ABSTRACT. Predictors of Nursing Home Placement for Home Care Consumers. By Ian Matthew Nelson

Transcription

1 ABSTRACT Predictors of Nursing Home Placement for Home Care Consumers By Ian Matthew Nelson The purpose of this study was to identify predictors of nursing home placement for individuals receiving home care services through a locally funded levy program. Utilizing the data gathered at baseline, newly enrolled individuals were followed for 2.5 years. Logistic regression analysis suggested that increased age, poor self-reported health, limitations in bathing and meal preparation, and higher medical expenses significantly increased the chances of nursing home placement. Recommendations for future research and policy implications are discussed.

2 Predictors of Nursing Home Placement for Home Care Consumers A Thesis Submitted to the Faculty of Miami University in partial fulfillment of the requirements for the degree of Master of Gerontological Studies Department of Sociology and Gerontology By Ian Matthew Nelson Miami University Oxford, Ohio 2003 Advisor Suzanne R. Kunkel, Ph.D Reader Robert A. Applebaum, Ph. D Reader Shahla Mehdizadeh, Ph.D

3 Table of Contents Chapter I-Introduction 1 Literature Review 5 Chapter II-Methodology 16 Sample 16 Predictor Variables 19 Dependent Variables 23 Chapter III- Results 25 Overall Sample Characteristics 25 Nursing Home vs. Remaining Active Comparisons 29 Analysis 33 Chapter IV-Discussion 39 Limitations 41 Future Recommendations 42 Summary & Conclusion 43 References 46 ii

4 List of Tables Table 1- Studies Examining Predictor Variables of Nursing Home Placement 7 Table 2- Studies Examining Predictors Contributing to Nursing Home Placement for Consumers Receiving Formal Care 11 Table 3- Individuals Enrolling in Program during January 1 st, 2000 And March 31 st, 2000, by Outcome 18 Table 4- Recoded Variables 21 Table 5- Description of Predisposing Characteristics, by Outcome 26 Table 6- Description of Enabling Characteristics, by Outcome 28 Table 7- Description of Need Characteristics, by Outcome 30 Table 8- Backwards Stepwise Logistical Regression: Final Model of Variables Predicting Nursing Home Placement 37 iii

5 Acknowledgments How great it feels to finally finish my thesis and the M.G.S. program. I could not have completed my journey through graduate school without the continued support of my family, friends, and colleagues. I thank my M.G.S. professors who, without hesitation, gave me a second chance. I will always be grateful to you for not giving up on me. I would like to also thank my great aunts, Ruth and Grace. Your love and support helped to nudge me in the direction of gerontology. You showed me how wonderful, yet trying and difficult at times, the experience of aging could be. Even though you are far away from me, you will always be in my thoughts. I would like to thank Dr. Kunkel for all of her guidance, insight, and patience. I would also like to thank the rest of my committee for allowing me to constantly interrupt their busy days with questions concerning this paper. A big thanks to Ken Wilson at the Council on Aging of Southwestern Ohio for helping me create the data set and discussing the practical applications of the findings. Finally, I would like to thank Dr. M.L. Klotz at Susquehanna University. iv

6 Chapter 1 Introduction Currently, there are 35 million older adults living in the United States, nearly 12% of the population. By 2030, this number will double to 70 million. One factor that will contribute to this growth is the increasing number of baby boomers reaching 65. Another factor will be increased life expectancy. The 85 and over population will double from 4.2 million in 2000 to 8.9 million by 2030 (Administration on Aging, 2002). As individuals age, there is an increased likelihood of disabilities or limitations. Assessing the effects of population aging, mortality, and disability, Kunkel & Applebaum (1992) proposed estimates of the future older disabled population. The most conservative model projection with the assumption of greater life expectancy and lower disability rates showed that by the year 2020 the population of older people with moderate to severe disability would increase by 80%. Older individuals who express the need for assistance with disabilities typically receive long-term care either through informal or formal care. According to Kane, Kane, and Ladd (1998), long-term care can be defined as assistance given over a sustained period of time to people who are experiencing long-term inabilities or difficulties in functioning because of disability (p. 4). Considering the future projections of moderate to severe disability, it is easy to surmise that there will be an increased number of frail older adults who will need some form of long-term care in order to remain independent in the community. This will create challenges for our current long-term care system. 1

7 Informal Care A majority of long-term care in the community is provided informally either by family members or friends (Mehdizadeh & Murdoch, 2003). According to Stone (2000) 95% of the non-institutionalized elders who have disabilities receive support from these sources. Caregivers provide personal care, homemaking, food preparation, companionship, and economic support in order to keep older adults at home. When the needs of older individuals become too great for caregivers, older individuals and their families can make a choice to access a variety of formal services to supplement informal care. Formal Care Formal long-term care includes services provided in institutions or at home by community agencies. At one time, vulnerable older adults received formal long-term care primarily in nursing homes. Over the past thirty years public expenditures on nursing homes have risen dramatically. Due to the increasing cost of nursing home care, along with preferences of older adults to receive care at home, states have attempted to create other alternatives (Lanspery, 2002; Stone, 2000). One alternative to nursing home care has been the implementation of home and community-based long-term care funded through a waiver of Medicaid. These alternatives allow individuals to receive supportive services congruent with their preferences to remain in the home. Older Individuals receive personal care, homemaking, home-delivered meals, and other services. Though home and community-based long-term care is essential in allowing frail older adults to remain in community, a drawback is that only a small percentage of 2

