ATTRIBUTIONS AND SYMPTOM REPORTS
OF COGNITIVE IMPAIRMENT: TESTING THE SYMMETRY RULE IN CAREGIVER THERAPY CLIENTS
by
LINDSAY NICHOLE ANDERSON M.S., North Dakota State University, 2007
A dissertation submitted to the Graduate Faculty of the University of Colorado at Colorado Springs
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy Department of Psychology
This dissertation for Doctor of Philosophy degree by Lindsay Nichole Anderson
has been approved for the Department of Psychology
by
Sara Honn Qualls, Chair
Robert Durham
Charles Benight
Molly Maxfield
Amy Silva-Smith
iii Anderson, Lindsay Nichole (Ph.D., Psychology)
Attributions and Symptom Reports of Cognitive Impairment: Testing the Symmetry Rule in Caregiver Therapy Clients
Dissertation directed by Professor Sara H. Qualls
Family caregivers (CGs) have a key role in early problem identification for
persons with cognitive impairment because CGs are more likely than the persons with the impairment to see symptoms and respond to them by seeking assistance. Researchers and clinicians need to understand which symptoms CGs recognize and how they interpret them because the CGs’ illness models will influence their caregiving behaviors. Given the ambiguities surrounding cognitive impairment presentation and recognition, the degree of symmetry between symptom identification and causal attributions in CGs is likely to be problematic for some portion of CGs and thus be an appropriate target for intervention. This project reports on the symptoms and attributions made by a sample of CGs at the time they sought counseling services in a program for aging families and caregivers. In addition to describing the CGs’ views of the care recipient (CR) problems, this study examines the extent and frequency of symmetry between different types of symptoms reported and the attributions for those symptoms. Results show that CGs report on a number of CR symptoms, and make multiple attributions for the problem. Our sample of CGs seemed to make informed attributions based on the behaviors of their CRs, but they did not hold a unidimensional frame for the CRs’ problems. Implications, limitations, and future directions are addressed.
DEDICATION
I would like to dedicate this dissertation to my family. It is our bond that has drawn me to want to help other families thrive, even in the most difficult of times. To my husband, Nate, your endless patience, support, and encouragement mean the world to me. To my son, Noah, I am blessed beyond measure to have you in my life. You’ve inspired me to finish strong so that I can be the mommy I want to be. To my mother, Carla, you taught me the value of faith, family, hard work, and never giving up—all things I’ve needed to draw on in this process. Thank you. To my dad, Dean, thank you for reminding me that I come from “good stock.” I’ve needed that reminder throughout this process! To my sister, Lacey, thank you for always finishing my sentences and reminding me that we’re family, no matter what. And to my brothers, Samuel and Matthew, thank you for giving me reasons to laugh all throughout the years, and for being so loyal even when I lived 1,000 miles away. Hey, I guess finishing this project means that moving to Colorado wasn’t a “waste” after all!
v
ACKNOWLEDGMENTS
I would like to thank my mentor, Dr. Sara Qualls for her advice and guidance with my research project. I would also like to thank my other committee members, Dr. Robert Durham, Dr. Molly Maxfield, Dr. Charles Benight, and Dr. Amy Silva-Smith for their time and assistance. I owe a great deal of gratitude to Dr. LeAnne Starr, as her help has been invaluable throughout my journey as a caregiving researcher and student: This dataset has now become useful! Your feedback and patience in helping with the “details” of the data could never be matched. Thank you to Dr. Ashley Williams, Renee Pepin, and Katie Kane for their contributions to this research project as well. Also, thank you to the various clinicians and staff at the CU Aging Center for all of their hard work and
commitment to serving caregivers of older adults in the community. And to the
caregivers who so willingly shared the details of their struggles through this data, thank you for all you do. Without you, this project would never have been possible.
TABLE OF CONTENTS CHAPTER
I. INTRODUCTION………...1
Symptoms and Attributions in Cognitive Impairment...………3
Leventhal’s Common-Sense Model..………...6
The symmetry rule: Attributions and symptoms...………8
The age-illness rule………11
Stage of Caregiving………12
Summary of Existing Literature……….13
Critique of Existing Literature………...15
Hypotheses……….16
II. METHODOLOGY………..19
Dataset………19
Participants and Design………..19
Cognitive Impairment Status.………20
Procedure………...23
Data Extraction and Management..………24
Missing Data………..24
Materials...25
vii
Intake Form………25
Behavior Problem Checklist………..26
Activities of Daily Living Scales………...28
Caregiver attribution checklist...………29
III. RESULTS………...31
Symptoms and Attribution Ratings: Descriptive Findings………31
Behavior problem checklist items..………31
Activities of daily living items...………33
Attributions………34
Symptom-Attribution Symmetry………...36
Correlational findings………...……….37
Regression findings………...………39
CG Stage: Group Differences………47
Relationship Impact ………..………49
Descriptives………53
Attributions by adult children and spouse CGs……….55
Symmetry: Correlation findings by adult children and spouse CGs………57
Symmetry: Regression findings by adult children and spouse CGs………60
IV. DISCUSSION………...61
CGs Report a Broad Range of Symptoms……….61
Symmetry...………67
Alignment was imperfect………...67
CGs held reasonably symmetric views………..71
Stage of Caregiving/Cognitive Decline……….74
Individual Characteristics Influencing CGs’ Schemas………..75
CR age………75
Relationship status……….77
Summary and Conclusions………79
Limitations and Future Directions……….81
Implications of Present Study………82
REFERENCES………..86
APPENDICES A. INFORMED CONSENT...………..93
B. IRB APPROVAL LETTER...………..96
C. INTAKE REPORT………..97
D. BEHAVIOR PROBLEM CHECKLIST………103
E. BASIC AND INSTRUMENTAL ACTIVITIES OF DAILY LIVING SCALE...………104
TABLES Table
1. Caregiver and Care Recipient General Demographics (All).………21
2. Caregiving Specific Demographics (All)..….………22
3. Behavior Problem Checklist: Scale Statistics………...………..27
4. Basic and Instrumental Activities of Daily Living: Scale Statistics...…29
5. Behavior Problem Checklist Items: Descriptives………....………..32
6. Basic and Instrumental Activities of Daily Living Items: Descriptives…………33
7. Correlations among Behavior Problem and ADL/IADL Subscales (All)……….35
8. Attribution Items: Descriptives..………35
9. Correlations among Attributions (All)………...37
10.Correlations between Behavior Problem Subscales, ADL/IADLs, and Attributions (All)………38
11.Summary of Hierarchical Multiple Regression Analysis for Variables Predicting Dementia Attribution……….41
12.Summary of Hierarchical Multiple Regression Analysis for Variables Predicting Depression Attribution...………42
13.Summary of Hierarchical Multiple Regression Analysis for Variables Predicting Medical Illness Attribution………44
14.Summary of Hierarchical Multiple Regression Analysis for Variables Predicting
Medication Attribution………...46
15.Summary of Hierarchical Multiple Regression Analysis for Variables Predicting Personality Attribution………...47
16.Summary of Hierarchical Multiple Regression Analysis for Variables Predicting Normal Aging Attribution………..49
17.CR Stage Group Comparisons (Means and Standard Deviations)………51
18.Adult Child Caregiver and Care Recipient General Demographics..………53
19.Spouse Caregiver and Care Recipient General Demographics..………54
20.Adult Child Caregiving Specific Demographics.………..55
21.Spouse Caregiving Specific Demographics..……….56
22.Correlations among Behavior Problem and ADL/IADL Subscales (Spouse)…...58
23.Correlations among Behavior Problem and ADL/IADL Subscales (Adult Children)……….…...58
24.Spouse vs. Adult Child Attribution Comparisons (Means and Standard Deviations)……….59
25.Correlations among Attributions (Spouse)………60
26.Correlations among Attributions (Adult Children)………60
27.Correlations between Behavior Problem Subscales, ADL/IADLs, and Attributions (Spouses Only)…..……….61
28.Correlations between Behavior Problem Subscales, ADL/IADLs, and Attributions (Adult Children Only)………62
CHAPTER 1
INTRODUCTION
Caregiving is a dynamic process marked by physical and emotional strain,
numerous transitions, and a great deal of ambiguity (Teel & Carson, 2003; Williamson et al., 2005). One of the early, most difficult tasks caregivers (CGs) face is interpreting the onset or change in symptoms in a care recipient (CR). For family CGs of older adults, this process is inherently complex, given the high incidence of cognitive impairment, physical decline, functional declines, and co-morbidities that accompany the aging process (Segal, Qualls, & Smyer, 2011). Determining whether changes are “normal,” part of a medical disease, a mental health issue, or a continuation of previous personality style makes the interpretation process especially hard, because all (or none) of these
explanations may apply. Nonetheless, CGs are in a unique position because they may be the only ones to recognize changes in day-to-day functioning (Prior, 2003).
