The Impact of the REACH II Intervention on Caregivers’ Perceived Health

Abstract

The primary goal of the current study was to examine the impact of a caregiving support program on caregivers’ perceived health at 6 months following baseline assessment in the Resources for Enhancing Alzheimer’s Caregiver Health II (REACH II) intervention. A composite measure of perceived health was established and incorporated self-rated health, change in self-rated health, and improvement in physical health. A total of 494 participants receiving the REACH II intervention or an education-only intervention were included in this study. Mixed effect linear regression analysis was performed to examine the effect of the intervention and the factors that mediate this relationship. Findings suggest that the enhanced supportive intervention led to significant improvement in caregivers’ overall perceived health at 6 months. This effect remains significant after controlling for positive satisfaction with caregiving. Improving caregivers’ stress and burden while fostering positive rewards and appraisals may provide indirect health benefits and maintain overall health in dementia caregivers.

Keywords

caregiving multisite intervention Alzheimer’s disease

More than 70% of individuals with Alzheimer’s disease (AD) are cared for by family caregivers and live in the community (Zhu et al., 2008). Caregivers of persons with AD are particularly vulnerable to the physical health effects of caregiving (Pinquart & Sorensen, 2003). The stress and burden associated with caregiving increases the risk of chronic health problems (e.g., heart disease, high blood pressure) as well as the onset of new illnesses (Navaie-Waliser et al., 2002; Schulz & Sherwood, 2008). Caregivers with poor physical health status are more likely to experience difficulty in providing care and may end the caregiving role prematurely (McCann, Hebert, Bienias, Morris, & Evans, 2004).

Aside from the negative effects of caregiving, caring also involves a rewarding component that leaves caregivers “feeling good” (sense of meaning), “useful” (returning care to a loved one), and “important” (Boerner, Schulz, & Horowitz, 2004; Grant & Nolan, 1993; Schulz et al., 2004). These positive benefits may protect caregiver physical health and enhance well-being (Semiatin & O’Connor, 2012). Yet the existing caregiving research predominantly focuses on the negative outcomes associated with caregiving and the impact on caregivers’ perceived health. This observation highlights the imbalance between negative and positive health impacts of caregiving and calls for a broader focus in caregiving research to better understand the role of positive experiences in caregiving (Semiatin & O’Connor, 2012), especially on the common outcome measure of self-rated health.

Self-rated health has been used in caregiving research as an outcome measure because of its potent predictability of future survival, morbidity, functional decline, and health care utilization (Banerjee, Perry, Tran, & Arafat, 2010; Deeg & Kriegsman, 2003; Emmelin et al., 2003; Ferraro & Kelley-Moore, 2001; Idler & Benyamini, 1997; Y. Lee, 2000; Robine, Jagger, & Euro-REVES 2 Group, 2002; Steiner et al., 2008). It is also one of the most comprehensively used, single measures of perceived current health (Goldman, Glei, & Chang, 2004; Leinonen, Heikkinen, & Jylhä, 2001). Each of these studies demonstrated persuasive evidence of the use of self-rated health as a physical health outcome measure to study the impact of a psychosocial intervention.

Despite the recognition of the positive aspects of the caregiving experience, existing randomized clinical trials of caregiving interventions that examined self-rated health have focused only on the negative health effects of caregiving. For example, Mittelman, Roth, Clay, and Haley (2007) examined the effects of New York University (NYU) Caregiver Intervention on the composite measure of self-rated health exclusively through the reduction in stress and depressive symptoms among spousal caregivers of people with AD. The only study based on the Resources for Enhancing Alzheimer’s Caregiver Health (REACH II) intervention data also investigated the impact of the intervention on caregivers’ self-rated health and sleep quality through reductions on measures of caregiving burden (Elliott, Burgio, & DeCoster, 2010). Both studies focused exclusively on negative effects of caregiving (i.e., stress, burden, and problem behaviors) on caregiver physical health. A lack of attention to positive aspects of caregiving (PAC) has therefore created a skewed perception of our understanding of how strengthening positive aspects could improve caregiver perceived health outcomes. A better understanding of positive effects of caregiving on caregiver perceived health may help health professionals not only to try to avoid negative problems but also to strengthen positive aspects in improving the health of dementia caregivers.

Data from The REACH II intervention provide an opportunity to investigate caregiver perceived health. The REACH II was a multisite randomized clinical trial funded by the National Institutes of Health designed to reduce stress and burden in caregivers of persons with AD and improve caregivers’ health. The intervention targeted multiple components of a stress-health model which was derived from the stress-process model used in many studies. This stress-process model provided a holistic approach in examining caregiver health and how each domain of the stress-health model influences several dimensions of caregiver health—including self-rated health (Blieszner & Roberto, 2010; Hilgeman et al., 2009; Son, Erno, Shea, Zarit, & Stephens, 2007). The specific therapeutic components included five domains linked to caregivers’ stress-health process: safety, social support, self-care, emotional well-being, and care recipients’ problem behaviors (http://www.edc.gsph.pitt.edu/Reach2/public/documents/SMOvol01sect03.pdf). According to the stress-health model, addressing these therapeutic components should reduce stressors (improving caregiver’s capacity to deal with stressors and negative emotional health), which in turn should decrease the risk of poor physical and mental health. Because of this inherent relationship of stress-health process and perceived physical health and the REACH II intervention’s direct link to each element of the stress-health process, we conceptualized that participation in the intervention would significantly improve caregivers’ perceived health.

