Predicting Longitudinal Changes in Caregiver Physical and Mental Health: A Stress Process Model

Transcription

1 Health Psychology Copyright 1998 by the American Psychological Association, Inc. 1998, Vol. 17, No. 2, /98/$3.00 Predicting Longitudinal Changes in Caregiver Physical and Mental Health: A Stress Process Model Kathryn T. Goode University of Alabama at Birmingham William E. Haley University of South Florida David L. Roth and Greg R. Ford University of Alabama at Birmi.ngham Alzheimer's family caregivers (N = 122) reported on physical and mental health, as well as stress process variables, at baseline and at a 1-year follow-up. Hierarchical regression analyses of stress process models revealed that increases in primary stressors (e.g., patient self-care and behavioral problems) did not directly affect changes in the mental and physical health outcome variables. However, analyses of models of direct, mediated, and moderated effects revealed that psychosocial resource variables (appraisals, coping responses, and social support) were related to caregiver outcomes over time through several mechanisms. In particular, benign appraisals of stressors, the use of approach coping, and greater levels of social support were associated with more positive caregiver health outcomes over time. Key words: Alzheimer's disease, appraisals, coping, social support Currently, up to 4 million Americans have Alzheimer's disease (AD) or another dementia, and projections indicate that the prevalence of dementia will continue to increase for the next several decades (Evans et al., 1990). Much of the burden for the care of these individuals falls on family members in the community. Many studies have documented both the physical and mental health consequences of the caregiving role. For example, several investigations suggest that the physical health of caregivers is poorer than that of noncaregivers (Fuller-Jonap & Haley, 1995; Kiecolt-Glaser, Dura, Speicher, Trask, & Glaser, 1991; Pruchno & Potashnik, 1989). It has also been reported that caregivers exhibit greater psychological distress and, in particular, elevated levels of depression (Dura, Stukenberg, & Kiecolt-Glaser, 1990, 1991; Gallagher, Rose, Rivera, LoveR, & Thompson, 1989; Russo, Vitaliano, Brewer, Katon, & Becker, 1995; Schulz, O'Brien, Bookwala, & Fleissner, 1995). Because of such negative health effects, increasing attention is being given to the stress experienced by these family caregivers. Caregivers commonly face numerous types of stressors in Kathryn T. Goode, David L. Roth, and Greg R. Ford, Department of Psychology, University of Alabama at Birmingham; W'dliam E. Haley, Department of Gerontology, University of South Florida. The present results are part of a larger, ongoing study, the University of Alabama at Birmingham Family Caregiver Research Project, supported by National Institute on Aging Grant P01AG Appreciation is expressed to Julie Forsythe and Kristine Donavan for their assistance with this project. Correspondence concerning this article should be addressed to Kathryn T. Goode, Department of Psychology, School of Social and Behavioral Sciences, University of Alabama at Birmingham, Birmingham, Alabama Electronic mail may be sent to kgoode@uab.edu. the course of caring for a relative with dementia (Haley, Brown, & Levine, 1987; Pearlin, Mullan, Semple, & Skaff, 1990; Schulz et al., 1995; Schulz & Williamson, 1991). The stressors that arise directly from caregiving have been termed pr/mary stressors to distinguish them from stress that may develop secondary to caregiving within work and family roles (Pearlin et al., 1990). These primary slressors are not unitary in nature, and there appear to be at least two different domains. One type of caregiving stress arises from the performance of direct patient care such as providing assistance with tasks that the care recipient is no longer able to perform independently. These self-care tasks include basic activities such as bathing and feeding, as well as more complex duties such as managing finances. Patient behavioral problems (e.g., angry, dangerous, or embarrassing behavior) and memory impairments (e.g., disorientation and asking repetitive questions) are another source of stress to the caregiver. Thus, providing assistance with patient selfcare and managing patient behavioral problems represent different stressors and place different demands on caregivers. For example, several studies of dementia caregivers suggest that behavioral problems are perceived as more stressful than patient self-care impairments (Haley et al., 1987; Schulz et al., 1995). Stress process models have been widely used to examine the relationship~ among caregiving stressors, possible psychosocial resources, and caregiver well-being. Haley and colleagues (Haley, Levine, Brown, & Bartolucci, 1987; Haley et al., 1996) have proposed a stress process model of caregiving adapted from the model of Lazarus and Folkman (1984). Several psychosocial resource factors that may mediate the effect of caregiving stressors on well-being have been identified. These psychosocial resource variables include caregiver appraisals of primary stressors, coping 190

