 7.1)
(ref.) 2.2 (2.1, 2.4)
1.4 (0.9, 2.2) 3.7 (2.9, 4.7)
2.6 (1.7, 4.1) 5.0 (3.8, 6.6)
4.4 (3.1, 6.3) 7.8 (6.0, 10.0)
(ref.) 1.9 (1.7, 2.0)
1.4 (0.9, 2.1) 3.2 (2.5, 4.0)
2.9 (2.0, 4.4) 4.0 (3.0, 5.4)
4.5 (3.2, 6.3) 6.4 (5.0, 8.2)
(ref.) 3.0 (2.7, 3.3)
1.7 (1.4, 2.1) 4.1 (2.4, 6.9)
2.6 (2.0, 3.3) 3.5 (1.7, 7.5)
4.0 (3.3, 5.0) 7.0 (3.7, 13.4)
Models control for all other caregiving responsibilities (care for a disabled or ill parent; disabled or ill other person; child, or grandchild), age, marital status, subject’s and husband’s level of education, living arrangement, chronic conditions (diabetes; rheumatoid arthritis), current cigarette smoking, and ethnicity.
ARTICLE IN PRESS 1254
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
Additionally, the study relied on self-report of caregiving time commitment and depressive symptoms. Participants with depressive symptoms may have differentially perceived and reported their informal care time commitments, thereby inflating the observed association between caregiving provision and depressive symptoms. Finally, the generalizability of this study may be limited because of the population sampled. All participants in the NHS were married registered nurses at the start of the study in 1976, and had, therefore, selfselected into a caregiving profession. However, the study has several strengths, including the large sample size, which allowed for examination of a range of caregiving roles and time commitments. The sample size also allowed us to examine, with power, the effects of care provision among women with different employment circumstances and type/number of social ties, while controlling for several social and health status variables.
adequate appraisal of why some informal care providers remain healthy and others experience depressive symptoms. In support of the expansion hypothesis, we observed that among women with no caregiving responsibilities, those who worked part-time or full-time outside the home were less likely than homemakers and retired women to report depressive symptoms (Table 2). In this situation, being engaged in employment outside the home—an expansion of social roles—is associated with lower risk of depressive symptoms. One caveat is that this observation may be an artifact of selection bias, with healthier women more likely than those with depressive symptoms to enter or remain in the paid labor market. The temporal association between employment status and depressive symptoms cannot be assessed in this cross-sectional study.
Employment status: scarcity vs. expansion
In this study, we observed that socially isolated caregivers, and especially spousal caregivers, experience a marked increase in depressive symptoms. However, as noted for employment status, above, we also cannot determine the temporal association between social ties and depressive symptoms. Social isolation could be either an antecedent of, or a reaction to, depressive symptoms. A further limitation of this assessment is that health status may confound the association between social ties and depressive symptoms. For instance, women who are physically ill or disabled may be less likely to engage in social activities or to attend religious services. This would result in lower SNI scores, and also could lead to depressive symptoms. We attempted to correct for this potential bias by excluding from analysis women with cancer, heart disease, and stroke, and by controlling for other chronic conditions (rheumatoid arthritis and diabetes). We observed strong independent associations between individual components of the SNI and depressive symptoms. The following factors were associated with increased risk: infrequent religious service attendance; lack of participation in social groups; having five or fewer close friends; having five or fewer close relatives; and not having a confidant. Each of these individual types of social ties may be a source of practical, material, or emotional support for caregivers (i.e., a resource for assistance with care-related tasks or for coping with grief or trauma surrounding family illness). In addition, we noted that caregivers in this sample were more likely than non-caregivers to report high levels of social ties, which implies that social integration serves to increase both social resources and responsibilities like care provision.
Conflicting results have been published in previous research on the interaction between caregiving and employment status. Some studies have supported the scarcity hypothesis, which holds that multiple roles present numerous stresses, thereby increasing risk of illhealth. Other studies have supported the expansion hypothesis, which states that multiple roles should confer extra benefits and protection against mental or physical illness. The current study suggests certain applications for—and shortcomings of—both the scarcity and the expansion hypotheses. For instance, we observed that, regardless of employment status, women with high caregiving time commitment reported heightened depressive symptoms. This observation may be consistent with the scarcity hypothesis, if the elevation in depressive symptoms results from carerelated demands on women’s limited time and energy. However, alternative explanations must be explored. For example, the care-related increase in depressive symptoms may result not from task-related burden, but from grief surrounding a loved one’s illness, or from loss of relational reciprocity, for example. One important observation from the current study is that spousal care is a much stronger predictor of depressive symptoms than is parent care, although similar tasks should be involved in both roles. This finding suggests that certain unmeasured characteristics of the caregiving relationship (e.g., intimacy, conflict, obligation, reciprocity, life stage) may underlie the observed association between caregiving time commitment and depressive symptoms. In other words, a scarcity model that examines only time commitment and number of roles may not elicit an
Depressive symptoms and social ties
ARTICLE IN PRESS C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
Conclusion We offer evidence that informal care provision is associated with increased depressive symptoms, especially among spousal caregivers. While employment status does not appear to alter the strength of that association, level of social ties does appear to be a strong predictor of caregivers’ mental health. Upon confirmation of this finding in longitudinal research, health care providers, employers, social service agencies, and family members should establish mechanisms for maintaining or increasing caregivers’ social ties. Socially isolated caregivers may be at greatest risk for depressive symptoms, which could jeopardize their own health and the wellbeing of care recipients. Provision of support through family, friends, social group participation, or religious service attendance may mitigate the depressive effects of caring for a disabled or ill family member.
Acknowledgements We thank the Nurses’ Health Study participants and staff for their commitment to advancing women’s health research. In addition to NHS grant CA 40356, the primary source of support for this study, the Nurses’ Health Study is supported for other specific projects by the following NIH grants: CA46475, AG12806, CA55075, CA67883, AG13842, CA65725, CA70817, DK46519, EY09611, DK45362, HL03535, HL34594, ES05947, CA75016, CA62252, CA66385, ES05947, CA62005, DK52866, CA08283, HL57871, AG15424, AR02074, CA80620. This work was conducted while C.C. Cannuscio was a doctoral candidate at the Harvard School of Public Health. Merck & Co., Inc. did not fund the research and will not benefit directly from the results reported.
