PRACTICE CONCEPTS
The Gerontologist Vol. 43, No. 6, 916–924
Copyright 2003 by The Gerontological Society of America
Evaluating a Practice-Oriented Service Model to Increase the Use of Respite Services Among Minorities and Rural Caregivers Julian Montoro-Rodriguez, PhD,1 Karl Kosloski, PhD,2 and Rhonda J. V. Montgomery, PhD3 Purpose: The goal of this study was to evaluate the practice-oriented model of service use (Yeatts, Crow, & Folts, 1992) relative to the more widely used behavioral model (Andersen, 1968) in its ability to explain the use of respite services by caregivers of Alzheimer’s patients. Unlike the behavioral model, which focuses primarily on characteristics of the service user, the practice-oriented model focuses primarily on characteristics of the service. Design and Methods: Interview data from 1,158 caregivers participating in the Alzheimer’s Disease Demonstration Grants to States program (Montgomery, Kosloski, Karner, & Schaefer, 2002) were analyzed. Separate regression models were estimated for adult day care and in-home respite, using the full information maximum likelihood procedure described by Arbuckle (1996), and ordinary least squares regression with listwise deletion of missing data. Results: The findings indicate that the factors related to respite use tapped by the practice-oriented model add significantly to explanatory models of service use over models that use only the factors typically represented by the behavioral model. Additional analyses, including a set of interactions with ethnicity, indicated that this improvement occurs primarily for White and Hispanic caregivers, and less so for African Americans. Implications: The findings are discussed in terms of their implications for enhancing the timely use of respite services and directions for future research. This project was supported, in part, by Contract 231-98-0010 from the Administration on Aging, Department of Health and Human Services, Washington, DC, and by a 2001 Summer Research Support grant by The Research Council of Kent State University, Kent, OH. Address correspondence to Julian Montoro-Rodriguez, Kent State University, School of Family and Consumer Studies, 142 Nixson Hall, Kent, OH 44242-0001. E-mail:
[email protected] 1 School of Family and Consumer Studies, Kent State University, OH. 2 Department of Gerontology, University of Nebraska at Omaha. 3 Gerontology Center, The University of Kansas, Lawrence.
916
Key Words: Service use, Respite, Ethnicity, Community-based services, Caregiving
The most frequently requested support service by caregivers of dementia patients is respite care (Gallagher-Thompson, 1994; Montgomery & Kosloski, 1994; Snyder & Keefe, 1985). Ironically, however, respite services, like most other communitybased support services, are often underutilized, even by caregivers who would appear to benefit from their use (Miller & Mukherjee, 1999; Strain & Blandford, 2002). After physical disability is controlled for, this is particularly the case for caregivers of dementia patients, who use even fewer community services than caregivers of nondementia patients (Pedlar & Biegel, 1999). As a result, providers of respite care are constantly confronted with the challenge of encouraging caregivers to use respite services at a therapeutic level. Clearly, understanding the conditions under which caregivers are likely to use supportive services is a requisite first step in meeting such a challenge. The most frequently used model for predicting and explaining the use of medical and social services is the behavioral model of Andersen and his colleagues (Aday & Andersen, 1974; Andersen, 1968, 1995; Andersen & Newman, 1973). The behavioral model identifies predisposing, enabling, and need factors as primary determinants of service use. Predisposing variables presumably do not cause service use directly but rather are included to acknowledge that some individuals have a greater propensity to use services than do others. Predisposing factors include such things as age, gender, marital status, race, family composition, and health beliefs. Enabling factors reflect the fact that, even when an individual is predisposed to use a service, certain conditions facilitate that use. Enabling conditions primarily include financial and community resources. With the assumption of a predisposition The Gerontologist
to use services and the ability to do so, a potential service user must also see a need for the service. In the case of medical services, need is generally assessed by the extent of illness or disease. In the case of many social services, or what Andersen referred to as ‘‘discretionary services,’’ the role of need is somewhat more complicated and interacts with predisposing factors (e.g., see Kosloski & Montgomery, 1994; Kosloski, Montgomery, & Karner, 1999). Empirical support for the behavioral model has been somewhat disappointing (see Wolinsky, 1990, for a comprehensive review). From a research perspective, the amount of variability in service use explained by the model has tended to be modest. From a practice perspective, the behavioral model has an additional drawback: its focus is almost exclusively on the characteristics of the service user. With the exception of health beliefs, virtually none of these individual characteristics is amenable to intervention by service providers. As a result (regardless of its statistical efficacy in explaining interindividual variation in service use), the behavioral model seems to be of very little practical use to service providers. Yeatts, Crow, and Folts (1992) attempted to remedy this latter shortcoming by introducing a ‘‘practice-oriented approach’’ that focused on factors that could be more readily manipulated by service providers in an effort to increase service use. This practice-oriented approach