The Gerontologist Vol. 42, No. 5, 603–612

Copyright 2002 by The Gerontological Society of America

Utilization Across the Continuum of Long-Term Care Services Evelinn A. Borrayo, PhD,1 Jennifer R. Salmon, PhD, 2 Larry Polivka, PhD, 2 and Burton D. Dunlop, PhD3

Purpose: This study presents an analysis of the influence of consumers’ predisposing, enabling, and need characteristics on the utilization of long-term care (LTC) services in nursing facilities (NFs), assisted living facilities, or home- and community-based services (HCBS). Design and Methods: Data were gathered through a record review of a cross-sectional sample of 1,968 consumers aged 60 years or older receiving formal LTC services. Results: Need contributes the most to differential LTC use. Those with the most need are more likely to be in a NF or HCBS. Enabling characteristics such as Medicaid eligibility and geopolitical region of the state were associated with higher use of NFs. Predisposing and enabling characteristics had a moderating influence on need. Implications: Although high need predicts care in NFs, some of the most impaired elders are also being cared for at home. Because Medicaid increases the likelihood of NF placement, public financing of LTC should reflect the capacity of the system to serve elders at home.

LTC services are typically needed when a person has an ongoing, long-term disability as indicated by increases in physical and cognitive impairments or loss of function as measured by impairments in activities of daily living (ADLs). Investigation of utilization across the continuum of services is important in understanding where these adults are likely to receive services. To date, most research on LTC utilization is limited to the study of predictors of institutional or community-based LTC services rather than simultaneously considering the entire LTC continuum, including assisted living facilities (ALFs). The purpose of this study is to provide an analysis of frail older adults’ utilization of nursing facilities (NFs), ALFs, and home- and community-based services (HCBS) on the basis of factors commonly associated with health care utilization (Andersen, 1995). This research adapts the behavioral model of health services use (Andersen, 1995; Andersen & Newman, 1973) for understanding how predisposing, enabling, and need factors affect use of LTC services (Bass & Noelker, 1987; Wolinsky, 1990). This model implies that both system factors, such as public policies, and population characteristics are associated with the use of formal LTC services (Wallace, Levy-Storms, Kington, & Andersen, 1998). Predisposing characteristics reflect the propensity to use services, independent of personal circumstances and experiences that may cause the need for service use. Typically, they are sociodemographic variables. Enabling characteristics explain differences in the resources available to the individual in using LTC services. Income and health insurance are likely to enhance service use while other sources of help from social support networks or publicly financed LTC may either facilitate or impede use of formal services (Miner, 1995). Need factors include perceived or evaluated physical, cognitive, functional, and mental health status. Need-based factors such as serious health problems and conditions are often recognized as the most immediate predictors of service utilization and have explained the largest share of variance in health care utilization (Andersen, 1995; Mercier & Shelley, 1997). Researchers have found that these characteristics

Key Words: Public policy, Nursing facilities, Assisted living facilities, Home care, Public financing, Long-term care utilization

The rapid aging of the population, especially increases in the oldest old (those aged 85 and older), along with stagnation or decline in the availability of federal and state resources for health and long-term care (LTC) services (G. D. Cohen, 1994; Rothman, 1993; Steinmo & Watts, 1995) has pressed states to address ways to serve the growing frail population in less expensive and less institutional settings. Overall, institutions such as nursing homes consume 75%– 85% of the public LTC dollars.

We thank Mary L. Oakley, MA, for her invaluable assistance with this study. Address correspondence to Evelinn A. Borrayo, Department of Psychology, Colorado State University, Fort Collins, CO 80523. E-mail: borrayo@ lamar.colostate.edu 1Department of Psychology, Colorado State University, Fort Collins. 2Florida Policy Exchange Center on Aging, University of South Florida, Tampa. 3The Center on Aging, Florida International University, North Miami.

Vol. 42, No. 5, 2002

603

need predictors of LTC utilization by frail older adults (Kemper, 1992; Morris, Sherwood, & Gutkin, 1988; Wolinsky, Callahan, Fitzgerald, & Johnson, 1993; Wolinsky & Johnson, 1991) and especially of nursing home placement (Egleston et al., 1999; Rudberg et al., 1996). As a critical need factor, ADL impairment frequently coexists with cognitive impairment, particularly later loss ADLs such as toileting, transferring, and eating. The compounding of risk factors places an individual at greater risk of LTC service use (Weissert & Cready, 1989). Persons with cognitive impairment and ADL limitations experience a two- to threefold increased risk of institutionalization (Foley et al., 1992), and the severity of cognitive impairment is invariably found to be a significant factor for NF admissions (Coughlin et al., 1990; Heyman et al., 1987; Murtaugh, 1994). Older adults with Alzheimer’s disease or dementia are institutionalized at significantly higher rates (e.g., 50%, Severson et al., 1994; 75%, Welch, Walsh, & Larson, 1992). Psychiatric and emotional problems increase the risk of LTC utilization. Depression, in particular, is a significant predictor of LTC placement (Steele, Rovner, Chase, & Folstein, 1990; Stern et al., 1997). The presence of chronic health conditions also increases the probability of nursing home use (Kart & Palmer, 1987; Mitchell, Matthews, & Hack, 2000), especially when associated with recent hospitalizations or Medicaid eligibility (Temkin-Greener & Meiners, 1995). However, private payers who experienced heart or respiratory disease or cancer were more likely to die in the community without having been admitted to a nursing home. In summary, predisposing, enabling, and need characteristics are important in the prediction of NF utilization and, to a lesser extent, of HCBS utilization by frail older adults. There has been much less systematic analysis regarding how these factors are associated with the full continuum of services available (NFs, ALFs, and HCBS; Polivka, Dunlop, & Brooks, 1997). Such analyses would aid policymakers in setting criteria for providing older adults who receive publicly funded LTC access to the full continuum of services on the basis of their needs and living arrangements. For example, if individuals with a high level of need prefer to live at home, then public funding of LTC should make home-based options as available as NFs are for low-income elders. This would be further demonstrated if need, rather than enabling factors such as income or medicaid eligibility, differentially predicted the utilization of the various LTC services available (Andersen, 1995). On the other hand, enabling factors are often the most mutable and can be affected by changes in public policy. Need is often considered to be less mutable. However, people’s perceived need for LTC may be changed through education about available programs or use of devices to maintain independence (Klerk, Huijsman, & McDonnell, 1997). In this study, our goal was to use the LTC continuum model as a conceptual framework to identify specific population characteristics that determine enrollment in LTC programs. In this effort, we tested three