References Abel, E. K. (1995). A terrible and exhausting struggle: Family caregiving during the transformation of medicine. Journal of the History of Medicine and Allied Sciences, 5, 478–506. Barnett, R. C., & Baruch, G. K. (1987). Social roles, gender, and psychological distress. In R. C. Barnett, L. Biener, & G. K. Baruch (Eds.), Gender and stress (pp. 122–143). New York: Free Press. Barnett, R. C., Marshall, N., & Singer, J. (1992). Job experiences over time, multiple roles, and women’s mental health: A longitudinal study. Journal of Personality and Social Psychology, 62, 634–644. Barnett, R. C., & Rivers, C. (1996). She works/he works: How two-income families are happier, healthier, and better-off. San Francisco: Harper San Francisco.
1255
Baumgarten, M., Hanley, J., Infante-Rivard, C., Battista, R. N., Becker, R., & Gauthier, S. (1994). Health of family members caring for elderly persons with dementia: A longitudinal study. Annals of Internal Medicine, 120, 126–132. Berkman, L. F., & Syme, S. L. (1979). Social networks, host resistance and mortality: A nine-year follow-up study of Alameda County residents. American Journal of Epidemiology, 109, 186–203. Berkman, L. F., Vaccarino, V., & Seeman, T. (1993). Gender differences in cardiovascular morbidity and mortality: The contributions of social networks and support. Annals of Behavioral Medicine, 15, 112–118. Berwick, D. M., Murphy, J. M., Goldman, P. A., Ware, J. E., Barsky, A. J., & Weinstein, M. C. (1991). Performance of a five-item mental health screening test. Medical Care, 29, 169–176. Brody, E. M., & Schoonover, C. B. (1986). Patterns of parentcare when adult daughters work and when they do not. Gerontologist, 26(4), 372–381. Cohen, C., Teresi, J., & Blum, C. (1994). The role of caregiver social networks in Alzheimer’s disease. Social Science & Medicine, 38, 1483–1490. Colditz, G. A., Manson, J. E., & Hankinson, S. E. (1997). The Nurses’ Health Study: A 20-year contribution to the understanding of health among women. Journal of Women’s Health, 6, 49–62. Eurelings-Bontekoe, E. H. M., Diekstra, R. F., & Verschuur, M. (1995). Psychological distress, social support, and social support seeking: A prospective study among primary mental health care patients. Social Science & Medicine, 40, 1083–1089. Gallagher, T. J., Wagenfeld, M. O., Baro, F., & Haepers, K. (1994). Sense of coherence, coping and caregiver role overload. Social Science & Medicine, 39, 1615–1622. House, J. S., Robbins, C., & Metzner, H. L. (1982). The association of social relationships and activities with mortality: Prospective evidence from the Tecumseh County Health Study. American Journal of Epidemiology, 116, 123–140. Kawachi, I., Colditz, G. A., Ascherio, A., Rimm, E. B., Giovannucci, E., Stampfer, M. J., & Willett, W. C. (1996). A prospective study of social networks in relation to total mortality and cardiovascular disease in men in the USA. Journal of Epidemiology and Community Health, 50, 245–251. Lieberman, M. A., & Fisher, L. (1995). The impact of chronic illness on the health and wellbeing of family members. The Gerontologist, 35, 94–102. McHorney, C. A., Ware, J. E., & Raczek, A. E. (1993). The MOS 36-Item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Medical Care, 31, 247–263. Moritz, D. J., Kasl, S. V., & Berkman, L. F. (1995). Cognitive functioning and the incidence of limitations in activities of daily living in an elderly community sample. American Journal of Epidemiology, 141, 41–49. Mui, A. C. (1992). Caregiver strain among Black and White daughter caregivers: A role theory perspective. The Gerontologist, 32, 203–212.
ARTICLE IN PRESS 1256
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
Oxman, T. E., Berkman, L. F., Kasl, S., Freeman, D. H., & Barrett, J. (1992). Social support and depressive symptoms in the elderly. American Journal of Epidemiology, 135, 356–368. Ozer, E. M. (1995). The impact of child care responsibility and self-efficacy on the psychological health of professional working mothers. Psychology of Women Quarterly, 19, 315–335. Rosenthal, C. J., Sulman, J., & Marshall, V. W. (1993). Depressive symptoms in family caregivers of long-stay patients. The Gerontologist, 33, 249–257. Skaff, M. M., & Pearlin, L. I. (1992). Caregiving: Role engulfment and the loss of self. The Gerontologist, 32, 656–664.
Stone, R., Cafferata, G. L., & Sangl, J. (1987). Caregivers of the frail elderly: A national profile. Gerontologist, 27(5), 616–626. Tausig, M. (1992). Caregiver network structure, support, and caregiver distress. American Journal of Community Psychology, 20, 81–96. Taylor, R., Ford, G., & Dunbar, M. (1995). The effects of caring on health: A community-based longitudinal study. Social Science & Medicine, 41, 1407–1415. Ware, J. E., Snow, K. K., Kosinski, M., & Gandek, B. (1993). SF-36 health survey: Manual interpretation and guide. Boston, MA: Nimrod Press.
