The impact of caring for a child with a developmental disability on the physical and mental health of Latina and Black American women was examined. We used the National Health Interview Survey to compare the health of older mothers who were co-residing with a child who had a developmental disability to the health of same age mothers without caregiving responsibilities. Findings show that for both groups, older adult caregivers were more likely to report having limitations from arthritis than their noncaregiving counterparts. Caregiving was associated with more depressive symptoms for Latinas, but this relationship was not found for Black American women. Findings suggest that physical and mental health of caregivers need more attention in research and practice.
Our focus in this study is on the impact of caring for a child with a developmental disability on Latina and Black American midlife and older mothers. We examined emotional and physical health outcomes of mothers co-residing with a child who has a developmental disability in comparison to women of the same age and race/ethnicity who do not have a child with a disability living in the household.
When researchers began to investigate the impact of lifelong caregiving on parents who have a child with a developmental disability, they theorized a “wear and tear” effect in which a build-up of stressors would impact the overall health and well-being of caregivers who were older and had been providing care throughout their child's lifetime (Seltzer & Krauss, 1989; Townsend, Noelker, Deimling, & Bass, 1989). In their longitudinal study of older mothers of adults with mental retardation, Seltzer and Krauss found that these mothers made positive adaptations and were, in fact, doing well relative to same age peers from other studies on outcomes such as self-reported health and depression (Seltzer & Krauss, 1989; Krauss & Seltzer, 1993). Although this research was based on volunteer samples, in a subsequent study, Seltzer, Greenberg, Floyd, Pettee, and Hong (2001) used a representative sample and compared midlife parents of children with developmental disabilities to same age parents who did not have a child with a disability. In this study, Seltzer et al. found that compared to mothers who did not have a child with a disability, the mothers who did were more likely to have fewer social contacts and to work less outside the home, suggesting that these mothers made accommodations in order to provide care for their child with a developmental disability. However, they found no differences in self-reported health and depression between the two groups of mothers, thus replicating previous findings from studies in which researchers used volunteer samples.
A limitation of these investigations is that they consisted of samples of primarily non-Latino White caregivers and, therefore, do not shed light on the impact of lifelong caregiving on families of color. The environmental and ecological contexts of families of color are often very distinct from those of non-Latino White families, and these disparities may produce differential effects on the impact of caring for a child with a developmental disability over the life course. Because families of color often experience many more hardships than do non-Latino White families (e.g., racism and discrimination, stressors of immigration, lower socioeconomic status), there may in fact be a negative effect on the health and well-being of caregivers of color.
Although it is important that researchers also focus on positive aspects of caregiving, recent research showing health disparities over the life-course between non-Latino Whites and Blacks and Latinos reminds us not to over look potential negative effects on health that caregivers of color may experience differently than do White caregivers. Latinos and Black Americans are more likely to be in poorer health and to have higher rates of chronic health conditions (D. Williams, 2002). People of color experience a chronic exposure to stressors, which through cumulative burden, may lead to physiological wear and tear on the human body, increasing the risk of disease (D. Williams, 2005). There is also evidence that persons who experience greater disadvantage not only experience more severe daily stressors, but are more vulnerable to stressors than those of higher socioeconomic status (Almeida, Neupert, Banks, & Serido, 2005).
Recently, researchers in the area of health inequalities across the life-course have referred to the theoretical concept of cumulative adversity (Zarit & Pearlin, 2005), which may be more applicable to caregivers of color than “wear and tear” theory. Cumulative adversity refers to any combination of three processes: (a) a chain of hardships, one surpassing the next over the life-course; (b) a layering or build-up of hardship effects; and (c) a single hardship or protective factor that persists over the life-course (Hatch, 2005). Caring for a child with a developmental disability may represent the latter by exposing parental caregivers to both hardships and protective factors over the life-course. For caregivers of color, the first and second processes within cumulative adversity may also contribute to the experience.