significantly contribute to older adults’ use of discretionary services such as in-home and community services and nondiscretionary services such as nursing home placement (when it is not the consumer’s choice; Calsyn & Winter, 2000; Temkin-Greener & Meiners, 1995; Wolinsky, Callahan, Fitzgerald, & Johnson, 1992). Briefly, among predisposing factors, advanced age predicts LTC utilization, especially NF use, although age loses its saliency as a predictor of LTC among high-risk individuals (Coughlin, McBride, & Lui, 1990; Mittleman, Ferris, Shulman, Steinberg, & Levin, 1996). Still, nursing home admission rates nearly double for each additional 5-year increase in age (Hanley, Alecxih, Wiener, & Kennell, 1990). Race and ethnicity discriminate between use of institutional and community-based LTC. Overall, African Americans and Hispanic older adults are significantly less likely than non-Hispanic Whites to use institutional LTC services and more likely to use both formal and informal community-based LTC services (Greene & Ondrich, 1990; Murtaugh, Kemper, & Spillman, 1990; Liu, McBride, & Coughlin, 1994; Wallace, Levy-Storms, Andersen, & Kington, 1997; Wallace, Levy-Storms, & Ferguson, 1995). This is not a function of other demographic characteristics, of enabling factors such as insurance, or even of need as indicated by functional status. Instead, race and ethnicity have a significant independent effect in LTC use that may be related to one or more social processes such as culture, class, institutionalized discrimination, and geopolitical differences (Wallace et al., 1998). Living arrangements, Medicaid eligibility, and geopolitical region are enabling characteristics identified as significant predictors for institutionalization, but are highly dependent on the nature of the study population. Impaired older adults who live with a spouse are half as likely to be placed in a nursing facility as those who live alone, are not married, or are cared for by another relative or friend (Boaz, Muller, & Hu, 1994; Freedman, Berkman, Rapp, & Ostfeld, 1994; Jette, Tennstedt, & Crawford, 1995; Montgomery & Kosloski, 1994; Weissert & Cready, 1989). Medicaid eligibility increases the risk of nursing home placement (Branch & Jette, 1982; M. A. Cohen, Tell, & Wallack, 1986; Kelman & Thomas, 1990; Mittleman et al., 1993). Individuals who are able to pay for LTC and not use Medicaid have a considerably lower probability of nursing home admission, even when faced with the same set of health conditions and cognitive impairment (Temkin-Greener & Meiners, 1995). Regional differences in utilization are due to variations in the supply of LTC options: number of nursing home beds, number of available slots for alternative LTC services, or both (Coughlin et al., 1990; Dwyer, Barton, & Vogel, 1994; Egleston, Rudberg, & Brody, 1999; Rudberg, Sager, & Zhang, 1996). For example, higher state Medicaid spending on HCBS was significantly associated with a longer time before NF admission among unmarried elders (Miller et al., 1998). ADLs, with the exception of mobility, are the best 604

The Gerontologist

For the HCBS population, a representative sample of 350 active cases from the Home Care for the Elderly, the Community Care for the Elderly (CCE), and the Medicaid Waiver programs was targeted. In addition, the following alternative demonstration programs were oversampled to ensure sufficient cases to determine program characteristics: the Channeling Program (n  175), the Medicaid Prepaid HMO Frail Elderly Option program (n  175) and the Medicaid Waiver program (n  300). There were no program exclusions for the home-care sample. Subpopulations were stratified geographically on the basis of the most recent state agency data for each region (north, central, west, and south Florida), and the sample was weighted to reflect the proportion of clients in each state region and in each program. For example, the CCE program is the largest (in terms of client census) of the HCBS programs, yet a larger sample of Medicaid Waiver clients than CCE clients was drawn. Thus, we weighted each CCE case to count more heavily, and the Medicaid Waiver cases were downweighted to adjust for the intentional oversampling of this relatively smaller population.

hypotheses. First, we tested whether need variables differentially contribute to explaining LTC utilization across the continuum of services. We expected that those with high need would be in the most institutional settings (NFs), followed by ALFs and HCBS. Second, we assessed whether enabling variables, especially mutable variables such as Medicaid eligibility and geographic region, significantly discriminate the utilization of LTC services, specifically because of geopolitical differences in the supply of NF beds and alternatives. Last, we tested whether the moderating influence of predisposing and enabling variables on need (e.g., race on cognitive impairment) significantly predicts the differential utilization of LTC services. This model has traditionally served to investigate how population characteristics influence LTC use. However, the utilization of different LTC services is a fairly complex issue that is related to the context of states’ policy and programs and the private LTC market. Although we were not able to directly measure these complexities, we discuss our findings within this relevant context and make recommendations accordingly.

Methods

Data Collection.—Trained graduate nursing students extracted data from standardized assessment instruments used by state agencies and providers to assess clients in all three settings. Data were drawn from (a) the admission Minimum Data Set (MDS) Resident Assessment Instrument in NFs, (b) the most recently completed Department of Elder Affairs Uniform Assessment Instrument for HCBS and OSSsupported ALF populations, and (c) the ALF admission screening instrument (Department of Elder Affairs Form 1823) for private-pay ALF residents. The data collection instrument included common elements from each of these assessment instruments and identified the location of each element on each standard form to minimize errors. Missing information was obtained using earlier assessments and complemented as necessary by interviews with administrators or staff regarding the characteristics of the residents.