Social Science & Medicine 58 (2004) 1247–1256
Employment status, social ties, and caregivers’ mental health Carolyn C. Cannuscioa,*, Graham A. Colditzb,c, Eric B. Rimmb,d, Lisa F. Berkmane, Camara P. Jonesf, Ichiro Kawachie a
Merck Research Laboratories, Department of Epidemiology, Blue Bell, PA 19422, USA b Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA c Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women’s Hospital, Boston, MA, USA d Department of Nutrition, Harvard School of Public Health, Boston, MA, USA e Department of Health and Social Behavior, Harvard School of Public Health, Boston, MA, USA f Division of Adult and Community Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA
Abstract The purpose of this study of mid-life and older women was to assess the relation between informal care provision and depressive symptoms, taking into account concurrent demands on women’s time (including multiple caregiving roles and employment outside the home) as well as participants’ access to potentially supportive social ties. This cross-sectional study included women ages 46–71, free from major disease, who provided complete health and social information in the 1992 Nurses’ Health Study follow-up survey (n ¼ 61; 383). In logistic regression models predicting depressive symptoms, we examined the interaction between employment outside the home and informal care provision for a disabled or ill spouse or parent. We also investigated level of social ties, measured with the Berkman– Syme Social Network Index, as a potential modifier of the association between informal care provision and depressive symptoms. In all analyses, higher weekly time commitment to informal care for a spouse or parent was associated with increased risk of depressive symptoms. This relationship persisted whether women were not employed outside the home, were employed full-time, or were employed part-time. Higher weekly time commitment to informal care provision was associated with increased risk of depressive symptoms whether women were socially integrated or socially isolated. However, both informal care provision and social ties were potent independent correlates of depressive symptoms. Therefore, women who reported high spousal care time commitment and few social ties experienced a dramatic elevation in depressive symptoms, compared to women with no spousal care responsibilities and many social ties (OR for depressive symptoms=11.8; 95% CI 4.8, 28.9). We observed the same pattern among socially isolated women who cared for their parent(s) many hours per week, but the association was not as strong (OR for depressive symptoms=6.5; 95% CI 3.4, 12.7). In this cross-sectional study, employment status did not seem to confer additional mental health risk or benefit to informal caregivers, while access to extensive social ties was associated with more favorable caregiver health outcomes. r 2003 Elsevier Ltd. All rights reserved. Keywords: Caregivers; Mental health; Depressive symptoms; Social ties; Employment; USA
Introduction
*Corresponding author. Fax: +1-148-43442992. E-mail address: carolyn [email protected] (C.C. Cannuscio).
With the aging of the population, and with women’s increasing labor market participation, juggling of work and family responsibilities is increasingly common. More than 60% of all women are expected to provide
0277-9536/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0277-9536(03)00317-4
ARTICLE IN PRESS 1248
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
care to elderly relatives at some point in their lives (Abel, 1995), and approximately 30–40% of informal caregivers in the United States are also salaried workers (Stone, Cafferata, & Sangl, 1987; Brody & Schoonover, 1986). Researchers and the popular press have debated the potential health effects of women’s multiple social roles, with essentially two reigning perspectives. The first view is the ‘‘scarcity hypothesis,’’ which argues that multiple roles deplete women’s limited energy and resources, resulting in adverse health outcomes (Barnett & Baruch, 1987; Mui, 1992). The second concept is the ‘‘expansion hypothesis,’’ which asserts that benefits accrue to people who operate in multiple roles or domains. With multiple roles come opportunities for prestige, recognition, and financial reward, all of which can bolster women’s self-concept and wellbeing (Skaff & Pearlin, 1992). The scientific evidence regarding multiple roles and health is equivocal. In support of the scarcity hypothesis, Barnett and Baruch (1987) found that the number of roles a woman occupies predicts both role demand overload and role strain. Those authors also reported that motherhood is rarely associated with psychological wellbeing and frequently associated with distress, and that employed women sacrifice their own leisure activities in order to meet work and family demands. In support of the expansion hypothesis, Barnett, Marshall, and Singer (1992) reported that employed women generally exhibit better mental health than do non-employed women. Ozer (1995) also offered evidence that employed mothers are healthier than non-employed mothers, although this may be an artifact of selection bias or ‘‘the healthy worker effect.’’ There is some empirical evidence that employed elder caregivers experience lower levels of depressive symptoms than do non-employed caregivers (Rosenthal, Sulman, & Marshall, 1993). Barnett and Rivers (1996) suggest in She Works, He Works that families are thriving in an era of multiple work and family care demands. In contrast, Taylor, Ford, and Dunbar (1995) noted no significant difference between the health effects of caregiving among non-employed vs. employed women.
(EPESE) cohort, presence of a confidant, frequency of contact with family and friends, and church or other group membership all were related to lower levels of depressive symptoms. Oxman’s study also demonstrated that the following factors were associated with increased depressive symptoms: loss of a spouse or confidant; decline in the number of children seen weekly; and decline in emotional or tangible support available. Although these shifts in social ties may predict mental health status, social ties and support appear fairly consistent over time (Eurelings-Bontekoe, Diekstra, & Verschuur, 1995). There is some evidence that supportive social ties moderate the negative health effects (e.g., depressive symptoms and physical distress) of elder care (Rosenthal et al., 1993; Moritz, Kasl, & Berkman, 1995; Tausig, 1992; Lieberman & Fisher, 1995; Baumgarten et al., 1994; Gallagher, Wagenfeld, Baro, & Haepers, 1994). Cohen, Teresi, and Blum (1994) reported that 47% of a caregiver sample attended church or synagogue on a regular basis, and that religious attendance had the most potent direct and interactive effects on caregiver psychological wellbeing of any social network variable studied. Access to tangible support also may predict caregiver wellbeing. In Rosenthal’s et al. (1993) study of caregivers for long-stay patients, lower depressive symptom levels (measured by the Center for Epidemiologic Studies Depression Scale) were reported by caregivers who felt that someone else could assume their role if necessary. In the current study, we examined the association between informal care and depressive symptoms, exploring employment status and social ties as potential modifiers of that relationship. We hypothesized that employment outside the home would attenuate the association between informal care and depressive symptoms, in accordance with the expansion hypothesis. We also postulated that, compared to socially isolated caregivers, those with extensive social ties would be less likely to suffer depressive symptoms. Specific social ties may be particularly protective (e.g., large family size may represent access to tangible support and potential relief from informal care responsibilities).