Existing research on midlife and older Black American and Latino parents who have a child with a developmental disability is extremely limited and has been focused more on emotional well-being as outcomes. Some investigators have compared Black American mothers to non-Latina White mothers on psychological appraisal and well-being outcomes and reported that Black American mothers had similar, if not better, outcomes (Pruchno, Patrick, & Burant, 1997; Valentine, McDermott, & Anderson, 1998). In contrast, research on Latino families caring for children and adults with mental retardation found that Mexican American, Puerto Rican, and Cuban American mothers experience high rates of depressive symptoms and burden (Blacher, Shapiro, Lopez, Diaz, & Fusco, 1997; Magaña, Seltzer, Krauss, Rubert, & Szapocznik, 2002) and higher rates of depression in comparison to non-Latino Whites (Magaña, Seltzer, & Krauss, 2004).
Similar to the initial studies of non-Latino parents caring for adults with mental retardation, researchers investigating families of color have relied on volunteer samples. Representative samples are needed to elucidate our understanding of the impact of caregiving on the health of caregivers of color. Furthermore, to understand the impact of caring for a child with a developmental disability, studies are needed in which researchers compare same age Black American and Latina mothers who have this caregiving responsibility to their counterparts who do not (S. Williams, Dilworth-Anderson, & Goodwin, 2003; Magaña, Greenberg, & Seltzer, 2004) rather than comparing caregivers of color to White caregivers.
In the present study we used a national probability sample to compare Black American and Latina midlife and older women who are co-residing with their child with a developmental disability to Black American and Latina women who do not co-reside with a child with a disability. We tested the following research questions: (a) Do midlife and older Latina and Black American mothers living with a child with a developmental disability differ in physical health and mental health outcomes from their age peers who do not live with a child with a developmental disability? (b) How do these comparisons vary by age group (i.e., midlife women vs. older adults)?
Sample and Procedure
The sample for the present study was drawn from the National Health Interview Survey (NHIS) and included 162 Latina and Black American women who co-reside with a child who has a developmental disability and 2,754 Latina and Black American women who do not co-reside with a child with a disability (we use the term Black American in our study rather than African American because there is ethnic variation in the Black population that could include those that identify as African Americans, Caribbean Blacks, or African Blacks; furthermore, the NHIS does not identify ethnic identities within the racial category Black).
The NHIS data are obtained through a complex sample design involving stratification, clustering, and multistage sampling. Each year's sample is studied once. Thus, the NHIS is a series of cross-sectional samples. We drew our sample from 3 years of data: 1999, 2000, and 2001. It was necessary to merge 3 years of data because although the NHIS oversamples Latinos and Black Americans, the number of midlife and older mothers caring for a child with a developmental disability within these ethnic groups was expected to be small. Further details on the NHIS are available at their website (National Center for Health Statistics, 2005a).
Our sample was selected according to the following criteria: the caregiver group included Latina (non-Black Latina) and Black American (U.S. born non-Latina Black) mothers who were 40 years and older, lived with a minor or adult child with a developmental disability, and did not co-reside with anyone else who needed care due to a physical or mental health condition. We adapted criteria for identifying children with mental retardation or a developmental disability from the work of Larson et al. (2001). In the person-level data file, we selected those household members who (a) were identified as a minor or adult child to the respondent and (b) had limitations in physical activity due to mental retardation, other developmental delay, or epilepsy. We then identified children from the sample child core who, according to respondent report, were diagnosed by a health professional to have mental retardation, Down syndrome, cerebral palsy, autism, or other developmental delay. We excluded cases in which there was a child in the household with a functional limitation due to a physical or mental illness, or someone else in the household who needed care due to a physical or mental health condition or needed assistance with daily living tasks.
Overall, we identified 83 Latina and 79 Black American mothers who were 40 and over and co-resided with a child who had a developmental disability. We were limited to mothers who co-resided with their child because the NHIS does not include data on children who live outside the household. A total of 170 minor and adult children were identified as having a developmental disability. Table 1 presents the characteristics of the children with disabilities. Developmental disability categories for children under the age of 18 included autism, cerebral palsy, Down syndrome, epilepsy, mental retardation, or other developmental disability. However, the majority of children were identified in the latter two categories. The children who were 18 years or older were limited to the two categories: mental retardation and other developmental disability. As a result, these frequencies of specific developmental disabilities are not representative of national prevalence rates. About 93% of the mothers had one child with a developmental disability; 7% had two or more children that met the criteria for having a developmental disability.