8 individuals are eligible to receive these services due to the strict financial and physical functioning requirements. These requirements were instituted due to policy-makers fear that older individuals would come out of the woodwork for home health care services if it were accessible to everyone (Lui et al, 1998). In Ohio, for example, individuals are assessed for program eligibility through preadmission screenings. An individual can only receive $ income per month and have roughly $ worth of assets in order to meet the income eligibility requirements. (Ohio Department of Aging, 2003). Furthermore, they must meet nursing home level of care; this requires the individual to have two Activities of Daily Living impairments and/or severe cognitive impairment. In addition to Medicaid-funded home care, Medicare provides some limited coverage of home and community based services. Many vulnerable older individuals are unable to qualify for Medicare home health because they do not have illnesses that require skilled care or care that requires medical intervention. Individuals who qualify for Medicare home health can only receive 100 visits of fully reimbursed services following a brief hospital stay. For those who have chronic illnesses or are in need of long-term care, this is not enough to remain independent. Those who are in need of services, but unable to qualify for the previously mentioned programs must either pay out-of-pocket for expensive services or place even greater pressures on the informal support network. These limited and costly options may be insufficient and could leave them with no other choice but to access nursing home care. Subsequently, older individuals who must pay out-of-pocket for nursing home care often spend down to meet Medicaid eligibility criteria. According to Weiner, 3

9 Sullivan, and Skaggs (1996), nearly 50% of residents admitted as private pay residents eventually spend down to Medicaid. A loss of community independence because of being unable to remain in their own homes also affects their quality of life (Kane, Kane, Ladd, 1998). Concerns over public expenditures and consumer preferences create a necessity to find ways to meet the needs of those who are at risk of unnecessary nursing home placements. Alternative Programs Some programs have been implemented in order to meet the needs of the vulnerable older population who do not qualify for state/ federal services. One innovative alternative is locally funded community based long-term care programs. Local programs that fund in-home services for older adults are unusual. In Ohio residents are given the opportunity to vote for tax levies or increases in property taxes in order to pay for programs not funded at the state or federal levels. Voters in an urban county in southwestern Ohio have chosen to fund in-home services for older adults among a variety of tax levies such as mental health services and school construction. Through tax-levies approved by county residents, older individuals who would otherwise receive no care except through informal means or high out-of-pocket expenditures receive necessary unskilled care such as homemaker services, home delivered services, and personal care services that help them to remain independent. One challenge of such programs is identifying older individuals who are at the greatest risk of losing their independence. By understanding the characteristics of individuals at entry, program administrators can consider targeting services such as 4

10 increased case management interventions and services towards older individuals in order to help them to continue to remain independent in the community. The purpose of this paper is to describe the characteristics of new enrollees of a locally funded long-term care program and to examine the trajectories for those receiving services. We will compare the characteristics of individuals who stay in the program to those who enter a nursing home. This research adds to literature on predictors of nursing home placement and contributes to the limited amount of research on alternative formal services for people not qualifying for state/federal assistance. Literature Review Identifying risk of nursing home placement has been an important topic for researchers and policymakers due to the high costs of nursing home care and older adults preferences to remain at home (Lanspery, 2002). Table 1 highlights a selection of the major studies and samples used to identify predictors. These studies have consisted of the general population of older adults (Cohen, Tell, Wallack, 1986; Wolinsky, Callahan, Fitzgerald, & Johnson, 1992; Jette, Branch, Sleeper, Feldman, & Sullivan, 1992; Freedman, 1996) and disabled community-living older adults (Lui, McBride, & Coughlin, 1994; Hanely, Alexcih, Weiner, & Kennel, 1990; McFall & Miller, 1992; Pearlman & Crown, 1992; Ondrich & Greene, 1990; Jette, Tennestedt, & Crawford, 1995). Researchers have also examined predictors of institutionalization for specific groups such as older individuals living in public housing (Black, Rabins, & German, 1999), living in urban or rural areas (Dwyer, Barton, & Vogel, 1994), and living with dementia (Smith, Kokmen, & O Brien, 2000). 5

11 A few studies have examined predictors of nursing home placement for samples specifically comprised of older individuals who were enrolled in home and community long-term care services; usually formal care usage was assessed through yes and no questions on national surveys (McFall & Miller, 1992; Hanely et al., 1990). Table 2 shows a selection of studies that have used samples constituted solely of long-term care clients. Tsuji, Whalen, & Finucane (1996) investigated reasons for nursing home 6

