Race and Gender Differences in Perceived Caregiver Availability for Community-Dwelling Middle-Aged and Older Adults

Abstract

Purpose

Informal family caregivers are increasingly recognized as critical for meeting the needs of individuals with chronic diseases associated with aging. This study examined race and gender differences in perceived informal caregiver availability for participants age 45 and over in a large national epidemiological study.

Design and Method

Cross-sectional data were collected in structured telephone interviews from 32,999 participants from the REasons for Geographic and Racial Differences in Stroke (REGARDS) study. Participants were asked if they believed that someone was available to provide care for them in the event of a serious illness or disability, and if so, to describe that person.

Results

Over 80% of the participants reported having an available caregiver. Variables associated with lower perceived caregiver availability from a multivariable logistic regression analysis included being female, White, unmarried, living alone, over age 85, and having worse self-rated health. Spouses were identified as the most likely caregivers for all race and gender groups except for African American women, who identified daughters as the most likely caregivers. African American women also showed the smallest differential in perceived caregiver availability between married and unmarried (82.8% vs. 75.7%), while White men showed the largest differential (90.9% vs. 60.4%).

Implications

Most individuals believe they have an informal caregiver available to them, but certain factors increase the risk of reporting no available caregiver. Increased efforts are needed to anticipate future caregiving needs, particularly for individuals who perceive a lack of available informal caregivers and may require more formal care services.

The majority of care for older adults with chronic conditions is provided in the community by informal caregivers, usually family members. Family caregivers generally want to provide care, but a number of pressures may ultimately affect the availability of family caregivers to provide this care in the future. These pressures include the aging of the population, which is resulting in more people in need of caregiving assistance, hospital policies that encourage cost savings via early discharges, and observed decreases in the utilization of nursing home care (Pastor, Makuc, Reuben, & Xia, 2002). Demographic trends (e.g., high divorce rates, lower birth rates, highly mobile society, more people living alone) also introduce pressures and work against needs for increased informal caregiver availability. Sociological research has shown that Americans have become increasingly isolated socially over the past few decades (McPherson, Smith-Lovin, & Brashears, 2006), and the merging of these pressures for more informal care with demographic trends of increased social isolation could lead to a geriatric health care crisis if informal caregivers are relied on to carry ever-increasing portions of the care burden but fewer family members and other social contacts are available, willing, or capable of providing such care.

Despite concerns about the future availability of informal caregivers, there are surprisingly little data on the perceived availability of informal caregivers for middle-aged or older adults who may, at some point in the near future, require such care. We identified only one large-scale study of perceived caregiver availability among the general (e.g., non-disabled) population, which was published over 10 years ago and based on a sample of young-old Mexican Americans and non-Hispanic Whites in the San Antonio, Texas area (Talamantes, Cornell, Espino, Lichtenstein, & Hazuda, 1996). In that study, men were more likely to identify their spouses as potential caregivers than women, and participants with multiple chronic conditions were less likely to report that they had someone to provide care for them than participants with none or only one chronic condition.

While little is known about perceived caregiver ability, more information is available about individuals who actually become caregivers and about disabled persons who lack informal caregiving assistance. A recent meta-analysis of 229 studies reported that 69% of the informal caregivers were women (Pinquart & Sörensen, 2006). Women have also been found to take on more caregiving tasks, to report more care recipient problems, and to experience more distress due to caregiving than male caregivers (Pinquart & Sörensen, 2006; Yee & Schulz, 2000). A similar meta-analytic review documented that White caregivers are more likely to be spouses, whereas African American families utilize adult daughters as caregivers to a greater extent (Pinquart & Sörensen, 2005). A recently published update of national statistics on caregiving found an increasing trend for declining levels of caregiving assistance and increasing numbers of disabled older adults reporting no formal or informal caregiving assistance (Wolff & Kasper, 2006). People receiving informal care were older and more severely impaired at the time of the most recent survey. Similarly, Lima and Allen (2001) found that about 21% of disabled older adults had either partially or completely unmet needs for care, and unmet needs for care were highest among disabled individuals who were female, African American or Hispanic, living alone, and divorced or separated.