hypothesized three sets of factors affecting service use: knowledge, access, and intent. Knowledge refers to information needed by the caregiver in order to use a service. This knowledge is of three types. First, the caregiver must perceive or recognize that he or she could benefit from a supportive service (e.g., respite). This perception of need must then be followed by an awareness that a service exists that could offset this need and that the service is accessible to him or her. Finally, a caregiver must have the knowledge of the explicit steps required to enroll or sign up for the desired service. Access refers to freedom from barriers that make service use difficult or impossible. Three types of access barriers are suggested in the practice-oriented approach: (a) transportation, (b) affordability, and (c) availability. Because respite is generally of two types—in-home respite or adult day care—transportation is primarily relevant to the use of adult day care. Affordability refers to the cost of the service relative to the individual’s ability to pay. Availability pertains to getting the respite service in the amounts and at the times needed. The caregiver’s intent to use a respite service is also influenced by three factors: (a) attractiveness of the service, (b) preferences for cultural similarity, and (c) attitudes toward receiving help. The notion of attractiveness captures all of those features of a program that make it desirable to the caregiver. Vol. 43, No. 6, 2003
917
With respect to cultural similarity, Yeatts and his colleagues have noted that service use is reduced if clients must come in contact with others who are markedly different from themselves. Finally, negative attitudes toward receiving help, such as feelings of humiliation or embarrassment, are expected to reduce the likelihood of service use as well. A key advantage of using the factors in the practice-oriented model as a tool to enhance appropriate service use is that virtually all of the elements hypothesized to affect service use are amenable to intervention by service providers, either by manipulating the manner in which services are offered and advertised or through programs designed to change the attitudes and beliefs held by clients. Currently, however, there are several disadvantages associated with the practice-oriented model as well. First, and foremost, the lack of empirical testing of the practice-oriented model requires the assumption that implementing the approach will actually result in increased service use. This can be a risky undertaking in a time of scarce resources. Second, as the authors of the practice-oriented approach originally noted, the differing aspects of the model are likely to have differential relevance to service use, depending on the unique characteristics of the service under consideration. Thus, exactly how to implement the model in any specific service context remains unclear. Third, because the focus of the practiceoriented approach is on perceived barriers by potential service users, the model is likely to operate differently for members of different cultural groups. The goal of the present study was to evaluate the practice-oriented model by assessing its utility for explaining the use of respite services by caregivers of dementia patients. Two different types of respite were examined: adult day care and in-home respite. The relevance of the practice-oriented model for each type of respite service was evaluated for each of three cultural groups: African Americans, Hispanics, and Caucasians. In addition, the behavioral model was used as a standard to judge the relative efficacy of the practice-oriented model. This comparison seemed appropriate for two reasons. First, Yeatts and his colleagues, in their original statement of the practice-oriented model, used the behavioral model as a conceptual frame of reference in the explication of their model. From this perspective, it made sense to use the behavioral model as an empirical frame of reference as well. Second, the behavioral model remains the dominant model for studying the use of supportive services by older adults (Calsyn & Winter, 2000), and the contribution of any new approach can be judged according to its ability to improve on the status quo. This was accomplished in the present study by determining whether the practice-oriented model explained variation in respite use over and above that explained by the behavioral model.
Methods Study Participants Data for the analyses were taken from 1,158 interviews with family caregivers participating in the Alzheimer’s Disease Demonstration Grants to States (ADDGS) program (Montgomery, Kosloski, Karner, & Schaefer, 2002). The goal of this program is to provide or expand support services for traditionally underserved or hard-to-serve Alzheimer’s patients and their caregivers, especially minorities and rural residents. The demonstration was conducted in South Carolina, Washington, DC, Maine, Puerto Rico, Michigan, Ohio, Florida, Maryland, Montana, California, Washington, North Carolina, and Hawaii. Although there was great diversity among participating states in programming and services offered through the demonstration, at least 50% of the grant award had to be used for respite services. Data were collected in 1996 through telephone interviews with 1,183 caregivers. A structured interview protocol was used to elicit caregivers’ attitudes about their use of respite services. Of 1,183 caregivers, 540 were White, 323 were Hispanic, and 320 were African American. The average age of the caregivers was 61 years and the average age of elders was 80. Of the caregivers, 36% (n¼422) were spouses, 51% (n ¼ 606) were adult children, and 13% (n ¼ 155) were others. The average health rating of caregivers was 2.8 on a scale ranging from 1 (poor health) to 5 (perfect health). The need for assistance on the part of elders with activities of daily living (ADLs) was moderate with a mean of 4.7 on a scale ranging from 0 (no need) to 10 (high need). There were 465 users of adult day care, 693 users of in-home respite, and an additional 25 caregivers who used residential (short stay) respite in nursing homes. Because of the small sample size, the latter group was not included in the analyses, leaving a total sample size of 1,158. Variable Measurement Conceptualization and Operationalization.