12 Table 1 Studies examining predictor variables of nursing home placement Studies Data Set Sample Characteristics Predictors Disabled Sample Lui, Mcbride, NLTCS 1 65 and over with age, living alone, Coughlin (1994) or more ADL or IADL cognitive impairment, impairments, community ADLs, formal care residents Hanley, Alecxih, NLTCS 65 or older 1 cognitive impairment, Weiner, & Kennel (1990) or more ADL or IADL formal care, not caregiver, impairments, community self-reported health, age, race resident McFall & Miller (1992) NLTCS age, race, IADLs, or older caregiver burden, Informal Caregiver Survey 1 or more ADL or IADL use of formal services impairments, community resident Pearlman & Crown (1992) NLTCS 65 or older 1 increasing age, or more ADL or IADL living alone, Race, impairments, community having ADL impairments, resident high levels of ADL impairments, caregiver is not spouse or child Ondrich & Greene (1990) NLTCCD 2 severely disabled over 65 advanced age, Caucasian, living alone, functional

13 Table 1 Studies examining predictor variables of nursing home placement Studies Data Set Sample Characteristics Predictors Sample of Jette, Tennestedt, Massachuesettes disabled and over 65, amount of informal care, Crawford (1995) Residents community resident males, living with caregivers General Population General Population Cohen, Tell, Medicare Beneficiaries 65 and over, Medicare age, confinement to bed, Wallack (1986) Recipients mobility issues, single, residents perceived health status Wolinsky, Callahan, LSOA 3 65 and over, increased age, lives alone, race national sample, Fitzgerald, Johnson (1992) community less informal support, Adls residents Jette, Branch, Sleeper, Sample of Massachusetts 65 and over prior admission to NH, Feldman, and Sullivan (1992) Residents regional sample, community ADL disability and Mobility issues residents Freedman (1996) New Haven EPESE 4 65 and over, marital status, national sample, community caregiver relationship residents

14 Table 1 Studies examining predictor variables of nursing home placement Studies Data Set Sample Characteristics Predictors Other samples Black, Rabins, Sample of Baltimore Public moderately disabled elders greater impairment in IADLs, German (1999) Housing Residents living in Public cognitive impairment, Housing medical conditions Smith, Kokmen, Dementia Patients/Mayo Clinic 65 or over, single, & O'Brien (2000) Minnesota with early onset change in assistance needed, dementia age of dementia onset Dwyer, Barton, NLTCS 65 or over with 1 or for urban clients: increased Vogel (1994) more ADL and IADL age, income, high ADL impairements impairments, less informal care 1 National Long-Term Care Survey 2 National Long-Term Channeling Demonstration 3 Longitudinal Study of Aging 4 Established Population for Epidemiological Studies of the Elderly

15 placement for a group of individuals receiving a mixture of medical care and long-term care from an outreach organization in Baltimore. Following individuals for 2.5 years, the researchers found that medical condition, and caregiver characteristics such as stress played a significant role in the utilization of nursing homes for this population. Bauer (1996) followed a group of Medicaid waiver program participants in Arizona for 2 years. The results showed that for individuals entering the program, increased age, Caucasian ethnic/group membership, and poorer cognitive functioning played a significant role in nursing home utilization. Hughes, Manheim, Edelman, and Conrad (1987) utilized a sample of vulnerable older adults who according to Medicare standards, were not considered skilled enough to receive home health care. In addition to medical supervision, the program provided low tech services such as meals, homemaking, companion, and emotional support to the clients in their homes. Regression analysis showed that after 4 years, marriage decreased the chances of nursing home placement, while advanced age and poorer cognitive functioning increased the chances of placement. Furthermore, impairments in bathing at baseline increased the chances of nursing home placement. Finally, Nocks, Learner, Blackman, and Brown (1986) compared characteristics of a sample of older adults receiving Home and Community based services with those receiving traditional services. HCBS clients were receiving case management and expanded services such as personal care, adult day care, home delivered meals among others. Findings suggested that being hospitalized before admission, living alone, and having higher ADL and IADL impairments contributed to nursing home placement for those receiving the expanded services. Overall, these studies suggest that there are different factors that either increase or decrease the 10

16 Table 2 Studies examining influence of formal services on outcomes of Nursing Home Placements Studies Data Set Sample Characteristics Predictors Tsuji, Whalen, Sample of Baltimore Area High level of disability number of medical conditions, Finucane (1996) Homebound clients caregiver characteristics such as Receiving Long-term care services stress, increased age Bauer (1996) Arizona Long-term Disability and income levels increased age, Caucasian, poorer Care Program cognitive functioning Nocks, Learner, South Carolina Must qualify for Medicaid prior hospitalization, living alone, Blackman, Brown (1986) Long-Term care Project waiver status higher level of ADL impairment, higher level of IADL impairment Hughes, Manheim, Chicago Hospital-based 60 or older, advanced age, poorer cognitive Edelman, Conrad (1987) Long Term Care Program with medical issues functioning, impairments in bathing