Social ties An extensive literature documents the impact of social ties on health status. Social networks and support predict both short- and long-term mortality, especially from cardiovascular disease (see especially Berkman & Syme, 1979; House, Robbins, & Metzner, 1982). This finding is particularly strong and consistent among men (Berkman, Vaccarino, & Seeman, 1993). Research also points to a protective effect of supportive social ties on mental health status. In Oxman, Berkman, Kasl, Freeman, and Barrett’s (1992) research in the Established Population for the Epidemiologic Study of the Elderly
Methods Sample The current study is a cross-sectional analysis of women’s caregiving responsibilities and mental health, based on data from the 1992 Nurses’ Health Study (NHS) follow-up questionnaire. Participants enrolled in the NHS cohort were 30–55 years old at baseline in 1976, and 46–71 in 1992. Eligible women were registered nurses who were married and living in one of 11 states at
ARTICLE IN PRESS C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
the time of enrollment. A detailed description of the population is available elsewhere (Colditz, Manson, & Hankinson, 1997). The original NHS cohort included 121,700 nurses. After losses to death, follow-up, and non-response, 104,064 participants returned the 1992 survey on health and disease status. Of those women, 75,453 completed an additional section on social variables, which was included in the first mailing to the cohort but not in subsequent attempts to obtain updated information from study participants. Only those women who provided complete information on both health/disease status and social variables were included in the current analysis. We also excluded women who provided no information on caregiving responsibilities, who reported in 1992 (and in conflict with the cohort’s original inclusion criteria) that they had never been married, and who had been diagnosed previously with cancer, heart disease, or stroke. The final sample includes 61,383 women. Caregiving exposure measure Participants self-reported the number of hours per week they typically spent caring for various people. The question was framed as follows: ‘‘Outside of your employment, do you provide regular care for any of the following? (Mark one response on each line. For people to whom you do not provide regular care, mark ‘Zero hours’.)’’ The survey queries participants regarding care to a disabled or ill spouse, disabled or ill parent, disabled or ill other person, child, or grandchild. Since ‘‘care’’ is not defined specifically in the survey question, types of informal care assistance (e.g., with bathing, dressing, cooking, etc.) were not assessed. Mental health outcome We measured depressive symptoms using the five-item Mental Health Inventory (MHI-5), one of the eight health-related quality of life scales included in the Medical Outcomes Study 36-Item Short Form Health Survey (SF-36). The MHI-5 captures four major dimensions of mental health: anxiety, depression, loss of behavioral/emotional control, and psychological wellbeing (Ware, Snow, Kosinski, & Gandek, 1993; McHorney, Ware, & Raczek, 1993). The MHI-5 is scored from 0 to 100 and can be treated either as a continuous measure of mental health or as a dichotomous measure indicating the presence or absence of depressive/anxious symptoms. Higher-scoring participants enjoy better mental health, while those who score o52 are likely to satisfy the clinical diagnostic criteria for depression and related disorders. Using ROC analysis, Berwick et al. (1991) found the MHI to be a viable tool for the identification of clinical depression,
1249
with high area under the curve (AUC=0.892). However, since the MHI-5 is not a clinical diagnostic instrument, women in this low-scoring range will be said to exhibit depressive or anxious symptoms (abbreviated ‘‘depressive symptoms’’), rather than depression per se. Employment status Participants reported whether they were currently retired, working as homemakers, working full-time (in nursing or another field), or working part-time (in nursing or another field). In all analyses, women were classified as not employed outside the home; employed full-time outside the home; or employed part-time outside the home. Measure of social ties Social ties were measured using the Berkman–Syme Social Network Index (SNI). The SNI measures marital status, number of close friends, relatives, and children and frequency of contact with these people, religious service attendance, and participation in social groups. Contact with a confidant also was reported. Using the method described by Kawachi et al. (1996), subjects were classified into four levels of social ties, ranging from low ties (Level I), to medium (Level II), to medium-high (Level III), to high (Level IV). Participants in Level I have few intimate contacts (not married; fewer than six close friends or relatives) and no religious service attendance or social group membership. Employment status and informal care provision Indicator variables were created for combinations of informal care type (to a disabled or ill spouse or parent) and hours per week (0; 1–8; 9–20; 21–35; 36+) and employment status (not employed outside the home; employed part-time; employed full-time). We created indicator variables, rather than multiplicative interaction terms, to maintain the integrity of original response categories and to avoid ‘‘force fitting’’ interaction models. Using logistic regression, odds ratios were estimated for each combination of employment and informal care responsibility. The outcome was presence (MHI-5 score o52) or absence (MHI-5 score >52) of depressive symptoms. Multivariate models controlled for additional caregiving responsibilities; age; marital status; subject’s and husband’s level of education; chronic disease (diabetes, rheumatoid arthritis); living arrangement; current cigarette smoking; and ethnicity. We examined the effect estimates (ORs) within strata of employment status (not working outside the home; working full-time outside the home; working part-time outside the home) in order to evaluate the presence/ absence of effect modification by employment status.
ARTICLE IN PRESS 1250
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
Social ties and care to a disabled/ill spouse or parent
Results
We used similar methods to evaluate the interaction between informal care and level of social ties. To investigate potential interaction between spousal care and social ties, we limited the set to married participants (n ¼ 49; 810), who were the only women ‘‘eligible’’ to provide spousal care (the exposure of interest). Therefore, for this analysis, the marital status indicator was removed from calculation of the SNI. In logistic regression models predicting the relative odds of depressive symptoms, we included indicator variables for every combination of spousal care hours per week (0; 1–8; 9–20; 21–35; 36+) and social ties (low; medium; medium–high; high). The reference group, including women presumed to have the lowest occurrence of depressive symptoms, was composed of married women with no spousal care reported and high levels of social ties. To analyze the interaction between parent caregiving and social ties, the analysis set included both married and unmarried women, but included only participants ‘‘eligible’’ to provide parent care—those who had at least one living parent when this variable was last assessed on the 1988 survey. Therefore, we used the latest information available regarding parents’ vital status. However, women who became parentless during the 1988–1992 follow-up period would be most likely to be classified as providing zero hours of parent care, thereby introducing potential bias. The reference group included women who had extensive social ties, at least one living parent, and no parent care responsibility. As in the case of spousal caregiving, we created indicator variables for every combination of parent care responsibility and social ties and used logistic regression to estimate the association between each combination and depressive symptoms. Multivariate models controlled for the following variables: additional caregiving responsibilities; age; subject’s and husband’s education; living arrangement; chronic illness (rheumatoid arthritis, diabetes); current cigarette smoking; ethnicity; employment status; and access to a confidant. The parent care model also controlled for marital status and number of parents alive in 1988. After exploring potential interactions between informal care and a composite measure of social ties, we also examined the individual components of the SNI, since previous reports indicate that certain social ties (e.g., religious participation) may be particularly beneficial. We used methods similar to those described above to examine the association with depressive symptoms for each element of the SNI, including religious service attendance, social group participation, number of close friends and relatives, and presence of a confidant.