2 SPECIAL ISSUE: CAREGIVER STRESS PROCESS MODEL 191 responses, and social support. Specifically, better caregiver well-being has been found to be associated with more benign appraisals of primary stressors (lower stressfulness ratings and higher self-efficacy ratings), greater social support, and coping responses characterized by more approach or problem-focused coping than avoidance or emotion-focused coping (Haley, Levine, et al., 1987; Haley et ai., 1996). Similar caregiving stress process models have been proposed and generally supported by other researchers studying dementia caregiving (Pearlin et al., 1990; Vitaliano, Russo, Young, Ted, & Maiuro, 1991). In a review of the dementia caregiving literature, Schulz et al. (1995) found that caregiver mental and physical health outcomes are related to multiple patient and caregiver characteristics, including patient problem behaviors, perceived stress, and social support. These authors pointed out that most studies of caregiving outcomes have focused on simple effects models and that future research should assess the interactions of multiple variables. Stress process models have rarely been assessed in a longitudinal design (see Aldwin, 1994). Because AD generally results in progressive deterioration in the patient over time, AD caregiving provides a unique opportunity to study the multiple possible influences of stress and resource variables over time. In the present project, we applied the previously developed stress process model (Haley, Levine, et al., 1987; Haley et al., 1996) to predicting changes in caregiver physical and emotional well-being over time in the face of worsening stressors. We examined whether initial psychosocial resources provided direct or stress-moderator benefits, and we examined changes in resources over time as a possible stress mediator. It is intuitively appealing to hypothesize that caregivers would show worsening physical and mental health over time. This "wear and tear" hypothesis (Haley & Pardo, 1989; Townsend, Noelker, Deimling, & Bass, 1989) would parallel the deterioration seen in the AD patient, However, longitudinal studies of caregiving have commonly shown that caregivers as a group exhibit relatively stable levels of mental and physical health problems over time (Kiecolt- Glaser et al., 1991; Townsend et al., 1989; Vitaliano et al., 1991). For example, in a longitudinal study of depression among caregivers (Schulz & Williamson, 1991), the majority of participants (59%) showed relatively stable levels of depression over a 2-year period. Related to physical health, Vitaliano et al. (1991) found that the number of illnesses reported by spousal caregivers at a month follow-up assessment had not changed from baseline. This overall group stability may mask considerable individual differences related not only to differences in patient decline over time but also to caregiver stress process variables. In particular, it is essential to closely examine the potential differential impact of worsening patient self-care problems versus patient behavioral problems. These different stressors undoubtedly require different types of adjustments by the caregiver. Self-care problems, such as the patient requiring greater assistance with bathing, dressing, or continence, can provide a direct physical strain on the caregiver but can generally be managed through the development of appropriate skills. Behavioral problems, such as patient agitation and wandering, are generally less predictable and less controllable and are commonly reported as more stressful by families. A number of processes could occur over time to explain changes in caregiver mental and physical health outcomes. These processes are shown graphically in Figure 1. First, changes in primary caregiving stressors could have a direct effect on changes in caregiver outcomes (as shown by Path a). Second, changes in caregiving stressors could produce changes in mediating variables such as caregiver psychosocial resources, including appraisals, coping, and social support (Path b), which could in turn subsequently affect caregiver outcomes (Path c). It is also possible that changes in caregiving psychosocial resources directly affect changes in outcomes but without being influenced by changes in stressors (Path c in the presence of a nonsignificant Path b). This would not qualify as a mediated effect but would simply indicate a unique direct effect for changes in Initial Caregiver Psyehosoclal Resources Changes in" Care~ver Psychosocial Resources Changes in Changes Caregiver Mental n and Physical Stressors Health Figure 1. Possible mechanisms predicting changes in caregiver physical and mental health through stressors and caregiver resources.