The initial investigation by Belle et al. (2006) found a significant impact of the REACH II intervention in improving several domains of quality of life of ethnically diverse dementia caregivers. Subsequent studies using the REACH II data examined impact of the intervention on a variety of outcome measures such as depression, anxiety, and burden and bother (Elliott et al., 2010; C. C. Lee, Czaja, & Schulz, 2010). In particular, Elliott et al. (2010) found a positive and significant impact of the intervention on various measures of caregivers’ perceived health status with improvement in caregivers’ depression identified as the main mediator of this relationship. Cross-sectional and longitudinal evidence suggests that positive and negative aspects of caregiving may impact caregivers’ perceived health outcomes differently (Boerner et al., 2004; Schulz et al., 2004). Moreover, positive aspects within the caregiving relationship may serve as a buffer against negative consequences and improve caregivers’ coping responses (Tarlow et al., 2004).

Because of the purposeful sampling procedures of the REACH II clinical trial, the sample had a strong representation of White (219), African American (211), and Hispanic (212) caregivers (Belle et al., 2006). The primary goal of the current study was to examine the impact of the REACH II intervention on the composite measure of perceived physical health of caregivers of persons with AD. For the current study, we established a composite measure of caregivers’ perceived health which incorporated information on how caregivers perceived changes in their physical health status in response to the chronic stress over time along with self-rated health at a single time point. Using this multidimensional measure of perceived health enabled us to examine the effectiveness of the intervention on the comprehensive assessment of caregivers’ perceptions about their overall health, as opposed to a single item measure which may not provide a sufficiently accurate reflection of caregivers’ overall health status.

In examining the intervention effect, we studied the roles of the PAC and the negative caregiving experience (depression, anxiety, burden and bother) as potential mediators in this relationship. We hypothesized that the intervention would improve caregivers’ perceived health status from baseline to 6-month follow-up. Improvement to PAC and reduction in caregiver stress was hypothesized to be the mediator of improvement of caregiver perceived health status. This rationale is based on current evidence that suggests interventions that are tailored to improve caregivers’ stress, depressive symptoms, and social support would also improve self-rated perceived health. This is because of the known association of caregiving outcomes with self-perceived health even when the intervention explicitly did not focus on health promotion (Mittelman et al., 2007). The REACH II intervention was specifically tailored to the needs of individual caregivers with the aim of improving the caregivers’ burden, depressive symptoms, self-care, and care recipients’ problem behaviors.

The current study extends the existing literature examining the impact of the REACH II intervention on caregivers’ outcomes. First, this study contributes to our understanding of the impact of PAC on caregivers’ perceived health in addition to known negative effects. Second, examining the multi-item composite measure of perceived health can provide a complete profile of perceived physical health status across the heterogeneous ethnically diverse caregivers in the REACH II clinical trial. Finally, the use of growth curve modeling allows for the examination of individual variability in the changes in perceived health scores from baseline to follow-up (i.e., estimating the effect size of the intervention).

Method

Study Design

Family caregivers included in this study were enrolled in the REACH II trial in June 2002 with follow-up ending August 2004. Eligibility criteria, recruitment processes, and baseline information for caregivers and care recipients are discussed in detail elsewhere (Belle et al., 2006). The intervention included 642 qualified participants who were randomly assigned to control or intervention groups. Participants were enrolled from five intervention sites across the United States (Birmingham, Alabama; Memphis, Tennessee; Miami, Florida; Palo Alto, California; and Philadelphia, Pennsylvania). For the purpose of the present study, only participants who completed the 6-month follow-up (and for whom data on their physical health status showed negative and positive measures) were included in the analysis. This study included 494 dementia caregiver and care recipient dyads (160 Whites, 165 African American, and 169 Hispanic), assigned to either the intervention or control group.

After the initial assessment (N = 995), caregivers were randomly assigned (n = 642) to the intervention or control groups using the block randomization technique. The intervention was delivered by certified interventionists over a 6-month period. It included 12 sessions (9 in-home [1.5 hr each] sessions and 3 telephone [half-hour each] sessions) and five structured telephone support group sessions. Each of the five domains (safety, social support, self-care, emotional well-being, and problem behaviors) was delivered with therapeutic skill building and support activities including reinforcement with practice and role-playing to reduce stress and burden associated with caregiving. To facilitate the intervention, caregivers were provided resource notebooks with educational materials.

Variables Used in the Analysis

Outcome variable

The composite measure of perceived health was the primary outcome of interest for this study. This measure was created using three components: self-rated health, change in self-rated health, and improvement in physical health from baseline to the 6-month follow-up. Because perceived health is a multidimensional measure of overall health, we expected individual items would be less informative compared with this composite measure. We therefore performed our analysis using this composite measure only. The components of a caregiver’s perceived health were based on responses to the set of questions asked in the intervention:

Self-rated health: In general, would you say your health is “excellent,” “very good,” “good,” “fair,” or “poor” (on a 5-point sale from 0-4; 0 = poor, 4 = excellent). This measure was created by recoding the direction of the scores in the survey.