—In evaluating the measurement of variables, there is an important caveat. Data for the analyses were collected as part of a broader study of client satisfaction with respite services in the ADDGS demonstration. The items were not designed specifically to evaluate either the practice-oriented or behavioral model. As a result, some of the variables may not represent the hypothesized constructs in these models as directly as might otherwise be the case. Nonetheless, all of the variables appear to be plausible representations of the relevant underlying constructs in the two models and thus offer at least a preliminary comparison of the relative efficacy of the two models in explaining respite use. Dependent Variable.—The dependent variable was the number of times in the past month that the 918
caregiver had used respite services in the ADDGS demonstration. These data were taken from actual service use records kept by the service providers participating in the demonstrations. Typically, caregivers used either adult day care or in-home respite. A small proportion of caregivers used both types of services. In the present study, separate models of service use are estimated for each type of respite service, because different supportive services are known to have different predictive models (Kosloski & Montgomery, 1994). On average, caregivers reported using adult day care and in-home respite approximately the same number of times (10.7) in the past month. The Behavioral Model.—The behavioral model included predisposing factors, enabling factors, and need factors. Predisposing factors in the behavioral model were represented by a set of five variables: ethnicity, relationship of the caregiver to the elder, caregiver’s age, caregiver’s gender, and caregiver’s education. Ethnicity was evaluated by using a set of three dummy variables comparing African Americans and Hispanics to Whites. Relationship to the elder was also represented by using a set of three dummy variables that compared spouses and children with ‘‘others’’ (i.e., nonspouses and nonchildren). Gender was coded 1 (women) or 0 (men). Education was coded into one of six categories ranging from 1 (less than high school) to 6 (graduate study) and was treated as an ordinal variable. The age of the caregiver was coded directly. Two variables representing enabling factors were available: income and geographic location. Income was coded by using five categories ranging from 1 (under $5,000) to 5 (over $50,000) and was treated as an ordinal variable. Geographic location contrasted residence in urban and suburban areas having total populations greater than 50,000 (1) with geographic residence in smaller population centers (,50,000) and rural locations (0). Need was assessed by using ADLs and instrumental ADLs (IADLs). ADLs included bathing, using the toilet, dressing, moving in and out of chairs, and eating. IADLs included shopping, using the telephone, preparing meals, taking medicine, doing housework, maintaining dental care, and using transportation. Item responses were coded from 0 (no help needed) to 2 (a lot of help needed–cannot do at all) and formed into additive composites to represent the ADL and IADL factors. Reliabilities (Cronbach’s alpha) of the two measures were .87 and .84, respectively. The Practice-Oriented Model.—The practice model included knowledge and access. In terms of knowledge, perceived need for the particular service was measured with a single item that contrasted the caregiver’s preference for using the respite service versus asking his or her family for help. Service knowledge or awareness that a useful service exists The Gerontologist
was measured with this item: ‘‘It is difficult to get the services my [relative] needs because I do not know where to find them.’’ Knowledge of how to acquire the particular service was measured by using this item: ‘‘The application process for the program is too difficult.’’ Each item was rated on a 5-point scale ranging from 1 (not at all–never) to 5 (extremely– always). Access refers to three separate factors in the practice-oriented approach: transportation, availability of the service, and affordability. In the case of respite, transportation for in-home respite is typically not a problem for the caregiver because the respite worker comes to the caregiver’s home. Therefore, users of in-home respite were not asked about transportation. With adult day care, however, the responsibility for getting the patient to the day care site often falls on the caregiver. For purposes of the present study, caregivers who used adult day care were asked two separate questions about transportation. The first was, ‘‘Does the adult day care program provide transportation for you if you need it?’’ Responses were coded 1 (yes) or 0 (no). The second was, ‘‘Is it a problem for you to get your relative to and from the adult day care center?’’ The latter item was coded on a 5-point scale ranging from 1 (always) to 5 (never). Because these two items tap different aspects of the transportation issue, their correlation was low (r ¼ .06) and they were treated as separate factors. The affordability of the respite service was assessed by using this question: ‘‘How reasonable is the service fee of the program for your family budget?’’ Reasonableness was coded from 1 (not at all) to 5 (extremely). The availability of respite was measured by asking the respondents to what extent they agreed that ‘‘we can get the amount of [respite] care that we need (number of hours and number of visits),’’ coded on a 5-point scale ranging from 1 (not at all) to 5 (completely agree). In the practice-oriented model, intent to use services encompasses three factors: attractiveness, cultural differences, and negative attitudes toward receiving help. Attractiveness of the service refers to those features of a program that make it appealing to the caregiver. A three-item composite was created to tap this issue: (a) ‘‘how important is this care to you?,’’ (b) ‘‘how helpful is the [program] to you?,’’ and (c) ‘‘the program is beneficial to me and my family.’’ Responses were coded on a 5-point scale from 1 (not at all) to 5 (extremely). Reliability (Cronbach’s alpha) of the composite variable was .83. The extent of cultural similarity was assessed by a single item, ‘‘the volunteers or workers are from cultures very different from mine,’’ which was assessed on a 5-point scale ranging from 1 (not at all) to 5 (completely). Negative attitudes toward receiving help were assessed by using two composite variables. The first Vol. 43, No. 6, 2003
919
variable tapped feelings of embarrassment or stigmatization that may affect the caregiver’s preference for using out-of-the-home versus in-home services. It was composed of four items: ‘‘it is embarrassing to take my [relative] out in public,’’ ‘‘it is easy for me to have guests in my home while my [relative] is there,’’ ‘‘I find myself being selfconscious when I am in public with my [relative],’’ and ‘‘I avoid inviting company to my home because of my [relative’s] condition.’’ The items were measured by using a 5-point scale ranging from 1 (not at all) to 5 (completely) and were combined to create a composite variable with a reliability of .64. The second variable focused on attitudes toward receiving help from the government. A two-item composite was formed from the following items: ‘‘the government should provide more money for [program] services’’ and ‘‘I believe the government should help families care for persons at home.’’ These two items were measured by using a 5-point scale ranging from 1 (not at all) to 5 (completely agree) and formed into a composite with an estimated reliability of .66. Analysis Strategy.—As a way to evaluate the utility of the practice-oriented approach, relative to the behavioral model, the variables representing the behavioral model were evaluated as a set, using multiple regression analysis. The variables representing the practice-oriented approach were then added to the model, as a set, to determine whether they improved the model. This procedure was performed for each type of respite use (i.e., in-home respite vs. adult day care). For the most part, the proportion of missing data for individual items was generally small. Aggregated across all of the items to be evaluated, however, the amount of missing data became considerable. Therefore, the models in this study were estimated two ways: first, by using the full information maximum likelihood (FIML) procedure described by Arbuckle (1996), and second, by using ordinary least squares regression with listwise deletion of missing data. Although there were a few minor discrepancies between the two approaches in the significance tests of individual predictors that were most likely due to sample size differences, the broader substantive findings were the same regardless of the approach used. Because the FIML approach has been shown to be superior to other approaches used to address missing data under a broad range of conditions (Arbuckle, 1996), the results from the FIML analyses are presented here. Results Adult Day Care A comparison of the relative efficacy of the practice-oriented model relative to the behavioral
Table 1. Hierarchical Regression of In-Home Respite Use on the Sets of Behavioral and Practice-Oriented Variables Adult Day Care (n ¼ 465)
Dependent Variable: Service Use Independent Variables Behavioral model Predisposing variables Hispanics African Americans Caregiver age Caregiver gender Spouse Adult children Caregiver education
b
Model I (SE) b
2.2 5.8 0.01 1.3 0.81 1.0 0.04
(1.2) .11 (1.0) .32* (0.03) .02 (0.91) .07 (1.3) .04 (1.1) .06 (2.6) .00
b
In-Home Service (n ¼ 693)
Model II (SE) b
2.7 5.7 0.00 1.1 0.85 0.41 0.00
(1.1) .14* (1.0) .31* (0.03) .01 (0.86) .05 (1.2) .04 (1.0) .02 (0.23) .00
b
Model I (SE) b
3.6 2.2 0.02 1.1 1.8 0.85 0.15
(0.99) .19* (0.96) .11* (0.03) .04 (0.78) .05 (1.1) .10 (1.0) .05 (0.23) .03
b
Model II (SE) b
3.1 1.4 0.02 1.5 2.0 0.75 0.26
(1.0) (0.95) (0.03) (0.78) (1.1) (1.0) (0.22)
.17* .07 .03 .07* .11 .04 .05
Enabling variables Caregiver income Geographic location
0.71 (0.45) .09 0.98 (1.0) .06
0.91 (0.44) .12* 1.2 (0.98) .07
0.90 (0.43) .10* 1.9 (0.85) .11*
0.77 (0.42) .08 1.1 (0.84) .06
Need variables Activities of daily living Instrumental activities of daily living
0.02 (1.7) .00 0.07 (0.14) .02
0.02 (0.16) .00 0.04 (0.13) .01
0.41 (0.13) .15* 0.19 (0.14) .07
0.23 (0.13) .09 0.08 (0.14) .03
Practice-oriented model Knowledge Of need Of services Of procedures Access Transportation by program Transportation by caregiver Affordability Availability Intent Attractiveness Cultural differences Attitudes: stigma Attitudes: government’s role Pseudo R2 v2 change
0.07 (0.24) .00 0.64 (0.28) .10* 0.12 (0.37) .01 2.2 0.49 0.18 0.30 0.86 0.03 0.02 0.13 .10 47.6*
(0.76) (0.32) (0.38) (0.28)
0.16 (0.21) .03 0.31 (0.27) .04 0.40 (0.33) .05
.13* .07 .02 .05
— — 0.55 (0.26)
(0.18) .23* (0.26) .05 (0.10) .01 (0.22) .02 .21 57.5*
0.37 0.37 0.26 0.40 .08 58.0*
.08*
(0.14) .10* (0.20) .07 (0.10) .10* (0.20) .07* .14 42.1*
*p , .05.