17 chances of nursing home placement for older individuals receiving home and community based long-term care. One limitation of analyzing predictors of nursing home placement is that different samples have been utilized for each study. The end result is difficulty confirming results and drawing broad conclusions about the relative effects of different risk factors (p. 787, Cohen, Tell, Wallack). Basically, it is hard to generalize the findings from one particular study to another. In addition, some attempt to categorize the wide range of variables used in all of these studies would be helpful. Anderson s Behavioral Model of Health Care Utilization: The Andersen Behavioral Model (1995) addresses the importance of social, economic, psychological, and medical conditions in health care utilization. Researchers have begun to use the model to identify factors that contribute to nursing home placement (Wolinsky et al.,1992; Borrayo et al., 2002; Jette et al., 1992; Greene & Ondrich, 1990). A test of the model is beyond the scope of this study, but the framework is useful for organizing variables and concepts used in this and previous studies. The model includes several categories of variables: predisposing, enabling, and need. These three categories provide a conceptual framework for existing literature and the current study. Predisposing factors Predisposing factors are items such as demographic characteristics of individuals that are separate from the medical conditions or resources that individuals have at the time of health care usage. These items are usually taken at baseline and do not change over time. The studies mentioned above have shown us that some 12

18 predisposing characteristics do play a role in nursing home placement. Increased age appears to be a strong predictor of nursing home placement regardless of whether or not the sample is disabled or receiving formal services. Findings on racial/ethnic group composition have shown mixed results. For samples of disabled older adults, racial ethnic group composition appears to be significant in explaining nursing home entry. Studies have shown that Caucasians are more likely to enter a nursing home than minorities. Researchers have suggested discrimination or family support as reasons for this. For instance, nursing home policies that discriminate against the entrance of minorities would lower the number of minorities in nursing homes. Next, it has been thought that minorities receive more informal support from family members than Caucasians. Only one of the studies using a sample of formal service clients found that Racial/Ethnic group factored in placement (Tsuji et al, 1996). Living arrangement also appears to be significant in the disabled population (Lui et al, 1994, Pearlman & Crown, 1994), but not as much so in the general population. Individuals who live with others are shown to have a decreased risk of placement probably because of the support provided on a daily basis at home whereas those who are alone have an increased risk due to potential lack of support when an event occurs. Enabling factors Everybody utilizing health care has different resources such as family supports and incomes that help them to access services. According to the model these resources are considered enabling characteristics. Marital status has not consistently been significant in predicting nursing home usage among disabled individuals. One study of disabled individuals shows that individuals with dementia and those unmarried 13

19 have a greater likelihood of nursing home placement (Smith et al., 2000). In another study of the general population of older adults, not married individuals had a greater chance of nursing home placement. It is suggested that marriage provides the disabled older adult a caregiver who will provide support. Information about the caregiver support network is also important, although the findings on these items are inconsistent. These findings appear to be dependent on sample characteristics with the presence of caregivers being more important for individuals with functional impairments. Individuals who have caregivers have a decreased likelihood of nursing home placement because social support from family and friends moderates the effects of declining health status and thus reduces institutionalization (Pearlman, 1994, p, 528). To the contrary, other studies have suggested that having caregivers who provide informal assistance increases the likelihood of nursing home placement when receiving formal services (Newman, Struyk, Wright, Rice, 1990). One reason for this could be that the caregivers are already experiencing a level of burden when accessing formal services. The formal services received are not enough to alleviate the caregiver burden. Need factors Functional disabilities or health problems that contribute to usage of services are considered need factors. Almost all studies have shown that Activities of Daily Living such as mobility and bathing and Instrumental Activities of Daily Living play a role in nursing home placement. Cognitive status has also been a factor. Individuals with dementia or Alzheimer s disease are more likely to receive care in a nursing home. Finally, self-reported health has been a factor that predicts nursing home placement 14

20 (Cohen et al., 1986, Hanley et al, 1990). Those who rate their health as poor have a greater chance of entering a nursing home. Purpose of Study From this literature review it is apparent that predictors contributing to nursing home placement vary depending on a range of study factors. Different samples (disabled vs. services received), methods (varying longitudinal time frames), predictor variables, and analyses (logistic vs. time-hazard) contribute to these inconsistent results. Utilizing the categories of variables derived from the Andersen s Behavioral Model, this study will add to the existing literature on nursing home placement by examining characteristics of those entering a locally funded long-term care program. We will investigate the outcomes of clients entering the program over a 2.5-year period. The goal of this study is to identify the factors that play a role in nursing home placement for program individuals. 15

