| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
Correspondence: Address correspondence to Kathleen A. McGinnis, University Center for Social and Urban Research, 121 University Place, Pittsburgh, PA 15260. E-mail: kathym{at}pitt.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Race or ethnicity • Concordance • Caregivers • CES-D • RMBPC •
Communication is an important aspect of intervention studies as well, and poor communication between study participants and study staff also may affect outcomes in intervention studies. A study on youth and their families in multisystematic therapy found that adherence ratings were higher when caregiver and therapist were ethnically matched (Schoenwald, Halliday-Boykins, & Henggeler, 2003). Another study found better youth outcomes when youth caregivers were ethnically matched to therapists in a multisystemic therapy (Halliday-Boykins, Schoenwald, & Letourneau, 2005). We are not aware of any studies that assess the effects of racial or ethnic concordance between interventionists and elderly participants on study outcomes. If racial or ethnic concordance does affect outcomes in intervention studies, then all participants may not be receiving the same benefits from study enrollment.
We analyzed data from the Resources for Enhancing Alzheimer's Caregiver Health (REACH) program, a study designed to test psychosocial interventions for maintaining the health and functioning of family caregivers for individuals with mild or moderate impairment from Alzheimer's disease (Schulz et al., 2003; Wisniewski et al., 2003). Because the caregiving experience in minority families is a particularly neglected research area, the REACH study emphasized the inclusion of African-American and Latino caregivers. Site-specific assessments and interventions were tailored to meet the needs of racially or ethnically diverse populations (Gallagher-Thompson et al., 2003; Schulz et al.). Sites recruiting Latino caregivers also recruited Latino interventionists so that Latino caregivers could be offered the intervention in Spanish or English. Interventionists in the REACH program provided various psychosocial interventions to caregivers in the intervention groups. The REACH program enrolled 1,222 caregiver–care recipient dyads from 1996 to 2000. Of the caregivers, 56.0% were White, 24.2% were African American, 19.0% were Latino, and 0.8% were other race or ethnicity (Wisniewski et al.).
This analysis involves the 694 caregivers in the REACH study who were randomized to an active treatment group and received a face-to-face intervention from one of the 36 interventionists at one of five REACH intervention sites. Our purpose in this article is to assess attrition and changes in caregiver depression and burden by racial or ethnic concordance between caregivers and interventionists. At the 12-month follow-up, we hypothesize that caregivers with an interventionist of the same race or ethnicity will have lower attrition rates and greater decreases in caregiver depression and burden than will caregivers with an interventionist of a different race or ethnicity.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Caregivers were randomized into either an intervention or control group at each site. Site-specific interventions typically lasted 6 months, with booster sessions added at some sites between 6 and 12 months. Interventions were designed to reduce caregiver depression and burden, the main outcomes of REACH (Wisniewski et al., 2003). All interventions were based on basic health-stress models with a goal of changing the nature of specific stressors, the caregivers' appraisal of stressors, or the caregivers' response to the stressors. Caregivers were encouraged to remain in REACH even after their care recipient was institutionalized or died.
Caregivers were not randomized to interventionists, but rather assigned on the basis of scheduling considerations (timing of interventions and availability of caregiver) and language preference. Bilingual staff was employed at Palo Alto and Miami with attempts made to match the race or ethnicity of interventionists and caregivers at those sites (Wisniewski et al., 2003). Cultural sensitivity training was provided to all interviewers and interventionists at all sites (Coon et al., 2004).
The research protocol was approved by the institutional review boards of all participating universities. Written informed consent was obtained from all caregivers who participated in this study.