Procedure The original study was conducted by the Florida Policy Exchange Center on Aging at the University of South Florida and The Center on Aging at Florida International University for the Commission on LongTerm Care in Florida. The data were collected through record review of LTC clients 60 years and older in NFs, ALFs, and HCBS. Sampling.—The representative cross-sectional sample was selected to achieve a sampling error of no more than 3% at the 95% confidence level. The total sample size included 600 NF residents, 650 ALF residents, and 1,000 clients from HCBS programs for a total of 2,250. Because NF and ALF populations reside in facilities, we used a two-stage random sampling procedure. First, facilities were randomly selected, and second, residents were randomly selected within facilities. Of the NF residents, long-stay residents were included in this analysis, and short-stay NF residents were excluded. The long-stay residents had been in the nursing home at least 90 days or, if admitted more recently, had no plans for discharge from the NF. The ALF sample included 350 private-pay residents plus an oversample of 300 ALF residents who received public funding through Optional State Supplementation. At the time, Optional State Supplementation was the only source of public financing of ALFs. Although NFs are standard across the states, ALFs have no standard definition. In 1995, the state of Florida officially changed the designation of adult congregate living facility to assisted living facility and defined it to be any building or part of a building that provides housing, meals, and one or more personal services for a period exceeding 24 hr to one or more adults not related to the owner or administrator. Vol. 42, No. 5, 2002

Sample The total LTC sample used for this analysis was 1,968 clients aged 60 and older, which included 474 residents from NFs, 599 residents from ALFs, and 895 HCBS clients. Table 1 displays sample characteristics by LTC setting. Overall, our 1995 Florida sample was less likely to be non-Hispanic White (75%) than were elders in the 1994 National Long-Term Care Survey (87%; Spector, Fleishman, Pezzin, & Spillman, 2000). The Florida sample was also more likely to be female (75%) than the national sample (71%) and not married (77% vs 66%). On average, the Florida sample had impairments in 2.65 ADLs, whereas 43% of the national sample had impairments in 3 or more ADLs. In Florida, elders scored 1.6 on a 0–3 scale of cognitive impairment, and 40% of the national 605

Table 1. Frequency of Predisposing, Enabling, and Need Characteristics of Older Consumers by Long-Term Care Setting

Characteristics Predisposing Ethnicity Non-Hispanic White Other ethnicity a Age 85 75–84 60–74 Gender Female Male Enabling Marital status Not married Married Insurance Medicaid Other (Medicare, other) Region North Central West South Need Alzheimer’s disease or other dementia Yes No Incontinence Often/always has accidents Never/occasionally has accidents Mobility Needs some help/can’t do at all Independent

ADL (0–5)b Chronic diseasesc (0–16) Cognitive impairment (0–3)d

NF (%; n  474)

ALF (%; n  599)

HCBS (%; n  895)

Total (%; N  1,968)

83 17

85 15

65 35

75 25

52 37 11

52 33 15

29 40 31

42 37 21

77 23

71 29

78 22

75 25

78 22

83 17

73 27

77 23

66 34

25 75

53 47

48 52

28 25 27 20

12 16 35 37

23 16 24 37

21 18 28 33

56 44

42 58

18 82

34 66

44 56

17 83

24 76

27 73

66 34

36 63

42 57

47 54

M (SD) 3.98 (1.32) 2.55 (1.56) 1.42 (1.05)

M (SD) 2.21 (1.82) 2.01 (1.25) 1.60 (1.12)

M (SD) 2.24 (1.81) 3.34 (1.80) 1.72 (1.23)

M (SD) 2.65 (1.86) 2.74 (1.68) 1.61 (1.16)

Notes: NF  nursing facilities; ALF  assisted living facilities; HCBS  home- and community-based services. aOther ethnicity  Hispanic, Asian/Pacific Islander, non-Hispanic Black, American Indian/Alaskan Native. bActivities of daily living (ADLs) entered in the analysis as a continuous variable that counted only ADLs that an individual couldn’t do at all or needed some help or supervision with. cChronic diseases counted any diseases that were recorded for a consumer: arthritis, emphysema/asthma, diabetes, gallbladder problems, gastritis/ulcers, high blood pressure, heart disease, cancer, Parkinson’s, stroke, osteoporosis, anemia, urinary tract infection (past 30 days), renal failure, hypothyroidism, and pneumonia. dCognitive impairment: 0  independence, 1 mild impairment, 2  moderate impairment, 3  severe impairment; based on weighted data.

dressing, 43% with transferring, 48% with using the toilet, 36% with eating, and 47% with mobility.

sample was cognitively impaired. These differences were expected given the demographic characteristics of Florida compared with the nation. In comparison to Florida’s frail population older than 65 (Polivka, Dunlop, & Rothman, 1996), our sample was comparable in age. In our sample, 37% were between 75 and 84 years old compared with 37% of the frail Floridians, and 42% were 85 years and older compared with 27% of the frail Floridians. Our sample consisted of 75% women versus 69% of the frail Floridians. The sample was also quite impaired. On average, they had 2.70 chronic diseases and needed assistance with 2.65 ADLs (out of 5). As shown in Table 2, 77% of the consumers needed help with bathing, 62% with

Conceptual Framework for Analysis Using the LTC continuum model as a conceptual framework, we analyzed how population characteristics determine utilization across the continuum of LTC services. We entered race/ethnicity, age, and gender as predisposing demographic characteristics that influence utilization of LTC services. On the basis of previous research, we expected that being non-Hispanic White and older would be associated with use of NF services over HCBS. Enabling resources available to the individual for 606

The Gerontologist

Table 2. Levels of Functioning in Activities of Daily Living by Long-Term Care Setting

Activities of Daily Living a Bathing Can’t do it at all Some help/supervision Independent Dressing Can’t do it at all Some help/supervision Independent Transferring Can’t do it at all Some help/supervision Independent Toileting Can’t do it at all Some help/supervision Independent Eating Can’t do it at all Some help/supervision Independent Walking/mobility Can’t do it at all Some help/supervision Independent

NF (%; n  474)

ALF (%; n  599)

HCBS (%; n  895)

69 27 4

5 64 31

33 40 27

33 44 23

47 44 9

4 48 48

19 35 46

21 41 38

41 33 26

2 28 69

14 22 65

17 26 57

47 36 18

4 36 60

16 18 66

20 28 52

17 41 42

2 26 72

8 21 71

8 28 64

36 30 34

3 33 63

17 25 57

18 29 54

Total (%; N  1,968)