The REasons for Geographic and Racial Differences in Stroke (REGARDS) study is a large national cohort study examining the causes of geographic and racial differences in stroke incidence and mortality. Community-dwelling African Americans and Whites over the age of 45 are being recruited from across the country to participate in a structured telephone interview and an in-home evaluation to assess known and suspected risk factors for stroke (Howard et al., 2005, 2006). As part of the intake interview, REGARDS participants are asked about the availability of an informal caregiver in the event of a serious illness or disability, and some related questions concerning the characteristics of the potential caregiver. These data provide a unique opportunity to study perceived caregiver availability and to assess predictors of caregiver availability in a large national sample. Our goals for the present manuscript include: 1. To describe the frequency of perceived caregiver availability in this large national sample; 2. To determine whether perceived caregiver availability varies by race, gender, age, marital status, or other demographic variables; 3. To determine the multivariable profile of individuals at greatest risk of not being able to identify a likely informal caregiver; and 4. To describe the characteristics of the individuals who are identified as potential caregivers.

Methods

Participants

Recruitment to the REGARDS study began in January of 2003 and is ongoing. Potential participants are selected from a commercially available nationwide list purchased through Genesys, Inc. and contacted by mail with a brief description of the project. The sampling, recruitment, and telephone interviewing procedures for REGARDS have been described in detail elsewhere (Howard et al., 2005). All interview procedures and informed consent procedures have been reviewed and approved by the Institutional Review Boards of each REGARDS study site.

A letter and brochure are mailed to each potential participant, and approximately 2 weeks later, a contact by telephone is attempted. The mailings are made in accordance with a stratified random sampling design, which calls for 50% of the sample to be obtained from the “stroke belt” and the remaining 50% to reside in the rest of the 48 contiguous states. The stroke belt consists of Southern states (AL, AR, GA, LA, MS, NC, SC, and TN), and within this region, 40% of the sample (20% of the overall sample) is being recruited from the “stroke buckle” that consists of the coastal plains region of NC, SC, and GA, The remaining stroke belt cases (30% of the overall sample) reside in the stroke belt states but not in the stroke buckle region.

Within each region, the stratified random sampling design calls for approximately 1/2 of the sample to be African American and 1/2 White, and within each region-race stratum, approximately 1/2 male and 1/2 female. Exclusion criteria include age less than 45, race other than African American or White, previous diagnosis for cancer requiring chemotherapy, a serious medical condition that would prevent long-term participation in the project, residence in or on a waiting list for a nursing home, and inability to communicate in English. In the initial telephone calls, the number of adults over age 45 residing in each household is determined, and in households with more than one such resident, one potential participant is randomly selected and invited to enroll in the project. A 44.7% response rate has been observed. This rate represents the percent of known and expected eligible candidates who have agreed to participate in the baseline telephone interview and is comparable to rates observed in other cohort epidemiologic studies (Morton, Cahill, & Hartge, 2006).

From January 2003 through May 1st of 2006, a total of 32,999 participants had enrolled in the REGARDS study and completed the initial telephone interview. Of these, 32,957 (99.9%) answered the interview question about whether they had an informal caregiver available to them if they were to experience a serious illness or become disabled (see below). These 32,957 participants constitute the sample for the analyses reported in this paper. Descriptive data about the sample are provided in Table 1.

Table 1.