21 Chapter 2 Methodology The clients for this study were drawn from a locally funded home and community based long-term care program administered by an Area Agency on Aging. The home and community based program is available to individuals 65 or older living in an urban area who meet physical and income eligibility requirements. Older individuals who have impairment in at least 2 Activities of Daily Living or Instrumental Activities of Daily Living are eligible. There are no income restrictions for individuals below 150% of the poverty threshold. Clients who have incomes between 150% and 400% are required to make co-payments based on a sliding scale. Clients who are at or above 400% of the poverty threshold are required to make 100% co-payments. Long-term services provided are case management, personal care, adult day care, homemaking, medical transportation, and meal preparation among others. Sample characteristics From the initial phone call to the start of service, information about predisposing, enabling, and need factors are gathered. For example, information about age, level of physical functioning, and income are documented. Those who qualify for the program based on screening information are scheduled to have a face-to-face home visit with a case manager. After completion of visits, case managers coordinate with home care agencies to arrange a start of service date for the clients. Data collected on each client are stored in the Area Agency on Aging s Management Information System. For this study, a query was written in Microsoft Access to create a data set of individuals enrolling from one of the AAA s urban 16

22 counties between Jan 1 st 2000 through March 31 st The query filtered out individuals who were not scheduled to receive case managed services during the threemonth time frame, such as those who would receive home-delivered meals only. Home delivered meal only clients are not case managed currently and not likely to be high need clients. The interval assessed for disenrollment outcome was approximately 30 months after start of service for each client. For example, the time frame of a client scheduled to receive services beginning January 1 st, 2000 would be assessed on July 31 st, Three hundred and forty-one consumers newly entered the program and were scheduled to begin services during the period of January 1 st through March 31 st, Analysis of outcomes shows that after 30 months, 31% remained active, 24% left for nursing home placement, 17% died, and 27% had other outcomes (See Table 3). 17

23 Table 3 Outcomes of Individuals Enrolling in Program during Jan 1 st, 2000 and March 31 st, 2000 Outcome Number % Nursing Home Placement Deceased Remaining Active No Longer Needs Services Transferred to Different Program Moved Reopened Other Not at Home Dissatisfied with the program Total

24 Predictor Variables Client data stored in the Management Information System are extensive. There is demographic, functional, and medical information. As specified in the literature review, there are several risk factors that have been associated with nursing home placement. The following paragraphs will specify those items and will describe how these items were coded in the Management Information System and any recodes that were made for purpose of this analysis. Table 4 highlights the recoded variables. Predisposing characteristics As noted in the literature, predisposing characteristics are defined as variables such as age and living arrangement that are used to predict healthcare utilization. The programs assessment for racial/ethnic group composition included Caucasian, African American, Hispanic Origin, Native American/Alaskan, Asian/Pacific Islander, Other, and Declined to answer. The racial/ethnic group category was collapsed to 0= Non- Caucasian, and 1=Caucasian serving as the reference category because literature has shown that Caucasians have a greater likelihood of nursing home placement. The composition of client s residence is also assessed, scored as lives alone, lives with spouse, lives with dependent child, or other. The variable was collapsed to 0=lives with others and 1=lives alone serving as the reference category because previous literature has suggested that those who live alone have a higher risk of leaving the community. Enabling characteristics Marital Status was assessed as Married, Widowed, Single, Separated, Divorced, and Unknown. This variable was collapsed to 0=not married, and 1= married (the 19

25 reference category) because marriage has been identified as a factor in reducing risk of nursing home placement. Through review of case manager notes, four variables were constructed to assess the caregiving role: whether clients had an informal caregiver, whether caregivers were relatives, what care relatives provided, and whether caregiver had some stress providing care. If a caregiver was mentioned anywhere within the notes, a score of 1 was given for having caregiver. If no information was found or if case manager notes specifically mentioned that there were no caregivers, a score of 0 20

26 Table 4 Variables taken from the Management Information System: Recodes Characteristics Recodes Race 0=Not Caucasian 1=Caucasian Living Arrangement 0=Lives with others 1=Lives alone Marital Status 0=Married 1=Not married Has Caregiver 0=No 1=Yes Caregiver is relative 0=No 1=Yes Caregiver provides assistance 0=No 1=Yes Caregiver Burden 0=No 1=Yes Activities of Daily Living 0=Not Impaired 1=Impaired Instrumental Activities of Daily Living 0=Not Impaired 1=Impaired Self reported Health 0=Not poor health 1=Poor health Dementia 0=No 1=Yes 21