Notes: NF  nursing facilities; ALF  assisted living facilities; HCBS  home- and community-based services. aActivities of daily living were entered in the analysis as a continuous variable that counted only ADLs that an individual couldn’t do at all or needed some help or supervision with; based on weighted data.

ida would be more likely to receive services in a nursing home, particularly in regions where the nursing home bed supply was above the state median and the availability of alternative LTC programs was more limited. Measures of need included Alzheimer’s disease, chronic bladder incontinence, assistance with mobility, ADL impairment, chronic diseases, and a separate measure of severity of cognitive impairment. Mobility was not included in the ADL score because it has a distinct predictive value from other ADLs in LTC models (e.g., Weissert & Cready, 1989). Chronic diseases were counted if they were recorded as a client’s diagnosis, whether or not these diseases were identified as the reason for receiving LTC services. Sixteen chronic diseases were included in the score: arthritis, emphysema/asthma, diabetes, gallbladder problems, gastritis/ulcers, heart disease, high blood pressure, cancer, Parkinson’s, stroke, osteoporosis, anemia, urinary tract infection (past 30 days), renal failure, hypothyroidism, and pneumonia. Severity of cognitive impairment was used to provide a more precise level of need that is not represented by physical health, a diagnosis of Alzheimer’s disease, or ADL measures (Wiener, Hanley, Clark, & Van Nostrand, 1990). A variable was constructed that matched the Department of Elder Affairs Uniform Assessment Instrument Mental Status Questionnaire’s score (severely impaired, moderately impaired, mildly impaired, and independent), based on the short Blessed Test, with the four-level scale of the Cognitive Skills for Daily Decision-Making scale from the MDS Resident Assessment Instrument for the NF and private-pay ALF residents.

LTC were measured by marital status, Medicaid eligibility, and geopolitical region of the state. The Medicaid eligible variable was coded “yes” if the individual was currently on Medicaid, eligible for supplemental security income, or participating in any welfare program. It was coded “no” if none of the conditions were true. Marital status was used to measure the enabling effects of social support. Individuals who are eligible for Medicaid and do not have a spouse are at higher risk for institutionalization. We expected to find these variables associated with NF utilization because Medicaid covers this option for low-income elders; it did not, however, cover ALFs at the time of the study. Region of the state (i.e., north, central, west, or south Florida) was included to represent geopolitical differences that affect LTC supply such as availability of nursing home beds and other LTC options. In a separate analysis of LTC supply and utilization data among Florida’s counties, a significant inverse relationship was found between the percentage of lowincome older adults in nursing homes and state expenditures on HCBS at the county level (Oakley, 2000). This negative correlation was strongest in the regions of Florida where nursing home bed supply was at or below the state median of beds per 1,000 Medicaid-eligible population aged 65 and older. Institutionalization rates were twice as high among Medicaid-eligible elders in the Florida counties with nursing home bed supply above the state median than in the counties with supplies below the state median (32% and 16% respectively). Overall, we expected that frail elders who lived in regions outside of South FlorVol. 42, No. 5, 2002

607

abling variables to .586 with the addition of the need variables. The interactions improved the model only slightly (.016; total R2  .602). It is not unusual for interaction variables to improve explained variance by as little as 1% in field studies (McClelland & Judd, 1993). The first hypothesis that need differentially contributes to explaining LTC utilization across the continuum of services was for the most part supported. Individuals with a diagnosis of Alzheimer’s disease or high levels of cognitive impairment were more likely to be in a NF or ALF than to be in HCBS, and between the two facility settings these individuals were more likely to be in a NF. The fact that the HCBS population was less likely to have a formal diagnosis of Alzheimer’s disease or other dementia probably says little about the prevalence of the disease in the population. Clients in publicly supported in-home care programs, compared with those in NFs and ALFs, are less likely to have access to a neurological evaluation that can lead to or rule out such a diagnosis. Evidently, the NF and ALF populations are more likely to have a diagnosis at all stages of the disease, including the earliest stages when cognitive impairment is minimal. Individuals who needed assistance with ADLs were more likely to be in a NF as well, but those with more chronic diseases were more likely to be in a NF or HCBS program than in an ALF. Those who needed assistance with mobility were more likely to be in HCBS. However, when NFs were compared with ALF settings, individuals who needed total assistance with mobility were more likely to be in a NF than in an ALF, and individuals who only needed some assistance were more likely to be in an ALF than a NF setting. Overall, individuals who had higher levels of need in terms of functional assistance were in NF and ALF settings, but those with higher numbers of chronic diseases and those in need of assistance with mobility were receiving HCBS. These findings are more likely to reflect program policies as most ALFs do not accept individuals with total impairment in mobility, whereas NFs are established to serve individuals with the highest functional impairment levels, including lack of mobility. The second hypothesis—that enabling factors, especially mutable variables such as Medicaid eligibility and public policies associated with geographic region, would be significantly associated with the differential utilization of LTC services—was confirmed as anticipated. Individuals who were not married were more likely to be in a NF or ALF than in HCBS, and they were equally as likely to be in the two facility settings. Individuals who were living in north, central, and west Florida were more likely to be in a NF than in an ALF or HCBS, and those in south Florida were more likely to be in HCBS programs than in either an ALF or a NF setting. Individuals who were Medicaid eligible were more likely to be in a NF or HCBS program and not in an ALF. These findings may confirm that there are very clear LTC supply differences between south Florida and the rest of the state (Oakley, 2000). For example, we believe that our findings reflect