Data
Out of 1,222 caregiver–care recipient dyads in the REACH study, we excluded 491 who were randomized into the control group or received only telephone-based interventions. Of the remaining 731 dyads, 705 had at least one face-to-face contact with an interventionist. We excluded 5 of these 705 caregivers who were of a race or ethnicity other than White, African American, or Latino, as well as 2 caregivers who had the same number of meetings with interventionists of the same and different race or ethnicity and 4 caregivers who died prior to the 12-month follow-up. Because some caregivers had more than one interventionist, we identified the interventionist who worked with each caregiver most often. Of the 694 caregivers in this analysis, 69.0% had the same interventionist for every contact, 23.5% had more than one interventionist but the interventionists were of the same race or ethnicity, and 7.5% had more than one interventionist and the interventionists were of differing race or ethnicity. Race or ethnicity of caregivers and interventionists is based on self-response to this question: "How would you describe your primary racial or ethnic group?" The possible responses were "White, Caucasian," "Black, African American," "Native American, Eskimo, Aleut," "Asian or Pacific Islander," or "Hispanic, Latino." We code racial or ethnic concordance of the caregiver and interventionist as 1 if the caregiver and interventionist are of the same race or ethnicity and 0 otherwise.
We considered caregivers to be lost to follow-up if they refused further participation in the study or did not have follow-up survey data at the 12-month follow-up, even if the caregivers' care recipient died or was institutionalized.
We measured depressive symptoms by using the Center for Epidemiologic Studies–Depression (CES-D) scale, a 20-item self-report measure that asks about the frequency of depressive symptoms (Radloff, 1977). CES-D scores range from 0 to 60, with higher values indicating more depressive symptoms. A positive change in CES-D (baseline to 12-month follow-up) indicates that depression decreased over time.
We assessed caregiver burden by using the Revised Memory and Behavior Problems Checklist (RMBPC), which characterizes memory and behavior problems in patients with dementia and the extent to which caregivers are bothered by them (Teri et al., 1992). Scores range from 0 to 96, with a higher score indicating that memory and behavior problems are more bothersome for the caregiver. A positive change in RMBPC (baseline to 12-month follow-up) indicates that burden decreased over time. The RMBPC was collected only at follow-up interviews at which the care recipient was still living at home, whereas the CES-D scale was used even when the care recipient was institutionalized or died. Care recipient dependence was assessed with the index for activities of daily living (ADL) at the time of study enrollment. ADL quantifies the level of dependence in performing the following tasks: bathing, dressing, using the toilet, transferring out of a bed or chair, maintaining continence, and eating. Scores range from 0 to 6, with higher scores indicating more dependence (Katz, Ford, Moskowitz, Jackson, & Jaffe, 1963).
Statistical Analysis
Descriptive Analysis
We compared baseline characteristics and 12-month outcomes by caregiver race or ethnicity by using chi-square tests for categorical variables, analysis of variance for approximately normally distributed variables, and Kruskal–Wallis tests for nonnormally distributed continuous variables. We compared baseline caregiver characteristics and 12-month outcomes by racial or ethnic concordance of interventionists and caregivers by using chi-square tests for categorical variables, t tests for approximately normally distributed variables, and Wilcoxon rank-sum tests for nonnormally distributed continuous variables.
Multivariable Modeling
We assessed the association of caregiver loss to follow-up at 12 months with racial or ethnic concordance by using a random effects logistic regression model. We compared changes in CES-D and RMBPC scores at 12 months between racial or ethnic concordance groups by using generalized linear random effects regression models (Rabe-Hasketh, Pickles, & Taylor, 2000) for those caregivers who were still actively caring for care recipients at 12 months (i.e., the care recipient had not died or been institutionalized). Unlike baseline CES-D and RMBPC scores, changes in CES-D and RMBPC scores were approximately normally distributed. We accounted for the clustering of caregivers within interventionists by specifying interventionist as a random effect. For each outcome, we included racial or ethnic concordance, caregiver race or ethnicity, interventionist race or ethnicity, and caregiver characteristics (age, race or ethnicity, gender, and relationship to care recipient) and care recipient baseline ADL score. In loss to follow-up models, we also included care recipient death or institutionalization prior to the 12-month follow-up, as caregivers with care recipients who died or were institutionalized may be more likely to be lost to follow-up. Analyses of change in CES-D and RMBPC scores at the 12-month follow-up were limited to active caregivers. As a sensitivity analysis, we also modeled change in CES-D scores including caregivers with care recipients who died or were institutionalized to assess whether the inclusion of those individuals would affect results.