The sample for the comparison group was determined by selecting all women who were at least 40 years old, did not have a child who met the criteria for a developmental disability, did not have a child with a functional limitation due to a physical or mental illness, did not live with anyone else who needed care due to a physical or mental health condition, and did not have an additional household member who needed assistance with daily living tasks. The sample sizes for the Latina and Black American comparison groups were 1,667 and 1,087, respectively.
Of the Latina mothers (caregivers and noncaregivers), 59.7% were of Mexican descent; 15.4%, Central or South American; 10.2%, Puerto Rican; 6.4%, Cuban; 4.2%, no specified ethnicity; and 4.1%, Dominican. The sample sizes of caregivers within the various Latina groups were not large enough to conduct within-group analyses of health outcomes between caregivers and noncaregivers. We aimed to limit the potential differences between the Latino ethnicities by controlling for whether or not they were born in the United States, level of education, employment, and income.
Table 2 presents the demographic characteristics of the Latina and Black American mothers and their comparison groups. For the Latina women, the only variable that was significantly different between caregivers and noncaregivers was percentage born in the United States. Almost 50% of the caregivers were born in the United States, while only 31% of the noncaregivers were born in the United States. For the Black American women, caregivers were significantly more likely to have incomes under $20,000 than were noncaregivers. Also, the caregiver group had more family members residing in the household than did the noncaregiver group in the Black American sample; however, this difference was only marginally significant.
Age was used as a continuous variable in the analyses for the first research question. However, in order to determine the effect of midlife versus older mothers, respondent's age was recoded from a continuous variable into a dichotomous variable, 0 (40 to 54 years) and 1 (55 and older). We recoded marital status of the respondent as 0 (not married) and 1 (married). The annual household income variable was coded in the original data as 1 (< $20,000) and 2 (> $20,000). The respondent's education level was coded as 0 (never attended school or kindergarten only), 1 through 12 (grades), 13 (high school graduate/ GED), 14 (some college, no degree), 15 (associates degree), 16 (bachelors degree), 17 (masters degree), 18 (professionals or doctoral degree). The size of the respondent's family living in the household was reported as a continuous variable. Whether or not the Latina respondents were born in the United States was coded as a 0 (born outside the United States) or 1 (born in the United States).
A dichotomous variable was constructed to determine the impact of being a caregiver on health outcomes: 0 (comparison group) and 1 (caregiver group).
This measure was adapted from the Composite International Diagnostic Interview-Short Form (Kessler, Andrews, Mroczek, Ustun, & Wittchen, 1998). This measure has been used with and validated for many populations including Latino and Black Americans (Andrews & Peters, 1998; Kessler et al., 1998). The measure consists of six items that rate the frequency of depressive symptoms over the past month. For example, the measure included statements such as “Over the past 30 days, I felt so sad that nothing could cheer me up.” The measure is scaled along a 5-point continuum ranging from 0 (none of the time) to 4 (all of the time). A total score was calculated by summing the total of the six items. The scale's alpha reliability was .89 for our total sample.
Diagnosed health conditions
These measures were created from responses to the question “Have you ever been told by a doctor or other health professional that you have ___ (health condition).” Six health conditions were listed, and the respondent answered yes or no. Those diagnosed health conditions that were present in 10% or more of any of the groups (caregiver or comparison group from either ethnicity or age group) were chosen for the present analysis: hypertension, heart problems, asthma, and diabetes. The heart problem variable was computed by identifying whether the respondent had at least one of the following heart problems: coronary heart disease, myocardial infarction, angina, or other heart condition. Dichotomous variables were created for each of the diagnosed health conditions and were coded 0 (condition not present) and 1 (condition present).