model in explaining the use of adult day care and inhome respite is shown in Table 1. With respect to use of adult day care, the variables from the behavioral model, when added as a first step (Model I), improved model fit over the null model significantly: v2(11) ¼ 47.6, p , .01. Only African-American status (relative to White status) emerged as a significant covariate of respite use in Model I. The overall fit of the model for adult day care use (Model II) improved significantly when the practice-oriented variables were added: v2(11) ¼ 57.5, p , .01. Knowledge of services, whether transportation was provided by the adult day care program, and the perceived attractiveness of the program were all significant correlates of adult day care use. In addition, once the practiceoriented variables were controlled (Model II), Hispanic status and caregiver income also emerged as significant covariates of adult day care use in the set of variables comprising the behavioral model. 920
In-Home Respite For use of in-home respite, the variables of the behavioral model, when considered separately (Model I), improved model fit over the null model significantly: v2(11) ¼ 58.0, p , .01. Both Hispanic and African-American ethnicity were positively related to in-home respite use relative to White status. The caregiver’s income, geographic location, and ADL status were also significantly related to inhome respite use. When the practice-oriented variables were added (Model II), the overall fit of the model for in-home respite use improved significantly: v2(9) ¼ 42.1, p , .01. The availability of the service, its attractiveness to the caregiver, the stigma experienced by the caregiver, and the caregiver’s beliefs about the government’s role in supporting family caregivers were all significantly associated with the use of in-home The Gerontologist
respite. In addition, when the practice-oriented variables were controlled, an additional behavioral variable, caregiver’s gender, emerged as a significant covariate of in-home use as well. The Relevance of the Practice-Oriented Model for Differing Cultural Groups Yeatts and his colleagues point out that the study of specific minority groups is necessary to identify their differing needs and preferences. This admonition suggests that the variables in the practiceoriented model are likely to perform differently for differing ethnic groups. As a way to test this hypothesis, a set of 22 interaction terms, representing interactions between ethnicity and the practiceoriented variables, was entered into each of the previous models as a third step for both the adult day care and in-home respite models. When the set of 22 interactions was added to the model of adult day care use, model fit improved significantly: v2(22) ¼ 52.8, p , .01. In the case of in-home respite, the set of interactions also improved the model significantly, v2(18) ¼ 38.2, p , .01, indicating that separate models for each cultural group would be more informative. The regression of adult day care use on the behavioral and practice-oriented variables is shown for each of the three ethnic groups separately in Table 2. In the case of African Americans, the behavioral variables, when entered first, improve model fit significantly: v2(9) ¼ 17.9, p , .05. The set of practice-oriented variables did not improve overall model fit: v2(11) ¼ 15.0, ns. Of the practiceoriented variables, only attractiveness of the service was significantly related to use of adult day care among African Americans. For Hispanic and White caregivers, the pattern was reversed. The behavioral variables did not improve fit over the null model: v2(9) ¼ 14.9 and 7.0, respectively. Adding the practice-oriented variables, however, improved the model fit for Hispanics and Whites, v2(11) ¼ 56.6 and 48.9, p , .05, respectively. For both Hispanics and Whites, the attractiveness of the day care program was significantly related to use. In addition, for Whites, the availability of the day care service was also significantly related to use. The regression of in-home respite use on the behavioral and practice-oriented variables is shown for each of the three ethnic groups in Table 3. Once again, the practice-oriented model performed poorly for African Americans. The set of behavioral variables, when entered first, improved fit over the null model, v2(9) ¼ 23.1, p , .05, but adding the practice-oriented variables did not improve the model significantly, v2(9) ¼ 6.4, ns. For both Hispanics and Whites, however, the behavioral variables and the practice-oriented variables were both significantly related to use of in-home respite. Vol. 43, No. 6, 2003
921
In the case of Hispanics, the chi-square contribution for the set of behavioral variables was v2(9) ¼ 38.2, p , .05, with a significant improvement added by the practice-oriented variables, v2(9) ¼ 31.3, p , .05. The pattern was similar for Whites with the set of behavioral variables improving fit over the null model significantly, v2(9) ¼ 21.3, p , .05, followed by a significant improvement once the practiceoriented variables were added, v2(9) ¼ 30.3, p , .05. Discussion The findings indicate that the practice-oriented model put forth by Yeatts and colleagues (Yeatts et al., 1992) may be a useful tool for developing intervention strategies to enhance the appropriate use of respite services for particular groups of caregivers. The results indicate that the factors related to respite use tapped by the practice-oriented model