This Article Abstract Full Text (PDF) Services Download to citation manager PubMed PubMed Citation

The Relative Benefits and Cost of Medicaid Home- and Community-Based Services in Florida

Correspondence: Address correspondence to Glenn Mitchell, PhD, State Data Center on Aging, School of Aging Studies, Florida Policy Exchange Center on Aging, University of South Florida, USF#30437, Tampa, FL 33620. E-mail: gmitchel{at}cas.usf.edu

	Abstract

TOP Abstract Methods Results Discussion References

 
Purpose: We compared inpatient days, nursing home days, and total Medicaid claims for five Medicaid-funded home- and community-based services (HCBS) programs for in-home and assisted living services in Florida. Design and Methods: We studied a single cohort of Medicaid enrollees in Florida aged 60 and older, who were enrolled for the first time in any of five Medicaid HCBS programs and who had at least one assessment (N = 6,014). In this 3-year longitudinal study, we used a two-stage probit regression and ordinary least squares regression in order to test the independent effects of explanatory variables on outcomes and cost. Results: After controlling for differences in frailty, chronic health conditions, presence of dementia, and available caregiver, we found that Medicaid HCBS programs had a differential effect on hospital and nursing home utilization and cost. Implications: Medicaid HCBS programs serve very impaired populations at a wide range of costs to Medicaid. The rates for Medicaid HCBS programs could be adjusted upward or downward in order to better reflect the level of need in each program. At the same time, providers could use titration techniques based on the relative costs of these needs in order to budget for the costs of meeting the needs of particular elders.

Key Words: Home- and community-based services • Long-term care • Managed care • Outcomes • Public policy

Most research on the cost and benefits of particular HCBS programs (e.g., adult day care, hospice, dementia care) using experimental or quasi-experimental designs has reported nonsignificant outcomes; the few outcomes that have been significant have also been positive (Weissert & Hedrick, 1994). The notable exception is reduction in unmet need, which was found in 21 out of 35 case-controlled studies (Weissert & Hedrick). The most widely recognized early study of HCBS, the National Long-Term Care Channeling Demonstration, which provided an array of services in the home, found little empirical support for cost effectiveness of HCBS programs, although there were high levels of consumer satisfaction with quality of care (Kemper, 1988). A reanalysis of the Channeling data identified the optimum use of HCBS, based on better targeting of enrollees who have greater need for assistance with personal care activities of daily living (ADLs) and instrumental ADLs, have cognitive impairments, live alone, have lower income, and have fewer potential caregivers (Greene, Lovely, Miller, & Ondrich, 1995).

In the 20 years since the evaluation of Channeling and other HCBS programs and services, 49 states have implemented 1915(c) Medicaid waivers for HCBS (such as Florida's Aged and Disabled Adult), and 41 states have implemented this waiver for assisted living (such as Florida's Assisted Living for the Elderly; Miller, 2003). In addition, some states have developed capitated-rate managed long-term-care programs, such as Florida's Nursing Home Diversion waiver, funded under a 1915(c) Medicaid waiver. For-profit managed care organizations are increasing their presence in these Medicaid long-term-care markets. One such provider, Evercare, is providing Medicaid managed long-term care in 16 states (United HealthCare Services, 2004). Partially in response to the growing challenges to the traditional aging network's role in providing HCBS, 44 states and territories have recently received funding for Aging and Disability Resource Centers, which integrate long-term-care resources through a single coordinated system based in the traditional not-for-profit aging network ("AoA, CMS seek to expand," 2005). In effect, these Aging and Disability Resource Centers could become part of the foundation for not-for-profit managed care systems.

With all of this activity, states may have learned to manage HCBS services more efficiently (Wiener et al., 2004), warranting case studies of individual state experiences of Medicaid HCBS programs. Arizona and Wisconsin are two such examples. Arizona contained long-term-care costs through consolidated policy and budget control over both HCBS and nursing homes and by using a managed long-term-care approach with local government, for-profit, and not-for-profit providers (Weissert, Lesnick, Musliner, & Foley, 1997). As a result, the state expanded HCBS and reduced nursing home use. Wisconsin's Family Care Program uses aging network-based managed long-term-care systems in order to provide a broad array of HCBS and nursing home care in several areas of the state (Alecxih, Neill, Zeruld, & Olearczyk, 2002).

Because most states are expanding these Medicaid-funded HCBS alternatives, the efforts of policy makers and researchers now focus increasingly on improving the efficiency of these programs (Grabowski, 2003). There is a need for methodologically rigorous research to determine the potential cost savings from recent waiver developments for state long-term-care systems (Wiener et al., 2004). One way to accomplish this is through a comparative analysis of cost and benefits of Medicaid HCBS programs within a single state system.