How do CGs determine the level or type of behavior change that should be
defined as a problem? Previous studies recount that in the earliest stages, CGs feel uneasy or believe that “something isn’t right” with their loved one, but often have a great deal of uncertainty about the cause, significance, or appropriate action to take regarding the vague symptoms they see (Teel & Carson, 2003). Prior (2003) summarized this by saying: “So what lay people recognize and report upon is change, and not disease” (p.
48). However, like anyone who encounters a change in behavior, CGs develop common-sense notions about what is causing the problem by asking a series of questions that inform action (e.g., Am I observing a symptom or is this normal? Does this person need care? What kind of care does he/she need? Can we treat the condition, or will it heal itself without intervention? What is causing the problem?). Simply put, CGs engage in a complex process of symptom identification and interpretation. This framework is critical to CGs accessing services in a useful way (Eustace et al., 2007; Hamilton-West, Milne, Chenery, & Tilbrook, 2010).
This study aims to understand both aspects of the CGs’ cognitive frame (symptom identification and interpretation) by reporting on the symptoms, and the attributions, made by a sample of CGs at the time they sought counseling services in a program for aging families and CGs. Although the program accepts CGs of persons with any type of impairment, this project examined only the subset in which the CR had cognitive impairment (approximately 70% of the total client population). Because of the unique struggles of this population (e.g., CRs’ lack of awareness of difficulties, higher rates of burden/physical health problems for CGs; Orfei, Robinson, Bria, Caltagirone, & Spalletta, 2008; Pinquart & Sorensen, 2007; Ory, Hoffman, Yee, Tennstedt, & Schulz, 1999), it is particularly important to understand these CGs’ frames for the problems. In addition to describing CGs’ views of the CRs’ difficulties, this study examined the extent and frequency of symmetry between different types of symptoms reported and the
attributions for those symptoms and whether the symmetry increases as the CRs’ cognitive impairment progresses.
3 Symptoms and Attributions in Cognitive Impairment
The role of family in early detection of dementia is significant because people with cognitive impairment often delay or avoid seeking treatment on their own. In a recent review, Orfei and colleagues (2008) described how pathological brain changes in patients with traumatic brain injury, stroke, dementia, psychosis, and mood disorders often lead to anosognosia, lack of concern, or lack of insight into the nature or severity of the problem (also see Lopez, Becker, Somsak, Dew, & DeKosky, 1994; Seltzer et al., 1997). Put another way, “unawareness of illness consists of the inability to recognize being ill and to assign a correct meaning to deficits, symptoms, and their functional implications” (Orfei et al., 2008, p. 203). These deficits often leave family members with the important task of recognizing symptoms and responding appropriately (Onor,
Trevisiol, Negro, & Aguglia, 2006).
Making sense of behavioral changes in a CR with cognitive decline is no easy task. Misattribution of symptoms is problematic for CGs for a number of reasons. Conditions that impair cognition are accompanied by a complex web of behavioral changes. Early symptoms often overlap with depressive symptoms, pre-existing personality features, or any number of other physical conditions (Archer et al., 2007; Buracchio & Kaye, 2009). In addition, CGs of older adults (commonly the recipients of care) must navigate through the normative changes of aging. CGs may see symptoms, but symptom progression is often gradual, nonspecific, and variable, leaving CGs in a period of “watching and waiting” during which CGs do not seek help, but instead try to make sense of the behavioral changes on their own (Knopman, Donohue & Gutterman, 2000). Regardless of their help-seeking trajectory, CGs engage actively in a search for meaning
regarding their CRs’ symptoms. In a qualitative study, one CG who was attempting to get a diagnosis for a CR put it this way, “We were just searching for something. Did she have Alzheimer’s, or was it something else?” (p. 47, Teel & Carson, 2003). Although CGs sometimes fear the ramifications of a diagnosis (Maguire et al., 1996; Meyers, 1997), studies have shown that diagnoses can help relieve uncertainty about symptom
interpretation, especially in cases of dementia (Husband, 2000). Thus, having an accurate label or “name” for the problem is a critical starting place for CGs as they approach their role (Hayslip, Han, & Anderson, 2008; Qualls & Noecker, 2009).
Despite the importance of “naming” the problem, research has shown there is great variability in how this task is accomplished by CGs. In a recent investigation of “trigger events” that prompt CGs to seek a diagnosis for their CR, Streams, Wackerbarth and Maxwell (2003) found that multiple CR changes (e.g., cognitive and behavioral) were needed in order for CGs to seek help on their behalf. Furthermore, Knopman et al. (2000) found that cognitive changes, per se, were not what prompted CGs to seek medical advice for their CR. Instead, it was change in behavior. In this same study, CGs reported significant delays in seeking help for their loved one’s difficulties. Of great interest for the present study is the fact that 47% of the sample did not seek help because they were unsure of the severity of the CR’s problems, and 37% attributed the symptoms they observed to normal aging processes. CG attributions for the problem were a major factor predicting how CGs went about the “naming” process.
Nichols and Martindale-Adams (2006) examined “decisive moments” where CGs recognized that their CR had a serious problem. They found that in the overwhelming majority of cases, CGs noticed early changes, but did not think the CR had a problem
5 until a major crisis occurred (e.g., driving accidents, leaving the stove on). These
“crises,” once again, were behavioral changes. Attributions for early symptoms were normal aging or medication use, factors that did not prompt appropriate evaluation or support seeking.
Further empirical work has documented that CGs often misdiagnose (or fail to recognize) cognitive impairment in their CRs. In a study by Eustace et al. (2007), 29% of family CGs did not recognize cognitive impairment in a CR who met criteria for a
diagnosis of dementia after a neuropsychological assessment. Even in those cases where cognitive impairment was recognized, 40% of the CGs misattributed symptoms to normal aging, and 25% believed they were simply “not a problem.” Similarly, 21% of Ross et al.’s (1997) total sample failed to recognize cognitive impairment in a loved one; this number increased to 52% when the CR had mild dementia.