In exploratory analyses, we fit caregiver race- or ethnicity-specific models to assess whether racial or ethnic concordance was associated with the 12-month outcomes within any of these groups considered separately. This allowed us to partially disentangle the effects of concordance from caregiver and interventionist race or ethnicity.
We conducted statistical analyses by using Stata statistical software, version 9 (StataCorp, 2005). We considered differences to be statistically significant if p 
.05, with no adjustment made for multiple comparisons.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
Baseline Caregiver Characteristics
Overall, mean age at baseline was 61.7 years (SD = 14.0); 83.4% of caregivers were female; 54.5% were White, 22.2% were African American, and 23.3% were Latino; and 47.7% were spouses of care recipients. Annual income was less than $20,000 for 38.9% of the caregivers, and at least $40,000 for 27.8% of them. Approximately 20% of the caregivers had less than a high school education; 23.6% had a high school education or equivalent; 32.7% had some college or technical school; and 23.9% were college graduates. The median care recipient ADL score was 3 (interquartile range = 1–5).
Caregiver Racial or Ethnic Comparisons
White caregivers tended to be older than African-American and Latino caregivers (65.2 years vs 57.2 and 57.9 years, p <.001; see Table 2). White caregivers were more likely than African-American and Latino caregivers to be a spouse of the care recipient, have an annual income of at least $40,000, and to have graduated from college (p <.001 for each). Care recipient ADL scores were similar between racial or ethnic groups. Median baseline CES-D and RMBPC scores were lower for African Americans, indicating less depression and caregiver burden, relative to White and Latino caregivers (p <.002 for each; see Table 2).
|
.10 for each; see Table 2). The majority of caregivers with an interventionist of the same race or ethnicity were White, and the majority of caregivers with an interventionist of a different race or ethnicity were African American (p <.001). Relatively more caregivers with an interventionist of the same race or ethnicity were caring for spouses (p <.001) or had an income of at least $40,000 (p =.05) than were caregivers with an interventionist of a different race or ethnicity. Median baseline CES-D scores were higher among caregivers with interventionists of same race or ethnicity (p =.001).
Caregiver Loss to Follow-Up at 12 Months
Of the 694 caregiver–care recipient dyads included in these analyses, 120 caregivers (17.3%) were lost to follow-up at 12 months, 21 (3.0%) had care recipients who died, and 14 (2.0%) had care recipients who were institutionalized by the 12-month follow-up. The remaining 422 caregivers were still active caregivers at the 12-month follow-up.
At 12 months, 17.2% to 17.5% of White, African-American, and Latino caregivers were lost to follow-up (p =.99; Table 3), and 16.5% of caregivers with racially or ethnically concordant interventionists were lost to follow-up compared with 19.2% of caregivers with racially or ethnically discordant interventionists (p =.38; see Table 3).
|
.27) or any of the other variables examined (p .08).
|
.50 for each; see Table 3). Caregivers with interventionists of the same race or ethnicity indicated somewhat more improvement in CES-D scores than did caregivers with interventionists of a different race or ethnicity (1.67 vs 0.12, p =.11; see Table 3).
Neither racial or ethnic concordance nor caregiver or interventionist race or ethnicity was statistically significantly associated with change in CES-D scores among active caregivers at the 12-month follow-up (p .35 for each; see Table 5). Increasing caregiver age was associated with an increase in depression, whereas a higher care recipient ADL score was significantly associated with decreases in depression over time (p .002 for each; see Table 5).
|
None of the variables examined, including racial or ethnic concordance, was significantly associated with change in RMBPC scores over time (p >.25 for each; see Table 5). In race- or ethnicity-specific analyses, change in RMBPC score did not vary by racial or ethnic concordance (p >.50) or any of the other variables examined.