3 192 GOODE, HALEY, ROTH, AND FORD psychosocial resource variables. Initial caregiving psychosocial resource variables might also provide direct protective effects on changes in outcomes over time (Path d). Finally, initial caregiving psychosocial resources might moderate the association between changes in stress and outcome variables over time, leading to protective or buffering effects (Path e). In the present project, we examined changes in primary caregiving stressors, appraisals, coping responses, social support, and mental and physical health outcomes over a 1-year period. In terms of hypotheses, we predicted that (a) patients as a group would demonstrate poorer functioning over time, as indexed by both self-care and memory and behavior problem measures; (b) caregivers as a group would show stability in both mental and physical health; (c) the degree of patient deterioration (in both self-care impairment and memory and behavior problems) would not be directly related to the degree of caregivers' worsening in physical and mental health; (d) changes in resource variables would mediate some of the impact of caregiving stressors on outcomes; and (e) initial caregiver psychosocial resources would affect changes in outcomes by moderating some of the impact of caregiving stressors. A mediation process would be supported by a finding that changes in stressors predict changes in psychosocial resource variables, which in turn predict changes in outcomes. Proposed moderator or stress-buffering effects between the degree of change in stressors and the initial value of psychosocial resource variables were tested by including interaction terms (see Aldwin, 1994; Baron & Kenny, 1986). Participants Method Caregivers and dementia patients were recruited through the Memory Disorders Clinic at the University of Alabama at Birmingham as part of a larger, longitudinal project focusing on stress and coping in African American and White caregiving families (Haley et al., 1995, 1996). Patients were diagnosed by Alzheimer's Disease Center neurologists as having AD or other progressive dementia. Caregivers were eligible to participate if they provided care to a community-dwelling patient and also met the following criteria: (a) They identified themselves as the family member who was responsible for the patient's care; (b) they were related to the patient either by blood or marriage; and (c) they resided within a 50-mile (80-kin) radius of the University of Alabama at Birmingham. Procedure Each caregiver who consented to participate completed an extensive structured interview and a series of questionnaires. The interview and questionnaire battery were administered at a time and place convenient for the participant, typically in the caregiver's home. Additional details of the procedure have been described in other project articles (Haley et al., 1995, 1996). After intake at Time 1 (T1), the same interview procedure was repeated at Time 2 (T2; 1 year after the initial interview). Attrition From Time 1 to Time 2 Of those meeting eligibility criteria, 197 (83%) consented to participate in the project (130 female caregivers and 67 male caregivers). These participants (123 White and 74 African American) all completed the initial interview. Over the next year, a number of participants became ineligible to complete the interview procedure at T2. Of the initial group, 30 caregivers had placed the patient in a nursing home during the year and so did not complete the T2 interview. Nineteen of the initial caregivers were ineligible to participate at T2 because they were no longer the primary caregiver. Six patients died during the year. A number of participants (20) simply declined to continue their participation in the project. At the 1-year follow-up (T2), the remaining 122 participants (86% of eligible participants) completed a second interview and were included in the present analyses. Measures Demographic and descriptive information. Caregiver demographic data (e.g., age and education) were assessed through self-report. Relevant patient data (e.g., age and diagnosis) were obtained from medical records. Caregiving stressors. Caregivers reported the total number of impairments experienced by their relative using the Activities of Daily Living Scale (ADL; Katz, Ford, Moskowitz, Jackson, & Jaffe, 1963). This scale measures a patient's need for assistance with six basic self-care activities (e.g., dressing, bathing, and feeding). In addition, caregivers completed the Instrumental Activities of Daily Living Scale (IADL; Lawton & Brody, 1969), which assesses patient disability in eight higher order self-care skills such as managing finances and doing household chores. For both the ADL and IADL scales, caregivers reported whether the patient had received assistance with each activity during the previous week using a scale ranging from 1 (no impairment) to 4 (most severe impairment). The ADL and IADL scales were summed to form a total self-care impairment score (or =.88). The Memory and Behavior Problem Checklist (MBPC; Zarit & Zarit, 1988) is a 30-item instrument that assesses the frequency of occurrence of memory and behavioral problems exhibited by the patient during the previous week. MBPC items (e.g., "asking repetitive questions" and "wandering") were scaled from 0 to 3, with 0 indicating no occurrence of the problem and 3 indicating occurrence of the problem on a daily basis or more often (ct =.86). Appraisals. For each endorsed item on the ADL, IADL, and MBPC, caregivers completed two additional rating scales. First, they were asked to rate their perception of the stressfulness of the problem, and, second, they were asked to describe their confidence in their ability to successfully manage the problem (self-efficacy). Stressfulness and self-efficacy items were rated on a scale ranging from 0 (not at all) to 3 (extremely). Average stressfulness ratings were computed separately for self-care problems and for memory and behavior problems by summing total appraised stressfulness within each problem area and then dividing by the number of problems. Similarly, self-efficacy ratings were summed within each problem area and divided by the number of problems. This approach, which we have used in previous analyses (Haley, Levine, et al., 1987; Haley et al., 1996), allows for computation of average appraisals of stressors independent of the number of stressors; however, alpha coefficients could not be computed because caregivers rated different items. In cases in which caregivers reported no self-care stressors (or no behavior problems), appraisal scores were coded as missing values because we could not divide by a denominator of zero.