Descriptive Statistics

	African American		White		Total Sample
Variable	Women	Men	Women	Men
N	9,456	6,071	8,677	8,753	32,957
Age (M ± SD )	64.4 ± 9.6	65.9 ± 9.3	66.0 ± 9.7	67.0 ± 9.2	66.1 ± 9.4
Marital Status (%)
Married	29.9	65.3	49.1	82.9	55.6
Divorced/Sep	27.0	17.7	18.4	6.6	17.6
Widowed	34.2	11.3	28.3	7.2	21.3
Never Married	8.9	5.9	4.3	3.3	5.6
Education (%)
< 12 years	24.6	24.6	9.4	8.5	16.3
12 years (HS grad)	28.1	27.4	29.5	23.1	27.0
Some college	25.0	24.0	29.1	24.7	25.8
College grad	22.3	24.0	32.0	43.7	30.9
Income (%)
< $20,000	39.3	24.8	22.0	9.5	23.8
$20,000 – $35,000	30.2	30.5	29.4	24.6	28.5
$35,000 – $75,000	24.0	31.4	32.7	40.5	32.2
> $75,000	6.5	13.3	15.9	25.4	15.5
Live Alone (%)	33.5	22.3	37.1	14.3	27.3
Self-Rated Health (%)
Excellent	8.5	12.0	18.6	20.7	15.1
Very Good	22.9	24.7	34.6	34.4	29.4
Good	39.9	39.4	31.5	31.8	35.4
Fair	23.7	19.8	11.9	10.2	16.3
Poor	5.0	4.1	3.4	2.9	3.9
Potential Caregiver? (%)
Yes	77.8	83.3	75.7	85.7	80.4
No	13.5	10.5	16.8	9.6	12.8
Don’t Know	8.7	6.2	7.5	4.7	6.9

It might be noted that the sample size of 32,957 exceeds the enrollment target of 30,000 participants for the REGARDS study, but that enrollment target is defined as those participants who complete both the telephone interview and a subsequent in-home exam. A subset of the entire sample has completed the telephone interview only. All participants who gave verbal consent to participate in the telephone interview and who provided data on the caregiver availability questions (as of May 1, 2006) were included in the analyses presented in this paper.

Procedures

Trained interviewers with the University of Alabama at Birmingham Survey Research Unit (SRU) made the telephone calls and first established eligibility for participation. Once eligibility was confirmed, the REGARDS study was further explained and verbal informed consent was obtained. The SRU interviewer then proceeded to administer a computer-assisted telephone interview (CATI) that obtained information on demographic variables, socioeconomic status (income, education), current living arrangement (e.g., number and age of other individuals living with the participant), medical history, self-rated health (excellent, very good, good, fair, poor), quality of life, social support, and potential caregiver availability.

Near the end of the CATI after obtaining all medical history and current health information, each participant was asked “if you had a serious illness or became disabled, do you have someone who would be available to provide care to you on an on-going basis?” Response options were “yes,” “no,” or “don’t know/not sure.” For those who responded “yes,” they were then asked what relationship this person had with them (e.g., spouse, daughter, son, sister, brother, etc.) and whether they currently lived in the same residence with this person.

A series of chi-square analyses and logistic regression analyses were conducted to examine age, gender, race, and health effects on participant responses to the caregiver availability question. Analyses were conducted that 1) compared “yes” responses to a combined reference category of both “no” and “don’t know/not sure” responses, and 2) compared “yes” to “no” responses (with “don’t know/not sure” responses coded as missing). Those two sets of analyses yielded very similar results, with slightly larger effect sizes observed for some variables using the second option. Consequently, in this manuscript, we report only the results that were based on the combined referent category, which tended to yield slightly more conservative effect size estimates and included data from all 32,957 participants who answered the caregiving availability question.

After simple bivariate analyses, multivariable logistic regression analyses were conducted that predicted perceived caregiver availability as a function of gender, race, age, marital status, education, living status, and self-rated health simultaneously. This analysis was based on 32,788 participants (99.4%) who did not have any missing data on the caregiver availability question or on any of these predictor variables. The multivariable analysis allowed an examination of unique predictive effects of each variable after controlling for the effects of the other correlated predictor variables in the model. For those who reported having a caregiver available to them, additional descriptive frequency analyses were conducted to examine group differences in the relationship between the participant and available caregiver.