Characteristics of the study population Among women with no informal care responsibilities, 5.2% scored less than 52 on the MHI-5 (indicating presence of depressive symptoms), as did 11.0% of spousal caregivers and 6.7% of parent caregivers (Table 1). Spousal caregivers were older and parent caregivers were younger, on average, than women without informal care responsibilities. Both spousal and parent caregivers were more likely than noncaregivers to report high levels of social ties, suggesting that, in addition to the often discussed benefits of social ties, social integration also carried responsibilities or role demands. Employment status and informal care provision The association between informal care provision and depressive symptoms is similar among women who are not employed outside the home, who are employed parttime, and who are employed full-time (Table 2). Independent of employment status, there is a dose– response relation between higher time commitment to care provision and greater risk of depressive symptoms. The magnitude of the association is similar in all employment categories. For example, among women who do not work outside the home, high-level spousal caregivers (36+ h/week) are 3.5 times more likely than non-caregivers to experience depressive symptoms (multivariate OR=3.5; 95% CI 2.7, 4.5). Among women who are employed full-time, the association between high-level spousal care and depressive symptoms is equally strong (multivariate OR=3.4, 95% CI 2.2, 5.1). While high time commitment to spousal caregiving is associated with an approximate three-fold increase in depressive symptoms (controlling for employment status), the association between parent care and depressive symptoms is not as strong. For example, among women who do not work outside the home, those who provide 36+ h/week of parent care are 1.5 times more likely than non-caregivers to report depressive symptoms (multivariate OR=1.5; 95% CI 1.0, 2.1). We observed similarly moderate associations between parent care and depressive symptoms among women employed part-time (multivariate OR=1.5; 95% CI 0.9, 2.3) and full-time (multivariate OR=1.8; 95% CI 1.3, 2.6). The similarity of the association between caregiving and depressive symptoms across strata indicates limited effect modification by employment status. For women without informal care responsibilities, employment outside the home is associated with somewhat lower depressive symptoms. Among women who are not informal caregivers, those who are employed part-time are approximately 20% less likely than
ARTICLE IN PRESS C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
1251
Table 1 Characteristics of the Nurses’ Health Study participants in 1992a Characteristic
Women with no informal elder care roles
Women who care for a disabled/ill spouse
Women who care for a disabled/ill parent
Women who care for a disabled/ill other person
n Depressive symptoms (%) Age (%) o50 50–54 55–59 60–64 65–69 70+
46,141 5.2
3123 11.0
8617 6.7
5450 6.4
13.4 21.0 21.3 19.0 19.7 5.5
7.0 13.0 19.0 21.5 29.4 10.3
14.4 27.1 26.2 19.6 10.9 1.8
10.4 18.4 21.0 20.4 23.3 6.5
Marital status (%) Married Divorced Separated Widowed
81.3 7.5 1.1 10.0
96.3 1.2 0.6 1.7
84.6 5.7 0.7 8.7
79.0 8.4 0.8 11.5
Social ties (%) Low Medium Medium–high High Confidant (yes) (%)
5.6 27.3 19.6 47.5 93.7
4.2 20.2 21.7 53.9 92.6
3.8 21.4 19.3 55.6 94.7
3.0 22.3 14.3 60.5 95.5
a
Age-adjusted. Participants could report multiple caregiving roles.
Table 2 Combinations of informal care and employment outside the home: relative odds of exhibiting depressive symptoms (MHI-5o52), with 95% confidence intervalsa Caregiving responsibilities
Among women who do not work outside the home
Among women who work part-time outside the home
Among women who work full-time outside the home
n No care to a disabled spouse
22,963 21,449 (ref.) 607 2.0 (1.5, 253 2.4 (1.6, 132 1.7 (0.8, 522 3.5 (2.7,
14,891 14,223 0.8 (0.7, 342 1.0 (0.6, 117 2.1 (1.1, 57 1.6 (0.6, 152 3.1 (2.0,
22,804 21,858 0.8 (0.7, 506 1.4 (1.0, 172 1.6 (0.9, 101 2.6 (1.5, 167 3.4 (2.2,
Disabled spouse 1–8 h/week Disabled spouse 9–20 h/week Disabled spouse 21–35 h/week Disabled spouse 36+ h/week
No care to a disabled parent Disabled parent 1–8 h/week Disabled parent 9–20 h/week Disabled parent 21–35 h/week Disabled parent 36+ h/week a
20,368 (ref.) 1527 0.9 (0.7, 502 1.3 (0.9, 166 1.2 (0.6, 400 1.5 (1.0,
2.7) 3.7) 3.3) 4.5)
1.1) 1.8) 2.1) 2.1)
12,541 0.8 (0.7, 1619 0.9 (0.7, 388 1.1 (0.8, 134 1.3 (0.7, 209 1.5 (0.9,
0.9) 1.6) 3.7) 4.2) 5.0)
0.9) 1.1) 1.6) 2.4) 2.3)
19,236 0.8 (0.7, 2530 0.9 (0.7, 580 1.0 (0.7, 177 1.3 (0.8, 281 1.8 (1.3,
0.9) 1.9) 2.7) 4.6) 5.1)
0.8) 1.0) 1.4) 2.2) 2.6)
Models control for all other caregiving responsibilities, age, marital status, subject’s and husband’s level of education, living arrangement, chronic conditions (diabetes and rheumatoid arthritis), current cigarette smoking, and ethnicity. Within each stratum of employment status, the trend of increasing risk with increasing time commitment was statistically significant.