This study compares the cost and benefits of five Medicaid HCBS programs available to elders in Florida. Florida is the fourth largest state and has the second largest population aged 65 and older. In addition, Florida has the highest proportion of elders in its population (17%), as well as nearly 20 years of experience with pilot programs that provide managed long-term-care services to an impaired population (Table 1).

The Agency for Health Care Administration (AHCA) funds and administers two programs. The Channeling Project began as part of a national demonstration project to evaluate the effectiveness of diverting potential nursing home residents to receive in-home services (Kemper, 1988). This program began in 1982 and was continued by the state in 1985 under a 1915(c) waiver. The Miami Jewish Home and Hospital for the Aged in Dade and Broward counties operates the program, which serves 1,724 individuals who are aged 65 or older, qualify for nursing home care, meet income and asset requirements for Supplemental Security Income or Institutional Care Program, and receive Medicaid services. Channeling has the least restrictive eligibility requirements (two or more unmet long-term-care needs) and has the lowest capitated rate (Table 1). The $843 per member per month provides case management and access to an array of HCBS, including most of the services provided by the other waiver programs, with the exception of nursing facility care. Caseloads for Channeling are around 60 per case manager. High-need clients make heavy use of adult day care.

The state has funded the Frail Elder Project since 1987, when it was operated by Mount Sinai Medical Center. It is now operated by United Healthcare and serves 4,302 clients who are aged 21 and older, are on Supplemental Security Income, and meet requirements for nursing home level of care. The Frail Elder Project is now funded as an optional "expanded service" within the state's standard Medicaid health maintenance organization (HMO) provider agreement (Dunlop, Oakley, & Seff, 2004). The minimum level of frailty for the Frail Elder Project is needing help with two or more ADLs. Its monthly rate of $1,190 pays for some but not all of the services that Channeling covers, but the contractor is at risk for up to one year of nursing home care. Caseloads for the Frail Elder Project are around 125 per case manager.

The AHCA funds three other Medicaid waiver programs that are administered by the Department of Elder Affairs. In 1997, the Florida legislature enacted the Long-Term Care Community Diversion Pilot Project Act in order to provide comprehensive acute and long-term-care services to individuals who are dually eligible for Medicare and Medicaid. The Medicare HMO capitated rate was not adjusted for the higher acuity levels of this population (as it is in the Program of All-inclusive Care for the Elderly), and the state required contractors to allow clients to opt out of enrolling in the contractor's Medicare HMO for acute care services. As a result, contractors have opted for the Medicaid and long-term-care services capitated rate and not the Medicare capitation. Contractors provide "coordinated" rather than "integrated" care. The State Fiscal Year (SFY) 1999–2000 capitated rate was $2,342 per member per month, three fourths of the average monthly Medicaid claim for nursing home care. Contractors are responsible for all long-term-care services (including nursing home care, prescription drugs, acute care copayments, and deductibles) as long as the client is enrolled in the program. (Salmon et al, 2001). The goals are broader than just cost effectiveness. Consistent with many HCBS programs, program administrators allow contractors considerable discretion in case management in order to encourage innovation. Eligibility for this program is the strictest and requires, at a minimum, needing help with three or more ADLs. Caseloads for this newest program were around 40 clients per case manager at the time of this study.

The Aged and Disabled Adult program began in 1982 and was the largest among the five Medicaid HCBS programs until 2004. This fee-for-service program is operated by not-for-profit aging network providers. Aged and Disabled Adult enrollees must demonstrate deterioration in function that places them at risk for nursing facility admission without the provision of HCBS. Caseloads for this program are around 80 clients per case manager. The Aged and Disabled Adult covers nearly all of the same services as Nursing Home Diversion with the exception of nursing home care (although skilled nursing is covered), behavior management, medication administration, and therapeutic and recreational services.

Assisted Living for the Elderly began in 1995 and enables a limited number of eligible elders to receive care at less cost than in nursing homes. Assisted Living for the Elderly pays for services but does not cover room and board. In Florida, assisted living facilities are predominantly for-profit independent businesses, although there are a number of not-for-profit providers. The program is capped at $1,065 per member per month, with up to $840 for assisted living services, $125 for incontinence supplies, and $100 for case management provided through the aging network. Caseloads for this program are around 80 clients per case manager. Assisted living services are similar to those of the other waiver programs, with the exception that Assisted Living for the Elderly does not provide adult day care or respite care because the clients are residents. The analyses presented here represent the first time researchers have compared all of Florida's Medicaid HCBS programs to one another in terms of total Medicaid costs and benefits.

We posed three main research questions for this study: (a) Are there important differences between these five Medicaid HCBS programs in hospital utilization? (b) Are there important differences between these five Medicaid HCBS programs in nursing home utilization? (c) Are there important differences between these five Medicaid HCBS programs in their overall costs to Medicaid?