Change in self-rated health: Compared with 6 months ago, how would you rate your health now? (on a 5-point scale; 0 = much worse now, 4 = much better now); this was also reverse coded.

Physical health: In the past 6 months, do you feel your physical health has improved? (measured by yes/no answer); for those answering “no,” a follow-up question was asked to rate the decline in health status from “no change” to “substantial decline” (on a 3-point scale; 1=no change, 2= minimal decline and 3= substantial decline), and for those who answered “yes,” another follow-up question was asked to determine whether the improvement was “minimal” or “substantial” (on a 2-point scale; 1 = minimal improvement, 2 = substantial improvement). Based on how respondents answered this question, responses from both follow-up questions were combined to create a measure of physical health on a 5-point scale from 0 to 4 (where 0 = substantial decline, 4 = substantial improvement).

The sum of these three health status components was used to measure the caregiver’s overall health status. Possible scores ranged from 0 (not healthy at all) to 12 (healthiest): a higher score representing a higher level of perceived physical health. The internal consistency of these three items was measured by calculating Cronbach’s alpha and the value was found to be .73. The alpha value of .70 is usually used to indicate a reliable set of items (de Vaus, 2002). The interitem correlation was also measured and found to be highly correlated (correlation coefficient r range = .29-.74; p < .001).

Background variables

Caregivers’ sociodemographic background collected at baseline included gender, age, primary racial/ethnic group, marital status, education level, household income, and the caregiver’s relationship with the care recipient. Evidence suggests that caregiving stressors may vary between spousal versus nonspousal caregivers and thus intervention may have different impact on perceived health of spouse caregivers and adult child caregivers (Pinquart & Sorensen, 2006).

Measures of Burden, Depression, Behavioral Bother, and Positive Satisfaction

For the purpose of this study, we measured caregivers’ burden, stress, and bother with the following measures: the brief 12-item version of the Zarit Subjective Burden Inventory, Behavioral Bother and Burden of the Revised Memory and Behavior Problem Checklist (RMBPC), and the Center for Epidemiologic Studies Depression Scale (CES-D).

Zarit Subjective Burden score

This is a 12-item modified Zarit Subjective Burden Inventory measured by caregivers’ responses to questions about physical and emotional stress and strain. Each question was measured on a 5-point scale ranging from 0 (never) to 4 (nearly always). The range of possible score was 0 to 44, with higher scores indicating greater levels of caregiving burden. This shorter version of the Zarit score represents good internal consistency and is highly correlated with the full measure (r range = .92-.97, and Cronbach’s α = .88; Bedard et al., 2001).

Behavioral Bother and Burden

Appraisal of caregivers’ burden, how bothered or upset caregivers became while providing assistance to care recipients with memory-related disruptive behavior, was measured with the RMBPC. The RMBPC is a brief 24-item list that assesses problem behaviors (7 memory-related, 8 depressive, and 9 disruptive) exhibited by care recipients. The instrument assesses presence or absence of a behavior as well as the intensity of bother or upset associated with the behavior. The overall upset score was calculated by totaling responses to each of these subscales, ranging from 0 to 96. A higher score indicated greater behavior bother and upset. The RMBPC scale is well validated and highly reliable in measuring problem behavior and burden (Cronbach’s α = 0.90; Roth et al., 2003).

CES-D

The 12-item version of the CES-D scale was used as a measure of caregivers’ depressive symptoms due to caring for a person with dementia. For each item, caregivers’ responses were recorded using the scale ranging from 0 (rarely or none of the time) to 3 (most or all of the time). Summarizing all individual scores yielded the total score, with higher scores representing a higher level of depression. These scores passed the sensitivity test with positive predictive values as compared with the original CES-D score (Irwin, Artin, & Oxman, 1999; Roth, Ackerman, Okonkwo, & Burgio, 2008). This short form of CES-D score has shown to have good predictive value (Cronbach’s α = .82; Czaja et al., 2009).

The caregiver’s positive satisfaction score was measured by the Positive Aspects of Caregiving scale (PAC score). The PAC can be identified in several ways, but they are typically defined as rewards and satisfaction derived from the caregiving relationship (Hilgeman, Allen, DeCoster, & Burgio, 2007). This score captures the positive feelings that caregivers often report as satisfaction and rewards derived from caregiving responsibilities. The 11-item PAC scale measures a caregiver’s mental state or overall feelings about providing care to persons with dementia. Responses were recorded on a 5-point scale that was used to measure caregivers’ perceptions of benefits in the context of caregiving, such as feeling good, feeling useful, feeling appreciated, and finding meaning. A higher score represents more positive appraisals, and internal consistency measured by the Cronbach’s alpha score was estimated as .89 (Tarlow et al., 2004). Table 1 describes the demographics of the intervention and control group caregivers.

Table 1.

Sample Demographic at Baseline (N = 494).