In clinical settings, CGs may seek help because they believe something is wrong, but have not formed a clear schema for the problem. Paton, Johnston, Katona and
Livingston (2004) used qualitative interviews to investigate family and professional CGs’ attributions for behavioral problems in CRs with known diagnoses of AD. Surprisingly, CGs were able to identify a myriad of behavioral symptoms they considered distressing (e.g., memory loss, apathy, irritability, wandering), but most made attributions that were not related to dementia. Misattributions to unrelated factors (e.g., medications, part of normal aging, attention seeking, or personality factors of the CR) were prevalent. Similar findings were reported in a study by Hinton, Franz, Yeo, and Levkoff (2005), as most of their sample made attributions that deviated from known medical causes of dementia (despite the existence of a dementia diagnosis for their CRs). Family CGs believed stress,
normal aging, mental illness, having a “tough life” or a “difficult personality” were causes of dementia, and many did not even acknowledge the medical label (p. 1407). What is more, CGs expressed unrealistic expectations given these frames (e.g., one man believed his wife would one day “snap out of it”; p. 1407).
Common behavioral difficulties such as forgetting names, losing keys, or difficulty concentrating can all be attributed to alternative causes because they are ambiguous and do not necessarily seem related to illness (Knopman et al., 2000). It is encouraging that CGs can recognize a health threat, because simply identifying that something is wrong is likely to promote help-seeking. However, when people lack an accurate understanding of the nature of the problem, they are at risk for seeking the wrong help or responding ineffectively to the person with cognitive impairment.
Given that misattribution of symptoms by CGs of persons with cognitive
impairment is a prevalent problem, clinicians and researchers may enhance their work by understanding the causal frames that CGs bring into therapy by utilizing a theoretical model developed with other illnesses: The Common-Sense Model of Self-Regulation of Health and Illness (CSM; Leventhal, Brissette, & Leventhal, 2003). A recent pilot study by Hamilton-West and colleagues (2010) found that the CSM may be a useful model for understanding perceptions of dementia held by relatives.
Leventhal’s Common-Sense Model
The experience of an illness involves cognitive processes, including symptom identification, symptom labeling, and assignment of causal attributions for the problem (Leventhal, Forster, & Leventhal, 2007). The CSM outlines five major classifications of illness beliefs that summarize how people think about disease, with strongest emphasis
7 on the two addressed in this study, symptom identity and causal attribution/explanation (Leventhal et al., 2003). Although originally derived from work on acute illnesses, these concepts have been extended to relate to chronic conditions as well (Leventhal,
Weinman, Leventhal, & Phillips, 2008), a fact that makes this model appropriate for the present study of CGs of older adults. Beliefs about illness symptoms is the first category of illness beliefs. This category is called identity, as the focus is on identifying the illness and putting a label to certain symptoms. This area is one of concern because there is often a discrepancy between what people believe are symptoms and the true symptoms of any given condition. For example, people may believe that they have cancer when they do not, or they may believe that they do not have cancer when they do, depending on symptom interpretation (Petrie & Pennebaker, 2004). The second category includes beliefs about the causes of disease, such as genes or certain behaviors (e.g., smoking causes lung cancer). Cause beliefs are important because they may produce or impair prevention efforts, and may lead to unhelpful or inappropriate behavior overall. As described above, cause beliefs are especially potent (and problematic) for many family CGs. The remaining dimensions (timeline, consequences, controllability) are secondary to these primary components of the illness schema (Leventhal et al., 2003, Leventhal, Meyer, & Nerenz, 1980).
Identity and causal beliefs are related dimensions, as the combination of
symptoms and a causal label has been identified as the “core” of illness representations (Leventhal et al., 2003; Leventhal et al., 1980). Either component alone may drive action (Petrie & Pennebaker, 2004), but together these beliefs form a strong illness frame. Although all five of the CSM dimensions are important for illness representations,
symptom and causal attributions play a foundational role in outcomes (Cameron, Petrie, Ellis, Buick, & Weinman, 2005) and thus were chosen as the priorities for examination. Specifically, the alignment of the identity and cause components of the model is the focus of this investigation. Consistent with the literature, a pilot study of CGs seeking
counseling (Anderson, Qualls, Hiroto, & Starr, 2009) showed that symptoms of cognitive impairment-related disorders dominated the symptom identity, whereas a broad range of causal attributions beyond cognitive impairment were endorsed.
The symmetry rule: Attributions and symptoms. The CSM includes a
symmetry rule that asserts that people seek to align their views of symptom identity and illness cause. Either a symptom or an attribution/label propels a search for the other. In the presence of symptoms, people search for labels or attributions to explain or give a frame for their illness (Leventhal et al., 1980). Once people observe an abnormality, they search the internet, talk to friends or call the doctor in order to find out what is wrong. Reciprocally, when people believe they have a certain condition (e.g., hypertension), they actively search for symptoms to confirm this belief (Leventhal et al., 1980). According to the literature, both search processes occur, making the relationship between symptoms and attributions a rich area of investigation for health promotion (Diefenbach &
Leventhal, 1996).
The CSM postulates that people are not passive in monitoring their bodies for health threats; instead, they take an active role in trying to determine the cause of symptoms and in confirming disease labels (Baumann, Cameron, Zimmerman, & Leventhal, 1989; Brewer, Hallman, & Kipen, 2008; Nakajima and Fleming, 2008). Although some studies show that labeling a disease (especially AD) can have negative
9 ramifications (e.g., blame, stigma, rejection, fear; Werner, 2008; Werner & Giveon, 2008; Werner, Goldstein, & Buchbinder, 2010), this has generally not been the case (although it may vary by disease; see Marlow, Waller, & Wardle, 2010 for a discussion of varying levels of blame by cancer type). Indeed, positive benefits of a label accrued in an investigation of Alzheimer’s disease (AD) and major depression attributions. Wadley and Haley (2001) discovered that having a label (especially for AD) led to more
compassion, a greater willingness to help, and less blame for behavioral problems. Furthermore, Nakajima and Fleming (2008) also posited that when people are educated about the possible effects of certain experiences (e.g., symptoms that are likely after surgery), they are better able to deal with symptoms. Instead of spending precious energy searching for the cause, they can direct their attention to coping. It has been argued that “a label facilitates transfer of meaning to others” (p. 1008; Anderson et al., 2005), a process that is likely to have positive effects for both patients and providers when the label shows symmetry with observed symptoms. In a sense, achieving symmetry creates an “anchor” by which people can make sense of abstract symptom experiences (p. 51; Leventhal et al., 2003). Helping people align the identity of symptoms with a causal attribution is considered a useful step in facilitating effective coping. This is clearly a need for CGs of people with cognitive impairment, as studies show a great deal of mismatch between symptoms and attributions in this population (Eustace et al., 2007; Knopman et al., 2000; Ross et al., 1997).
A clear understanding of causal attributions for illness is important because treatment compliance is predicted by alignment between the patient’s attribution and the healthcare provider’s attribution that determined the choice of intervention. Leventhal
and colleagues (2008) noted that “patients resist treatments that are inconsistent with their illness representations” (p. 489). In other words, physicians or other health professionals can make clear recommendations for dealing with symptoms or treating underlying disease, but unless these recommendations match the individual’s illness schema, they are unlikely to follow through. In studies of blood pressure, Baumann and colleagues (1989) reported that people were more likely to buy into treatment when they could tie their specific symptoms (identity) to a label (cause) that was then tied to the treatment approach. Moreover, Cameron et al. (2005) discovered that causal attributions are stable over time, making it important to examine these beliefs and shape them appropriately as soon as possible during interventions. Recent research has begun to test intervention protocols aimed at influencing, correcting, and shaping illness schemas (see McAndrew et al., 2008 for descriptions of this work in diabetes and asthma).