4 SPECIAL ISSUE: CAREGIVER STRESS PROCESS MODEL 193 Social support. The Social Support Questionnaire, Short Form- Revised (Sarason, Sarason, Shearin, & Pierce, 1987) is a 12-item instrument in which individuals are asked to list the people (up to nine) in their social network they perceive as helpful and dependable in different areas (e.g., providing emotional support). Caregivers are also asked to rate their satisfaction with the level of support in each of six areas. Separate scores are obtained for total number of social supports (et =.73) and level of satisfaction with social support (et =.86). This instrument correlates highly with the original Social Support Questionnaire (Sarason et al., 1987). Coping strategies. The Coping Responses Inventory (Moos, 1988) was used to assess caregiver coping responses. This scale assesses eight domains of coping, each of which is measured by six items. Individuals rate their reliance on each of the coping items on a 4-point scale ranging from 0 (not at all) to 3 (fairly often). The instrument provides subscales for approach and avoidance coping. Approach coping (et =.88) subscales include Logical Analysis, Positive Reappraisal, Seek Guidance and Support, and Take Problem-Solving Action. Avoidance coping (et =.80) is measured by four subscales: Cognitive Avoidance, Acceptance-Resignation, Seek Alternative Rewards, and Emotional Discharge. As suggested by Vitaliano, Maiuro, Russo, and Becker (1987), these scores were converted to a relative approach coping percentage by dividing approach coping totals by the total coping score. Caregiver mental health. The Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977) is a 20-item self-report measure of depression (et ). It is commonly used in studies of caregiving, stress, and coping (Schulz et al., 1995). Caregiver physical health. The Cornell Medical Index (CMI; Brodman, Erdmann, Wolff, & Miskovitz, 1986) is a self-report measure of physical health symptoms that has been used in research on stress and aging (Aldwin, Spiro, Levenson, & Bosse, 1989; Costa & McCrae, 1985). This instrument was scored to yield a total for health symptoms (et =.90). Results Characteristics of the Sample We have provided detailed demographic and descriptive information on the complete sample in several previous articles (Haley et al., 1995, 1996), but the current sample differed slightly given that it included only caregivers and patients who continued in-home care over the 1-year follow-up period. Patients averaged years of age (SD = 8.48) at the initial assessment, with initial Mini- Mental State Examination (Folstein, Folstein, & McHugh, 1975) scores of (SD = 8.16). All patients had been diagnosed with dementia, and 87% had a diagnosis of AD. At the time of the initial interview, participants had been icaregivers for an average of about 3 years. Of these participants, 70% were women (n = 85), 30% were men (n = 37), 60% were White (n = 73), and 40% were African American (n = 49). Forty-nine percent were spouses (n = 60), 30% were daughters (n = 36), and 21% were other relatives (n = 26) of the patient. Average caregiver age was years (SD = 14.13), and median educational attainment was a high school diploma. Because caregiver gender, age, and relationship to the patient were somewhat correlated in our sample, and because our longitudinal sample size was too small to accommodate including these factors as separate predictor variables, data were collapsed across these groups. In previous analyses of the sample (Ford, Goode, Barrett, Harrell, & Haley, 1997; Haley et al., 1995), we have found relatively few effects of gender and relationship to the patient on variables included in the present analyses. In t tests comparing those caregivers who completed both interviews (n = 122) with those who completed only the first interview (n = 75), no significant differences were revealed on any of the stressors, psychosocial resource variables, or outcome variables obtained during the first interview. In addition, the two groups of caregivers did not differ significantly by age. Chi-square analyses of race, gender, and educational attainment also were not significantly different by group. Mean Changes Over Time Descriptive information on patient and caregiver variables is shown in Table 1, including T1 and T2 mean scores, test-retest stability coefficients, and t tests of the significance of the changes from T1 to T2. Related to our first hypothesis, patients showed significant worsening over time on both self-care problems and memory and behavioral problems. An additional indicator of the worsening status of patient stressors over time is that the average number of hours per week spent caregiving increased from (SD = 50.69) to (SD = 56.14), t(121) = 5.28, p <.01. Descriptive data on the psychosocial resource variables of appraisal, social support, and coping are also displayed in Table 1. These variables remained relatively stable, with only the total number of social supports showing significant change (decreasing) over time. Table 1 also shows that there were no significant mean changes in caregiver physical or mental health from baseline to the 1-year follow-up. In fact, for both dependent variables, there was considerable stability over time. Thus, initial CES-D scores accounted for 56% of the variance in follow-up CES-D scores, and initial CMI scores accounted for 74% of the variance in follow-up CMI scores. Standardized Residualized Change Scores Because both the outcome variables and stress process variables represented changes over time, residualized change scores were used in all subsequent analyses. This removed the linear effects of initial scores on the index of change that was analyzed. In addition, these residualized change scores were transformed to z scores to remove scale of measurement concerns and simplify the analyses. Correlational Analyses Pearson product-moment correlation coefficients were computed between the z-transformed residualized change scores (see Cohen & Cohen, 1975) for the caregiver outcome variables and the stress process model variables. As can be seen in Table 2, changes in stressors showed no significant relationships with changes in mental or physical