Variables	Intervention	Control
Age, M (SD)	61 (12)	60 (13)
Female (%)	81	85
White, non-Hispanic (%)	37	30
African American, not Hispanic (%)	30	37
Hispanic (%)	33	33
Education, M (SD)	13 (3)	12 (3)
Relationship
Spouse (%)	40	40
Nonspouse (%)	60	60
Income
<20,000 (%)	37	37
20,000-30,000 (%)	30	33
>30,000 (%)	33	30
Overall health, M (SD)	5.79 (.13)	5.65 (.14)

Mixed effects individual growth curve analysis that modeled participants’ perceived health scores at the two time points under consideration was conducted using Stata’s (STATA Corporation, College Station, TX) xtmixed procedure. The growth curve modeling technique does not assume homogeneity of treatment effect (rate of change) across all participants making it desirable to use in our study as it was unreasonable to expect equal rates of change over time for all caregivers (Figueredo, Brooks, Leef, & Sechrest, 2000). Therefore, this analysis technique can be considered as the optimal analysis strategy for this type of study in which the assumption of homogeneity of treatment effect across all caregivers is theoretically implausible. Another advantage is that growth curve technique may be more sensitive to emerging difference between treatment groups (intervention and control) compared with simple pre and post comparisons when the follow-up period is relatively short (Leff, Mulkern, Rabb, & White, 1996). Finally, individual growth curve can be fit for each caregiver based on amount of data that participant had provided. While frequently applied to a design with more than two data points, it is an accepted technique for two data points when the homogeneity of treatment effects cannot be assumed. In this study, individual growth curve parameters were modeled as a function of treatment groups (treatment vs. control) and other covariates of interest. In addition, individual variability in unequal treatment effects over time was captured by including a random effect for time (Roesch, Noman, Villodas, Sallis, & Patrick, 2010).

We began by estimating a reference model in which the overall health score was modeled as a function of time–invariant covariates: age (expressed as deviations from the baseline mean for all participants), gender, and baseline self-rated health. Three time-dependent covariates included in this model were time, baseline self-rated health by time interaction effect, and baseline age by time interaction. To gain some understanding about the size of the intervention effect on overall health, we calculated the reductions in residual variance that remained unexplained by each model, and proportion of variance accounted for using the procedure described by Singer and Willett (2003). We then added the intervention group (treatment vs. control) and the group by time interaction effect as predictors in Model 1. The interaction effect tests whether the effect of treatment significantly increased the perceived health from baseline to 6-month follow-up. In Model 2, we added other demographic characteristics (race/ethnicity, education, relationship, and income) to examine how the impact of the intervention was moderated. Model 3 included caregivers’ depressive symptoms and burden (CES-D, Zarit burden score, and RMBPC score) in addition to the variables in Model 2. Finally, Model 4 included caregivers’ positive satisfaction with caregiving. These analyses allowed us to examine the respective impacts of caregiver depression and burden, and positive caregiving experiences as potential mediators of any intervention effect identified in the reference model.

Results

Out of 494 caregivers in the REACH II sample, there were 249 caregivers in the intervention group and 245 caregivers in the control group (assigned randomly). Demographic characteristics at baseline are presented in Table 1. On average, the majority of caregivers were female, and intervention and control groups exhibited the same demographic characteristics at baseline.

Table 2 summarizes differences in burden, depression and PAC scores between intervention and control groups at baseline and 6-month follow-up. It also represents significant differences, if any, in all measures from baseline to the 6-month follow-up between as well as within treatment and control groups. There were no significant baseline differences between treatment and control group caregivers in overall perceived health, depression, and burden scores. However, caregivers in the treatment group showed significant improvements in overall perceived health as well as burden and depression levels from baseline to the 6-month follow-up, compared with caregivers in the control group. These unadjusted mean differences were examined by the two-sample t tests from baseline to the 6-month follow-up for both groups. Although control group caregivers also experienced significant improvements in burden and Behavior and Bother scores, no significant improvements were observed in perceived health from baseline to the 6-month follow-up.

Table 2.

Within- and Between-Group Differences in Caregiving Stress and Burden.

	Intervention M (SD)		Control M (SD)		p values for group differences
	Intervention M (SD)		Control M (SD)		Within		Between
Caregiving measures	Baseline	Follow-up	Baseline	Follow-up	Intervention	Control	Baseline	Follow-up
Zarit burden	18.62 (9.4)	16.38 (9.8)	18.32 (10.2)	17.18 (11.2)	<.001	.004	.75	.33
CES-D	11.20 (5.6)	9.7 (4.8)	11.21 (5.1)	10.08 (6.5)	<.001	.09	.98	.02
RMBPC	17.63 (13.4)	13.3 (11.2)	18.23 (14.3)	15.6 (13.9)	<.001	<.001	.63	.03
PAC	31.04 (10.9)	32.4 (10.5)	31.81 (10.8)	32.3 (13.4)	.005	.72	.43	.90
Overall health	5.8 (2.1)	6.3 (2.3)	5.6 (2.3)	5.7 (2.2)	<.001	.49	.46	.005

Note. Bolded numbers indicate results are statistically significant at less than 5% level. Zarit burden = burden scores; CES-D = Center for Epidemiologic Studies Depression Scale; RMBPC = Revised Memory and Behavior Problem Checklist; PAC = positive aspects of caregiving.