	Methods

TOP Abstract Methods Results Discussion References

 
This longitudinal study analyzes benefits (fewer hospital inpatient and nursing home days) and cost (Medicaid claims) for consumers who were enrollees in any of Florida's five Medicaid HCBS programs during SFY 1999–2000 (N = 6,014).

Data Sources
The data for this study were from a single cohort of Medicaid HCBS enrollees in Florida who were aged 60 and older; who enrolled during SFY 1999–2000 for the first time in Aged and Disabled Adult, Assisted Living for the Elderly, Channeling, Nursing Home Diversion, or Frail Elder Project; and who were followed for three years through SFY 2001–2002.

Hospital inpatient, nursing home, and Medicaid claims data from acute and long-term-care providers are reported to AHCA. Comprehensive enrollee assessment data are reported to the Florida Department of Elder Affairs and captured in two administrative databases, CARES and CIRTS. These databases include assessments of functional status, mental impairment, chronic health conditions, nutrition, and social support for Florida elders who are candidates for long-term care in a nursing home (CARES) or in the community (CIRTS). The Florida Medicaid monthly claims file aggregates claim activity in terms of the number of units and the dollar value of claims paid for commonly used expenditure categories such as inpatient hospital care and nursing home care.

Services that are included under the capitated rates for Channeling, Nursing Home Diversion, and Frail Elder Project do not generate additional claims. In the case of Nursing Home Diversion and Frail Elder Project, their monthly Medicaid claims cover copayments for Medicare inpatient and outpatient care, short-term nursing home stays, prescriptions, and long-term nursing home care. Due to inconsistencies in self-reported data for inpatient days from the managed care organizations, we substituted data from the State Center for Health Statistics at AHCA.

The State Data Center on Aging in the School of Aging Studies at the University of South Florida maintains a data warehouse in order to assist Florida policy makers with outcome studies and program evaluations. State Data Center on Aging holdings include statewide Medicaid claims and the complete CARES and CIRTS databases. We linked administrative records from AHCA claims files with CARES and CIRTS records by using probabilistic record matching, a statistically sound technique for matching data across different databases (Baldwin, Acheson, & Graham, 1987; Newcombe, 1988). The result was an analytic file yielding monthly Medicaid eligibility and claims data for enrollees in the five Medicaid HCBS program along with needs assessments.

There were a total of 7,336 new enrollees during SFY 1999–2000 in at least one of the five Medicaid HCBS programs. Of these, 1,322 (18%) did not have a single assessment during the entire study period (1999–2002); we excluded them from these analyses. The excluded enrollees were more likely to be female, less likely to be White, and more likely to be in Channeling or Frail Elder Project. These two programs do not have a way to electronically input their updated assessment information into the state's databases (M. Walzer, Miami Jewish Home, personal communication, September 19, 2005). In terms of outcomes, people without an assessment cost the State of Florida less and spent less time in a hospital or a nursing home.

Measures
We computed participants' age at time of admission into one of the five programs. We recorded race from White, African American, Hispanic, and other to White and non-White. Other data included gender; caregiver availability; living in the community (including an assisted living facility); assessment in a nursing home or hospital; needing help with medications; number of chronic health conditions (0–13); presence of 4 conditions of interest for one or more models in the study: bed sores, dementia, diabetes, and stroke; ADLs (0–6), including bathing, dressing, eating, transferring, toileting, and walking; and IADLs (0–8).

From the ADL and chronic health conditions information, we computed a new variable: meets Diversion frailty criteria. Of the five programs, Nursing Home Diversion receives the highest monthly Medicaid claims for HCBS services, and its threshold for admission is the highest: (a) some assistance with at least five ADLs, or (b) some assistance with four ADLs and assistance with medications, or (c) a diagnosis of Alzheimer's or dementia and some assistance with three or more ADLs, or (d) total assistance with two or more ADLs, or (e) a degenerative or chronic condition requiring daily nursing services. We coded individuals who met the Nursing Home Diversion eligibility criteria as "more impaired" and those who did not as "less impaired." Although one would expect all Nursing Home Diversion enrollees to meet these criteria, the fifth criterion (requiring daily nursing assistance) is not included in the CARES or CIRTS administrative databases and therefore could not be included in any determination of who met these criteria. Other programs use this criterion as well. We dichotomized ADLs and IADLs, with 1 = total help, some help, or coaching, and 0 = environmental adaptation, assistive device, or no assistance.

We included all five Medicaid HCBS programs in these analyses. We did not exclude a baseline category, because enrollment is not mutually exclusive. About 10% of enrollees switched programs at some point during the three years of this study. Enrollees in Medicaid HCBS programs often spend some months in Aged and Disabled Adult while waiting for an open slot in another waiver or demonstration project.