Although past research has suggested that older people with very similar health and functioning profiles live in all three kinds of settings, the LTC system is structured to provide a higher level of care in NFs, followed by ALFs and HCBS. We expected that those with higher impairments in functioning and health would be in one of the two facility settings compared with HCBS. Methods of Analysis A multinomial logistic regression model was used to estimate the effects of differentiating predisposing, enabling, and need variables on utilization of services in one of the three settings (NFs, ALFs, or HCBS) when comparing each with one another. Interaction terms were included to improve the model’s ability to predict service utilization (Kosloski & Montgomery, 1994). Some predisposing and enabling variables have maximum impact on service use under conditions of high need (Calsyn & Winter, 2000). Thus, we tested the interactions for race/ethnicity, gender, and marital status with need factors of cognitive impairment, functional status, and chronic diseases. The interaction between incontinence and cognitive impairment was also tested. Before computing the interaction terms, we created center scores (Aiken & West, 1991) for continuous variables (cognitive impairment, functional status, and number of chronic diseases) to avoid problems with multicollinearity. All other variables in the interactions were dichotomies (female  1, non-Hispanic White  1, married  1). Interactions were entered into the regression equation as the product of the two variables. Significant interactions were retained in the final analysis reported here. Dropping nonsignificant interactions simplified the final analysis and further decreased the collinearity inherent with interaction terms (Bass, Looman, & Ehrlich, 1992). The same multinomial logistic regression was run twice, first to compare NFs versus HCBS and ALFs versus HCBS and then to compare NFs versus ALFs. A hierarchical approach allowed us to compare the predictive power of the predisposing variables (race/ ethnicity, age, gender), enabling variables (marital status, insurance, region of the state), need variables (functional disability, chronic diseases, and cognitive impairment), and significant interaction terms (e.g., chronic diseases and gender) on LTC utilization across the continuum of care.

Results Table 3 is a display of the multinomial logistic regression results showing the probability of receiving LTC in NF versus HCBS settings, ALF versus HCBS settings, and NF versus ALF settings. There was a good model fit when all the variables were entered: deviance 2(3,524, N  1,968)  2,655.34, p  1.000. Hierarchical comparison of the models as variables showed considerable improvement in explained variance, from .272 with the predisposing and en608

The Gerontologist

Table 3. Predictors of Long-Term Care Services Utilization by Setting NF vs HCBS Predictors Based on the Behavioral Model Predisposing Non-Hispanic White (reference  other ethnicity)a Age (reference  60–74) 85 75–84 Female Enabling Marital status: not married Medicaid eligible (reference  Medicare, other) Region (reference  South Florida) North Florida Central Florida West Florida Need Alzheimer’s disease or other dementia Incontinence Mobility (reference  independent) Can’t do it at all Some help/supervision Activities of daily living (0–5 ADLs) b Chronic diseases (0–16 diseases) Cognitive impairment (severity from 0–3) Interactions with chronic disease Female Not married Interactions with cognitive impairment Female Non-Hispanic White (reference  other ethnicity) a Not married Interactions with ADLs Not married Intercept

b

Exp(b)

ALF vs HCBS b

NF vs ALF

Exp(b)

b

Exp(b)

1.226

3.408***

0.976

2.654***

0.250

1.284

1.417 0.804 0.559

4.126*** 2.234*** 0.572**

1.074 0.339 0.482

2.928*** 1.404 0.618**

0.343 0.464 0.077

1.409 1.591 0.926

0.567 1.093

1.763** 2.983***

0.976 0.842

2.653*** 0.431***

0.409 1.935

0.665 6.924***

0.693 1.317 0.660

1.999** 3.731*** 1.936**

0.647 0.431 0.064

0.523** 0.650* 1.066

1.340 1.747 0.597

3.819*** 5.739*** 1.816**

1.905 0.052

6.722*** 0.949

1.325 0.349

3.760*** 0.706

0.581 0.297

1.788** 1.345

0.281 0.748 0.658 0.122 1.216

0.755 0.473*** 1.930*** 0.886 3.374***

1.533 0.146 0.049 0.489 0.661

0.216*** 1.157 1.050 0.613*** 1.936***

1.251 0.894 0.609 0.368 0.556

3.495*** 0.409*** 1.838*** 1.444** 1.743*

0.103 0.256

0.902 0.774

0.217 0.353

1.242 0.703**

0.320 0.097

0.726* 1.102

0.322 0.855 0.390

0.725 0.425*** 1.476*

0.277 0.552 0.153

0.758 0.576*** 1.165

0.045 0.302 0.237

0.956 0.739 1.267

0.269 1.308* 4.183***

0.103 1.108 1.620***

0.166 1.181 2.562***

Notes: NF  Nursing Facility; ALF  Assisted Living Facility; HCBS  Home- and Community-Based Services. N  1,968, NF (n  474), ALF (n  599), HCBS (n  895). 2 LL  2,701.03, p  .001; deviance 2(3,524)  2,655.34, p  1.000; Nagelkerke pseudo R2  .586; correctly classified: NF 62%, ALF 63%, HCBS 76%, and overall 69%. aOther ethnicity  Hispanic, Asian/Pacific Islander, Black, American Indian/Alaskan Native. bSee Table 2 for a list of ADLs entered for the analysis. *p  .05; **p  .01; ***p  .001; based on weighted data.

with severe cognitive impairment, were less likely to use NF or ALF services than they were to use HCBS. Similarly, there were three significant interactions between enabling and need variables. Individuals who were not married and who had more chronic disease were less likely to be in an ALF than in HCBS. However, if they were not married and had severe cognitive impairment or needed assistance with ADLs, they were more likely to be in a NF than in HCBS. In other words, with the exception of women with chronic diseases and individuals who were not married and in need of functional assistance, most of the impaired LTC consumers with physical and mental health needs were likely to receive HCBS rather than 24-hr care in an ALF or NF.

geopolitical differences in the supply of NF beds and alternatives. In addition, the limited supply of affordable ALFs and the limited scope of public programs that support the assisted living option constrain access to ALFs among low-income elders. All other variables being equal, individuals who were Medicaid-eligible or who lived in regions outside of south Florida were more likely to receive care in a NF than in an alternative setting. The last hypothesis—that the moderating influence of predisposing and enabling factors on need is significantly associated with the differential utilization of LTC services—was also supported. There were three significant interactions between predisposing and need variables. First, women who had more chronic diseases, compared with men with more chronic diseases, were more likely to be in a NF than they were to be in an ALF. When a comparison of NF and ALF with HCBS was done, women with chronic diseases were as likely to be in the two facility settings as they were to be at home. The other two interactions revealed that non-Hispanic Whites who had severe cognitive impairment, compared with ethnic minorities Vol. 42, No. 5, 2002