A key question about CGs of persons with cognitive impairment who have taken the step of seeking help is whether they already have achieved alignment of the symptom identity and causal understanding. If not, this may be an appropriate target for
intervention. The fact that our sample sought help from a CG counseling program does not allow inference about their causal explanations for behavior problems because counseling is used for a broad range of problems. Counselors may benefit from knowing how well symptoms and attributions are aligned because an intervention that presumes a causal attribution not endorsed by the client is unlikely to succeed.
Furthermore, the distress that prompted help-seeking could be caused by or exacerbated by misalignment of symptoms and attributions. Misalignment creates
11 to CR behavior problems (Knight, 2004; Williamson et al., 2005). Specifically, studies show that CG attributions to stable factors (e.g., personality) can cause diminished physical health and well-being in CRs (Hinrichsen, Adelstein, & McMeniman, 2004; Fopma-Loy & Austin, 1997; Polk, 2005). When CGs of persons diagnosed with
dementia retain inaccurate attributions for specific problem behaviors (e.g., she does that on purpose, not because of her dementia), negative effects on the CG are also evident (e.g., criticism, hostility, distress; see Tarrier et al., 2002; Hayslip et al., 2008; Martin-Cook, Remakel-Davis, Svetlik, Hynan, & Weiner, 2003). However, when attributions align with symptoms, both CGs and CRs are more likely to experience improvements in well-being and happiness (Jampel, 1995). Both symptom identification and labeling play important roles in how CGs understand and react to loved ones, making it pertinent to review the symmetry between these processes in caregiving situations.
The age-illness rule. With advancing age, a logical alternative attribution for many early signs and symptoms of illness is aging of the body, leading Leventhal and colleagues to specify within the CSM an age-illness rule as a factor that complicates symmetry between symptoms and illnesses. Because normal aging brings with it a host of physical changes, older adults (and those close to them) often struggle to determine if they are sick or if they are just getting old. Despite greater prevalence of disease in aging populations, many health problems, including cognitive impairment, are not an inevitable part of aging. Nonetheless, aging attributions may deter individuals from taking
appropriate action, as they believe there is nothing that can be done. Rakowski and Hickey (1992) found that older adults who believed they had good health but had impairments in activities of daily living (ADL) attributed their functional limitations to
normal aging. This is problematic, in that decline in ADL functioning is not necessarily associated with chronological age, but may be indicative of some underlying disease process (Folquitto et al., 2007). Additionally, according to this literature, older
individuals tend to seek help when their symptoms are new and/or sudden, but are less likely to seek help when symptoms are mild and have a gradual onset (Diefenbach & Leventhal, 1996; Prohaska, Keller, Leventhal, & Leventhal, 1987). In the case of gradual change, older adults may simply believe their difficulties are due to normal aging
processes and may miss opportunities for early intervention.
Because the current study involves CGs of older adult CRs, attributions of observed behavior problems to normal aging seems likely if the behaviors have slow, gradual onset, as is the case with cognitive impairment. Thus, this CG population is at increased risk for misunderstanding the diagnosis and/or misaligning their attributions with the identified symptoms.
Stage of Caregiving
As dementia progresses, behavior problems and functional declines are evident in more basic domains of daily life, increasing the probability that CGs will view these problems as consequences of the disease (Ross et al., 1997). As outlined in the Global Deterioration Scale (GDS) developed by Reisberg, Ferris, de Leon, and Crook (1982), individuals with dementia move through seven stages of decline, ranging from “normal” (stage 1) to “very severe” (stage 7). Although cognitive deficits are the hallmark of dementing diseases, Reisberg et al. and others (see Eisdorfer et al., 1992) have found that functional deficits increase as CRs move toward the latter stages. The operational
13 daily living (IADL) and later impairment in basic ADLs. Specifically, needing substantial assistance with two or more ADLs is a commonly used criterion for eligibility for nursing home placement (Hendrickson & Kyzr-Sheeley, 2008). The Eisdorfer et al. (1992) study showed that a wide range of deficits occur throughout the disease, but by stage 4
(moderate decline), one-third of their sample had impairments in three ADLs. Although dementia progression is a heterogeneous process, dysfunction in daily personal care is a hallmark that requires more assistance from others. Thus, in the present study, the presence of marked problems in ADL function should lead CGs to attribute cognitive problems to dementia with more certainty.
Summary of Existing Literature
The present study extends previous work on symmetry of symptom identity and causal attributions in two ways. First, CGs are making attributions about another person, an area that has been understudied in research using the CSM, which has primarily focused on understanding illness in oneself. Second, in CRs with cognitive impairment, the symptoms that CGs see are often very ambiguous changes in everyday behavior, leading to great difficulty identifying the cause of changes in their loved ones (Polk, 2005). Instead, observers become aware of the fact that something is merely different. In the present study, it was expected that this finding would manifest itself through reports of multiple behavioral problems with less clarity about the cause of the CRs’ difficulties.
In short, the pathway through which CGs arrive at action involves identifying symptoms, making sense of those symptoms, and ultimately creating a coherent cognitive frame that implicates an appropriate course of behavior (whether that be help-seeking, providing care, decision-making, etc.). Previous studies indicate that CGs’ beliefs about
what causes behavioral problems play a central role in what they eventually decide to do (Fopma-Loy & Austin, 1993; Fopma-Loy & Austin, 1997). Multiple people can see similar symptoms, but each will end up with different proclivities to action if they frame the problem differently. Furthermore, CGs make many, often inconsistent, attributions for problem behaviors in CRs. Therefore, they may not be acting proactively or effectively on the CR’s behalf if causal explanations are inconsistent with the perceived symptom identity. Moreover, they may not be able to cope or embrace their caregiving role in a healthy way because of these incorrect beliefs.
In a sense, symmetry could be viewed as analogous to a “differential diagnosis” process in which the goal is to align symptoms with one explanation that is correct. Existing research has shown that CGs have difficulty with attributions, but they can identify pertinent symptoms when asked. The symptoms of dementia are broad and encompass multiple domains, making complex schemas a likely possibility for CGs. According to the CSM, achieving symmetry is a motivational process. Thus, it is expected that CGs who have sought counseling are somewhere in the midst of that process (but are not 100% accurate in their appraisals).
CG intervention programs may benefit from exploring and ultimately shaping CGs’ schemas in a way that aligns with appropriate action. Armed with knowledge of CGs’ symptom identity and causal explanations, clinicians can structure interventions more effectively by targeting misattributions and providing education around “naming” the problem correctly. As stated by Anderson et al. (2005), “knowing existing mental models is a starting point for reframing them in positive ways” (p. 1019). The present study provides data on the frames CGs hold at the outset of counseling. These frames
15 were sufficient to prompt help-seeking in the form of CG counseling, thus offering the opportunity to inform clinicians and counselors about the starting point for help-seekers.
Critique of Existing Literature
Although research has examined the diagnostic process for older CRs, to date no studies have assessed the symmetry between CGs’ beliefs about the symptoms and their common-sense attributions for the problem. Moreover, most research to date has focused on symptom reports by CGs invited to participate in research studies (e.g., through Alzheimer’s Association support groups, lists of dementia CGs assembled by physicians) or CGs who have taken their CR for a memory screen (e.g., Streams et al., 2003). Less is known about CGs who seek help in the form of counseling. In addition, much of the existing literature is qualitative in nature, which affects the generalizability of findings. Furthermore, most studies rely on retrospective CG reports of symptoms. In other words, participants’ CRs have already received a dementia diagnosis at study onset, and CGs are asked to “think back” on what symptoms they observed early on. However, knowing this information up front has tremendous benefits for tailoring CG interventions in
counseling, and for guiding CGs to seek help from the appropriate sources.