27 was given (See Table 4). Typically, a relative or friend or neighbor was specified in the notes when client had caregiver. This variable was coded as 1=caregiver was relative or 0= caregiver not relative. The notes also mentioned what kind of support the caregiver provided. A client was coded 1 if caregiver provided some type ADL such as bathing and IADL help such as shopping for client. If a client received ADL help from a caregiver, but IADL help was not mentioned, the client was also coded as 1. Those who received only IADL help or no help, were scored 0. Any mention of caregiver stress in notes was coded 1. If no stress was mentioned a 0 was given. Need Factors Functional information from Activities of Daily Living included questions on mobility-inside, stair climbing, transferring to and from bed/chair, take bath or shower, appearance and hygiene, dress, gets to and uses toilet, continence, and eating. Items included from Instrumental Activities of Daily Living were: manages money, can get to places within walking distance, shops for groceries, prepares meals, completes housework, does laundry, and uses telephone. All of the functional ability items were scored by case managers as 1=able with help, 2=needs mechanical assistance, 3=needs verbal assistance, 4=needs hands on assistance, 5=unable. For this study, clients who needed hands on assistance or were unable to perform activity were considered impaired in that activity. The scoring was 0=not impaired and 1=impaired serving as the reference category. Medical expenses are dollar amounts spent on medical care, insurance, and pharmacy bills for a one-month period per client. This variable is included as a need variable factor, because it indirectly represents the amount of medical care received by the client. The variable self-reported health was 22

28 recorded as an open-ended question during assessment. Clients were asked to describe their general health. For this study, 1= poor self reported health served as the reference group. If clients mentioned an illness or said that their health was poor, the comment was coded as 1. All responses that were ambiguous, for instance, I would feel better if my arm didn t hurt, were also coded as 1. Clients who mention that everything is fine were given a 0, health not poor. Finally, dementia was also scored as a dichotomy: 1 if the client had dementia or Alzheimer s disease and 0 if they did not as identified by the case managers. Dependent Variable The dependent variable for this study was whether the older individuals receiving formal services left the program for nursing home placement during the 2.5-year time interval. This was coded as 0=not leaving for nursing home, 1=leaving for nursing home. The did not enter a nursing home group was so heterogeneous; it included individuals who died, left the program because they no longer needed services, and those who were still enrolled in the program. Those who died or left the program for reasons other than nursing home placement were filtered out before multivariate analysis for two reasons. For one, the program s intervention did not provide skilled services that potentially could have played a role in preventing premature death. The second reason not to include these individuals in the analysis was due to the varied nature of the outcomes. For instance, individuals who disenrolled because they no longer needed services were quite different from those who moved or transferred to a 23

29 different program. Basically, the other category was potentially too heterogeneous to define or analyze. 24

30 Chapter III Results Descriptive statistics such as frequencies and means were utilized to examine the overall sample and to make comparisons between the nursing home placement and remaining active groups. Finally, logistic regression analysis was used in order to predict what factors might be associated with nursing home placement. Characteristics of Overall Sample Seven out of every ten enrollees into the program were over the age of 75; the mean age of the sample was (see Table 5). The typical client was Caucasian, female who is not married. Just over half of the clients lived alone. The majority of the sample had a monthly income of over $ and reported paying nearly 20% of their income in medical expenses (See Table 6). A large majority of clients had caregivers who provided some assistance with ADLs or IADLS. Twenty percent were receiving help with both ADLS and IADLS. Usually, caregivers were family members, although nearly 20% of clients received support from others outside the family unit. At entry, 1 out of every 3 caregivers expressed some form of burden either with clients health, their own health, or time restraints due to work and/or own family. 25

31 Table 5 Description of Predisposing Characteristics per Outcome Nursing Remaining Characteristics Total Home Active (N= 341) (N = 82) (N =106) p 1 Age 0.006* Age (Mean) * Gender Female Race Caucasian * Living arrangement Lives Alone Lives with Others * significant p <.05 26

32 Consumers entering the program were quite functionally impaired (See Table 7). Fifteen percent of the consumers had 4 or more impairments in Activities of Daily Living As for Instrumental Activities of Daily Living, 8 out 10 consumers had four or more impairments. Analysis of self-reported health ratings showed that nearly half of the group mentioned having poor health. Finally, 12% of those entering the program had dementia. 27

33 Table 6 Description of Enabling Characteristics per Outcome Nursing Remaining Total Home Active (N=341) (N = 82) (N =106) p Marital Status Married Total Monthly income (Mean) 1, Client has caregiver 0.036* yes Caregiver family member yes Caregiver Burden yes How caregiver helps 0.00* iadls/none mentioned iadls and adls * significant at p <.05 28

34 Comparing those who leave for nursing home with those remaining in the community In order to understand what triggers entrance into nursing homes for home health clients, it was helpful to compare the characteristics of those who left the program for nursing home placement (NHP) with individuals who remained in the program for the duration (RA). As mentioned above, individuals who left for other reasons were excluded from analyses (N=163), since this group included a broad range of outcomes, from mortality to improved health. 29

35 Table 7 Description of Need Characteristics per Outcome Nursing Remaining Total Home Active (N=341) (N = 82) (N =106) p % Impaired in ADLs Mobility-inside * Stair climbing * Get in and out of bed * Take bath or shower * Appearance/hygiene * Dress * Gets to and uses toilet * Continence * Eating # of ADL Impairments 0.00* or more ADL (mean) * % Impaired in IADLs Manages Money * Walking Distance * Shops for Groceries * Prepares Meals * Completes housework Does Laundry * Uses Telephone * IADLs 0.00* or more IADL (mean) * * significant at p <.05 (Continued) 30