5 194 GOODE, HALEY, ROTH, AND FORD Table 1 Descriptive Information on Caregiving Stressors, Caregiver Stress Process Variables, and Caregiver Mental and Physical Health (N = 122) Time 1 Time 2 t test Stability Variable M SD M SD coefficient t df Caregiving stressors Self-care ** 8.61" 121 Memory and behavioral problems ** 3.03* 121 Appraisals Self-care stressfulness a ** Self-care self-efficacy a ** MBPC stressfulness b ** MBPC self-efficacy b ** Social support SSQSRtotal ** -2.08* 121 SSQSR satisfaction ** Coping Approach percentage ** Mental health CES-D ** Physical health CMI total ** Note. MBPC = Memory and Behavior Problems Checklist; SSQSR = Social Support Questionnaire, Short Form-Revised; CES-D = Center for Epidemiological Studies Depression Scale; CMI = Cornell Medical Index. an=120, bn=119. *p <.05. **p <.01. health outcomes. When considered in isolation, changes in appraisals of stressfulness were positively related to changes in depression and physical health symptoms. Furthermore, it was found that increases in approach coping percentage were associated with decreased depression. As one might expect, changes in depression were positively correlated with changes in health symptoms. In addition, we found a small but significant correlation between changes in self- care stressors and changes in behavioral problems (r =.24, p <.05). Regression Analyses Hierarchical regression analyses (Aldwin, 1994; Cohen & Cohen, 1975) were used to test the hypotheses concerning the effects of initial psychosocial resource variables, changes Table 2 Correlation of Changes in Caregiver Outcomes With Stress Process Variables (N = 122) Variable I1 Changes in caregiving stressor 1. Self-care 2. MBPC.24* -- Changes in appraisals 3. Self-care stressfulness a * Self-care self-efficacy a MBPC stressfulness b *.50* MBPC self-efficacy b * Changes in coping 7. Approach percentage Changes in social support 8. SSQSRtotal SSQSR satisfaction * ".20* Changes in health outcomes 10. CES-D " " " I1. CMI * * " "-- Note. MBPC = Memory and Behavior Problems Checklist; SSQSR = Social Support Questionnaire, Short Form-Revised; CES-D = Center for Epidemiological Studies Depression Scale; CMI = Comell Medical Index. an=120, bn=119. *p <.05.

6 SPECIAL ISSUE: CAREGIVER STRESS PROCESS MODEL 195 in psychosocial resources, and changes in caregiving stressors on the outcome variables (changes in physical and mental health). All predictor variables, including residualized change scores, were transformed to standardized (i.e., z) scores before analysis. Hierarchical regressions were conducted separately for each of the outcome variables (CES-D and CMI) and for each caregiving stressor (MBPC and self-care). Also, separate analyses were conducted for each of the psychosocial resource variables (appraisals, coping strategies, and social support measures). The same procedure was followed in each case. In Step 1, the standardized residualized change score for caregiver stressor, the initial psychosocial resource variable (also transformed to a z score), and the standardized residualized change score for the psychosocial resource variable were entered. This analysis is depicted by Paths a, c, and d in Figure 1. In the second step, the change in Caregiver Stressor Initial Psychosocial Resource interaction term was added (Path e). A separate regression equation was used to examine the potential effect of changes in caregiving stressors on changes in psychosocial resource variables (Path b). These analyses allowed tests of the significance of direct, mediated, and moderated effects on health outcome variables. Direct effects of change in stressors. Degree of change in self-care stressors and degree of change in memory and behavioral problems were not significantly related to residualized change scores for any of the dependent variables. Thus, there was no evidence that patient deterioration over the year directly led to caregiver decline in either physical or mental health on average. Mediated effects on physical and mental health outcomes. Analyses revealed direct unique effects of changes in stressfulness appraisals on changes in physical health (CMI) and depression (CES-D) in the context of both self-care and memory and behavior problems (Path c; betas ranged from.19 to.37, all ps <.05). These results indicate that as stressfulness appraisals increased, physical health symptoms and depression worsened. There was some evidence that changes in stressfulness appraisals mediated changes in memory and behavior problems for both physical health and depression. That is, memory and behavior problems significantly predicted changes in stressfulness appraisals (Path b; 13 =.19, p =.03), which in turn predicted changes in physical health symptoms (Path c; 13 =.37, p <.01) and depression (Path c; 13 =.20, p =.04). Changes in approach coping percentage were also directly related to changes in depression in the context of both self-care stressors and memory and behavior problems (Path c; 13s = -.29 and -.31, respectively, ps <.01), indicating that as relative levels of approach coping increase, depression decreases. Moderated effects on physical health outcomes. The