The present study aimed to understand the cognitive frames held by a large, help-seeking sample of informal (unpaid, primarily family) CGs who were at various stages of the caregiving career. Specifically, this project investigated the nature of, and symmetry between, CGs’ symptom identity and causal attributions for the CRs’ problems at the time they sought help from an established CG counseling program, the Aging Families and Caregiver Program, housed within a community-based mental health clinic. CGs’ descriptions of the problem (e.g., the CRs’ behavior problems and capabilities) and the
labels CGs use (e.g., their attributions for the source of the problem) were explored. Results address the following questions: In family CGs who seek help from a community mental health clinic, how are the CRs’ behavioral difficulties described? What do family CGs believe is the source of the problems that concern them? What is the degree of symmetry between symptom identity (behavior descriptions) and causal attributions? How do CGs’ views differ in cases of mild or more advanced CR cognitive impairment?
Hypotheses
Study predictions included:
1) CGs for CRs with cognitive impairment will report that the CR has problems with cognitive function along with multiple related problems with mental health, personality change, and decline in daily functioning (as well as ADLs and IADLs) that are common in dementias.
2) CGs will make multiple attributions for the CR’s problem, including dementia, normal aging, personality problems, depression, medications, and medical illness.
3) The relationship between symptoms and causes will be evaluated to test whether this sample of CGs have achieved symmetry in their illness representation for their CR. I hypothesize that the behavior and
ADL/IADL subscales will correlate with multiple attributions, but most strongly with the symmetric attribution. Specifically:
a) Cognitive problems will correlate with dementia attributions more strongly than with others.
17 b) Mental health problems will correlate with depression attributions
more strongly than with others.
c) Health and functioning problems will correlate with medication and medical illness attributions more strongly than with others. d) ADL dysfunction will correlate with medical illness and dementia
attributions but not other attributions.
e) IADL dysfunction will correlate with dementia attributions but not other attributions.
f) Personality attributions will relate similarly to all behavioral descriptions.
4) The symmetric symptoms will be significant predictors of the symmetric attribution even after variance from other symptoms has been accounted for. Specifically:
a) The cognitive subscale, ADLs, and IADLs will add significantly to a regression equation predicting dementia attribution ratings after other behavior problem subscales are entered.
b) The mental health subscale will add significantly to a regression equation predicting depression attribution ratings after other behavior problem subscales are entered.
c) The health/daily functioning and ADL subscales will add significantly to a regression equation predicting medical illness attributions.
d) Normal aging and personality attributions will not relate systematically to any of the behavior problem subscales.
5) Finally, I hypothesize that more advanced stages of cognitive decline (operationalized as lower levels of daily functioning) will elicit stronger endorsement of dementia attributions and a reduced likelihood of multiple
CHAPTER 2
METHODOLOGY
Dataset
This study utilizes data derived from a de-identified client dataset of demographic and intake information collected from the files of clients served through the Aging
Families and Caregiver Program at the CU Aging Center (CUAC) from 2005 to present. Data collection and storage was in compliance with HIPAA regulations, and was
completed in collaboration with the Psychology Department under the approval of the Institutional Review Board at the University of Colorado at Colorado Springs (UCCS). Faculty and graduate researchers are permitted access to this dataset for individual and ongoing projects.
Participants and Design
Participants included 218 family members of older adults who sought services within the Aging Families and Caregiver Program at the CUAC with concern about an older adult CR. The selection process is described below. The CUAC is a community-based mental health clinic serving older adults in the Colorado Springs area. Participants were referred to this clinic from agencies or service providers [e.g., physicians, senior citizens center, the local Area Agency on Aging (AAA)], or were self-referred. All participants included in this dataset signed a consent form agreeing to the use of their
data for research purposes (see Appendix A). They were not actively recruited to participate in this study; instead, upon consenting for services they were given the opportunity to participate by allowing their de-identified information to be used to expand our understanding of caregiving processes and shape clinical interventions. Due to the archival nature of this dataset, information is not available on the number of CGs who declined to participate.
General demographic information is presented in Table 1. The “typical” CG in our study was an adult child who was female, 58 years old (range for entire sample 26-87 years) and had completed at least some college. Most CGs were married (n = 158; 72%), and many were still employed outside of the home (n = 89; 41%). CGs had an average annual income of $45,338 (range $0-$500,000). The “typical” CR was female, 79 years old (range 34-101 years) and had a high school education. The majority of CRs were married (n = 90; 41%) or widowed (n = 83; 38%) and a minority were single or divorced [n = 6 (3%) and n = 26 (13%), respectively]. Almost all CGs and CRs were Caucasian.
In terms of the caregiving-specific variables, 61% of the CGs were adult children, 32% were spouses, and 6% were other relatives. On average, CGs had been providing care for 2.5 years (M = 31 months, SD = 44.5; range 1 month-40 years). Forty-five percent of the CRs were living with the CG, while 38% lived at home and 15% were residing in a facility (see Table 2 for demographic data specific to the caregiving situation).
Cognitive impairment status. Participants were selected for this study based on evidence in the clinical chart that cognitive impairment was present. A two-phase process was used to select cases from the broader client pool (N = 273 in full CG dataset). First,
21 Table 1
Caregiver and Care Recipient General Demographics (All)
CG CR
N % N %
Education
High School or Less 18 18 % 36 47 %
Some College/Trade 33 33 % 10 13 % College Degree 34 34 % 19 25 % Advanced Degree 16 16 % 12 16 % Ethnicity Euro-American 186 92 % 187 94 % African Am 6 3 % 8 4 % Hispanic 7 4 % 4 2 % Other 3 2 % 1 1 % Sex Female 176 80 % 135 62 % Male 43 20 % 83 38 % Employment status Employed 89 41 % 5 2 % Non-employed 128 59 % 208 98 % Marital status Married/Cohabitating 158 73 % 90 44 % Divorced/Separated 29 13 % 26 13 % Widowed 9 4 % 83 41 % Never married 22 10 % 6 3 % M (SD) Range M (SD) Range Age 58 years (10.89) 26-87 years 79 years (10.20) 34-101 years Income $45,339 (44,796) $0-$500,000 n/a n/a
Note. CG education, CG ethnicity, and CG income information was missing for a subset of the sample. Data reported here n = 101 for CG education, n = 202 for CG ethnicity, and n = 194 for CG income. CR age, CR education, CR ethnicity, and CR marital status also had reduced sample sizes: n = 192 for CR age, n = 77 for CR education, n = 200 for CR ethnicity, and n = 205 for CR marital status. Income data was not available for CRs. All other variables: N = 218.
cases were coded yes/no based on whether the clinician’s intake report noted concerns about the CR displaying signs of cognitive impairment (based on interviews with the Table 2
Caregiving Specific Demographics (All)
N % CR Living Arrangement At home 83 39 % With CG 97 46 % Nursing Home 5 2 % Assisted Living 27 13 % CG Relationship to CR Spouse 69 32 % Adult Child 132 61 % Other Relative 12 6 % Other 5 2 % Stage of Caregivinga Early 76 52 % Mid/late 71 48 % Referral Source AAA 64 33 %
Physician/Health Care Provider 26 13 %
Senior Center 8 4 %
Self-Referral 43 22 %
Mental Health Provider 9 5 %
Agency (Silver Key) 4 2 %
Other 36 18 %
M (SD) Range
CG Timeline 30.5 months
(44.50)
1 month-40 years
Note. CG Timeline, Referral Source, and Stage of caregiving information was missing for a subset of the sample. Data reported here n = 186 for CG Timeline, n = 196 for Referral Source, and n = 147 for Stage of Caregiving. All other variables: N = 218. a
Stage of Caregiving was computed based on CG reports of CR ADL difficulties. Those who had 0 or 1 ADL difficulty (indicated by a rating of ≥ 4) were labeled early, and those with 2 or more ADL difficulties were labeled mid/late.