36 Table 7 Description of Need Characteristics per Outcome (continued) Nursing Remaining Characteristics Total Home Active (N=341) (N = 82) (N =106) p Self Reported Health not poor poor Dementia reported 0.013* yes no Total monthly Medical Expenses (Mean) * * significant at p <.05 31

37 The data suggest similarities and differences in predisposing, enabling, and need characteristics at baseline between home health care clients leaving for nursing home placement or remaining active throughout the 2.5-year interval. As for predisposing characteristics, the two groups significantly differed in age with the NHP group on average nearly 4 years older than the RA group. Furthermore, there were nearly 2 times as many individuals 85 and over in the NHP group than the RA group (See Table 5). The majority of clients within both groups were female. Differences in racial/ethnic status existed with a larger proportion of Caucasian individuals in the NHP group than for those who remained in the program. Differences were noted for enabling characteristics. There were significant differences in having or not having an informal caregiver who provided assistance. NHP clients were more likely to have an informal caregiver. This finding is somewhat counterintuitive, since previous literature suggested that presence of a caregiver may delay nursing home placement. For this population, caregivers may signal greater impairment. However, out of all clients who had caregivers, there were no differences in the proportion who have family caregivers versus non-family caregivers (See Table 6). As for the type of informal care received, the NHP group was significantly more likely to receive help with both ADLs and IADLs than the RA group. There was a slight but not significant difference in caregiver-burden for the two groups with 10% more NHP caregivers expressing problems with care. Individuals who entered the nursing home during the 2.5 years differed in need factors when compared to those who remained in the program. There were significant differences in mobility, bathing, transferring, grooming, eating, and incontinence. One- 32

38 third of NHP clients had 3 or more Activities of Daily Living Impairments; this was nearly 5 times that of RA clients (See Table 7). As for Instrumental Activities of Daily Living, NHP clients differed significantly from RA clients in all categories except completing housework; in this category both groups were equally impaired. Over 90% of NHP clients had 4 or more IADL impairments, a proportion that is 30 percent higher than RA for clients. These significantly higher levels of functional impairments could have been related to the increased number of NHP clients who had dementia. More than 3 times as many NHP clients have dementia in comparison to RA clients. As for self-reported health, nearly one-half of those who enter a nursing home report at program entry that their health was poor. Though not significantly different, almost 13% more of the RA group reported that health was not poor. Finally, there was a significant increase in the amount of medical expenses paid by NH clients. Analysis In order to assess what characteristics help to predict the home care clients who are more likely to leave for nursing home placements, logistic regression analysis was utilized. In the initial phase of the analysis, the data were examined for two factors that could affect the outcome of regression equations: normal distribution of cases and multicollinearity. Certain variables had too few responses in certain categories. For this study, any variables cross-tabulated by nursing home placement or not which yielded cells less than 4 were removed from the equation. For example, no clients in the RA group were impaired in activities related to mobility-inside. Keeping the variable in its present form could have caused problems in the analysis because it did not truly vary. 33

39 When possible, variables that were similar conceptually were collapsed together to avoid the no variability problem. Mobility inside, gets in and out of bed/chair, and stair climbing were collapsed together into the variable Mobility; clients impaired in any of the three activities were considered impaired for Mobility. Variables that did not vary- had too few cases in response category- and could not be collapsed together with another variable conceptually were removed before running the regression. These variables are: toileting, continence, eating, shopping. Multicollinearity is when the correlation among the predictor variables is very high (Neter, Kutner, Nachtsheim, & Wasserman, 1996, p. 285). This problem can have a suppressing effect on the explanatory variables. For instance, two variables highly correlated that are placed into the same equation do not explain as much as each would when used in separate equation. Before performing logistic regression analysis, correlations among all the predictor variables to be included in the analysis were examined. Certain IADL variables were strongly correlated. Variables such as dressing, hygiene, and bathing were all highly correlated at.60 or higher. Collapsing the dressing and hygiene variable still produced a strong relationship with bathing. Since bathing yielded a higher correlation (r =.41) with nursing home prediction, the variable remained in the equation, while the other two variables were removed. A logistic regression is utilized when the dependent variable is binary instead of continuous; for instance, the event happened or did not happen. This statistical technique can be used to identify what variables or group of variables help to explain and predict an event. For this study, logistic regression was used to investigate 34