Discussion This cross-sectional study showed that predisposing, enabling, and need characteristics significantly differentiated LTC utilization. Need, as anticipated, contributed the most variance to the differential utilization 609

was spent on NF care (Polivka, 2001). Our findings imply that public financing of LTC does not reflect the capacity of the system to serve frail elders at home.

of LTC services. Need characteristics, such as more impairment in ADLs and diagnosis of Alzheimer’s disease, were associated with utilization along the standard continuum of care. That is, frail older adults with the highest need were more likely to receive care in the most skilled setting (NFs) and then in ALFs rather than at home. It appears, however, that differential utilization also occurred because of program policies. For example, ALFs do not accept individuals with total impairment in mobility. Medicaid eligibility and region of the state were two enabling variables that were associated with utilization without regard to specific need. Medicaid primarily covers nursing home services and, to a lesser extent, in-home care through the HCBS Medicaid Waiver. After data collection was completed, the state instituted a small-scale Medicaid Waiver program for ALFs. Although we expected that Optional State Supplementation–supported clients in assisted living who were on supplemental security income (and, generally, Medicaid eligible) might improve the impact of Medicaid on utilization across the full continuum, they did not. This is most likely because the Optional State Supplementation program is also relatively small compared with Medicaid support of NFs. Regional differences in nursing home bed supply and the availability of publicly supported alternatives to nursing home care also determine differential LTC program utilization. Coughlin and colleagues (1990) found in a national sample that a person living in an area with a higher than average nursing home bed supply was almost twice as likely to have a permanent nursing home stay than a person living in an area with a lower bed supply. Some very functionally impaired individuals in the sample were being cared for at home and not in NFs or ALFs. Individuals who were totally impaired in mobility or had multiple chronic diseases were in HCBS. Unmarried elders with chronic diseases were at home rather than in an ALF, but if they had severe cognitive impairment or needed assistance with ADLs, they were likely to be in a NF. A plausible explanation is that spousal caregivers are an important source of support in order to remain in the community (Jette et al., 1995; Weissert & Cready, 1989). Finally, although non-Hispanic Whites were more likely to be in NFs or ALFs, those with higher levels of cognitive impairment were more likely to be in HCBS. This may be less of a difference by race/ethnicity than a difference in access to LTC options when faced with caring for a family member with severe cognitive impairment (Mitchell et al., 2000). Overall, our findings suggest that a similar tendency might exist nationwide in the population of consumers of publicly funded LTC services that have these characteristics; however, it is important to remember that not all consumers follow this pattern of LTC use. In summary, inconsistent with state financing of LTC, highly functionally impaired individuals were also in HCBS settings, although reimbursement for HCBS is typically one-sixth of the NF reimbursement, and 89% of the 1995–1996 budget for LTC in Florida

Limitations This cross-sectional field study made practical use of existing client records across the full continuum of LTC services for one state. Florida does not represent the country, but it is the bellwether state for aging policy issues in the country. Rather than developing a standardized instrument for data collection, we used common elements from three existing standardized client and resident assessment instruments. In a few instances, source variables were similar but not an identical match (e.g., Medicaid eligibility), and new variables were created for analysis that were based on reasonably equivalent measures (e.g., eligibility for supplemental security income and welfare were proxies for Medicaid eligibility). In addition, the data available to us included admission data for NF and private-pay ALF residents, and admission data from the most recently completed Department of Elder Affairs Uniform Assessment Instrument for HCBS residents and for publicly supported (Optional State Supplementation) ALF residents. Although we relied on trained data collectors, there were different types of professionals who completed the assessments that provided the source of data for our study. Generally, these professionals included registered nurses in the NFs who completed the Minimum Data Set, state social worker/case managers in the HCBS and Optional State Supplementation programs who completed the Department of Elder Affairs Uniform Assessment Instrument, and licensed physicians or nurse practitioners who completed the Department of Elder Affairs Form 1823 required for assisted living. Future research that focuses on the utilization of LTC services across the continuum should consider collecting data by means of a single instrument and commonly trained evaluators to enhance assessment precision and data comparability. As a cross-sectional study, this research demonstrates associations between predictors and setting but does not necessarily predict utilization. Our sample consisted of clients who were already receiving LTC services in one of three settings and not from the point when they first needed assistance. Follow-up research with the same study population would clarify how these predictors work. More research that follows individuals through the disablement process and LTC decision making is still needed. Recommendations The expansion and structuring of LTC programs is needed to attain a more equitable distribution of services by minimizing the influences on service use of predisposing factors such as ethnicity and enabling factors such as geographic region of the resident, while maximizing the influence of need factors such 610

The Gerontologist

admission policies of ALFs, which often specialize in dementia care and primarily serve clients who are mobile and continent and who can, for the most part, self-finance this care rather than depend on public funding. In essence, our analysis allows for a comparison of important factors that distinctly differentiate consumers’ utilization of formal services along the LTC continuum.

as functional disability, chronic diseases, and cognitive impairment (Andersen & Newman, 1973; Spector & Fleishman, 1998). Consistent with our findings, the absence of a clear policy-driven relationship between the level of services provided and the needs (e.g., functional impairment) across long-term programs is probably characteristic of LTC services in most states. The development of a closer link between services and needs will become increasingly important for policymakers as the population needing LTC services grows and the pressure on state and federal fiscal resources mounts over the next several years. This has important fiscal implications. We have evidence that targeting works (Greene, Lovely, Miller, & Ondrich, 1995). We can spend more resources (such as are available through Medicaid waivers) on individuals who have higher impairments and save state and federal dollars. Yet, if targeting based on impairment is abandoned in favor of targeting based on financial eligibility, then we are not making effective use of these funds. In addition, more effective targeting should not neglect individuals who have fewer ADL impairments and need assistance in order to remain out of institutions. Targeting should also not neglect the need for General Revenue–funded programs that do not require a strict income and asset test. The process of targeting could be informed by regional needs assessments that are conducted at regular intervals and include a wide range of health, functional status, and socioeconomic and service capacity information that goes well beyond the demographic data that are commonly available from secondary sources. This information, in combination with the latest findings on the relative cost-effectiveness of LTC programs, could be used to adjust resource allocation formulas, including those governing the nursing home certificate of need determinations. Costs of routinely conducted regional needs assessments could be more than offset by the cost savings generated from the rigorous tailoring of LTC programs to those who are seriously impaired and with relatively few economic and social support resources.

References Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. Andersen, R. M. (1995). Revisiting the behavioral model and access to medical care: Does it matter? Journal of Health and Social Behavior, 35, 1–10. Andersen, R. M., & Newman, J. F. (1973). Societal and individual determinants of medical care utilization in the United States. The Milbank Fund Quarterly, 51, 95–124. Bass, D. M., Looman, W. J., & Ehrlich, P. (1992). Predicting the volume of health and social services: Integrating cognitive impairment into the modified Andersen framework. The Gerontologist, 32, 33–43. Bass, D. M., & Noelker, L. S. (1987). The influence of family caregivers on elders’ use of in-home services: An expanded conceptual framework. Journal of Health and Social Behavior, 28, 184–196. Boaz, R. F., Muller, C. F., & Hu, J. (1994). Predicting the risk of “permanent” nursing home residence: The role of community help as indicated by family helpers and prior living arrangements. Health Services Research, 29, 391–414. Branch, L. G., & Jette, A. M. (1982). A prospective study of long-term care institutionalization among the aged. American Journal of Public Health, 72, 1373–1379. Calsyn, R. J., & Winter, J. P. (2000). Predicting different types of service use by the elderly: The strength of the behavioral model and the value of interaction terms. The Journal of Applied Gerontology, 19, 284–303. Cohen, G. D. (1994). Health care reform and older adults: Introduction. The Gerontologist, 34, 584–585. Cohen, M. A., Tell, E. J., & Wallack, S. S. (1986). Client-related risk factors of nursing home entry among elderly adults. Journal of Gerontology, 41, 785–792. Coughlin, T. A., McBride, T. D., & Lui, K. (1990). Determinants of transitory and permanent nursing home admissions. Medical Care, 28, 616– 631. Dwyer, J. W., Barton, A. J., & Vogel, W. B. (1994). Area of residence and the risk of institutionalization. Journal of Gerontology: Social Sciences, 49, S575–S584. Egleston, B. L., Rudberg, M. A., & Brody, J. A. (1999). Prior living arrangements and nursing home resident admission ADL characteristics: A study of two states. Journal of Gerontology: Social Sciences, 54B, S203–S206. Foley, D. J., Ostfeld, A. M., Branch, L. G., Wallace, R. B., McGloin, J., & Cornoni-Huntley, J. C. (1992). The risk of nursing home admission in three communities. Journal of Aging and Health, 4, 155–173. Freedman, V. A., Berkman, L. F., Rapp, S. R., & Ostfeld, S. R. (1994). Family networks: Predictors of nursing home entry. American Journal of Public Health, 84, 843–845. Greene, V. L., Lovely, M., Miller, M., & Ondrich, J. (1995). Reducing nursing home use through community long-term care: An optimization analysis. Journal of Gerontology: Social Sciences, 59B, S259–S268. Greene, V. L., & Ondrich, J. I. (1990). Risk factors for nursing home admissions and exits: A discrete-time hazard function approach. Journal of Gerontology: Social Sciences, 45, S250–S258. Hanley, R. J., Alecxih, L. M. B., Wiener, J. M., & Kennell, D. L. (1990). Predicting elderly nursing home admissions: Results from the 1982– 1984 National Long-Term Care Survey. Research on Aging, 12, 199– 228. Heyman, A., Wilkinson, W. E., Hurwitz, B. J., Helms, M. J., Haynes, C. S., Utley, C. M., et al. (1987). Early-onset Alzheimer’s disease: Clinical predictors of institutionalization and death. Neurology, 37, 980–984. Jette, A. M., Tennstedt, S., & Crawford, S. (1995). How does formal and informal community care affect nursing home use? Journal of Gerontology: Social Sciences, 50B, S4–S12. Kart, C. S., & Palmer, N. M. (1987). How do we explain differences in the level of care received by the institutionalized elderly? The Journal of Applied Gerontology, 6, 53–66. Kelman, H. K., & Thomas, C. (1990). Transitions between community and nursing home residence in an urban elderly population. Journal of Community Health, 15, 105–122. Kemper, P. (1992). The use of formal and informal home care by the disabled elderly. Health Service Research, 27, 421–451.

Conclusions Findings from this study were in line with what was expected on the basis of other LTC utilization research. Essentially, the literature and our findings indicate that the best predictors of nursing home placement are functional impairment (i.e., ADLs), Alzheimer’s disease, marital status (i.e., not married), and ethnicity (i.e., non-Hispanic White). There is also consistency in predicting that disadvantaged individuals are more likely to receive formal LTC services in their home, but we provide further insight into the severity of impairment of these consumers. This study adds value to earlier research because we considered all the LTC programs available to frail elders in need of care. In other words, a novel contribution is the inclusion of ALFs to the common comparison of nursing homes to home- and communitybased programs. Our findings clearly reflect the distinct Vol. 42, No. 5, 2002