23 CGs). Second, a validation process was conducted to ensure the accuracy of clinician reports of cognitive impairment and inclusion in this study. In this second phase, all client files were examined for the date of intake, clinician rating of CR cognitive impairment (yes/no), whether the CR had a neuropsychological evaluation, and the result of the assessment. This process was completed on 199 cases. Of these, 43 had copies of the CR’s neuropsychological evaluation in the file; the remaining 155 were not included in the validation process either because they did not have neuropsychological data on the CR, or made mention of an evaluation but did not have a copy of the report in the chart. Of the 43 with reports in the chart, sixteen of the CRs completed the assessment before intake and thus had neuropsychological data at the time of intake. In the other 27 cases, the CR completed the assessment after the intake session, providing a subsample from which to determine clinician accuracy in predicting cognitive impairment. Twenty-six of the 27 CRs were found to have cognitive impairment, thus confirming clinician ratings. In one case, the clinician rated the CR as having cognitive impairment, but the
neuropsychological report found no evidence of cognitive deficits. Based on this subsample, I estimate that clinicians were 96% accurate in their assessments of CR cognitive impairment at intake, making these ratings a useful criterion for selection into this study.
Procedure
Clients in the Aging Families and Caregiver program initially scheduled an appointment to meet with a CG therapist through the scheduling coordinator at the CUAC. At the intake appointment, clients completed a consent form authorizing
be used for research. Initial assessment forms were completed, and an in-person intake interview assessed the CG’s views of the CR, the caregiving situation, and the CG’s own responses to the situation. The portion of the assessment data that was used in this study are detailed below. Intake procedures took approximately two to three hours with a trained, supervised student clinician. CGs were not compensated for research
participation, but up to six counseling sessions were available at no cost funded through the Area Agency on Aging, after which sessions were billed on a fee-for-service basis at rates determined by a sliding scale based on income. All procedures were approved by the Institutional Review Board at UCCS as part of a de-identified archival dataset (see Appendix B for IRB approval).
Data extraction and management. Data were accessed and entered from paper charts (2003-09) and an electronic record system used by the CUAC (2009-11) by undergraduate lab assistants trained in clinical data entry. Each file was examined individually and culled for all relevant data, including the symptom scales, attributions, and demographic variables. Data integrity was assured through both double data entry by two independent enterers, and spot checks were completed periodically for dataset accuracy by the graduate-level data manager and the Aging Families and Caregiver program director.
Missing data. Mean substitution was conducted for participants missing small amounts of data. Specifically, the mean of each individual’s responses on a specific subscale were substituted for missing items if no more than 20% of items on that subscale were missing. Of those missing data, means were substituted for one or two items (M =
25 1.57, SD = .81). Participants with missing data that surpassed this rule were excluded from analyses.
However, several participants were missing all of the ADL/IADL ratings, yielding a reduced sample size for this measure (n = 147) because it was added to the clinical protocol after data collection had begun. All analyses that include this scale were conducted with this reduced sample. T-tests examining differences between those who were missing data (n = 152) and those whose data were complete (n = 66) on the
Behavior Problem Checklist, ADL/IADL scales, and attributions revealed no significant differences (all ps > .05).
Materials
CGs completed several questionnaires regarding their experiences and perceptions of the CR’s problems:
Informed consent. At intake, CGs were given an informed consent sheet to read and sign indicating that they agreed to participate in CG therapy. Within this, CGs also consented to their clinical data being used for research purposes (Appendix A).
Treatment was not dependent on participation in research, and this was communicated explicitly to all CGs seeking help at the CUAC. CGs were given the opportunity to decline having their data used for research.
Intake form. CGs described demographic characteristics of themselves and their CRs, including age, sex, level of education, marital status, ethnicity, employment status, living arrangement, relationship between the CG and the CR (e.g., spouse, adult child), income, referral source, and their help-seeking timeline (e.g., how long they were
engaged as a CG before seeking therapy services). Although not used in the present study, the intake also included a narrative description of their primary concerns, reasons for seeking services at this time, family involvement and relationships, current
community and health contacts of the CR, and the clinician’s concerns for both the CG and CR (Appendix C).
Behavior problem checklist. CGs rated their perception of problem intensity for 34 symptoms, activities and behaviors on a 7-point Likert-type scale (1 = no problem; 7 = frequent problem; see Appendix D). This scale is a modification of the Revised Memory and Behavior Problem Checklist (Teri, Truax, Logsdon, Uomoto, Zarit, & Vitaliano, 1992), with additional behavior items and altered response format to a single rating for each item (extent to which it is a problem). Instructions for the questionnaire asked, “In what areas do you find your family member having difficulty? Please rate the degree of problems your family member is experiencing by circling the appropriate number in each of the following areas on a scale from 1 (no problem) to 7 (frequent problem or intense problem). Place a check beside the areas of functioning that have changed within the past four to six months.” There was an additional space for CGs to note other behaviors that are not listed and to make the same problem intensity rating for this item as well.
Table 3 presents the results of an exploratory factor analysis conducted on this measure using a principal components analysis (PCA) of the initial CG dataset as a pilot study (N = 273; see Pallant, 2010 for approach to data analysis). The data were
appropriate for factor analysis, given the correlation matrix (many correlations >.3), Kaiser-Meyer-Olkin value of .83, and the statistical significance of Bartlett’s Test of Sphericity (p < .001). Initial PCA revealed the presence of nine components (eigenvalues
27 Table 3
Behavior Problem Checklist: Scale Statistics
Subscale/Items Scale α Item-total R α without item Factor Factor loading Cognition .86 Planning .78 .80 1 .85 Memory .73 .82 1 .81
Follow through on plans .69 .82 1 .76
Concentration .75 .81 1 .83 Appointments .50 .89 1 .56 Mental Health .89 Anxiety/Worry .61 .88 2 .73 Mood .73 .88 2 .59 Depression .66 .88 2 .56 Sadness .62 .88 2 .59 Irritability .70 .88 2 .75 Apathy .56 .89 2 .40 Withdrawal .56 .89 2 .45 Personality Changes .60 .88 2 .64 Suspiciousness .58 .88 2 .68 Inappropriate Behavior .60 .88 2 .69 Aggressive Behavior .57 .88 2 .73
Health & Daily Functioning .87
Energy Level .61 .86 3 .70 Medical Problems .53 .86 3 .64 Isolation .53 .86 3 .48 Driving .39 .87 3 .46 Household Tasks .56 .86 3 .65 Falls/Balance .62 .86 3 .73 Safety Issues .58 .86 3 .58 Sleep .51 .86 3 .53 Self-care/Hygiene .58 .86 3 .53 Medical Care .53 .86 3 .54 Nutrition .62 .86 3 .71 Appetite .56 .86 3 .68 Incontinence .48 .87 3 .46
Items Deleted in Final Model Suicidal Thoughts
Homicidal Thoughts Social Relations Decision-Making Finances
Note. PCA conducted on the full clinic sample of CGs (N = 273). Scale reliabilities conducted on CG sample for this study (N = 218).