ARTICLE IN PRESS 1252
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
homemakers and retired women to experience depressive symptoms (multivariate OR=0.8; 95% CI 0.7, 0.9). A similar decrease in depressive symptoms was observed among non-caregivers when women employed full-time were compared to women not employed outside the home (multivariate OR=0.8; 95% CI 0.7, 0.9). Social ties and informal care provision We found a dose–response relation within each category of social ties, so that a higher number of weekly hours of spousal care provision is associated with higher risk of depressive symptoms (Table 3). However, at any given level of time commitment to caregiving, women with few social ties are more likely to exhibit depressive symptoms than are women with many social ties. Because of the strong independent associations with depressive symptoms for both informal care provision and social ties, women with high time commitment to informal care and few social ties represent an extremely high-risk group. For example, women with high spousal caregiving responsibility (36+ h/week) and few social ties experience a dramatic elevation in depressive symptoms, compared to women with no spousal care responsibilities and many social ties (multivariate
OR=11.8; 95% CI 4.8, 28.9). The same pattern holds in the case of parent care, although the magnitude of the association is not as strong as it is for spousal care. Again, the greatest elevation in depressive symptoms is observed among high-level parent caregivers with few social ties (multivariate OR=6.5; 95% CI 3.4, 12.7). Despite the marked elevation in depressive symptoms observed in these subgroups of caregivers, there is limited evidence of statistical interaction between spousal care provision and social ties. In other words, the incremental increase in depressive symptoms with increasing hours of care provision is equally strong among women with low, medium, medium–high, or high levels of social ties. Within each category of social ties, women who report 36+ h/week of spousal care are approximately three to four times more likely than women with no care responsibilities to experience depressive symptoms. In the case of parent care, high-level caregivers are approximately two times more likely than non-caregivers to report depressive symptoms, at most levels of social ties. The only exception appears to be among women with the most social ties, who do not experience a care-related increase in depressive symptoms. In this socially integrated group, even women who report
Table 3 Combinations of spousal/parent care and social ties in the Nurses Health Study: relative odds of exhibiting depressive symptoms (MHI-5 o52)a Caregiving responsibilities
Among women with high levels of social ties
Among women with medium–high levels of social ties
Among women with medium levels of social ties
Among women with low levels of social ties
n (married women) No care to a disabled spouse
27,901 26,312 (ref.) 756 1.1 (0.8, 1.7) 428 2.6 (1.8, 3.8) 405 4.1 (3.0, 5.7)
10,812 10,176 1.6 (1.4, 289 2.9 (1.9, 175 4.0 (2.5, 172 4.5 (2.8,
8343 7803 2.6 (2.4, 239 3.8 (2.5, 148 3.8 (2.3, 153 8.3 (5.5,
1389 1280 3.7 (3.1, 4.4) 53 6.2 (3.1, 12.6) 30 8.5 (3.5, 20.5) 26 11.8 (4.8, 28.9)
18,248 14,005 (ref.) 2870 1.0 (0.9, 1.3) 1004 1.3 (0.9, 1.7) 369 1.1 (0.7, 1.9)
7323 5784 1.5 (1.3, 1039 1.9 (1.5, 340 2.5 (1.7, 160 3.4 (2.1,
Disabled spouse 1–8 h/week Disabled spouse 9–35 h/week Disabled spouse 36+ h/week
n (women with X1 parent alive) No care to a disabled parent Disabled parent 1–8 h/week Disabled parent 9–35 h/week Disabled parent 36+ h/week a
1.8) 4.4) 6.4) 7.2)
1.8) 2.5) 3.6) 5.6)
9096 7425 2.4 (2.1, 1089 3.0 (2.4, 380 3.3 (2.4, 202 5.1 (3.4,
2.9) 5.6) 6.3) 12.5)
2.7) 3.8) 4.6) 7.5)
1845 1540 3.8 (3.2, 178 3.9 (2.5, 67 3.8 (1.9, 60 6.5 (3.4,
4.6) 6.1) 7.6) 12.7)
To assess interaction between spousal care and social ties, set is limited to married women; for parent care the set is limited to women who have at least one living parent. In both cases, models control for all other caregiving responsibilities, age, subject’s and husband’s education, living arrangement, chronic illness (rheumatoid arthritis, diabetes), current cigarette smoking, ethnicity, employment status, and access to a confidant. Parent care model also controls for marital status and number of parents alive in 1988. Within each stratum of social ties, the trend of increasing risk with increasing caregiving time commitment was statistically significant.
ARTICLE IN PRESS C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
36+ h/week of parent care are no more likely than noncaregivers to report depressive symptoms (multivariate OR=1.1; 95% CI 0.7, 1.9).
Elements of the SNI Specific SNI components were assessed and are summarized in Table 4. As we found in analysis of the composite index, we again observed that less socially connected women with high care responsibilities consistently experienced the most extreme elevation in depressive symptoms. Several types of social ties were found to alter the association between spousal caregiving and depressive symptoms. For instance, women who provide 36+ h/week of care to a disabled spouse and rarely/never attend religious services experience a dramatic elevation in depressive symptoms (multivariate OR=5.7; 95% CI 3.9, 8.3), compared to women who provide no spousal care and regularly attend religious services. The increase in depressive symptoms was less pronounced among women with high spousal care responsibilities and frequent religious service attendance (multivariate OR=3.7; 95% CI 2.9, 4.6). We observed particular elevations in depressive symptoms among women with high caregiving time commitment and: no regular participation in social group activities; five or fewer close family ties; five or fewer close friends; and no confidant. Again, we observed similar patterns, but weaker associations, among women who provided care to disabled or ill parents (data not shown).
1253
Discussion This study reports several key findings. First, there is a dose–response relation between higher informal care time commitment and increased risk of depressive symptoms, and the association is especially strong among spousal caregivers. Second, the association between informal care and depressive symptoms is not modified by employment status. The likelihood of exhibiting depressive symptoms is similar among caregivers who are not employed outside the home, who are employed part-time, and who are employed full-time. Third, there are strong independent associations with depressive symptoms for both informal care and level of social ties. Therefore, women with high time commitment to caregiving (especially for a spouse) and few social ties experience a marked elevation in risk of depressive symptoms. Strengths and limitations The primary limitation of this study is its crosssectional design, which thwarts our ability to infer causality in the associations between informal care, employment status, social ties, and depressive symptoms. The study is also limited by potential response bias: healthier women may have been more likely to return the 1992 follow-up questionnaire, including the social variables section analyzed in this paper. If this type of bias is operative, the observed associations between care provision and depressive symptoms may underestimate the true relationship between these factors.