Results

Descriptive Data

Table 1 presents descriptive data for the four race-gender groups. Due to the large sample size, women differed significantly from men and African Americans differed significantly from Whites on each variable (ps < .0001). Women were slightly but significantly younger than men, were less likely to be married, had less education and income, were more likely to live alone, and were less likely to rate their health favorably than men. Similarly, African American participants were younger, were less likely to be married, had less education and income, were more likely to live alone, and less likely to rate their health favorably than White participants.

Univariate Predictors of Caregiver Availability

Of the 32,957 participants interviewed, 26,483 (80.4%) answered “yes” to the question about whether they had believed that they had someone to provide care to them if they experienced a serious illness or became disabled. A total of 4,216 (12.8%) responded “no” to this question and the remaining 2,258 (6.9%) were unsure. Men were more likely to report having a potential caregiver than women (84.7% vs. 76.8%, p < .0001). The overall difference by race was very small and not statistically significant (80.0% vs. 80.7% for African American and White participants, respectively, p = .09). Age was analyzed as a categorical variable with 5 categories (45-54, n=3,202; 55-64, n=12,796; 65-74, n=10,689; 75-84, n=5,439; and 85 or older, n=823) and was found to be highly related to perceived caregiver availability (p < .0001). Adults aged 65-74 reported the highest likelihood of having a caregiver and adults over the age of 85 reporting the lowest likelihood (79.4%, 81.2%, 81.8%, 77.7%, and 71.6% for the 5 age groups, respectively).

Marital status, education, and income were each coded into 4 categories as summarized in Table 1, and each variable was significantly related to perceived caregiver availability (ps < .0001). For marital status, 87.9% of the married participants indicated that they had a caregiver available to them compared to 69.4% of the divorced or separated participants, 74.5% of the widowed participants, and 63.4% of those who never married. Income and education were positively associated with caregiver availability. An available caregiver was reported for 89.2% of participants with incomes greater than $75,000 per year compared to 84.3%, 79.6%, and 72.8% of those with annual incomes of $35,000 to $75,000; $20,000 to 35,000; and less than $20,000; respectively. The effects of education were smaller, but still statistically significant, with college graduates (81.8%) reporting the highest caregiver availability rate compared to 80.1% 80.2%, and 78.6% of those with some college, a high school diploma only, and less than a high school education, respectively, answering “yes” to the caregiver availability question.

Participants who lived alone were much less likely to report having a potential caregiver available to them than those who lived with at least one other person (68.5% vs. 84.8%, respectively, p < .0001). Self-rated health was also highly related to caregiver availability (p < .0001). Only 69.2% of those in “poor” health responded “yes” to the caregiver availability question compared to 84.9%, 83.6%, 80.0%, and 74.3% of those who characterized their health as excellent, very good, good, or fair, respectively.

Multivariable Model

The omnibus type III tests from the multivariable logistic regression model indicated that every predictor variable significantly predicted perceived caregiver availability in the multivariable model (ps < .0001) with the exception of education (p = .85). The adjusted odds ratios and 95% confidence intervals for the specific contrasts within each predictor variable after accounting for the other predictors in the model are presented in Table 2. These tests were generally consistent with the univariate effects reported above with two notable exceptions. First, as mentioned above, education was no longer associated with perceived caregiver availability after accounting for the other predictors in the model. Second, for race, African American participants were found to be more likely to report having a potential caregiver available to them than White participants after adjusting for racial differences on the other predictor variables (adjusted OR = 1.309, p < .0001). As before, women, participants older than age 85, participants who were divorced, widowed, or never married, participants who lived alone, and participants with ratings of general health poorer than “very good” were significantly less likely to report having a potential caregiver available to them than their respective reference groups.

Table 2.