Health problems that limit activities
These measures were created from a series of two questions. The first question was (a) “By yourself, and without using any special equipment, how difficult is it for you to ___ (e.g., carry something as heavy as 10 lbs., walk up 10 steps without resting, stand, or be on your feet for 2 hours)?” If an affirmative response was reported for any of these activities, the second question was “What conditions or health problems cause you to have difficulties with these activities?” The respondent was provided with a list of 19 health problems. Health problems that limit activity were selected for the current analysis if 10% or more of the caregivers or comparison group reported that the health problem limited their activities. Only three health problems met this criterion: arthritis, hypertension, and diabetes. Dichotomous variables were used for each of the physical health problems and were coded 0 (health problem does not limit activity) and 1 (health problem limits activity).
We used frequency weights provided in the NHIS data to accurately reflect the number of persons represented by any given case after adjusting for the probability of being included in the sample. Even with the use of weighted data, the methods of clustering and stratification will result in biased variance estimates; therefore, we conducted all analyses with SUDAAN 9.0.1, a statistical package that calculates robust variances and corrects for standard errors for clustering and stratification.
Research Question 1
Logistic and ordinary least squares (OLS) linear regression methods were used in order to determine the association between caregiving versus noncaregiving and each of the health measures, controlling for demographic characteristics (age, education, marital status, income, employment status, family size, and born in the United States vs. foreign born for the Latina sample). Separate models were run for each ethnic group. Logistic regressions were used to determine the association between caregiving and the dichotomous outcome variables (i.e., diagnosed health conditions and health problems that limit activity). The OLS regressions were used to examine the association between caregiving and depressive symptoms.
Research Question 2
In order to determine whether the association between caregiving and health was different for midlife versus older adult caregivers, we ran separate regressions within each ethnicity and within each age group. Because the caregiver groups were considerably small, especially among the older adult samples, an alpha level of .10 was used for statistical tests; differences reported between the .05 and .10 levels are described as being marginally significant.
The group level mean was imputed for those cases with missing data on the education variable (Allison, 2001). The group level mean was calculated by controlling for race, age group, and caregiver status. The SUDAAN 9.0.1 was used to generate predicted probabilities for the dichotomous variables with missing data (marital status, employment status, income) by regression of each dichotomous variable on race, age group, caregiving status, education level, family size, and if the respondent was born in the United States. The missing data were coded 1 if the predicted probability was estimated as .50 or greater and 0 if the predicted probability was estimated as less than .50 (Allison, 2001).
In Table 3, the diagnosed health conditions, health problems that limit activity, and mean depressive symptoms score are presented to show their prevalence within the Latina and Black American caregiving and comparison groups and by age group. In the overall sample, it is notable that among the diagnosed conditions, hypertension is the most prevalent for both Latinas and Black Americans, whereas among conditions that limit activity, arthritis is the most prevalent for both groups. When examining the prevalence rates for the midlife and older adult groups, we found that several of the health conditions were more prevalent in the older adult groups than in the midlife groups (e.g., hypertension, heart problems, diabetes, and arthritis). Depressive symptoms were more prevalent in the younger groups and among Latinas.
All of the multivariate analyses were conducted controlling for the demographic characteristics listed in Table 2. The variable percentage born in the United States was used as a control variable in the Latina sample only. In the first research question we asked, “Do midlife and older Latina and Black American mothers living with a child with a developmental disability differ in physical and mental health outcomes from their same age peers who do not live with a child with a developmental disability?” Table 4 shows that Latina caregivers were 2.1 times more likely to have been diagnosed with a heart problem than were those in their comparison group. Latina caregivers were 2.3 times more likely to report that arthritis limited their physical activity. Table 4 also shows that among Latinas, being a caregiver significantly predicted higher levels of depressive symptoms. There were no statistically significant differences between Latina caregivers and noncaregivers in hypertension, asthma, or diabetes. In our examination of Black American women, our results did not indicate any significant differences in health or depressive symptoms between Black American caregivers and their noncaregiving comparison group (see Table 4).