40 variables that predict whether a client entering the program left for a nursing home or not. A logistical regression with backwards elimination was used. This technique can be used to explore associations between predictor variables and dependent variables. Since results from studies investigating predictors for nursing home placement produced such a wide range of results, all variables expected to have an association with an outcome were entered into the equation. Variables that had a significance level greater than.10 were sequentially removed from the equation at each step. The equation was complete when all variables remaining had a significance level less than.1. The predisposing characteristics age, gender, racial/ethnic group composition, and living arrangement were initially entered into the equation. Also, the enabling characteristics, marital status, having caregiver, how caregivers helps, whether caregivers experienced burden, and clients income were entered into the equation. Finally, the need characteristics of dementia, self-reported health, impairment in mobility, taking bath or shower, preparing meals, homemaking, doing laundry, and using telephone were entered. The final model, as shown in Table 8, correctly predicted 71.6% of nursing home placement. Age was a significant predictor for nursing home placement (OR=1.064; 95% CI=1.006,1.12) when the other variables were held constant. Every one-year increase in age increased the likelihood of placement by nearly seven percent. A 10- year increase in age would increase the chances by approximately 70%. Self-reported health also significantly increased placement (OR=2.53; 95% CI=1.05, 6.12). Those 35

41 who stated that there health was poor upon entry had a nearly 2.6 times greater likelihood of entrance into a nursing home. Higher medical expenses per month were also a factor (OR=1.003; 95% CI=1.001, 1.005). For every one dollar increase in expenses, an individual s chances of entering a nursing home increased by only.3%. However, it is more intuitive to state that for every additional 100 dollars of expenditures, a client s chance would increase by 30%. ADLs and IADLs were associated with nursing home placement. Clients who were impaired in preparing meals upon entry into the program were five times more likely than those who were not (OR=5.00; 95% CI=1.93,12.96) to enter a nursing home. Having difficulty taking a bath or shower was also a factor in nursing home placement 36

42 Table 8 Final Model for Variables that Predict Nursing Home Placement 95% C.I.for Exp(B) Sig. Exp(B) Lower Upper Age 0.031* Impaired in Preparing Meals 0.001* Impaired in Housework Impaired in Bathing 0.003* Medical Expenses 0.014* Self-Reported Health 0.038* note: chi square value = 50.34, df = 7, p =.000 successfully predicts 71.6% of cases leaving for nursing home * significant at <.05 37

43 (OR=3.96; 95% CI=1.60,9.79). Clients impaired in this category were nearly 4 times more likely to enter a nursing facility. Those who were impaired in homemaking had a non-significant decrease in the likelihood of nursing home placement. It is apparent from these results that some variables are more important in making future predictions for nursing home placement than others. The findings suggest that predisposing and need factors such as age, medical expenses, and functional abilities, and self-reported health appear to increase the chance of placement for individuals who enter into a home care program. 38

44 Chapter IV Discussion The purpose of this study was to follow newly enrolled home care recipients throughout a 30-month period in order to identify variables that trigger nursing home placement. Using an extensive data set from a case managed home care program, this study was designed to include variables suggested as important by the Andersen Behavioral Model and by previous research. Those who left for nursing homes differed in predisposing, enabling, and need characteristics at baseline in comparison to those who remained in the community. Those who were placed in nursing homes during the 2.5 year period were significantly older, Caucasian, more functionally impaired, and more likely to have a caregiver upon entry into the program. Five variables were significant in predicting greater risk of nursing home placements. The predisposing factor, age, was associated with nursing home placement. Individuals who entered the program at age 90 had a markedly greater likelihood of entering a nursing home than individuals who entered at 65. This finding clearly could be related to the different changes that are experienced when growing older, including an increased prevalence of chronic conditions, greater frequency of mental status and behavioral changes in response to physical illness, and a decrease in the number of close family members and friends available to provide informal support (Murtaugh, Kemper, and Spillman, 1990, p. 957). Similar to other findings on risk of nursing home placement, the need characteristic of self-reported health also played a role in nursing home placement 39

45 (Hanley at al, 1990, Cohen et al, 1986). Individuals who expressed poor health when entering the program were at greater risk of leaving for nursing home. In previous literature, researchers coded Activities of Daily Living and Instrumental Activities of Daily Living in a variety of ways. For this study, the independent items included bathing, toileting, and eating among others. Although some ADL and IADL variables were not included in the analysis due to distribution and multicollinearity issues, certain impairments in ADL and IADL functioning were associated with leaving for nursing home placement. Similar to findings in other longterm care programs, impairments in bathing and meal preparation were the two strongest predictors of nursing home placement (Hughes et al., 1987; Nokes et al. 1986). This would suggest that for these individuals the amount of services received after enrollment might not have been enough, hence leading to nursing home placement. These findings suggest that an increase in the basic services provided might be needed for a select group of clients in order for them to remain independent and in the community. There is a need for further research on these two impairments in order to assess what they signify. The need variable, medical-expenses, also played a significant role in nursing home placement. Individuals who paid more money for prescriptions, medical insurance, and hospital bills, had a greater likelihood of nursing home placement. This variable might be significant in its own right and because it potentially acts as proxy for medical or chronic conditions. Other studies have included medical conditions in the analysis. The more severe the chronic condition, the more money paid out in expenses per month, and the greater the likelihood of nursing home placement. 40