When a categorical variable with k categories is converted into a set of dichotomous variables, it is customary to make k-1 dummy variables. The excluded category serves as a baseline when all of the dichotomous independent variables in a regression equation equal zero. Excluding a baseline category prevents perfect collinearity (the only other alternative being to exclude the intercept from the regression model altogether).

However, we did not construct the Medicaid HCBS dummy variables from a categorical variable. Using k-1 dummies was unnecessary in this study because Medicaid enrollees were free to move among the various waivers and demonstration projects during the study period. Although it would have been possible to force the data to fit mutually exclusive categories by excluding people who had switched programs or by lumping all of the switchers together in a single category, we would have lost valuable information on the various combinations of HCBS waiver experience. For example, we would expect an enrollee who had spent time in both Aged and Disabled Adult waiting for an opening and then in Nursing Home Diversion to have lower per-member-per-month (PMPM) total Medicaid claims (assuming all else remained the same) compared with an enrollee in Nursing Home Diversion who did not transition into or out of that program during the study period.

We calculated average PMPM services by summing a variable and dividing by the number of Medicaid case months. We counted months with zero utilization in the denominator, provided the enrollee was Medicaid eligible during that month. PMPM inpatient days is a marker of health deterioration and also of the ability of Medicaid HCBS programs to help prevent worsening outcomes. PMPM nursing home days is a measure of the extent to which these programs improve the quality of life of enrollees by preventing or delaying the need for nursing home care. PMPM claims is the average total Medicaid claims per enrollee for each program. We expected that the more service inclusive managed care programs would absorb most of the costs to Medicaid through their capitation rates. Most enrollees had Medicaid inpatient, nursing home, or other Medicaid claims in the 6-month period before they enrolled in one of the five Medicaid HCBS programs; we used these benefits and costs when measuring pre-program experience.

We included dummy variables for the 11 planning and service areas used by Florida agencies. The largest planning and service area (Region 11) serves South Florida (Dade and Monroe counties) and is the excluded baseline reference region. We included the Medicaid bed occupancy rate for nursing homes in the enrollee's county in the model predicting nursing home days, because the availability of nursing home beds can be a predictor of low use of HCBS alternatives (Mollica, 2003).

Method of Analysis
We used bivariate and multivariate analyses in order to answer the three research questions. We used ordinary least squares (OLS) regression in order to estimate total PMPM Medicaid claims. We conducted a two-stage probit regression (Alverez & Glasgow, 1999; Borjas & Sueyoshi, 1993) on a set of simultaneous equations in order to test the independent effects of explanatory variables on the two outcome variables: PMPM inpatient hospital days and PMPM nursing home days. A probit model is often used when the dependent variable is skewed by cases with a zero. In this case, many enrollees had no hospitalizations or nursing home.

	Results

TOP Abstract Methods Results Discussion References

 
Enrollee Characteristics
Enrollees in Florida's Medicaid HCBS programs were aged, frail, and female (see Tables 2 and 3). The average age was 81, and two fifths were African American, Hispanic, or of other ethnicity. Enrollees typically needed some or total help with three or more ADLs (out of six) and seven IADLs (out of eight). In addition, HCBS enrollees were chronically ill—they had been diagnosed, on average, with at least three chronic health conditions. Just 44% of these enrollees had a caregiver available, and 37% had a diagnosis of dementia.

Inpatient Care
The need for hospitalization was a relatively rare event. More than one third of the enrollees had no inpatient days during the 3-year period covered by this study. Although the enrollees in HCBS waivers and demonstration projects were frail, predictions of hospitalization at the level of the individual for a fixed study period would contain a large measure of stochastic error (R² = 0.09; Table 4).

Inpatient days prior to program enrollment, meeting Nursing Home Diversion frailty criteria, having been assessed in a hospital or nursing home, presence of an available caregiver, and race had a more moderate effect relative to enrollment in Nursing Home Diversion.

There was scant empirical evidence for regional differences in expected inpatient days after controlling for race, gender, and the other factors in Table 4. Although the coefficients were not zero, they were generally so close to zero that they merit little comment here.

One might be tempted to attribute the small coefficients for region to the effects of multicollinearity. It is an empirical fact that program participation and region are correlated. For example, Nursing Home Diversion did not (and still does not) have statewide enrollment. Multicollinearity affects only estimates of the standard errors, biasing them upwards and potentially leading a reader to conclude erroneously that an effect is not statistically significant. OLS point estimates are unbiased by multicollinearity (Johnston, 1984; Judge, Griffiths, Carter Hill, Lutkepohl, & Lee, 1985; Kmenta, 1986; Pindyck & Rubinfeld, 1991). This study used population data, rather than a random sample. We therefore do not report statistical significance (McCloskey, 1985; Shafer, 1993; Schmidt, 1996; Ziliak and McCloskey, 2004). Because it is easy to confuse statistical significance and substantive significance, we will avoid the use of the word significant when describing OLS estimates (McCloskey, 1985; Ziliak and McCloskey, 2004).