add significantly to explanatory models of service use over models that use only the factors typically represented by the behavioral model. This was true for both adult day care and in-home respite. Additional analyses, however, indicated that this improvement occurs primarily for White and Hispanic caregivers. The value of the practice-oriented model lies in the fact that, unlike the predisposing, enabling, and need factors of the behavioral model, the factors included in the practice-oriented model have the potential to be manipulated to encourage service use. For example, with respect to adult day care, the availability of the respite service (i.e., preferred times and amounts) was a significant determinant of use among Whites. Thus, programs with large constituencies of White caregivers might benefit by offering greater flexibility in the times and amount in which respite is offered. In addition, the attractiveness of the program, defined in terms of its perceived importance and helpfulness to the caregiver, was positively related to service use for all three ethnic groups. One limitation of the present study is that the specific elements that may serve to define ‘‘attractiveness’’ for each ethnic group are relatively undefined. The finding does, however, suggest an avenue for future research. In the case of in-home respite, timing and availability of respite in desired amounts was once again important for White caregivers. Knowledge of need for a formal respite service, defined as a preference for the demonstration respite service over asking one’s family for help, was also important for White caregivers, suggesting that the choice of formal versus informal assistance is more salient for them than the other groups. In contrast, knowledge about the specific procedures necessary to acquire the respite service was more important for Hispanics. This latter finding is consistent with previous research indicating that Latinos face numerous
Table 2. Regression of Adult Day Care Service Use on the Practice-Oriented Variables by Ethnicity Dependent Variable: Service Use Independent Variables Behavioral model Predisposing variables Caregiver age Caregiver gender Spouse Adult children Caregiver education
African Americans (n ¼ 122) b
0.06 2.5 0.78 0.06 0.05
(SE)
b
(0.06) .11 (1.7) .13 (2.3) .04 (1.8) .04 (0.53) .01
Hispanic Americans (n ¼ 103) b
0.14 0.02 2.0 1.2 0.85
(SE)
b
(0.08) .21 (2.1) .00 (3.3) .09 (3.1) .06 (0.58) .13
White Americans (n ¼ 240) b
0.02 0.48 1.4 0.66 0.16
(SE)
b
(0.04) .04 (1.0) .03 (1.5) .10 (1.4) .04 (0.28) .03
Enabling variables Caregiver income Geographic location
1.7 (0.86) .24* 1.5 (1.9) .08
0.08 (2.0) .00 0.77 (9.3) .00
0.60 (0.50) .09 0.09 (1.0) .06
Need variables Activities of daily living Instrumental activities of daily living
0.18 (0.34) .05 0.39 (0.22) .18
0.04 (0.47) .01 0.64 (0.38) .19
0.09 (0.19) .03 0.17 (0.16) .08
0.10 (0.48) .02 0.36 (0.65) .05 0.57 (0.76) .07
0.35 (0.62) .05 0.22 (0.63) .03 0.62 (0.82) .07
0.41 (0.28) .09 0.26 (0.35) .04 0.84 (0.59) .10
Access of services Transportation by program Transportation by caregiver Affordability Availability
1.7 1.1 0.00 0.54
(1.5) .10 (0.69) .15 (0.85) .00 (0.59) .09
0.39 1.2 0.55 1.2
(2.3) .01 (0.73) .19 (0.89) .07 (0.89) .22
1.4 0.36 0.61 1.4
(0.84) .10 (0.44) .05 (0.52) .08 (0.31) .29*
Intent of services Attractiveness Cultural differences Attitudes: stigma Attitudes: government’s role
0.81 0.01 0.17 0.80
(0.41) .20* (0.59) .00 (0.28) .06 (0.45) .15
(0.41) (0.62) (0.30) (0.71)
0.68 0.15 0.01 0.15
(0.20) .21* (0.30) .03 (0.11) .00 (0.24) .03
Practice-oriented model Knowledge of services Of need Of services Of procedures
v2 (behavioral variables) on 9 df v2 increment (practice-oriented variables) on 11 df Cumulative model v2 on 20 df
1.7 0.43 0.21 1.1
.46* .07 .07 .16
17.9*
14.9
7.0
15.0 32.9*
56.6* 71.5*
48.9* 55.9*
*p , .05.
barriers in accessing services, including low incomes and language barriers (Andersen, Giachello, & Aday, 1986; see also Solis, Marks, Garcia, & Shelton, 1990; Wallace, Levy-Storms, Kington, & Andersen, 1998). In the present study, however, income is controlled directly and language barriers are controlled by asking whether the providers are from different cultures. Thus, it seems more likely that the difficulties in accessing services faced by Hispanics in the present sample are related to an element of acculturation, specifically to knowledge of how to navigate the social service system. This hypothesis is supported by the finding that procedural knowledge was important in accessing in-home respite but not adult day care. In the ADDGS, adult day care is usually characterized by a single point of entry. In contrast, in-home services are more complicated and may require accessing different types of providers for different types of needs (e.g., home health agencies 922
for certain types of ADL limitations vs. companionship programs for others), all of which may be regulated by local statute. As a result, where service use was predicated on higher levels of procedural knowledge, Hispanics were especially disadvantaged. Finally, for Whites, negative attitudes and the stigmatization associated with Alzheimer’s disease was also a deterrent to in-home respite use. Each of these variables offers insights for service providers regarding intervention strategies that might enhance the timely use of respite care. For example, it may be useful to stress the advantages associated with seeking help from formal rather than informal caregivers, when efforts are made to reach White caregivers. The findings also suggest that White caregivers will benefit from greater knowledge about normative behavioral changes associated with Alzheimer’s disease that produce problematic behaviors. In contrast, informational programs about the The Gerontologist