In our second research question, we asked “How do these comparisons vary by age group (i.e., midlife women vs. older adults)?” Table 5 shows that older adult Latina caregivers were 5.1 times more likely than older noncaregivers to be diagnosed with a heart problem. Older adult Latina caregivers were also 3.8 times more likely than older noncaregivers to report arthritis as a limitation. However, for the midlife Latina mothers, there were no differences in physical health between caregivers and noncaregivers. In contrast, we found that being a caregiver was predictive of depressive symptoms only among the midlife Latinas and not among the older adults. Similar to the Latina women, we found that older adult Black American caregivers were 4.1 times more likely to have limitations in their activity due to arthritis than were the older noncaregivers, yet this difference was not found among the midlife women. However, unique to the Black American women, older adult caregivers were 4 times more likely to report limitations in activity due to diabetes than were older noncaregivers (this finding was marginally significant). Caregiving did not predict higher levels of depressive symptoms for either Black American age group.
Researchers examining midlife and older mothers caring for children with developmental disabilities have found that these mothers made adaptations to their lives but did not find mental or physical health effects (Krauss & Seltzer, 1993; Seltzer et al., 2001). One limitation of this research is that it is based on samples of primarily non-Latina White mothers. In the present study we extended this line of research to women of color by using a representative study to compare midlife and older adult Latina and Black American mothers who co-reside with a child who has developmental disabilities to those who do not.
Our findings suggest that there may be a cumulative adverse effect on Latina and Black American women who provide care to a child with a developmental disability across the life-course. We found partial support for our hypothesis that Latina and Black American caregivers would have more physical health conditions than noncaregivers. Further, older adult caregivers were more likely to report limitations due to arthritis than were older adult noncaregivers for both ethnic groups. Older adult Black American caregivers were more likely to report limitations due to diabetes than were older noncaregivers (significant at a trend level), and older adult Latina caregivers were more likely to report heart problems than were older noncaregivers in our study. We did not find differences between caregivers and noncaregivers in hypertension, limitations due to hypertension, and asthma. This is somewhat surprising because these are chronic conditions that are very prevalent in these two ethnic populations. One explanation is that those who continue to co-reside with their child with developmental disabilities into old age may remain relatively healthy, whereas those who are ill may be more likely to arrange a residential placement for their son or daughter.
As hypothesized, we found that Latina mothers co-residing with a child who has a developmental disability exhibited higher levels of depressive symptoms than did their noncaregiving counterparts. This finding is consistent with previous research showing that Latina mothers of children and adults with mental retardation have higher levels of depressive symptoms than do Latina mothers with no children who have disabilities, and non-Latina White mothers co-residing with adults with developmental disabilities (Blacher et al, 1997; Magaña et al., 2002). The relationship between depression and caregiving was found among the midlife Latina mothers in our study. However, we found no relationship between depression and caregiving among the older Latina mothers. Midlife Latina mothers may face multiple roles and different life-course issues that are related to greater levels of emotional distress than do the older adult Latina mothers. In addition, the midlife mothers have younger children, who may present more behavioral and other challenges than do adult children.
Our strongest finding was that both Latina and Black American older adult caregivers were more likely to report arthritis as a limitation in their activities than were noncaregivers. This is consistent with findings from a study by Magaña, Greenberg et al. (2004), who compared older adult Black mothers who cared for a son or daughter with schizophrenia to those who do not have a child with mental illness. They also found that the caregivers were more likely to report having arthritis. There are several reasons why there might be a relationship between caregiving and arthritis among Latina and Black American women. First, researchers have found that Black Americans have higher rates of diagnosed arthritis compared to Whites, and both Black Americans and Latinos have greater levels of arthritis-attributable limitations in activity (Centers for Disease Control, 2005; Kington & Smith, 1997). Stress from caregiving may have a greater impact on those areas of health in which the caregiver is already vulnerable (Schulz & Beach, 1999; Wallsten, 2000). Furthermore, stress can impact the perception of pain for those with arthritis (Zautra, Hamilton, Potter, & Smith, 1999). Second, there is a strong relationship between low levels of vigorous physical activity and functional decline due to arthritis (Dunlop et al., 2005). Latinos and Black Americans with arthritis were found to have more functional decline over time in activities of daily living than did Whites (Dunlop et al., 2005). Extensive caregiving responsibilities may interfere with the ability of caregivers to engage in vigorous physical activity. A third explanation for the Latina caregivers is that the influence of depression may be related to high levels of limitations due to arthritis. The Latina caregivers in our study have higher levels of depressive symptoms than do noncaregivers, and depression has been associated with an increased sensitivity to pain and functional impairment from disability and with arthritis (Dunlop et al., 2005; Lin et al., 2003). Furthermore, in a study examining arthritis and depressive symptoms among Latinos and non-Latino Whites, Escalante, del Rincón, and Mulrow (2000) found that Latinos with arthritis had higher levels of depressive symptoms than did non-Latino Whites.