> 1) accounting for 69% of the variance. However, the scree plot showed three major components, and thus three factors were retained. Using an oblique rotation, three factors accounted for 45% of the variance. However, five items were then removed due to the fact that they did not load on any factor (or they loaded on more than one factor) after suppressing values > .40: decision-making problems, suicidal thoughts, homicidal thoughts, social relations, and finances. All remaining item loadings were > .40. In the final model, the three components accounted for 50% of the variance, with component one contributing 31%, component two contributing 10%, and component three
contributing 9% (see Peterson, 2000, for meta-analysis of the amount of variance commonly accounted for in PCA). These constructs are represented as the following subscales: Health/daily functioning (13 items, Cronbach’s α = .87), cognition (5 items, Cronbach’s α = .86), and mental health (11 items, Cronbach’s α = .89).
The “appointments” item on the cognition subscale appeared to be a potent variable, as the subscale alpha decreased when this item was removed. Recent research has identified that the ability to remember appointments is “characteristic of clinically significant cognitive impairment” (Brown, Devanand, Liu, & Caccappolo, 2011, p. 617), perhaps making this variable an especially important one for CGs in describing their CRs’ difficulties.
Activities of daily living scales. Instrumental (8 items, Cronbach’s α = .87) and basic (7 items, Cronbach’s α = .92) activities of daily living items were rated for the amount of assistance needed by CRs on a 7-point Likert-type scale (1 = no assist; 7 = full assist; see Appendix E). Instructions read, “Please rate the degree of problems your family member is experiencing by circling the appropriate number in each of the
29 following areas on a scale from 1 (no assistance) to 7 (full assistance). Place a check beside the areas of functioning that have changed within the past four to six months.” Instrumental items included financial management, transportation, medication
management, and other skills needed to live independently. Basic activities included bathing, dressing, ambulation, toileting, and other fundamental self-care skills.
Again, PCA was conducted on this scale (see Table 4) with the larger sample (N = 273). The data were appropriate for factor analysis, given the correlation matrix
contained many correlations >.3, Kaiser-Meyer-Olkin value of .88, and the statistical significance of Bartlett’s Test of Sphericity (p < .001). Initial PCA revealed three factors, but the scree plot revealed two major components. Thus, two factors were retained. A Varimax rotation was used, and revealed each variable loaded on only one factor with strong loadings (all > .60). The two factors explained a total of 60% of the variance, with factor one contributing 46% and factor two contributing 14%. The data support the use of the ADL and IADL items as separate subscales.
Caregiver attributions checklist. CGs rated their degree of certainty that specific processes (e.g., normal aging; dementia/Alzheimer’s disease; medical illness) caused their CR’s problems (see Appendix F). The response format for this measure was a 7-point Likert-type scale (1 = unlikely; 7 = almost certain). Instructions read, “Please give your opinion about what you think is creating these problems.” CGs are given the opportunity to rate all six causes, making multiple attributions for the problem possible.
A rating of ≥ 4 was chosen as a cut-off for many findings using the Behavior Problem Checklist, ADL/IADL Scale, and the Caregiver Attributions Scale. Although this midpoint value may be somewhat ambiguous, CGs who rate their CR as ≥ 4 on
Table 4
Basic and Instrumental Activities of Daily Living: Scale Statistics
Subscale/Items Scale α Item-total R α without item Factor Factor loading ADLs .92 Bathing .82 .90 1 .76 Grooming .76 .91 1 .77 Ambulation .67 .92 1 .66 Dressing .83 .90 1 .82 Transfers .77 .91 1 .85 Eating .63 .92 1 .71 Toileting .81 .91 1 .91 IADLs .87 Heavy Chores .61 .86 2 .65 Transportation .62 .86 2 .89 Financial Management .58 .86 2 .62 Shopping .70 .85 2 .82 Food Preparation .71 .85 2 .78 Medication Administration .57 .87 2 .60 Laundry .72 .85 2 .78 Telephone .56 .87 2 .61
Note. PCA conducted on the full clinic sample of CGs (N = 273). Scale reliabilities conducted on CG sample for this study (N = 147).
symptoms are indicating that the CR has at least somewhat frequent problems with that symptom. These “somewhat frequent” symptoms are likely influencing CGs’ views, as they are not rare or uncommon occurrences in day-to-day life with the CR. In terms of the attribution ratings, the cut-off of ≥ 4 was again chosen because these ratings suggest that the CG is at least somewhat certain that the attribution is a cause. Because those attributions rated ≤ 3 are endorsed less strongly, they may be less potent causal labels for CGs.
CHAPTER 3
RESULTS
Statistical analyses were done using SPSS for Windows (Version 19). Preliminary examination of the data revealed that items on the Behavior Problem Checklist,
ADL/IADL scales, and the Caregiver Attribution Checklist were appropriate for further analysis.
Symptoms and Attribution Ratings: Descriptive Findings
Descriptive analyses summarize the symptom reports (behavior problem scales and activities of daily living scales shown in Tables 5 and 6) and attribution data (Table 8). Correlations among symptoms and attributions are in Tables 7 and 9, respectively.
Behavior problem checklist items. CGs endorsed many symptoms as problematic for their CRs (see Table 5). The highest ratings were assigned to CR problems on the Cognition subscale (all item Ms rated ≥ 4). Specifically, problems with planning (M = 5.4, SD = 1.7), follow-through (M = 5.3, SD = 1.7), memory (M = 5.3, SD = 1.7), and concentration (M = 5.1, SD = 1.7) were endorsed as especially problematic for our sample of CRs. In fact, one-fourth of the sample used the extreme score on each of these items, indicating they were a frequent problem for the CRs. Seven of the 11 (64%) mental health items including mood, anxiety/worry, irritability, sadness, depression,
Table 5
Behavior Problem Checklist Items: Descriptives
Subscale/Items M SD Range % ≥ 4 % extreme response (7) Cognition 5.14 1.42 1.40-7 80% 11% Planning 5.45 1.69 1-7 86% 36% Memory 5.34 1.62 1-7 86% 30%
Follow through on plans 5.30 1.66 1-7 85% 31%
Concentration 5.14 1.66 1-7 84% 24% Appointments 4.49 2.20 1-7 67% 27% Mental Health 4.11 1.38 1-7 51% 1% Anxiety/Worry 4.95 1.83 1-7 76% 24% Mood 4.88 1.81 1-7 78% 25% Depression 4.64 1.93 1-7 74% 20% Sadness 4.60 1.86 1-7 72% 20% Irritability 4.40 1.91 1-7 66% 16% Apathy 4.22 2.06 1-7 65% 17% Withdrawal 4.15 2.20 1-7 62% 21% Personality Changes 3.73 2.02 1-7 53% 12% Suspiciousness 3.61 2.05 1-7 51% 12% Inappropriate Behavior 3.30 2.16 1-7 45% 11% Aggressive Behavior 2.76 2.03 1-7 34% 8%
Health & Daily Functioning 4.15 1.34 1.08-7 58% 1%
Energy Level 4.94 1.85 1-7 79% 26% Medical Problems 4.63 2.00 1-7 71% 24% Isolation 4.62 2.04 1-7 72% 24% Driving 4.50 2.48 1-7 64% 39% Household Tasks 4.45 2.15 1-7 67% 24% Falls/Balance 4.24 2.18 1-7 63% 21% Safety Issues 4.19 2.07 1-7 62% 18% Sleep 4.06 2.11 1-7 62% 16% Self-care/Hygiene 3.96 2.17 1-7 57% 17% Medical Care 3.71 2.13 1-7 56% 14% Nutrition 3.70 2.17 1-7 53% 13% Appetite 3.54 2.22 1-7 50% 14% Incontinence 3.44 2.23 1-7 45% 13%
33 apathy, and withdrawal were rated above the midpoint for the scale (all Ms ≥ 4). CGs also endorsed 69% (9 of 13) of the health and daily functioning items as problematic for their CRs, including safety issues, household tasks, self-care/hygiene, driving, medical
problems, falls/balance, sleep, energy level, and isolation (all Ms ≥ 4). Interestingly, 39% of our sample used the extreme score for reporting that driving was a frequent problem (score of 7 on the 7-point scale).