Table 3. Regression of In-Home Respite Use on the Practice-Oriented Variables by Ethnicity Dependent Variable: Service Use Independent Variables Behavioral model Predisposing variables Caregiver age Caregiver gender Spouse Adult children Caregiver education Enabling variables Caregiver income Geographic location
African Americans (n ¼ 187) b
0.12 2.9 2.3 2.2 0.52
(SE)
b
(0.06) .19* (1.6) .14 (1.8) .13 (1.7) .13 (0.50) .10
Hispanic Americans (n ¼ 219) b
0.02 0.72 1.7 1.5 0.66
(SE)
b
(0.05) .03 (1.4) .03 (1.7) .09 (1.6) .08 (0.40) .12
0.84 (0.89) .09 0.96 (1.4) .05
2.0 (1.0) .16* 4.7 (1.5) .23*
0.47 (0.24) .18* 0.19 (0.25) .07
0.23 (0.24) .08 0.29 (0.28) .09
0.46 (0.39) .09 0.70 (0.55) .10 0.32 (0.59) .04
0.54 (0.36) .09 0.15 (0.48) .02 1.7 (0.60) .23*
Access of services Transportation by program Transportation by caregiver Affordability Availability
— — 0.49 (0.64) .06 0.10 (0.42) .08
— — 0.66 (0.36) .12 0.84 (0.45) .13
Intent of services Attractiveness Cultural differences Attitudes: stigma Attitudes: government’s role
0.08 0.15 0.12 0.27
0.21 0.33 0.41 0.64
Need variables Activities of daily living Instrumental activities of daily living Practice-oriented model Knowledge of services Of need Of services Of procedures
v2 (behavioral variables) on 9 df v2 increment (practice-oriented variables) on 9 df Cumulative model v2 on 18 df
(0.29) .02 (0.41) .03 (0.23) .04 (0.40) .05
(0.26) .06 (0.36) .06 (0.22) .13 (0.56) .06
White Americans (n ¼ 287) b
0.14 0.89 1.2 2.3 0.30
(SE)
b
(0.05) (1.0) (1.9) (1.7) (0.31)
.24* .04 .07 .12 .06
0.26 (0.51) .03 3.0 (1.4) .13* 0.09 (0.19) 0.02 (0.19)
.03 .00
0.68 (0.32) .12* 0.11 (0.38) .01 0.02 (0.53) .03 — — 0.30 (0.52) 1.1 (0.32) 0.16 0.23 0.24 0.33
.03 .22*
(0.23) .04 (0.29) .04 (0.12) .12* (0.24) .08
23.1*
38.2*
21.3*
6.4 29.5*
31.3* 69.5*
30.3* 51.6*
*p , .05.
procedures necessary to acquire in-home respite would appear to be most useful to Hispanic caregivers. In interpreting these findings, it is important to keep in mind two issues with respect to the representation of the underlying constructs of the practice-oriented model. The first issue deals with the extent to which the constructs are uniformly well represented by the measured variables. For example, in the practice-oriented model, an important construct, ‘‘service knowledge,’’ refers to awareness that a useful service exists. In the present study, this construct seems well represented by this item: ‘‘it is difficult to get the services my relative needs because I do not know where to find them.’’ In contrast, ‘‘affordability’’ in the practice-oriented model refers to the cost of a service relative to the user’s income. Affordability was less directly measured by asking the respondent this question: ‘‘how reasonable is the Vol. 43, No. 6, 2003
923
service fee of the program for your family budget?’’ Clearly, reasonableness is not the same thing as affordability, although both attempt to take into account available family resources. The essential point is that valid operationalization of constructs is a critical element in a rigorous test of the practiceoriented model. The second issue is somewhat broader in that, regardless of how adequately the measured variables in this study represent the practice-oriented model, the fact remains that they explain interindividual variation in respite use. For example, as affordability (i.e., reasonableness) increases, so does the use of inhome respite services (Table 1). From that viewpoint, the findings are important from a practice perspective because they are potentially amenable to manipulation. Whereas family income, an enabling variable in the behavioral model, is outside the control of a service provider, affordability is not
(e.g., using sliding fee-for-service scales). Consequently, even if the measured variables in the present study do not fall within the parameters of the practice-oriented model, they have practical significance nonetheless. Several other considerations seem relevant in interpreting the present findings as well. First, these findings are based solely on data from dementia caregivers. Respite, though, is a general service intended for all informal caregivers. Moreover, the behavioral and practice-oriented models of service use are also quite general in that they are not disease specific and do not target specific populations of caregivers. Nonetheless, the caregiving experience is likely to differ depending on the disease trajectory and care requirements of the care recipient. As a result, the factors related to respite use could very well be moderated by the disease state and care needs of the elder, and caution must be used in generalizing the results of this study to nondementia populations. Second, in addition to interacting with disease state, the variables in the practice-oriented model may also be related to respite use differently, depending on the relationship of the caregiver to the elder. For example, the behavioral model has been shown to explain service use differentially, depending on whether the caregiver is the elder’s spouse or adult child (Kosloski & Montgomery, 1994). Taking into account the relationship