Table 4 Spousal caregiving among women with and without specific social ties: relative odds of depressive symptoms (MHI-5o52)a
Religious service attendance Regular Infrequent/never Group social activities Regular None Number of close family ties Six or more Five or fewer Number of close friends Six or more Five or fewer Confidant Yes No a
Spousal care zero (h/week)
Spousal care 1–8 h/week
Spousal care 9–35 h/week
Spousal care 36+ h/week
(ref.) 1.4 (1.3, 1.6)
1.4 (1.1, 1.8) 3.4 (2.4, 4.8)
2.1 (1.6, 2.8) 4.4 (3.0, 6.5)
3.7 (2.9, 4.6) 5.7 (3.9, 8.3)
(ref.) 1.8 (1.7, 2.0)
1.6 (1.2, 2.2) 2.9 (2.2, 3.9)
2.7 (2.0, 3.7) 3.8 (2.7, 5.3)
4.6 (3.5, 6.0) 5.3 (4.0, 7.1)
(ref.) 2.2 (2.1, 2.4)
1.4 (0.9, 2.2) 3.7 (2.9, 4.7)
2.6 (1.7, 4.1) 5.0 (3.8, 6.6)
4.4 (3.1, 6.3) 7.8 (6.0, 10.0)
(ref.) 1.9 (1.7, 2.0)
1.4 (0.9, 2.1) 3.2 (2.5, 4.0)
2.9 (2.0, 4.4) 4.0 (3.0, 5.4)
4.5 (3.2, 6.3) 6.4 (5.0, 8.2)
(ref.) 3.0 (2.7, 3.3)
1.7 (1.4, 2.1) 4.1 (2.4, 6.9)
2.6 (2.0, 3.3) 3.5 (1.7, 7.5)
4.0 (3.3, 5.0) 7.0 (3.7, 13.4)
Models control for all other caregiving responsibilities (care for a disabled or ill parent; disabled or ill other person; child, or grandchild), age, marital status, subject’s and husband’s level of education, living arrangement, chronic conditions (diabetes; rheumatoid arthritis), current cigarette smoking, and ethnicity.
ARTICLE IN PRESS 1254
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
Additionally, the study relied on self-report of caregiving time commitment and depressive symptoms. Participants with depressive symptoms may have differentially perceived and reported their informal care time commitments, thereby inflating the observed association between caregiving provision and depressive symptoms. Finally, the generalizability of this study may be limited because of the population sampled. All participants in the NHS were married registered nurses at the start of the study in 1976, and had, therefore, selfselected into a caregiving profession. However, the study has several strengths, including the large sample size, which allowed for examination of a range of caregiving roles and time commitments. The sample size also allowed us to examine, with power, the effects of care provision among women with different employment circumstances and type/number of social ties, while controlling for several social and health status variables.
adequate appraisal of why some informal care providers remain healthy and others experience depressive symptoms. In support of the expansion hypothesis, we observed that among women with no caregiving responsibilities, those who worked part-time or full-time outside the home were less likely than homemakers and retired women to report depressive symptoms (Table 2). In this situation, being engaged in employment outside the home—an expansion of social roles—is associated with lower risk of depressive symptoms. One caveat is that this observation may be an artifact of selection bias, with healthier women more likely than those with depressive symptoms to enter or remain in the paid labor market. The temporal association between employment status and depressive symptoms cannot be assessed in this cross-sectional study.
Employment status: scarcity vs. expansion
In this study, we observed that socially isolated caregivers, and especially spousal caregivers, experience a marked increase in depressive symptoms. However, as noted for employment status, above, we also cannot determine the temporal association between social ties and depressive symptoms. Social isolation could be either an antecedent of, or a reaction to, depressive symptoms. A further limitation of this assessment is that health status may confound the association between social ties and depressive symptoms. For instance, women who are physically ill or disabled may be less likely to engage in social activities or to attend religious services. This would result in lower SNI scores, and also could lead to depressive symptoms. We attempted to correct for this potential bias by excluding from analysis women with cancer, heart disease, and stroke, and by controlling for other chronic conditions (rheumatoid arthritis and diabetes). We observed strong independent associations between individual components of the SNI and depressive symptoms. The following factors were associated with increased risk: infrequent religious service attendance; lack of participation in social groups; having five or fewer close friends; having five or fewer close relatives; and not having a confidant. Each of these individual types of social ties may be a source of practical, material, or emotional support for caregivers (i.e., a resource for assistance with care-related tasks or for coping with grief or trauma surrounding family illness). In addition, we noted that caregivers in this sample were more likely than non-caregivers to report high levels of social ties, which implies that social integration serves to increase both social resources and responsibilities like care provision.
Conflicting results have been published in previous research on the interaction between caregiving and employment status. Some studies have supported the scarcity hypothesis, which holds that multiple roles present numerous stresses, thereby increasing risk of illhealth. Other studies have supported the expansion hypothesis, which states that multiple roles should confer extra benefits and protection against mental or physical illness. The current study suggests certain applications for—and shortcomings of—both the scarcity and the expansion hypotheses. For instance, we observed that, regardless of employment status, women with high caregiving time commitment reported heightened depressive symptoms. This observation may be consistent with the scarcity hypothesis, if the elevation in depressive symptoms results from carerelated demands on women’s limited time and energy. However, alternative explanations must be explored. For example, the care-related increase in depressive symptoms may result not from task-related burden, but from grief surrounding a loved one’s illness, or from loss of relational reciprocity, for example. One important observation from the current study is that spousal care is a much stronger predictor of depressive symptoms than is parent care, although similar tasks should be involved in both roles. This finding suggests that certain unmeasured characteristics of the caregiving relationship (e.g., intimacy, conflict, obligation, reciprocity, life stage) may underlie the observed association between caregiving time commitment and depressive symptoms. In other words, a scarcity model that examines only time commitment and number of roles may not elicit an
Depressive symptoms and social ties
ARTICLE IN PRESS C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
Conclusion We offer evidence that informal care provision is associated with increased depressive symptoms, especially among spousal caregivers. While employment status does not appear to alter the strength of that association, level of social ties does appear to be a strong predictor of caregivers’ mental health. Upon confirmation of this finding in longitudinal research, health care providers, employers, social service agencies, and family members should establish mechanisms for maintaining or increasing caregivers’ social ties. Socially isolated caregivers may be at greatest risk for depressive symptoms, which could jeopardize their own health and the wellbeing of care recipients. Provision of support through family, friends, social group participation, or religious service attendance may mitigate the depressive effects of caring for a disabled or ill family member.