Because Nursing Home Diversion and Frail Elder Project are both managed care waivers, one might reasonably expect their enrollees to experience fewer hospitalizations (in the two-stage probit analysis referenced in the Method of Analysis section, Nursing Home Diversion and Frail Elder Project also had the highest hospitalization rate, all else being the same or similar) and shorter lengths of stay. This finding of higher hospital utilization for the managed care HCBS programs bears further research. One possible explanation is cost shifting from HCBS contractors to Medicare. All Nursing Home Diversion enrollees are dually eligible for Medicaid and Medicare services. This is a program requirement. Frail Elder Project does not have a dual eligibility requirement, although the majority of their enrollees are dually eligible. The financial risk for hospitalizations is Medicare deductibles and copayments. This contrasts with the larger financial risk for non-Medicare services like respite care, wherein the HCBS contractor is at risk for the entire cost of care.

Nursing Home Care
The model to explain PMPM nursing home days (Table 5) had a much better fit than the model for inpatient days (R² = 0.25).

Requiring some or total assistance with medications (+2.03 days), meeting Nursing Home Diversion frailty criteria (+2.03 days), and Medicaid nursing home bed occupancy rate (–3.10 x Rate) also had a moderately strong effect on estimates of PMPM nursing home days.

Regional differences in PMPM nursing home days were evident. Compared with residing in Miami Dade or Monroe counties (the Region 11 baseline category), residing in another region of the state increased expected PMPM nursing home days by one half to one full day (the one exception being Broward county, where expected nursing home days were reduced by a half day). This likely reflects the uneven access to nursing home care alternatives in Florida. It is not sufficient to be enrolled in an HCBS program when an enrollee can no longer live independently. There must also be available alternatives accessible to the enrollee that can divert the enrollee from nursing home care (e.g., assisted living facilities, adult health day care, transportation services, and the like). The penetration of such services into rural areas of Florida is a continuing challenge for Florida Medicaid and the Florida Department of Elder Affairs.

Enrollment in any of the HCBS waiver programs reduced PMPM days in a nursing home, with enrollment in Assisted Living for the Elderly or Nursing Home Diversion showing the largest reduction (–2.54 and –1.27 days, respectively).

Total Medicaid Claims
The OLS coefficients for enrollment in the five Medicaid HCBS programs describe the average difference in PMPM total Medicaid claims after controlling for other relevant variables (Table 6). The fit of the data to the regression line, as evidenced by R² = 0.62, is remarkable for individual-level health service data.

There was considerable variation in the effect of enrollment among the five Medicaid HCBS programs. Enrollment in Nursing Home Diversion had the second largest impact on PMPM total Medicaid claims. Controlling for other factors, enrollment in Nursing Home Diversion increased expected total Medicaid claims by $952 monthly. This contrasted with the other four Medicaid HCBS programs, which ranged from a reduction in expected claims of $139 per month for Frail Elder Project and $57 for Aged and Disabled Adult to increased claims of $88 per month for Channeling and $229 for Assisted Living for the Elderly. Managed care programs anchored both ends of the PMPM total Medicaid claims experience.

Enrollees who died had lower total PMPM Medicaid claims by $252.

There is empirical evidence of important regional differences in the PMPM total Medicaid claims shown in Table 6. Enrollees from the eastern panhandle, for example, had $365 less per month in predicted Medicaid claims, compared to enrollees in the baseline region (Miami Dade and Monroe counties).

Age had an inverse relationship to total PMPM claims, once other factors were controlled. The effect was moderately strong (–$10 per year). We also tested an alternative specification, namely that the relationship between age and Medicaid claims was a parabola (i.e., the inclusion of both age and age²). The fit of the model was not appreciably improved with the curvilinear specification, so we prefer the more parsimonious model with age being a linear term.

Requiring assistance with the medications (as measured by the IADL for medications) increased PMPM Medicaid claims by $93. Meeting the Nursing Home Diversion frailty criteria added $80 per month. Each chronic health condition added $9 per month. The presence of dementia added $23 PMPM.

Aspects of care that one might reasonably expect to reduce total Medicaid claims actually proved to increase costs, assuming other factors stayed the same. Living in the community added $76 PMPM. Having an available caregiver added a meager $9 per month.

Frailty-Adjusted Service Utilization and Claims
There were stark differences in service use and costs between the five Medicaid HCBS programs when level of disability was controlled (Table 7). Among the enrollees who met the Nursing Home Diversion criteria ("more impaired"), Nursing Home Diversion enrollees spent four times longer in the hospital compared with enrollees of all other programs (14.6 days vs 3.7 days). This is 8 times longer than Channeling enrollees and 2.6 times longer than Frail Elder Project enrollees of similar impairment. Aged and Disabled Adult and Assisted Living for the Elderly enrollees spent an average amount of time in the hospital, but they spent the most days in a nursing home (117 and 97 days, respectively) compared with other enrollees of similar impairment who spent 63–64 days (Frail Elder Project and Channeling) to 86 days (Nursing Home Diversion) in a nursing home.