Given that the REACH II intervention trial included equal numbers of White, African American, and Hispanic caregivers, we examined whether the main treatment effect varied among these ethno-racial groups. Using the factorial ANOVA, we found no significant difference between treatment and control groups across all racial and ethnic groups (results are available from the authors upon request). We also failed to detect any statistically significant treatment effect between spousal versus nonspouse caregivers.

Results from the longitudinal analyses predicting change in overall perceived health are presented in Table 3. A highly significant likelihood ratio test statistic value of 43.50, χ²(1) = 43.50, with p < .001, suggests that it is important to allow a random intercept in this particular model. There was a significant caregiver age-by-time interaction effect. The negative estimate indicated that controlling for other covariates in the model, older caregivers’ perceived health deteriorates faster than younger caregivers. Moreover, there was a significant negative interaction between baseline self-rated health and time indicating that those who had poor health at baseline had lower rates of decline over time, largely due to floor effect. Over and above the impact of these covariates, caregivers in the intervention group had significantly better overall perceived health than those in the control group at the 6-month follow-up (Model 1). Analysis of the residual variance within participants indicated that 3% of the variance in overall health that was unexplained by covariates in the reference model could now be explained by the intervention group-by-time interaction effect. About 70% of variation in overall health is attributable to differences within individual participants and the remaining 30% is attributable to differences across participants.

Table 3.

Change in Overall Perceived Health From Baseline to 6-Month Follow-Up.

Variables	Estimates	SE	z	p
Reference model
Age (centered)	−0.001	0.006	−0.15	.87
SRH (baseline-centered)	2.09	0.09	21.53	<.001
Female	−0.19	0.17	−1.13	.12
Time (0 = baseline, 1 = follow-up)	−0.76	0.48	−1.57	.11
Age × Time	−0.03	0.007	−3.46	.001
SRH × Time	−1.23	0.11	−10.67	<.001
Model 1
Age (centered)	−0.001	0.006	−0.15	.87
SRH (baseline-centered)	2.09	0.09	21.53	<.001
Female	−0.17	0.17	−1.02	.3
Time (0 = baseline, 1 = follow-up)	−0.96	0.48	−1.97	.04
Age × Time	−0.02	0.007	−3.64	.001
SRH × Time	−1.24	0.11	−10.79	<.001
Group (1 = treatment, 0 = control)	0.08	0.15	0.53	.59
Group × Time	0.51	0.18	2.70	.007
Model 2: Impact of demographics in addition to Model 1^a
Time	−0.97	0.48	−1.97	.004
Group	0.08	0.15	0.53	.59
Group × Time	0.51	0.18	2.70	.007
College	0.33	0.23	1.46	.14
High school	0.02	0.22	0.10	.91
White	−0.10	0.13	−0.76	.44
Hispanic	−0.31	0.014	−2.21	.03
Spouse	−0.13	0.15	−0.81	.41
Income	0.07	0.02	3.00	.02
Model 3 : Effect of depression and burden score in addition to Model 2^b
Group × Time	0.43	0.18	2.34	.01
Zarit burden	0.04	0.007	−6.23	<.001
CES-D	−0.03	0.01	−2.39	.02
RMBPC	−0.005	0.004	−1.19	.23
Model 4: Effect of positive aspect of caregiving in addition to Model 3^c
Group × Time	0.40	0.18	2.21	.03
PAC score	0.17	0.005	3.15	.002

Note. SRH = self-rated health; Zarit burden = burden scores; CES-D = Center for Epidemiologic Studies Depression Scale; RMBPC = Revised Memory and Behavior Problem Checklist; PAC = positive aspects of caregiving.

This model controls for other variables in Model 1 (estimates are not shown here).

Adjusts for all other variables in Model 2.

Adjusts for everything in Model 3.

The group by time interaction effect was not attenuated by inclusion of caregivers’ sociodemographic characteristics (as indicated that the estimate of the interaction term remained the same in Model 2). Depressive symptoms and caregiving burden had significant effects on perceived health (Model 3), with higher depression and burden being associated with worse perceived health. The intervention (group by time interaction) still had a significant impact on perceived health, although slightly attenuated. Of the residual variance explained by the intervention effect in Model 1, 6% of the within-participant variation in overall health was mediated by depression and burden. Finally, Model 4 indicated that a higher positive experience in caregiving was associated with better perceived overall health, along with a significant treatment effect. Of the residual variance explained by the intervention-by-time interaction effect, 9% within-participant variation was mediated by the PAC.