Activities of daily living items. CGs provided substantial assistance with IADLs, as evidenced by mean scores of four and above on all items of this subscale (see Table 6).
Table 6
Basic and Instrumental Activities of Daily Living Items: Descriptives
Subscale Items M SD Range % ≥ 4
% extreme response (7) ADLs 2.76 1.74 1-6.86 28% 1% Bathing 3.16 2.41 1-7 41% 17% Grooming 3.03 2.12 1-7 42% 8% Ambulation 2.96 2.07 1-7 36% 8% Dressing 2.90 2.21 1-7 39% 11% Transfers 2.47 2.07 1-7 19% 9% Eating 2.40 1.85 1-7 28% 3% Toileting 2.39 2.03 1-7 28% 5% IADLs 5.24 1.59 1-7 79% 12% Heavy Chores 5.73 2.06 1-7 83% 63% Transportation 5.70 2.16 1-7 83% 65% Financial Management 5.60 2.07 1-7 84% 57% Shopping 5.50 2.07 1-7 82% 55% Food Preparation 5.24 2.12 1-7 78% 46% Medication Administration 5.16 2.30 1-7 76% 48% Laundry 4.88 2.36 1-7 72% 43% Telephone 4.09 2.35 1-7 59% 24%
Most assistance was provided to CRs with transportation (M = 5.7, SD = 2.2), heavy chores (M = 5.7, SD = 2.1), and financial management (M = 5.6, SD = 2.1), but many CGs also helped their loved one with shopping, food preparation, and administering medications (all Ms > 5).
Means for the ADL subscale were lower, indicating that CRs in our sample
needed less help in these basic domains. Over one-fourth of our sample reported that their CR did not need any assistance with ADLs of ambulation, bathing, dressing, and
grooming, while over one-third of the CGs did not provide assistance with transfers, toileting, or eating (as indicated by a score of 1 on the 7-point scale). However, 33% (n = 71) of the CRs needed substantial help with ADLs (indicated by mean scores ≥ 4 on 2 or more ADL items) of whom 19% were placed in a nursing home or assisted living as compared with only 8% of those with minimal ADL needs. However, this difference in placement was not statistically significant (χ² < 1).
Correlational analyses of the symptom reports showed some relatively strong relationships, suggesting shared variance among the subscales for the sample as a whole (see Table 7). The strongest relationships were found between mental health and
health/daily functioning (r = .60, p < .01), ADL and IADL (r = .56, p < .01), and cognition and IADL (r = .49, p < .01). The weakest relationships were between mental health scores and activities of daily living (both ADL and IADL; r = .21, p = .01 and r = .14, p = .10, respectively).
Attributions. Table 8 displays the means and standard deviations for the attribution ratings. A rating ≥ 4 indicates that the CG believes with some degree of certainty that the attribution in question is causing the impairment in the CR. The most
35
Table 7
Correlations among Behavior Problem and ADL/IADL Subscales (All)
Cognition Mental Health
Health/Daily
Functioning ADL IADL
Cognition 1 .39** .40** .35** .49** Mental Health 1 .60** .21* .14 Health/Daily Functioning 1 .47** .30** ADL 1 .56** IADL 1 Note. * p < .05; ** p < .01. N = 218 Table 8
Attribution Items: Descriptives
Attribution Item M SD Range % ≥ 4
% extreme response (7) Dementia/AD 4.89 2.31 1-7 73% 41% Depression/Anxiety 4.60 2.13 1-7 71% 26% Normal Aging 4.55 2.20 1-7 71% 29% Medical Illness 4.17 2.40 1-7 61% 29% Medication 3.44 2.23 1-7 47% 13% Personality Problem 3.43 2.28 1-7 45% 17% Multiple Attributions (≥ 4) 3.67 1.49 0-6 54% 1 Attribution Rating n = 16 7% 2 Attribution Ratings n = 35 16% 3 Attribution Ratings n = 49 23% 4 Attribution Ratings n = 49 23% 5 Attribution Ratings n = 38 17% 6 Attribution Ratings n = 30 14%
Note. All items rated from 1 (unlikely) to 7 (very likely) except for Multiple
Attributions which is the number of attribution items with a rating of 4 or more. N = 218.
common attribution for observed problems was dementia/AD (M = 4.9, SD = 2.3), with 41% of respondents feeling certain about that cause (indicated by a rating of 7 on the 7- point scale; 72% made ratings ≥ 4). Normal aging (M = 4.6, SD = 2.2),
depression/anxiety (M = 4.6, SD = 2.1), and medical illness (M = 4.2, SD = 2.4) were also commonly endorsed as causing the problems, with approximately one-fourth of the sample certain that each of those was a cause (rating of 7). Least commonly endorsed attributions for the problems were medication and personality problems, with lower means (Ms = 3.4). Indeed, a third of the CGs were certain that these two factors did not create the problems identified. However, 13% were certain that medications and 17% were certain that personality problems caused the CRs’ problems (rating of 7).
The vast majority of CGs endorsed more than one cause as probable (rating > 4 on 7 point scale of certainty). The mean number of causes endorsed was 3.7 (SD = 1.5), with 54% of CGs rating more than three items. Almost one-third of the CGs (n = 68; 31%) rated five or all six of the causes as highly likely.
The attribution ratings were also correlated with each other though only modestly (see Table 9). The strongest relationship was found for medical and medication
attributions (r = .48, p < .01). Personality correlated with all of the other attributions.
Symptom-Attribution Symmetry
Correlation analyses were used to determine whether a reciprocal relationship between symptom reports and attributions exists in this sample of CGs. Symptom reports include the three behavior subscale scores and reports of problems in daily functioning (ADL and IADL subscale scores). Before testing our symmetry hypotheses, four outliers
37 were identified using Mahalanobis distance. These cases were subsequently removed for all symmetry analyses.
Table 9
Correlations among Attributions (All) Normal
Aging Dementia Depression Medical Medication Personality
Normal Aging 1 -.14* .10 .06 .01 .14* Dementia 1 .16* -.19** -.09 .17* Depression 1 .08 .30** .35** Medical 1 .48** .29** Medication 1 .29** Personality 1 Note. * p < .05; ** p < .01. N = 218.
Correlational Findings. Symptom reports were predicted to correlate with multiple attributions (Hypothesis 3) but have the strongest relationship with the symmetric attribution. In support of the prediction that many attributions would be
related to symptoms, a broad pattern of relationships were evident among the correlations shown in Table 10. Dementia attributions were significantly correlated with mental health problems (r = .17, p = .01) and IADL difficulties (r = .23, p = .005) in addition to cognitive symptoms. Depression attributions were significantly correlated with cognitive symptoms (r = .17, p = .01) and problems with health/daily functioning (r = .31, p < .01) in addition to mental health symptoms. Medical and medication attributions were
correlated with reports of mental health problems (r = .14, p = .05 and r = .21, p = .002, respectively) in addition to health and daily functioning difficulties. Normal aging was