Logistic Regression Results Predicting Caregiver Availability

Variable/Effect	Adjusted Odds Ratio	95% CI	p
Female vs. Male	0.819	0.769 – 0.872	<.0001
African American vs. White	1.305	1.228 – 1.388	<.0001
Age
55-64 vs. 45-54	1.130	1.021 – 1.250	.0183
65-74 vs. 45-54	1.191	1.071 – 1.324	.0013
75-84 vs. 45-54	0.994	0.883 – 1.119	.9198
≥ 85 vs. 45-54	0.788	0.653 – 0.952	.0132
Marital Status
Divorced/Separated vs. Married	0.408	0.373 – 0.446	<.0001
Widowed vs. Married	0.582	0.530 – 0.638	<.0001
Never Married vs. Married	0.313	0.278 – 0.352	<.0001
Education
HS grad vs. < 12	0.994	0.909 – 1.086	.8946
Some college vs. < 12	0.967	0.883 – 1.059	.4733
College grad vs. < 12	0.972	0.887 – 1.066	.5486
Live Alone vs. Living with Others	0.664	0.616 – 0.715	<.0001
Self-Rated Health
Very Good vs . Excellent	0.922	0.837 – 1.017	.1038
Good vs. Excellent	0.724	0.659 – 0.795	<.0001
Fair vs. Excellent	0.533	0.480 – 0.593	<.0001
Poor vs. Excellent	0.427	0.368 – 0.496	<.0001

Income was not included in the multivariable model presented in Table 2 because 13.8% of the participants failed to answer the income question. However, results from an alternative multivariable analysis that also included income as a predictor were examined for this smaller sample of 28,295 participants, and the findings were very similar to those reported in Table 2. The adjusted odds ratio for race, for example, after adding income to the other covariates listed in Table 2, was slightly stronger and still highly significant statistically (adjusted OR = 1.367, p < .0001).

The descriptive statistics in Table 1 indicated that the race and gender groups differed substantially in marital status, and additional multivariable logistic regression models that included race*marital status and gender*marital status interaction terms were conducted to determine if marital status affected perceived caregiver availability differently by race or gender. Those analyses revealed that the marital status effects on perceived caregiver availability were significantly stronger for Whites than for African Americans (p < .0001) and for men than for women (p < .0001). The percentages of married and unmarried (divorced/separated, widowed, and never married) participants who indicated an available caregiver for each of the four race-gender groups are displayed in Figure 1. For African American women, for example, 82.8% of the married participants and 75.7% of the unmarried participants reported having a caregiver available to them, resulting in a marriage effect of 7.1% (82.8% - 75.7%). This marriage effect was more than twice as strong for African American men (16.8%: 89.0% - 72.2%) and White women (17.5%: 84.5% - 67.0%), and more than four times as strong for White men (30.5%: 90.9% - 60.4%).

Figure 1.

Perceived Caregiver Availability by Race, Gender, and Marital Status

Characteristics of Potential Caregivers

Table 3 summarizes the relationship between the participant and the potential caregiver for the 26,483 participants who indicated that they did have someone available to provide care to them in the event of a serious illness or disability. For these participants, a spouse was most often listed as the available caregiver (47.5%) followed by a daughter (27.7%), a son (10.2%), and a sister (4.3%). Over 96% of participants who reported an available caregiver specified a family member or relative as the likely caregiver, with 3% specifying a partner, friend, or neighbor, and less than 1% refusing to identify the relationship of the likely caregiver. A majority of the potential caregivers lived with the participant (56.8%), and most potential caregivers were women (71.8%).

Table 3.