Discussion

Our findings suggest that the REACH II intervention led to significant improvement in caregivers’ perceived overall health from baseline to 6-month follow-up. The impact of the intervention may have been mediated by an improvement in caregiver depression and burden from baseline to the 6-month follow-up. This result may not seem surprising, because the intervention had an explicit emphasis on coping with depression and burden associated with caregiving. Our findings are consistent with previous research showing that interventions that improve caregiver depression and burden can also improve self-rated health (Callahan et al., 2005). However, our results are unique in two ways. First, we demonstrated that an improvement in positive experience in caregiving and satisfaction may have health-enhancing effects on caregivers’ perceived health status, even when the intervention did not have an explicit emphasis on improving positive experiences of caregiving. Caregivers who identified positive experience within the caregiving relationship were able to lower depression and caregiving burden that might serve as buffers against negative consequences and improve overall health (Cohen, Colantonio, & Vernich, 2002).

The use of the composite outcome measure allows for evaluation of the intervention’s ability to improve perceived health of dementia caregivers. Research has shown that self-rated health can capture the wide variability of disease severity, and complexity of illness, and allows individuals to assess their perceptions of physical health status. Moreover, the composite measure used in this analysis not only includes the measure of physical health at one time point but also captures the dynamic change in caregivers’ physical health. In short, the composite measure utilized in the current study captures the multidimensional aspect of physical health which is a powerful indicator of caregivers’ perceived overall health.

Second, we were able to quantify the variation in overall health explained by the treatment and time interaction effect at the 6-month follow-up. Although, the effects on overall health in the current study are relatively small (3% of variance in overall health was explained by the intervention group-by-time interaction effect at 6-month follow-up), this is similar to the NYU caregiving intervention reported in the literature (Mittelman et al., 2007). The public health impact of this intervention could be substantial, given the number of dementia caregivers in the United States. Although REACH II had only one 6-month follow-up, the effect size of this intervention suggests that a longer follow-up period may offer greater treatment impact if implemented over a longer period of time with multiple follow-up contacts. We have also found that control group participants showed significant improvement in Burden and Bother scores from baseline to follow-up, which undoubtedly decreased the apparent differential impact of the intervention.

This study has several limitations that merit discussion. The first limitation is that it includes only one 6-month follow-up, which does not allow us to account for the variability in perceived health for longer periods of time with multiple follow-ups. Multiple follow-ups would certainly provide more insight into longitudinal changes in positive and negative experiences of caregiving and their mediation effects on caregivers’ perceived overall health status. The second limitation is that measures of perceived health outcomes are self-reported by caregivers and involve caregivers’ subjective judgment about their own health status. The use of objective measures such as health care utilization may provide stronger evidence on how this type of multicomponent intervention can improve objective health outcomes and reduce health care costs. But the lack of health and health care utilization outcomes from current evidence-based interventions deters the assessment of objective health outcomes such as hospitalization or physician-based visits in the current caregiving research arena, and represents one of the important focus areas of future caregiving interventions. There is a real need for future caregiving interventions to target these objective outcome measures. The third limitation involves evidence that the spousal caregiver may rate health lower than an adult child caregiver, and more than 40% of the caregivers in this study were spousal caregivers—increasing the possibility of underestimation of overall perceived health status measured by the self-rated health. However, using the composite measure may capture true differences in health status, compared with a single measure of self-rated health. The impact of relationship between caregiver and care recipient on perceived health is unlikely captured because the grouping of spouse and adult children in REACH II data set was based on the homogeneity within these two groups. In addition, the intervention effects on perceived health that remained unexplained in this study may have been due to the measures on change in the other core intervention components (safety, social supports, self-care, and emotional well-being) that are not adjusted in the present study. Finally, while the findings of our study suggest a significant improvement in caregivers’ perceived health from baseline to 6-month follow-up in the REACH II intervention, it is difficult to support clinical significance of our result in terms of perceived health, given the public health focus of the intervention.

In this study we used a novel approach in examining the change in caregivers’ perceived health status in a longitudinal perspective using the growth curve modeling technique, indicating that caregivers are able to incorporate health and health-related information which can ultimately improve their perceived health. Moreover, this study provides support that both negative and positive aspects of caregiving are important in influencing caregivers’ perceived health status. In conclusion, a better understanding of the PAC could help health interventionists not only to focus on reducing negative problems associated with caregivers but also to strengthen positive experience which may be equally important in enhancing health of family caregivers.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Scott & White Research Grant Programs (R8000 R3544BASU).

Author Biographies

Rashmita Basu, PhD, is an assistant professor in the Department of Internal Medicine, Scott & White Healthcare and Texas A&M University Health Science Center. Her primary research goals are to better understand economic aspects of chronic disease prevention and family caregiving. Her research also covers life-course perspective of onset and progression of chronic diseases and implementation and evaluation of community-based health intervention programs to enhance health of older adults and their family caregivers. She is interested in studying economic and health impacts of family caregiving in developed versus developing countries. She completed her PhD from the School of Economic Science at Washington State University in 2009.

Angie K. Hochhalter, PhD, is a research scientist in System Quality & Safety at Scott & White Healthcare in Texas. She completed graduate work at the University of Minnesota. After finishing her PhD in cognitive and biological psychology and in gerontology at the University of Minnesota, she spent 2 years as a postdoctoral fellow in applied gerontology at the University of Alabama at Birmingham. Her work is focused on delivery science and engagement of patients and communities in health and healthcare.