results showed that initial approach coping percentage had a significant negative association with changes in physical health (CMI) in the context of changes in both self-care stressors and memory and behavior problems (Path d; both 13s = -.23, p =.01). That is, higher initial levels of approach coping percentage exerted a protective effect on health symptoms. A similar association was found with initial levels of satisfaction, with social support having a protective effect on changes in health symptoms in the context of both self-care stressors and memory and behavior problems (Path d; 13s =-.23 and -.22, respectively, ps <.02). Also, in the context of changes in memory and behavior problems, initial total social support showed a significant negative relationship with changes in health symptoms (Path d; 13 = -.21, p =.02), indicating a protective effect of social support. Initial total social support interacted significantly with change in self-care stressors (Path e; 13 =.19, p <.05) in predicting changes in physical health. As a means of interpreting the interaction, the regression equation was solved to generate lines for those high (mean + 1 SD) and low (mean - 1SD) in social support (Aldwin, 1994; Cohen & Cohen, 1975). Figure 2 shows that those caregivers who reported the greatest increases in health symptoms were those who had low initial social support and reported low changes in self-care stressors. Moderated effects on mental health outcomes. Initial stressfulness appraisals interacted significantly with change in self-care stressors in predicting depression as measured by the CES-D (Path e; 13 =.19, p =.03). Again, regression equations were solved to generate lines for those high and low in stressor appraisals. As shown in Figure 3, an increase in self-care stressors was positively associated with increased depression for those with high initial appraisals of stressfulness. Low initial stress appraisals buffered the relationship between self-care stress and depression. Discussion The results of these longitudinal analyses generally supported the hypotheses advanced in the context of the stress process model. As predicted, dementia patients evidenced significant worsening over time, reflected in higher levels of primary caregiving stressors over the 1-year period. Patients showed increased levels of both self-care impairments and memory and behavioral problems, Despite this progression of dementia, caregivers as a whole did not show worsening O ~0 (R E ~:~ 0.2 >, 00 - m 0 a) "r"._c (o.21 O1 g f- o (0.4) ~ Low Changes in Self-Cam Stressors (z-score) High Figure 2. Effects of social support and self-care stressors on physical health symptoms.

7 196 GOODE, HALEY, ROTH, AND FORD t- O -N o.2 a Q. J= o (0.2) (o.4) i LOW Low 8treufulness Apprsisals High Stressfulness Appraisals.,." oo,".. ' -" '" o, o -.. " Changes in Self-Care Stressors (z-score). y" o. High Figure 3. Effects of stressfulness appraisals and self-care stressors on depression physical and mental health, consistent with a number of previous studies (Kiecolt-Glaser et al., 1991; Townsend et al., 1989; Vitaliano et al., 1991). In addition, the degree of patient deterioration over time did not directly predict changes in the mental or physical health outcome variables. These results are consistent with those of Schulz and Williamson (1991), who found that changes in caregiving stressors did not independently predict changes in depression over time. The present findings confirm that overall declines in caregiver physical and mental health are not generally found and that worsening patient stressors are generally not predictive of individual differences in caregiver functioning. More complex models that incorporate the protective influences of psychosocial resource variables are needed to understand caregiver adjustment over time. In the present study, evidence was found to indicate that resource variables are related to individual differences in mental and physical health changes for caregivers. In terms of a mediating process, it was found that changes in one domain of caregiving stress, memory and behavior problems, did produce changes in one of the psychosocial resource variables, stressfulness appraisals. Changes in stressfulness appraisals were then positively associated with changes in depression and health symptoms. These results suggest that the appraised stressfulness of memory and behavior problems mediates the relationship between these problems and caregiver health outcomes. Another manner in which psychosocial resource variables might influence caregiver outcomes over time is through a direct effect of initial resource levels. In the present study, it was found that initially higher proportions of approach versus avoidance coping predicted better health over time. This effect was independent of the negligible (i.e., nonsignificant) effects for change in stressor on change in health and suggests that relatively greater use of approach coping may be useful, in general, for optimizing health over time in caregivers. Initial levels of social support also provided a protective effect on physical health changes over time. Those caregivers who reported higher initial levels of social support resources actually showed improved health over time. In addition, in the context of changes in memory and behavior problems, initial satisfaction with level of social support provided the same beneficial effect in terms of preventing physical health problems or promoting health improvements. A stress moderator effect was found for initial level of social support. This resource variable was found to interact with self-care stressors in predicting changes in physical