Relationship of Potential Caregiver by Race and Gender

	African American				White				Total Sample
	Women		Men		Women		Men
Relationship	N	%	N	%	N	%	N	%	N	%
Spouse	1,206	16.4	2,974	58.8	2,646	40.3	5,750	76.7	12,576	47.5
Daughter	3.537	48.1	859	17.0	2,159	32.0	770	10.3	7,325	27.7
Son	1,068	14.5	332	6.56	822	12.5	471	6.3	2,693	10.2
Sister	544	7.4	281	5.6	237	3.6	83	1.1	1,145	4.3
Brother	61	0.8	82	1.6	52	0.8	47	0.6	242	0.9
Daughter-in-law	22	0.3	8	0.2	58	0.9	13	0.2	101	0.4
Son-in-law	2	0.0	1	0.0	5	0.1	2	0.0	10	0.0
Other Relative	678	9.2	292	5.8	285	4.3	154	2.1	1,409	5.3
Partner/Boyfriend/ Girlfriend	54	0.7	93	1.8	100	1.5	89	1.2	336	1.3
Friend or Neighbor	126	1.7	106	2.1	150	2.3	82	1.1	464	1.8
Not Sure/Refused	60	0.8	30	0.6	51	0.8	41	0.5	182	0.7

The relationship between the participant and the potentially available caregiver varied significantly by race (p < .0001) and by gender (p < .0001). Spouses were most commonly reported as the likely caregiver by White men (76.7%), African American men (58.8%), and White women (40.3%), but not for African American women, who identified a daughter as the most likely caregiver (48.1%) and identified a husband as the likely caregiver in only 16.4% of the cases. Of course, only married participants could identify their spouses as likely caregivers, but even among the married participants who identified potential caregivers, less than half of the African American women (47.7%) indicated that their spouse would be this person. In contrast, 80.7% of the married African American men, 70.3% of the married White women, and 85.9% of the married White men who identified a potential caregiver listed their spouse as that person. Over 90% of the potentially available caregivers were women for the male participants (African American men = 90.5%, White men = 92.6%), compared to just over half for the female participants (African American women = 64.5%, White women = 41.2%).

Chi-square tests also revealed significant race (p < .0001) and gender (p < .0001) effects on whether the potential caregiver resided with the participant. Potential caregivers were more likely to reside with White men (79.4%) and African American men (66.6%) compared to White women (48.6%) and African American women (34.3%). The above statistics apply only to those participants who identified an available caregiver, and when the other participants without caregivers were also considered, only 26.6% of the African American women reported residing with a person who was designated as a likely caregiver for them, compared to 36.8% of the White women, 55.5% of the African American men, and 68.0% of the White men.

Discussion

This investigation provides national data on the perceived availability of informal caregivers for middle-aged and older adults and identifies key demographic and health correlates of perceived caregiver availability. Overall, over 80% of the participants responded that they have someone available to provide care to them if they were to become seriously ill or disabled. Even among the divorced, widowed, and participants who live alone, clear majorities indicate a belief that they have a family member who would be available to provide care to them. The age of the participants showed an interesting nonlinear relationship with perceived caregiver availability such that those 65 to 74 years of age were most likely to identify an available caregiver, but strong majorities of the relatively young and healthy individuals, who are presumably less likely to worry about their health, nonetheless also reported that they have someone available to provide care for them if they were to need it.

This study also highlights subgroups of people with lower expectations of having an informal caregiver available to them. Participants over 85 years of age reported the lowest level of perceived caregiver availability of any age group, and participants who rated their health as poor also reported a low rate of caregiver availability. Because older adults in poor health are probably at greatest risk of requiring caregivers in the near future, this finding uncovers a potential disparity between imminent caregiving needs and the perception of having a caregiver to fulfill of those needs. This finding that decreases in perceived health are associated with decreases in perceived caregiver availability is consistent with the earlier findings of Talamantes and colleagues (1996), who found that the number of chronic conditions was inversely related to perceived caregiver availability. Our findings are also consistent with the recently identified trends that increasing numbers of disabled older adults lack caregiving assistance (Wolff & Kasper, 2006). Several interpretations might be responsible for these findings. First, the oldest participants and those with poor self-rated health may be less likely to have spouses and other family members available in good enough health to provide care. Second, these groups may have responded on the basis of actual experiences with care provision, and may be less likely than younger persons and those in good health to believe that potential caregivers will actually make the sacrifices necessary to assume the caregiving role and provide adequate assistance.