Alan B. Stevens, PhD, holds the Vernon D. Holleman-Lewis M. Rampy Centennial Chair in Gerontology at Scott & White Healthcare and is a professor of medicine and public health at the Texas A&M University System Health Science Center. He serves as the director of the Center for Applied Health Research (CAHR), a joint endeavor of Scott & White Healthcare, Texas A&M College of Medicine, and the Central Texas Veterans Health Care System. CAHR conducts and facilitates collaborative projects in the areas of translational and outcomes research. He also heads the Program on Aging and Care, which develops and implements evidence-based clinical interventions for older adults and their caregivers, and he is the director of the NIH-funded Community Research Center for Senior Health. In his administrative leadership role within the Division of Academic Operations, he serves as Scott & White’s representative to HMO Research Network (HMORN) Board of Governors. In the Spring of 2012, he was appointed for a 3-year term to the Board of Directors of the Texas Institute for Health Care Quality and Efficiency. He completed his graduate training at the University of New Orleans, earning a master’s degree and a PhD in applied developmental psychology. Prior to joining Scott & White Healthcare and Texas A&M in 2005, he was an associate professor of medicine at the University of Alabama at Birmingham.

References

Banerjee

Perry

Tran

Arafat

(2010). Self-reported health, functional status and chronic disease in community dwelling older adults: Untangling the role of demographics. Journal of Community Health, 35, 135-141.

Bedard

Molloy

D. W.

Squire

Dubois

Lever

J. A.

O’Donnell

(2001). The Zarit Burden Interview: A new short version and screening version. The Gerontologist, 41, 652-657.

Belle

S. H.

Burgio

Burns

Coon

Czaja

S. J.

Gallagher-Thompson

, . . . Resources for Enhancing Alzheimer’s Caregiver Health (REACH) II Investigators. (2006). Enhancing the quality of life of dementia caregivers from different ethnic or racial groups: A randomized, controlled trial. Annals of Internal Medicine, 145, 727-738.

Blieszner

Roberto

K. A.

(2010). Care partners responses to the onset of mild cognitive impairment. The Gerontologist, 50, 11-22.

Boerner

Schulz

Horowitz

(2004). Positive aspects of caregiving and adaptation to bereavement. Psychology and Aging, 19, 668-675.

Callahan

C. M.

Kroenke

Counsell

S. R.

Hendrie

H. C.

Perkins

A. J.

Katon

, . . . IMPACT Investigators. (2005). Treatment of depression improves physical functioning in older adults. Journal of American Geriatric Society, 53, 367-373.

Cohen

C. A.

Colantonio

Vernich

(2002). Positive aspects of caregiving: Rounding out the caregiver experience. International Journal of Geriatric Psychiatry, 17, 184-188.

Czaja

S. J.

Gitlin

L. N.

Schulz

Zhang

Burgio

L. D.

Stevens

A. B.

. . . Thompson

D. G.

(2009). Development of the Risk Appraisal Measure (RAM): A brief screen to identify risk areas and guide interventions for dementia caregivers. Journal of American Geriatric Society, 57, 1064-1072.

Deeg

Kriegsman

(2003). Concepts of self-rated health: Specifying the gender difference in mortality risk. The Gerontologist, 43, 376-386.

10.

de Vaus

(2002). Analyzing social science data (Vol. 1). London, England: Sage.

11.

Elliott

A. F.

Burgio

L. D.

DeCoster

(2010). Enhancing caregiver health: Findings from the resources for enhancing Alzheimer’s caregiver health II intervention. Journal of American Geriatric Society, 58, 30-37.

12.

Emmelin

Weinehall

Stegmayr

Dahlgren

Stenlund

Wall

(2003). Self-rated ill-health strengthens the effect of biomedical risk factors in predicting stroke, especially for men-an incident case referent study. Journal of Hypertensions, 21, 887-896.

13.

Ferraro

K. F.

Kelley-Moore

J. A.

(2001). Self-rated health and mortality among black and white adults: Examining the dynamic evaluation thesis. Journal of Gerontology B: Psychological Sciences and Social Sciences, 56, S195-S205.

14.

Figueredo

A. J.

Brooks

A. J.

Leef

H. S.

Sechrest

(2000). A meta-analytic approach to growth curve analysis. Psychological Reports, 87, 441-465.

15.

Goldman

Glei

D. A.

Chang

(2004). The role of clinical risk factors in understanding self-rated health. Annals of Epidemiology, 14, 49-57.

16.

Grant

Nolan

(1993). Informal carers: Sources and concomitants of satisfaction. Health and Social Care in the Community, 1, 147-159.

17.

Hilgeman

M. M.

Allen

R. S.

DeCoster

Burgio

L. D.

(2007). Positive aspect of caregiving as a moderator of treatment outcome over 12 months. Psychology of Aging, 22, 361-371.

18.

Hilgeman

M. M.

Durkins

D. W.

Sun

DeCoster

Allen

R. S.

Gallagher-Thompson

Burgio

L. D.

(2009). Testing a theoretical model of stress process in Alzheimer’s caregivers with race as a moderator. The Gerontologist, 49, 248-261.

19.