The variance components associated with interventionists were virtually zero for the outcomes of lost to follow-up and change in CES-D-score, and small (interventionist-level variance = 1.78 with SD = 2.07) for the outcome of change in RMBPC score. This indicates very little variation in the overall level of response for caregivers assigned to different interventionists that is not accounted for by the predictors in the respective models.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Change in depression was not significantly different between racial or ethnic concordance groups, either unadjusted or adjusted for baseline characteristics, either overall or for White and Latino caregivers. The unexpected finding that depression and caregiver burden appear to increase over time for active caregivers with interventionists of the same race or ethnicity may be an artifact of the difficulty of isolating the concordance effect in the multivariable models. Most concordant pairs involved White caregivers, whereas most discordant pairs involved African American caregivers. Furthermore, income, education, whether the caregiver was a spouse all varied by the race or ethnicity of the caregiver.
However, among African Americans, caregivers with an interventionist of the same race or ethnicity reported significantly greater decreases in depression over time than did caregivers with an interventionist of a different race or ethnicity. This finding has to be interpreted with caution because there were only two African American interventionists. Should our findings for African Americans be corroborated, suggesting that that racial or ethnic concordance may impact study outcomes differently for caregivers in different racial or ethnic groups, the racial or ethnic relationship between participants and interventionists could provide another source of variation in studies involving interventions delivered by means of personal interaction. If the effectiveness of the intervention varies depending on the degree of racial or ethnic concordance, outcomes of a wide variety of intervention studies possibly could be enhanced by careful matching of study participants with interventionists.
Plausible mechanisms of an effect of racial or ethnic concordance on study outcomes should be investigated in future studies. For example, trust and communication have been identified as factors likely to be associated with the positive effects of racial or ethnic concordance in patient–physician relationships (Ashton et al., 2003) and should also be examined in participant–interventionist relationships. Cultural sensitivity training may be beneficial in overcoming mistrust and communication issues. Although cultural sensitivity in research studies has been emphasized in recent years as an important component in successful interventions for minorities (Gallagher-Thompson et al., 2003), there is no evidence supporting the conclusion that cultural sensitivity training is equivalent to providing racially or ethnically concordant interventionists. Systematically assessing the differential impact of racial or ethnic concordance and cultural sensitivity training could be done for relatively little additional cost in future intervention studies.
The National Institutes of Health (NIH) Revitalization Act of 1993 issued guidelines for including minorities in clinical trials funded by the National Institutes of Health in order to determine whether interventions affect members of minority groups differently than they do other participants in the trial (Freedman et al., 1995). In studies in which White participants are more likely to have an interventionist of the same race or ethnicity, and possibly better communication with the interventionist than do participants of other race or ethnic groups, White participants may receive greater benefits from study participation. Ensuring that all racial or ethnic groups are benefiting equally from study participation could play a role in improving not only study participation and retention of minority racial or ethnic groups, but also the health of these groups.
An important limitation of this secondary analysis is that the REACH study has limited power to detect race or ethnicity concordance effects because of the relatively small number of discordant caregiver–interventionist pairs for White and Latino caregivers, the small number concordant pairs for African-American caregivers, and the small number of African-American interventionists. Furthermore, site was confounded somewhat with caregiver race or ethnicity, particularly for Latino caregivers. In addition, caregivers were not randomized to interventionists, so that potential selection biases cannot be ruled out.
Despite these limitations, this analysis has documented differing patterns of racial or ethnic concordance with interventionists for White and Latino caregivers relative to African-American caregivers. Although these initial findings do not provide conclusive evidence that racial or ethnic concordance is associated with intervention outcomes for caregivers of Alzheimer's disease patients, these results, along with the paucity of research studies evaluating the effectiveness of racial or ethnic concordance, suggest that the issue should be systematically examined in future studies.
|
Footnotes |
|---|
1 University Center for Social and Urban Research, University of Pittsburgh, PA.
2 Department of Biostatistics, Graduate School of Public Health, University of Pittsburgh, PA.
3 Epidemiology Data Center, Graduate School of Public Health, University of Pittsburgh, PA.
Decision Editor: Linda S. Noelker, PhD
Received for publication March 17, 2005. Accepted for publication March 13, 2006.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