In addition to age and health effects, we also found that participants who were divorced, separated, or never married were much less likely than married participants to identify an available caregiver. The impact of marital status on perceived caregiver availability, however, varied substantially by race and gender, with White men being more likely than any other group to identify their spouses as their potential caregivers. Overall, the marital status effects are consistent with actual marital differences in availability of care. Lima and Allen (2001), for example, studied a sample of adults needing assistance with one or more activities of daily living (ADL) or instrumental activities of daily living (IADL) and found that divorced, separated, and never married participants were much less likely to be receiving help than the married participants.

The raw difference in perceived caregiver availability between African American and White participants was minimal. However, after accounting for marital status and other demographic differences, African American participants were much more likely than their White counterparts to identify an available caregiver. Interestingly, African Americans and women were also significantly less likely to currently reside with their potential caregivers than Whites and men, respectively. African American women were the least likely to reside with the person they identified as their available caregiver, suggesting a greater ability to call on extended family members outside the home regardless of marital status or current living situation. The impact of marital status on perceived caregiver availability was significantly smaller for African American women as well. Not only were African American women less likely to be married than the other race-gender groups, but less than half of the African American women who were married identified their husbands as their likely caregivers. Because many wives can be expected to outlive their husbands, older married women may have to adjust their plans for informal care as they age. Cultural factors and demographic influences appear to further affect this planning process for African American women compared to White women. The different characteristics of likely caregivers by race are consistent with the meta-analysis findings of Pinquart and Sörensen (2005), who showed that minority caregivers are less likely to be spouses than White caregivers and often report stronger beliefs about obligations to provide care to family members than White caregivers.

Of the 80% of our participants who affirmed their belief in the availability of a caregiver, the vast majority of them identified a close family member as that person. Only 3% of our participants anticipated receiving care from an unmarried partner, friend, or neighbor. Although this suggests that non-family sources of care may be relatively uncommon, in actuality, “nonkin” sources of informal care may be quite important and could be strengthened. Based on qualitative interviews, Barker (2002) estimated that nonkin caregivers were important for sustaining community living for about 10% of frail older adults. Sources of support from outside the person’s family may be needed as secondary sources of informal support that provide much needed respite to primary family caregivers, and nonkin sources of informal care may also be an important source for support for the 20% of our participants who did not identify an available caregiver from within their existing family relationships.

Among the available caregivers identified by our participants, 72% of them were women. This finding is consistent with the gender composition of existing informal caregiving populations (Pinquart & Sörensen, 2005). The consistency of these findings — that respondents from our large and diverse sample frequently expect a female family member to assume a caregiving role and the ubiquitous finding that women do in fact constitute the majority of informal caregivers — is informative for theoretical explanations of observed gender differences in caregiving, such as gender-role socialization and gender-role expectation frameworks (Gilligan, 1982; Barusch & Spaid, 1989). However, our findings also show that men are perceived to be available for caregiving roles, and that a majority of the potential caregivers identified by White women were men. For married White women, over 70% listed their husbands as their perceived caregiver. Thus, race and marital status are important moderating factors that should be incorporated in theoretical approaches to gender role differences and caregiving expectations.

Potential participants were excluded if they reported being on a waiting list for admission to a nursing home. Beyond this question, REGARDS participants were not asked about their expectations for formal care in the future. It is possible that some individuals who did not identify an available informal caregiver might be anticipating nursing home care in the event of a disabling illness in the future. A recent study demonstrated that individuals’ expectations regarding nursing home placement are generally rooted in their personal risk profiles and are strongly associated with actual placement in the future (Akamigbo & Wolinsky, 2006). Alternative programs do exist for individuals who need care but have no informal caregiver available to them. Long-term care insurance, which has the potential to fund in-home formal services, can help disabled older adults remain at home (Cohen, Miller, & Weinrobe, 2001) as can publicly paid in-home care (Li, 2005). These programs are also appropriate for many persons who have informal caregivers, and informal caregivers usually remain actively involved in care even with such formal support. Private or publicly funded formal care, however, may be an especially important alternative to institutionalization for aging individuals without good prospects for informal support from their families or existing social networks.