Idler

E. L.

Benyamini

(1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior, 38, 21-37.

20.

Irwin

Artin

K. H.

Oxman

M. N.

(1999). Screening for depression in the older adult: Criterion validity of the 10-item Center for Epidemiological Studies Depression Scale (CES-D). Archives of Internal Medicine, 159, 1701-1704.

21.

Lee

C. C.

Czaja

S. J.

Schulz

(2010). The moderating influence of demographic characteristics, social support, and religious coping on the effectiveness of a multicomponent psychosocial caregiver intervention in three racial ethnic groups. Journal of Gerontology B: Psychological Sciences and Social Sciences, 65, 185-194.

22.

Lee

(2000). The predictive value of self assessed general, physical, and mental health on functional decline and mortality in older adults. Journal of Epidemiology & Community Health, 54, 123-129.

23.

Leff

H. S.

Mulkern

V. M.

Rabb

White

(1996). Outcome trends for severely mentally ill persons in capitated and case managed mental health programs. Administration and Policy in Mental Health, 24, 3-11.

24.

Leinonen

Heikkinen

Jylhä

(2001). Predictors of decline in self-assessments of health among older people—A 5-year longitudinal study. Social Science & Medicine, 52, 1329-1341.

25.

McCann

J. J.

Hebert

L. E.

Bienias

J. L.

Morris

M. C.

Evans

D. A.

(2004). Predictors of beginning and ending caregiving during a 3-year period in a biracial community population of older adults. American Journal of Public Health, 94, 1800-1806.

26.

Mittelman

M. S.

Roth

D. L.

Clay

O. J.

Haley

W. E.

(2007). Preserving health of Alzheimer’s caregivers: Impact of a spouse caregiver intervention. American Journal of Geriatric Psychiatry, 15, 781-789.

27.

Navaie-Waliser

Feldman

P. H.

Gould

D. A.

Levine

Kuerbis

A. N.

Donelan

(2002). When the caregiver needs care: The plight of vulnerable caregivers. American Journal of Public Health, 92, 409-413.

28.

Pinquart

Sorensen

(2003). Differences between caregivers and noncaregivers in psychological health and physical health: A meta-analysis. Psychology and Aging, 18, 250-267.

29.

Pinquart

Sorensen

(2006).Helping caregivers for persons with dementia: Which intervention works and how large its effects. International Psychogeriatrics, 18, 577-595.

30.

Robine

J. M.

Jagger

, & Euro-REVES 2 Group. (2002). Creating a coherent set of indicator to monitor health across Europe. European Journal of Public Health, 13, 6-14.

31.

Roesch

Noman

J. G.

Villodas

Sallis

J. F.

Patrick

(2010). Intervention-mediated effects of adult physical activities: A latent growth curve analysis. Social Science Medicine, 71, 494-501.

32.

Roth

D. L.

Ackerman

M. L.

Okonkwo

O. C.

Burgio

L. D.

(2008). The four-factor model of depressive symptoms in dementia caregivers: A structural equation model of ethnic differences. Psychology and Aging, 23, 567-576.

33.

Roth

D. L.

Burgio

L. D.

Gitlin

L. N.

Gallagher-Thompson

Coon

D. W.

Belle

S. H.

. . .Burns

(2003). Psychometric analysis of the Revised Memory and Behavior Problems Checklist: Factor structure of occurrence and reaction ratings. Psychology and Aging, 18, 906-915.

34.

Schulz

Belle

S. H.

Czaja

S. J.

McGinnis

K. A.

Stevens

Zhang

(2004). Long-term care placement of dementia patients and caregiver health and well-being. JAMA, 292, 961-967.

35.

Schulz

Sherwood

P. R.

(2008). Physical and mental health effects of family caregiving. Journal of Social Work Education, 44, 105-113.

36.

Semiatin

A. M.

O’Connor

M. K.

(2012). The relationship between self-efficacy and positive aspects of caregiving in Alzheimer’s disease caregivers [PMID: 22360626]. Aging & Mental Health, 16, 683-688.

37.

Singer

J. D.

Willett

J. B.

(2003). Applied longitudinal data analysis: Modeling change and event occurrence. New York, NY: Oxford University Press.

38.

Son

Erno

Shea

D. G.

Zarit

Stephens

(2007). The caregiving stress process and health outcomes. Journal of Aging and Health, 19, 871-887.

39.

Steiner

Pierce

Drahuschak

Nofziger

Buchman

Szirony

(2008). Emotional support, physical help, and health of caregivers of stroke survivors. Journal of Neuroscience Nursing, 40, 48-54.

40.

Tarlow

B. J.

Wisniewski

S. R.

Belle

S. H.

Rubert

Ory

M. G.

Gallagher-Thompson

(2004). Positive aspects of caregiving, contributions of the REACH project to the development of a new measure for Alzheimer’s caregiving. Research on Aging, 26, 429-453.

41.

Zhu

Leibman

McLaughlim

Scarmeas

Albert

Brandt

. . . Stern

(2008). The effects of patient function and dependence on costs of care in Alzheimer’s disease. Journal of American Geriatric Society, 56, 1497-1503.