Acknowledgements We thank the Nurses’ Health Study participants and staff for their commitment to advancing women’s health research. In addition to NHS grant CA 40356, the primary source of support for this study, the Nurses’ Health Study is supported for other specific projects by the following NIH grants: CA46475, AG12806, CA55075, CA67883, AG13842, CA65725, CA70817, DK46519, EY09611, DK45362, HL03535, HL34594, ES05947, CA75016, CA62252, CA66385, ES05947, CA62005, DK52866, CA08283, HL57871, AG15424, AR02074, CA80620. This work was conducted while C.C. Cannuscio was a doctoral candidate at the Harvard School of Public Health. Merck & Co., Inc. did not fund the research and will not benefit directly from the results reported.
References Abel, E. K. (1995). A terrible and exhausting struggle: Family caregiving during the transformation of medicine. Journal of the History of Medicine and Allied Sciences, 5, 478–506. Barnett, R. C., & Baruch, G. K. (1987). Social roles, gender, and psychological distress. In R. C. Barnett, L. Biener, & G. K. Baruch (Eds.), Gender and stress (pp. 122–143). New York: Free Press. Barnett, R. C., Marshall, N., & Singer, J. (1992). Job experiences over time, multiple roles, and women’s mental health: A longitudinal study. Journal of Personality and Social Psychology, 62, 634–644. Barnett, R. C., & Rivers, C. (1996). She works/he works: How two-income families are happier, healthier, and better-off. San Francisco: Harper San Francisco.
1255
Baumgarten, M., Hanley, J., Infante-Rivard, C., Battista, R. N., Becker, R., & Gauthier, S. (1994). Health of family members caring for elderly persons with dementia: A longitudinal study. Annals of Internal Medicine, 120, 126–132. Berkman, L. F., & Syme, S. L. (1979). Social networks, host resistance and mortality: A nine-year follow-up study of Alameda County residents. American Journal of Epidemiology, 109, 186–203. Berkman, L. F., Vaccarino, V., & Seeman, T. (1993). Gender differences in cardiovascular morbidity and mortality: The contributions of social networks and support. Annals of Behavioral Medicine, 15, 112–118. Berwick, D. M., Murphy, J. M., Goldman, P. A., Ware, J. E., Barsky, A. J., & Weinstein, M. C. (1991). Performance of a five-item mental health screening test. Medical Care, 29, 169–176. Brody, E. M., & Schoonover, C. B. (1986). Patterns of parentcare when adult daughters work and when they do not. Gerontologist, 26(4), 372–381. Cohen, C., Teresi, J., & Blum, C. (1994). The role of caregiver social networks in Alzheimer’s disease. Social Science & Medicine, 38, 1483–1490. Colditz, G. A., Manson, J. E., & Hankinson, S. E. (1997). The Nurses’ Health Study: A 20-year contribution to the understanding of health among women. Journal of Women’s Health, 6, 49–62. Eurelings-Bontekoe, E. H. M., Diekstra, R. F., & Verschuur, M. (1995). Psychological distress, social support, and social support seeking: A prospective study among primary mental health care patients. Social Science & Medicine, 40, 1083–1089. Gallagher, T. J., Wagenfeld, M. O., Baro, F., & Haepers, K. (1994). Sense of coherence, coping and caregiver role overload. Social Science & Medicine, 39, 1615–1622. House, J. S., Robbins, C., & Metzner, H. L. (1982). The association of social relationships and activities with mortality: Prospective evidence from the Tecumseh County Health Study. American Journal of Epidemiology, 116, 123–140. Kawachi, I., Colditz, G. A., Ascherio, A., Rimm, E. B., Giovannucci, E., Stampfer, M. J., & Willett, W. C. (1996). A prospective study of social networks in relation to total mortality and cardiovascular disease in men in the USA. Journal of Epidemiology and Community Health, 50, 245–251. Lieberman, M. A., & Fisher, L. (1995). The impact of chronic illness on the health and wellbeing of family members. The Gerontologist, 35, 94–102. McHorney, C. A., Ware, J. E., & Raczek, A. E. (1993). The MOS 36-Item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Medical Care, 31, 247–263. Moritz, D. J., Kasl, S. V., & Berkman, L. F. (1995). Cognitive functioning and the incidence of limitations in activities of daily living in an elderly community sample. American Journal of Epidemiology, 141, 41–49. Mui, A. C. (1992). Caregiver strain among Black and White daughter caregivers: A role theory perspective. The Gerontologist, 32, 203–212.
ARTICLE IN PRESS 1256
C.C. Cannuscio et al. / Social Science & Medicine 58 (2004) 1247–1256
Oxman, T. E., Berkman, L. F., Kasl, S., Freeman, D. H., & Barrett, J. (1992). Social support and depressive symptoms in the elderly. American Journal of Epidemiology, 135, 356–368. Ozer, E. M. (1995). The impact of child care responsibility and self-efficacy on the psychological health of professional working mothers. Psychology of Women Quarterly, 19, 315–335. Rosenthal, C. J., Sulman, J., & Marshall, V. W. (1993). Depressive symptoms in family caregivers of long-stay patients. The Gerontologist, 33, 249–257. Skaff, M. M., & Pearlin, L. I. (1992). Caregiving: Role engulfment and the loss of self. The Gerontologist, 32, 656–664.
Stone, R., Cafferata, G. L., & Sangl, J. (1987). Caregivers of the frail elderly: A national profile. Gerontologist, 27(5), 616–626. Tausig, M. (1992). Caregiver network structure, support, and caregiver distress. American Journal of Community Psychology, 20, 81–96. Taylor, R., Ford, G., & Dunbar, M. (1995). The effects of caring on health: A community-based longitudinal study. Social Science & Medicine, 41, 1407–1415. Ware, J. E., Snow, K. K., Kosinski, M., & Gandek, B. (1993). SF-36 health survey: Manual interpretation and guide. Boston, MA: Nimrod Press.