health. In this case, those caregivers who reported the greatest increases in health symptoms were those who had low initial social support and reported few changes in self-care stressors. This was a rather unexpected pattern of results, and additional investigation would be necessary to further clarify the nature of this moderating effect. The interaction of appraisals and self-care stressors in depression demonstrates the protective effect of low stressfulness appraisals, even if the patient worsens over time. In this moderator effect, those caregivers who reported initially higher stress appraisals for self-care stressors showed a positive relationship between increases in stress and increases in depression. In other words, low stress appraisals attenuated the relationship between change in stress and change in depression. These results, taken together, support the utility of the stress process model for considering changes in a longitudinal context. Psychosocial resource variables (appraisals, coping responses, and social support) predict longitudinal changes in caregiver mental and physical health, whereas objective changes in the severity of patient impairment do not predict caregiver changes over time. Furthermore, the psychosocial resource variables may exert their influence through differing paths or mechanisms. The present findings also support the discriminative use of stressors (i.e., patient self-care vs. behavioral problems). Results show that although the changes in stressors were correlated, they provide independent contexts in which to examine predictors of health outcomes in caregivers. The present investigation had several important strengths. Caregivers were responsible for severely impaired relatives who worsened over time, providing an ideal context to study adaptation to stress. In addition, research that examines stress and coping longitudinally is needed (Aldwin, 1994). The patients in this sample were carefully diagnosed and characterized. Furthermore, the caregivers constituted a very diverse population. A limitation of this project arises from the attrition of participants over time. As is characteristic of the progression of AD, many patients were either deceased or institutionalized at the follow-up evaluation. Caregivers of these patients are being followed through the project to assess prospectively the predictors and consequences of patient death and nursing home placement. However, it is likely that some of these patients were placed in nursing homes as a result of the very processes under investigation, such as worsening patient self-care problems and memory and behavior problems. Thus, some caregivers may have been excluded from these analyses because of worsening patient and caregiver functioning. However, it was not appropriate to include

8 SPECIAL ISSUE: CAREGIVER STRESS PROCESS MODEL 197 these other caregivers in the present analyses, because the key variables of changes in primary stressors could not be comparably assessed at the longitudinal follow-up if the patient were deceased or no longer in the direct care of the family caregiver as a result of nursing home placement. In conclusion, the present findings demonstrate the protective effects of psychosocial resource variables for dementia caregivers. An additional avenue for future research would be to examine the utility of clinical interventions designed to strengthen these psychosocial resource variables, particularly with high-risk populations. Such caregiver interventions might focus on increasing the relative use of approach coping, modifying stressfulness appraisals, and increasing social support. This type of caregiver training should be beneficial for maximizing long-term mental and physical health outcomes for caregivers. References Aldwin, C. M. (1994). Stress, coping, and development: An integrative perspective. New York: Guilford Press. Aldwin, C. M., Spiro, A., Levenson, M. R., & Bosse, R. (1989). Longitudinal findings from the Normative Aging Study: Does mental health change with age? Psychology and Aging, 4, Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, Brodman, K., Erdmann, A. J., Wolff, H. G., & Miskovitz, P. F. (1986). The Cornell Medical Index Health Questionnaire manual. New York: Cornell University Medical College. Cohen, L, & Cohen, P. (1975). Applied multiple regression/ correlation analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum. Costa, P. T., & McCrae, R. R. (1985). Hypochondriasis, neuroticism, and aging: When are somatic complaints unfounded? American Psychologist, 40, Dura, J. R., Stukenberg, K. W., & Kiecolt-Glaser, J. K. (1990). Chronic stress and depressive disorders in older adults. Journal of Abnormal Psychology, 99, Dura, J. R., Stukenberg, K. W., & Kiecolt-Glaser, J. K. (1991). Anxiety and depressive disorders in adult children caring for demented parents. Psychology and Aging, 6, Evans, D. A., Scherr, P. A., Cook, N. R., Albert, M. S., Funkenstein, H. H., Smith, L. A., Hebert, L. A., Wetle, T. T., Branch, C. G., Chown, M., Hennekens, C. H., & Taylor, J. O. (1990). Estimated prevalence of Alzheimer's disease in the United States. Milbank Quarterly, 68, Folstein M. F., Folstein, S. E., & McHugh, P. R. (1975). Minimental state: A practical method for grading the cognitive state of patients for clinicians. Journal of Psychiatry Research, 12, Ford, G. R., Goode, K. T., Barrett, J. J., Harrell, L. E., & Haley, W. E. (1997). Gender roles and caregiving stress: An examination of subjective appraisals of specific primary stressors in Alzheimer's caregivers. Aging and Mental Health, 1, FuUer-Jonap, E, & Haley, W. E. (1995). Mental and physical health of male caregivers of a spouse with