	Discussion

TOP Abstract Methods Results Discussion References

 
This study evaluated two health benefits (fewer inpatient and nursing home days) and cost (total Medicaid claims) for five Medicaid HCBS programs in Florida. These programs are similar to programs in most other states that make use of the 1915(c) waiver for home care (N = 49) and assisted living services (N = 41) and to programs in a growing number of states that are using managed long-term-care providers for Medicaid HCBS. Lessons from this study can be applied to states that serve equally impaired individuals through a variety of Medicaid HCBS programs and to HCBS providers interested in understanding the relationship between client characteristics, service use, and costs. As described earlier, all of these programs cover costs for case management, skilled nursing, and emergency response systems, as well as a variety of typical services (e.g., adult day care, companionship, meals, physical therapy, and consumable supplies), depending on the program. We focused this research on the program funding source, not the utilization of individual services, although their costs are included in the total claims.

Policy Implications
These findings suggest changes in the rates paid to Medicaid HCBS providers in Florida and to similar programs in other states. Florida pays $1,074 more per member per month for enrollees in Nursing Home Diversion compared with enrollees in Frail Elder Project. This is more than the other three programs that serve very impaired elders. Yet, Nursing Home Diversion and Frail Elder Project are virtually identical on most measures of long-term-care need (race, gender, ADLs, IADLs, and chronic conditions), and Nursing Home Diversion enrollees are more likely to have a caregiver available and less likely to have dementia or need help with medications, three elements that are a cost savings to the program. The corollary could be that Frail Elder Project receives $1,074 less than what it needs to serve elders who have higher levels of need. Yet, Frail Elder Project does a better job than Nursing Home Diversion with fewer dollars. Frail Elder Project has fewer inpatient days compared with Nursing Home Diversion, although inpatient utilization in Frail Elder Project is higher than the other three programs. Enrollment in Frail Elder Project increases the likelihood of spending time in a nursing home, but this effect is less than the corresponding effect for Nursing Home Diversion. The state could reward Frail Elder Project for doing a better job with fewer resources.

Partly in response to the findings reported here, which a separate actuarial analysis conducted for the Department of Elder Affairs subsequently confirmed, the Department reduced the monthly capitation rate for Nursing Home Diversion in January 2004 from $2,342 to a variable rate based on region of the state. Rates now range from $1,699 in the Florida panhandle to $2,231 for Miami Dade and Monroe counties in southern Florida. The average rate is $1,881. It is expected that there will be another round of cuts in July 2006. The state's willingness to negotiate capitated rates for HCBS based on geographic differences in costs and to consider inflation corrections is a big step in the right direction. Nursing homes have enjoyed these considerations for many decades.

The purpose of Medicaid HCBS programs is to improve the quality of life for elders who prefer to receive long-term care at home rather than in a nursing home. If their care is coordinated better, the hope is that they will experience fewer days in the hospital. As was noted at the beginning of this section, Frail Elder Project and Nursing Home Diversion perform the worst on both counts, controlling for all client characteristics (Tables 4 and 5). But when just frailty is controlled, the average number of nursing home days is highest for Aged and Disabled Adult and Assisted Living for the Elderly (Table 7), although their number of inpatient days are average. The explanation for this may be different for each program.

Although only a small percentage of Aged and Disabled Adult enrollees meet Nursing Home Diversion frailty criteria, those that do are more likely to have a dementia diagnosis, less likely to have a caregiver available, and equally as likely to have a diagnosis of diabetes as enrollees in Nursing Home Diversion. The greater health care needs and lower caregiver resources may put Aged and Disabled Adult enrollees at high risk for nursing home placement even though they need help with fewer ADLs than do Nursing Home Diversion enrollees (2.8 vs 3.9, respectively). As a result, the state is not meeting its desire to defer nursing home placement for individuals in this program, although it is a relatively cost-neutral program. The Aged and Disabled Adult program might benefit from additional funding that would put it in line with funding levels for Frail Elder Project. These resources should be targeted to closer case management, given that the most impaired of these enrollees frequently lack a caregiver in the home. In fact, the Aged and Disabled Adult program might benefit from disease-specific case management given the relatively high level of diabetes among its enrollees (Fisher & Raphael, 2003; Frazier, 2003). In addition, some of these clients, especially those with dementia and without a caregiver, may be better served in Assisted Living for the Elderly.