Higher levels of depressive symptoms may also be related to the greater likelihood of the Latina caregivers to report being diagnosed with heart problems. Patel, Black, and Markides (2003) found a relationship of exertional chest pain and higher levels of depressive symptoms among Mexican Americans. Furthermore, results of research on the presence of coronary heart disease among older adults show that Latinos tend to have a higher prevalence of heart problems or risk factors than do non-Latino Whites but have similar mortality rates from heart disease as Whites (Federal Interagency Forum on Aging Related Statistics, 2004; Mitchell, Hazuda, Haffner, Patterson, & Stern, 1991; Patel et al., 2003).
There are limitations to the interpretation of the findings from this study. First, because of the small sample size, especially with respect to older adult caregivers, our findings may underestimate the relationship between caregiving and physical health of Latina and Black American caregivers. Second, the cross-sectional nature of this study does not allow the implication of causality with respect to caregiving and health outcomes. Longitudinal research is needed to determine the causal manner in which caregiving contributes to these outcomes. Third, the NHIS does not include data on children living outside of the household; thus, it is possible for mothers included in the comparison group to have children with a disability who are living outside of their household. This may lead to a self-selection bias in which those who are more or less healthy may have elected to continue co-residency with their son or daughter. However, it is important to note that the large majority of Latino and Black American adults with developmental disabilities are likely to live with their parents (Blacher, 2001; Heller & Factor, 1988; Heller, Markwardt, Rowitz, & Farber, 1994).
In conclusion, our findings suggest that the impact of caregiving on the physical health of women of color warrants special attention among researchers and service providers. There may be different processes at work for women of color, who already suffer from higher prevalence of chronic health conditions, than for non-Latina White women. As cumulative adversity theory suggests, there may be a layering affect of the build-up of hardships over time on women of color, which contributes to greater sensitivity to the persistent daily effort required in caregiving. Although it is important to acknowledge positive aspects of caregiving and the contributions that a family member with a developmental disability may make to caregivers, it is equally important not to ignore potential adverse affects. Our findings suggest that caregivers of color have needs that may be different from non-Latino White caregivers, particularly with respect to their own health and well-being. Thus, practice implications that are specific to the populations served may be important. Developmental disability service providers may want to develop specific programs for parental caregivers of color that address their physical and health needs. Often in the course of providing life-long care, caregivers deemphasize their own needs. Services that attend to these needs may help caregivers to understand the practical importance of taking care of their own health in order to be able to care for others. Developmental disability service providers may be able to adopt caregiver programs used in other fields. For example, service providers in the field of aging have developed support groups or psychoeducational models for caregivers of family members who have dementia or Alzheimer's disease. These programs often address issues of health and wellness for the caregivers themselves. Thus, the developmental disability service system may wish to expand the service focus beyond the needs of the person with developmental disabilities to include services that address physical and mental health needs of parental caregivers, particularly when serving Latino and Black American families.
Support for this manuscript was provided by the John A. Hartford Foundation Geriatric Social Work Faculty Scholars Program, National Institute on Disability and Rehabilitation Research funded Rehabilitation Research and Training Center on Aging with Developmental Disabilities at the University of Illinois at Chicago, and the Waisman Center at the University of Wisconsin–Madison.
Authors: Sandra Magaña, PHD (MAGANA@Waisman.Wisc.Edu), Assistant Professor, and Matthew J. Smith, MSW, Doctoral Candidate, 1500 Highland Ave., Madison, WI 53705