Alzheimer's disease. Journal of Aging and Health, 7, Gallagher, D., Rose, J., Rivera, E, Lovett, S., & Thompson, L. W. (1989). Prevalence of depression in family caregivers. Gerontologist, 29, Haley, W. E., Brown, S. L., & Levine, E. G. (1987). Family caregiver appraisals of patient behavioral disturbance in senile dementia. Clinical Gerontologist, 6, Haley, W. E., Levine, E. G., Brown, S. L., & Bartolucci, A. A. (1987). Stress, appraisal, coping, and social support as predictors of adaptational outcome among dementia caregivers. Psychology and Aging, 2, Haley, W. E., & Pardo, K. M. (1989). Relationship of severity of dementia to caregiving stressors. Psychology and Aging, 4, Haley, W. E., Roth, D. L., Coleton, M. I., Ford, G. R., West, C. A. C., Collins, R. P., & Isobe, T. L. (1996). Appraisal, coping, and social support as mediators of well-being in Black and White family caregivers of patients with Alzheimer's disease. Journal of Consulting and Clinical Psychology, 64, Haley, W. E., West, C. A. C., Wadley, V. G., Ford, G. R., White, F. A., Barrett, J. J., Harrell, L. E., & Roth, D. L. (1995). Psychological, social, and health impact of caregiving: A comparison of black and white dementia family caregivers and noncaregivers. Psychology and Aging, 10, Katz, S., Ford, A. B., Moskowitz, R. W., Jackson, B. W., & Jaffe, M. (1963). Studies of illness of the aged: The index of ADL, a standardized measure of biological and psychosocial function. Journal of the American Medical Association, 185, Kiecolt-Glaser, J. K., Dura, J. R., Speicher, C. E., Trask, O. J., & Glaser, R. (1991). Spousal caregivers of dementia victims: Longitudinal changes in immunity and health. Psychosomatic Medicine, 53, Lawton, M. P., & Brody, E. M. (1969). Assessment of older people: Self-maintenance instrumental activities of daily living. Gerontologist, 9, Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal and coping. New York: Spinger. Moos, R. H. (1988). Coping Response Inventory manual Palo Alto, CA: Stanford University Medical Center and Veterans Administration Medical Center. Pearlin, L. I., MuUan, J. T., Semple, S. J., & Skaff, M. M. (1990). Caregiving and the stress process: An overview of concepts and their measures. Gerontologist, 30, Pruchno, R. A., & Potashnik, S. L. (1989). Caregiving spouses-- Physical and mental health in perspective. Journal of the American Geriatrics Society, 37, Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, Russo, J., Vitaliano, P. P., Brewer, D. D., Katon, W., & Becker, J. (1995). Psychiatric disorders in spouse caregivers of care recipients with Alzheimer's disease and matched cbntrols: A diathesis-stress model of psychopathology. Journal of Abnormal Psychology, 104, Sarason, I. G., Sarason, B. R., Shearin, E. N., & Pierce, G. R. (1987). A brief measure of social support: Practical and theoretical implications. Journal of Social and Personal Relationships, 4, Schulz, R., O'Brien, A. T., Bookwala, M. S., & Fleissner, K. (1995). Psychiatric and physical morbidity effects of dementia caregiving: Prevalence, correlates, and causes. Gerontologist, 35, Schulz, R., & Williamson, G. M. (1991). A 2-year longitudinal study of depression among Alzheimer's caregivers. Psychology and Aging, 6, Townsend, A., Noelker, L., Deimling, G., & Bass, D. (1989).

9 198 GOODE, HALEY, ROTH, AND FORD Longitudinal impact of interhousehold caregiving on adult children's mental health. Psychology andaging, 29, Vitaliano, P. P., Maim'o, R. D., Russo, J., & Becker, J. (1987). Raw versus relative scores in the assessment of coping strategies. Journal of Behavioral Medicine, 10, Vitaliano, P. P., Russo, J., Young, H. M., Ted, L., & Maiuro, R. D. (1991). Predictors of burden in spouse caregivers of individuals with Alzheimer's disease. Psychology andaging, 6, Zarit, S. H., & Zarit, J. M. (1988). The Memory and Behavior Problem Checklist-1987R and the Burden Interview. University Park: Pennsylvania State University, Department of Individual and Family Studies. Call for Nominations The Publications and Communications Board has opened nominations for the editorships of Experimental and Clinical Psychopharmacology, Journal of Experimental Psychology: Human Perception and Performance (JEP:HPP), Journal of Counseling Psychology, and Clinician's Research Digest for the years Charles R. Schuster, PhD, Thomas H. Carr, PhD, Clara E. Hill, PhD, and Douglas K. Snyder, PhD, respectively, are the incumbent editors. Candidates should be members of APA and should be available to start receiving manuscripts in early 1999 to prepare for issues published in Please note that the P&C Board encourages participation by members of underrepresented groups in the publication process and would particularly welcome such nominees. Self-nominations are also encouraged. To nominate candidates, prepare a statement of one page or less in support of each candidate and send to Joe L. Martinez, Jr., PhD, for Experimental and Clinical Psychopharmacology. Members of the search committee are Conan Kornetsky, PhD; Irwin Lucki, PhD; and Alice M. Young, PhD. Lyle E. Bourne, Jr., PhD, for JEP:HPP. Members of the search committee are Margaret J. Intons-Peterson, PhD; David E. Myer, PhD; and Rose Zacks, PhD. David L. Rosenhan, PhD, for Journal of Counseling Psychology. Carl E. Thoresen, PhD, for Clinician's Research Digest. Members of the search committee are Lizette Peterson-Homer, PhD; Laura S. Brown, PhD; and Maria P. P. Root, PhD. Send all nominations to the appropriate search committee at the following address: Karen Sellman, P&C Board Search Liaison Room 2004 American Psychological Association 750 First Street, NE Washington, DC First review of nominations will begin December 8, 1997.