More than half (54%) of Assisted Living for the Elderly enrollees meet Nursing Home Diversion frailty criteria. Relative to the other programs, they are also the most likely to have dementia and least likely to have a caregiver. This is usually a recipe for the 24-hour assistance that can be provided in assisted living facilities. In terms of quality of life, assisted living facilities are often preferred to nursing homes by frail seniors because they provide more autonomy and privacy. Yet the more impaired Assisted Living for the Elderly enrollees spent 11 more days in a nursing home during the course of the study than did Nursing Home Diversion enrollees. This anomaly may be explained by cost shifting. The Frail Elder Project and Nursing Home Diversion capitated rates include Medicare copayments for hospital care and for the first 90–100 days of nursing home care because they serve dually eligible individuals. Medicare covers the rest of these costs. Therefore, inpatient or short-term nursing home stays are potentially a way for managed care programs to substitute for the higher cost in-home long-term care for which they are responsible for paying. In fact, although these enrollees are more likely to spend time in a nursing home compared to fee-for-service programs, they will spend, on average, few than 90 days there (Table 7). When enrollees in the Aged and Disabled Adult and Assisted Living for the Elderly programs move into nursing homes, it may be because it is a necessary permanent stay.

Holding program enrollment constant, the next biggest contributor to cost is region of the state (and, by extension, region of the country). Labor, energy, real estate (especially for assisted living facilities), and other costs are relatively higher in southern Florida and its neighboring counties than in other regions of the state, especially northern Florida. Policymakers should consider the same geographic and inflation adjustments that are currently in place for nursing homes and Nursing Home Diversion for other Medicaid HCBS programs that provide services in the same types of markets.

Practice Implications
These analyses also suggest ways for long-term-care providers to plan for the costs of their enrollees. Among new enrollees in Medicaid HCBS who are on average 81 years old, increased age and being Caucasian slightly reduce overall Medicaid claims, and gender has no effect. Of more importance is the availability of a caregiver and living in the community at the time of assessment. Although one would expect these two factors to reduce total Medicaid claims, they actually contribute $42 and $79 PMPM. Dementia contributes $30 each month, each inpatient day adds $68 per month, and each nursing home day adds $88 per month on average, controlling for program enrollment. This type of analysis and these titration techniques can generate a more informed understanding of the relationship between costs and benefits, which in turn can help states and providers better plan for, and make decisions about, how to use Medicaid's increasingly scarce resources (Weissert, Hirth, Chernew, Diwan, & Kim, 2003). States can use the methods described here in order to combine their State Unit on Aging assessment data with their Medicaid claims data for better accountability and planning. As HCBS programs are privatized, this accountability will be particularly critical in containing costs and demonstrating positive outcomes.

Limitations
As was reported in the Data Sources section, we excluded 18% of the potential study population due to lack of a recent assessment. These clients were slightly more likely to be female, less likely to be White, and more likely to be in Channeling or Frail Elder Project. They had lower total claims and spent less time in a hospital or nursing home. These findings may be less representative of the two oldest programs and the lowest cost clients.

We also excluded Medicaid nursing home enrollees from this analysis, although other researchers have included them as a comparison group. This study instead compared the five Medicaid HCBS programs that had been developed as alternatives to nursing homes in order to better understand the relative efficiency of different HCBS strategies (Grabowski, 2003).

Conclusion
Perhaps the most fundamental message conveyed by the findings of this study is that the traditional aging network providers who administer the Aged and Disabled Adult and Channeling programs in Florida have the demonstrated capacity to efficiently provide an array of community-based long-term-care services. This does not mean, however, that aging network providers can assume that the future of community-based long-term care will be a replication of the past. In many states, providers are likely to be challenged by for-profit managed care organizations, which are increasingly prepared to administer long-term care through integrated plans. In Florida, for example, by the end of 2005, for-profit managed care organizations were administering Nursing Home Diversion, the state's largest Medicaid waiver-funded program (in terms of relative funding levels).

In response to this emerging challenge, aging network providers should seriously consider the development of their own not-for-profit managed long-term-care programs designed to achieve greater efficiencies in the use of resources and to improve the fit between consumer preferences and services provided. The aging network has demonstrated its value to elders for more than 25 years, but the future of the aging network in many communities may depend on its capacity to compete with for-profit managed care organizations.

One potentially productive method of strengthening the aging network and maximizing its cost-effectiveness would be to create aging-network–based managed long-term-care systems that could include risk for nursing home care within the capitation rate, as is currently done in the Wisconsin Family Care Program (Alecxih et al., 2002). This initiative would begin with Area Agencies on Aging and aging network providers who are best prepared in terms of fiscal and administrative capacities but could be gradually expanded based on lessons learned from initial evaluations of these capitated rate programs (including the current study).