611

Severson, M. A., Smith, G. E., Tangalos, E. G., Petersen, R. C., Kokmen, E., Ivnik, R. J., et al. (1994). Patterns and predictors of institutionalization in community-based dementia patients. Journal of the American Geriatrics Society, 42, 184–185. Spector, W. D., & Fleishman, J. A. (1998). Combining activities of daily living with instrumental activities of daily living to measure functional disability. Journal of Gerontology: Social Sciences, 53B, S46–S57. Spector, W. D., Fleishman, J. A., Pezzin, L. E., & Spillman, B. C. (2000). The characteristics of long-term care users. Rockville, MD: Agency for Healthcare Quality. Steele, C., Rovner, B., Chase, G. A., & Folstein, M. (1990). Psychiatric symptoms and nursing home placement of patients with Alzheimer’s disease. American Journal of Psychiatry, 147, 1049–1051. Steinmo, S., & Watts, J. (1995). It’s the institutions, stupid! Why comprehensive national health insurance always fails in America. Journal of Health Politics, Policy, and Law, 20, 329–383. Stern, Y., Tang, M.-X., Albert, M. S., Brandt, J., Jacobs, D. M., Bell, K., et al. (1997). Predicting time to nursing home care and death in individuals with Alzheimer’s disease. Journal of the American Medical Association, 277, 806–812. Temkin-Greener, H., & Meiners, M. R. (1995). Transitions in long-term care. The Gerontologist, 35, 196–206. Wallace, S. P., Levy-Storms, L., Andersen, R. M., & Kington, R. S. (1997). The impact by race of changing long-term care policy. Journal of Aging and Social Policy, 9(3), 1–21. Wallace, S. P., Levy-Storms, L., & Ferguson, L. R. (1995). Access to paid in-home assistance among disabled elderly people: Do Latinos differ from non-Latino whites? American Journal of Public Health, 85, 970–975. Wallace, S. P., Levy-Storms, L., Kington, R. S., & Andersen, R. M. (1998). The persistence of race and ethnicity in the use of long-term care. Journal of Gerontology: Social Sciences, 53B, S104–S112. Weissert, W. G., & Cready, C. M. (1989). Toward a model for improved targeting of aged at risk of institutionalization. Health Services Research, 24, 485–510. Welch, H. G., Walsh, J. S., & Larson, E. B. (1992). The cost of institutional care in Alzheimer’s disease: Nursing home and hospital use in a prospective cohort. Journal of the American Geriatrics Society, 40, 221– 224. Wiener, J. M., Hanley, R. J., Clark, R., & Van Nostrand, J. F. (1990). Measuring the activities of daily living: Comparisons across national surveys. Journal of Gerontology: Social Sciences, 45, S229–S237. Wolinsky, F. D. (1990). Health and health behavior among elderly Americans: An age-stratification. New York: Springer. Wolinsky, F. D., Callahan, C. M., Fitzgerald, J. F., & Johnson, R. J. (1992). The risks of nursing home placement and subsequent death among older adults. Journal of Gerontology: Social Sciences, 47, S173–S182. Wolinsky, F. D., Callahan, C. M., Fitzgerald, J. F., & Johnson, R. J. (1993). Changes in functional status and the risks of subsequent nursing home placement and death. Journal of Gerontology: Social Sciences, 48, S94– S101. Wolinsky, F. D., & Johnson, R. J. (1991). The use of health services by older adults. Journal of Gerontology: Social Sciences, 46, S345–S357.

Klerk, M. M. Y., Huijsman, R., & McDonnell, J. (1997). The use of technical aids by elderly persons in The Netherlands: An application of the Andersen and Newman model. The Gerontologist, 37, 365–373. Kosloski, K., & Montgomery, R. J. V. (1994). Investigating patterns of service use by families providing care for dependent elders. Journal of Aging and Health, 6, 17–37. Liu, K., McBride, T., & Coughlin, T. (1994). Risk of entering nursing homes for long versus short stays. Medical Care, 32, 315–327. McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detecting interactions and moderator effects. Psychological Bulletin, 114, 376–390. Mercier, J. M., & Shelley, M. C. (1997). Access to health care among three cohorts of older Americans residing in a rural state. Policy Studies Journal, 25, 140–156. Miller, S. C., Prohaska, T. R., Furner, S. E., Freels, S., Brody, J. A., & Levy, P. S. (1998). Time to nursing home admission for persons with Alzheimer’s disease: The effect of health care system characteristics. Journal of Gerontology: Social Sciences, 53B, S341–S359. Miner, S. (1995). Racial differences in family support and formal service utilization among older persons: A nonrecursive model. Journal of Gerontology: Social Sciences, 50B, S143–S153. Mitchell, J., Matthews, H. F., & Hack, K. A. (2000). Differences by race in long-term care plans. The Journal of Applied Gerontology, 19, 424– 440. Mittleman, M. S., Ferris, S. H., Shulman, E., Steinberg, G., & Levin, B. (1996). A family intervention to delay nursing home placement of patients with Alzheimer’s disease. Journal of the American Medical Association, 276, 1725–1731. Mittleman, M. S., Ferris, S. H., Steinberg, G., Shulman, E., Mackell, J. A., Ambinder, A., et al. (1993). An intervention that delays institutionalization of Alzheimer’s disease patients: Treatment of spouse-caregivers. The Gerontologist, 33, 730–740. Montgomery, R. J. V., & Kosloski, K. (1994). A longitudinal analysis of nursing home placement for dependent elders cared for by spouses vs adult children. Journal of Gerontology: Social Sciences, 39, S562–S574. Morris, J. N., Sherwood, S., & Gutkin, C. E. (1988). Inst-risk II: An approach to forecasting relative risk of future institutional placement. Health Sciences Research, 23, 511–536. Murtaugh, C. M. (1994). Discharge planning in nursing homes. Health Services Research, 28, 751–769. Murtaugh, C. M., Kemper, P., & Spillman, B. (1990). The risk of nursing home use in later life. Medical Care, 28, 952–962. Oakley, M. L. (2000). Controlling nursing home bed supply must be part of the plan to promote aging in place. Aging Research and Policy Report, 9, 18–23. Polivka, L. (2001). Home and community-based services. In L. Polivka & J. R. Salmon (Eds.), Informational report of the task force on availability and affordability of long-term care for the Florida legislature in response to House Bill 1993: Vol. 2. Research (pp. 317–337). Tampa: Florida Policy Exchange Center on Aging. Polivka, L., Dunlop, B. D., & Brooks, M. (1997). Project two: The Florida Long-Term Care Elder Poplation Profiles survey. Tampa: Florida Policy Exchange Center on Aging. Polivka, L., Dunlop, B. D., & Rothman, M. B. (1996). Long-term care for the frail elderly in Florida: Expanding choices, containing costs. Tampa: Florida Policy Exchange Center on Aging. Rothman, D. J. (1993). A century of failure: Health care reform in America. Journal of Health Politics, Policy, and Law, 18, 271–286. Rudberg, M. A., Sager, M. A., & Zhang, J. (1996). Risk factors for nursing home use after hospitalization for medical illness. Journal of Gerontology: Medical Sciences, 51A, M189–M194.

Received October 30, 2001 Accepted April 16, 2002 Decision Editor: Laurence G. Branch, PhD

612

The Gerontologist