of the caregiver to the elder may also enhance the explanatory power of the factors represented by the practice-oriented approach and remains an avenue for future research. Third, although the practice-oriented model adds substantially to the explanatory power of the behavioral model, both models seem to neglect other important factors, especially attitudinal factors and respite outcomes. For example, trust in service providers is related to service use (Pedlar & Biegel, 1999), and caregiving mastery and preparedness is related to overall confidence in services (Miller & Mukherjee, 1999). Exactly how to position such variables conceptually into the behavioral and practice-oriented models remains a topic for future consideration. Finally, any enthusiasm for the practice-oriented approach must be tempered by the finding that it does not explain variability in respite use for African Americans beyond that offered by the behavioral model. This does not mean, however, that the practice-oriented model is completely irrelevant to African Americans. When the practice-oriented variables are entered as a first step in the model predicting use of adult day care, the variables, as a set, improve the null model significantly: v2(11) ¼ 20.5, p , .05. Because the practice-oriented variables appear to offer greater opportunity for intervention, they may deserve closer attention, even by adult day care providers serving African-American constituen924
cies. In contrast, in the case of in-home respite, the practice-oriented variables, as defined in this study, were largely irrelevant for African Americans in that they did not even improve the null model: v2(9) ¼ 10.4, ns. Thus, the behavioral model remains the model of choice in explaining the use of in-home respite among African-American caregivers. References Aday, L., & Andersen, R. (1974). A framework for the study of access to medical care. Health Services Research, 9, 208–220. Andersen, R. M. (1968). A behavioral model of families’ use of health services. Chicago: Center for Health Administration Studies. Andersen, R. M. (1995). Revising the behavioral model and access to medical care: Does it matter? Journal of Health and Social Behavior, 36, 1–10. Andersen, R., Giachello, A., & Aday, L. (1986). Access of Hispanics to health care and cuts in services: A state-of-the-art overview. Public Health Reports, 101, 238–252. Andersen, R., & Newman, J. (1973). Societal and individual determinants of medical care utilization in the United States. Milbank Memorial Fund Quarterly, 5, 95–124. Arbuckle, J. (1996). Full information estimation in the presence of incomplete data. In G. A. Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling: Issues and techniques (pp. 243–277). Mahwah, NJ: Erlbaum. Calsyn, R., & Winter, J. (2000). Predicting different types of service use by the elderly: The strength of the behavioral model and the value of interaction terms. Journal of Applied Gerontology, 19, 284–303. Gallagher-Thompson, D. (1994). Direct services and interventions for caregivers: A review of extant programs and a look to the future. In M. H. Cantor (Ed.), Family caregiving: Agenda for the future (pp. 102– 122). San Francisco: American Society on Aging. Kosloski, K., & Montgomery, R. J. V. (1994). Investigating patterns of service use by families providing care for dependent elders. Journal of Aging and Health, 6, 17–37. Kosloski, K., Montgomery, R., & Karner, T. (1999). Differences in the perceived need for assistive services by culturally diverse caregivers of dementia patients. Journal of Applied Gerontology, 18, 239–256. Miller, B., & Mukherjee, S. (1999). Service use, caregiving mastery, and attitudes toward community services. Journal of Applied Gerontology, 18, 162–176. Montgomery, R., & Kosloski, K. (1994). Outcomes of family caregiving: Lessons from the past and challenges for the future. In M. H. Cantor (Ed.), Family caregiving: Agenda for the future (pp. 123–136). San Francisco: American Society on Aging. Montgomery, R., Kosloski, K., Karner, T., & Schaefer, J. (2002). Initial findings from the evaluation of the Alzheimer’s Disease Demonstration Grants to States Program. Home Health Care Services Quarterly, 21, 5– 32. Pedlar, D., & Biegel, D. (1999). The impact of family caregiver attitudes on the use of community services for dementia care. Journal of Applied Gerontology, 18, 201–221. Snyder, B., & Keefe, K. (1985). The unmet needs of family caregivers for frail and disabled adults. Social Work in Health Care, 10, 1–14. Solis, J., Marks, G., Garcia, M., & Shelton, D. (1990). Acculturation, access to care, and use of preventive services by Hispanics: Findings from HHANES 1982–84. American Journal of Public Health, 80(Suppl.), 11– 19. Strain, L., & Blandford, A. (2002). Community-based services for the taking but few takers: Reasons for nonuse. Journal of Applied Gerontology, 21, 220–235. Wallace, S., Levy-Storms, L., Kington, R., & Andersen, R. (1998). The persistence of race and ethnicity in the use of long-term care. Journal of Gerontology: Social Sciences, 53B, S104–S112. Wolinsky, F. (1990). Health and health behavior among elderly Americans: An age stratification perspective. New York: Springer. Yeatts, D., Crow, T., & Folts, E. (1992). Service use among low-income minority elderly: Strategies for overcoming barriers. The Gerontologist, 32, 24–32.
Received May 2, 2002 Accepted October 11, 2002 Decision Editor: Eleanor S. McConnell, RN, PhD
The Gerontologist