Limitations to our study include the fact that it was based on a cross-sectional survey with only a limited number of questions on caregiver availability and future plans in the event of a disabling illness. The primary dependent variable in this analysis was a single question about one’s perception of informal caregiver availability. Although questions about such perceptions are often used in other areas of research (e.g., on advance directives), responses to hypothetical scenarios are obviously not perfect predictors of actual events or future decisions. Some family members who were identified as potential caregivers may not be able and willing to provide such care, and actual caregiving decisions may be further complicated by the actual degree of impairment experienced and caregiving required. Pillemer and Suitor (2006) found that older women who were asked to identify potential caregivers from among their adult children may have ignored some factors that could be important in the actual decisions of their children as to who will provide care. Such factors may lead some individuals to overestimate the availability of certain potential family caregivers and to underestimate the availability of other family members.

More predictive validity evidence is needed for questions concerning perceived caregiver availability and other sources of support. Akamibgo and Wolinsky (2006) found that older adults’ expectations concerning the likelihood of nursing home care was predictive of actual use of nursing home care during a five year follow up period, and in our own ongoing research, we are actively recruiting and following the actual caregivers of the REGARDS participants who subsequently experience strokes. Thus, we will be able to determine empirically how frequently the perceived or anticipated caregiver identified in the REGARDS intake interview actually turns out to be the primary informal caregiver following a stroke event.

Identifying a likely caregiver and discussing possible future caregiving needs with family members and friends is an important health promotion and health maintenance activity, especially for middle-aged and older adults who are at higher risk for needing such care. Social workers and case managers often document caregiver availability when an individual is released from an acute hospitalization, but it is not routine practice for health care practitioners to inquire about the availability of caregivers before a situation arises in which they are needed. In preventive health care, individuals are often urged to be screened for diseases and to improve their health behaviors such as diet, smoking, and physical activity, and there has been considerable attention paid to the importance of having advance directives that identify decision makers for care if an individual becomes unable to communicate his or her preferences (Ditto, Hawkins, & Pizarro, 2005). However, much less emphasis has been placed on the far more likely possibility that individuals may need informal caregiving from a family member or friend, and it is uncommon for individuals to be advised by their health care providers to make advance arrangements with key family members or friends in anticipation of future caregiving needs. Pillemer and Suitor (2006) have recommended that practitioners who work with older adults and their families should pay increased attention to older adults’ perceptions of potential caregivers, including during the pre-caregiving stages, and that some older adults may have misperceptions about the willingness of certain family members to provide this care. Bromley and Blieszner (1997) have also discussed the importance of family discussions of caregiving expectations and planning before caregiving is necessary, and note that these discussions are an important part of advance care planning beyond the legal arrangements that are made.

Clinicians are uniquely positioned to address caregiving needs as a component of comprehensive preventive care, and additional progress is needed, not only to anticipate future caregiving needs, but also to strengthen the support that families receive for providing informal care to older adults once these needs are evident. As Polivka (2005) points out, the sacrifices that informal family caregivers make constitute the bulk of the long-term care that is provided in this country, and these efforts are estimated to be worth nearly $300 billion annually. As our society ages over the next few decades, many more individuals will be asked to provide informal caregiving services to family members and friends, but it is unlikely that this demand for more capable and willing informal caregivers can be met unless crucial public policy changes are made. Social insurance programs and long-term care plans could be augmented with compensation incentives for family members who provide critical care services at home, and more formal services should be made available to better complement the important contributions of informal family caregivers.