At a minimum, policy makers, taxpayers, and long-term-care consumers are likely to benefit from a variety of strategies to provide long-term care more efficiently over the next several years. These strategies include managed long-term-care systems, the current fee-for-service approach where managed care strategies are not feasible (e.g., in many rural areas), and community-residential options such as assisted living facilities.

Future analyses could include Medicaid nursing home claims and outcomes, managed care provider costs, and Medicare claims, as enrollees are dually eligible and cost shifting between Medicaid and Medicare may be occurring in the inpatient and short-term nursing home stays, as reported here. Future models should include consumer satisfaction and more sensitive measures of changes in health and functional status (Kane, 1999) in order to provide a fuller understanding of the benefits and costs of providing Medicaid-funded HCBS.

	Footnotes

 
¹ State Data Center on Aging, School of Aging Studies, Florida Policy Exchange Center on Aging, University of South Florida, Tampa.

² School of Aging Studies, Florida Policy Exchange Center on Aging, University of South Florida, Tampa.

³ Florida Department of Elder Affairs, Tallahassee.

Decision Editor: Linda S. Noelker, PhD

Received for publication June 7, 2005. Accepted for publication March 24, 2006.

	References

TOP Abstract Methods Results Discussion References

 

• Alecxih, L. M. B., Neill, C., Zeruld, S., & Olearczyk, B., (2002). Wisconsin family care implementation process evaluation: Final report III. Falls Church, VA: Lewin Group.
• Alverez, R. M., & Glasgow, G., (1999). Two-stage estimation of non-recursive choice models. Political Analysis, 8 (2), 147-165.
• AoA, CMS seek to expand network of one-stop resource centers. (2005, May 6) Older Americans Report, 29, 137–138.
• Baldwin, J. A., Acheson, E. D., & Graham, W. J., (1987). Textbook of medical record linkage. New York: Oxford University Press.
• Borjas, G. J., & Sueyoshi, G. T., (1993, November). A two-stage estimator for probit models with structural group effects (Technical Working Paper No. 146). Cambridge, MA: National Bureau of Economic Research.
• Dunlop, B. D., Oakley, M. L., & Seff, L. R., (2004, January). Evaluation of Florida Medicaid's Frail Elder Option Program: Prepared for the Agency for Healthcare Administration (Policy Brief). North Miami: Center on Aging of Florida International University.
• Fisher, H. M., & Raphael, T. G., (2003). Managed long-term care: Care integration through care coordination. Journal of Aging and Health, 15 223-245.[Abstract/Free Full Text]
• Frazier, A. P., (2003). Health management strategies: Coordinating care for high risk populations. Continuing Care, 22(4), p. 23.
• Grabowski, D., (2004, Fall). The cost-effectiveness of home- and community-based long-term care services: Review and synthesis of the most recent evidence (Policy Brief). New York: Center for Home Care Policy and Research, Visiting Nurse Service of New York.
• Greene, V., Lovely, M., Miller, M., & Ondrich, J., (1995). Reducing nursing home use through community long-term care: An optimization analysis. Journal of Gerontology: Social Sciences, 50B S259-S268.[Abstract]
• Johnston, J., (1984). Econometric methods (3rd ed.). New York: McGraw-Hill.
• Judge, G., Griffiths, W. E., Carter Hill, R., Lutkepohl, H., & Lee, T., (1985). The theory and practice of econometrics (2nd ed.). New York: Wiley.
• Kane, R. L., (1999). Examining the efficiency of home care. Journal of Aging and Health, 11 322-340.[Abstract/Free Full Text]
• Kemper, P., (1988). The evaluation of the national long term care demonstration: 10. Overview of the findings. Health Services Research, 23 (1), 161-173.[Medline]
• Kmenta, J., (1986). Elements of econometrics (2nd ed.). New York: Macmillan.
• McCloskey, D. N., (1985). The rhetoric of economics. Madison: University of Wisconsin Press.
• Miller, N. A., (2003, July). State strategies to support community based long-term care. Paper presented at Leveraging Resources for Home and Community Based Services (HCBS) in a Time of Fiscal Retrenchment, New York, NY.
• Mollica, R., (1999, April 26). Extended Remarks by Robert L. Mollica, EdD, Deputy Director, National Academy for State Health Policy (U.S. Senate, Special Committee on Aging). Portland, ME: The Academy.
• Mollica, R., (2003). Coordinating services across the continuum of health, housing, and supportive services. Journal of Aging and Health, 15 165-188.[Abstract/Free Full Text]
• Newcombe, H. B., (1988). Handbook of record linkage: Methods for health and statistical studies, administration, and business. New York: Oxford University Press.
• Pindyck, R., & Rubinfeld, D., (1991). Econometric models and econometric forecasts (3rd ed.). New York: McGraw-Hill.
• Salmon, J. R., Mitchell, G., Chen, H., Albert, L., & Robb, C., (2001). Preliminary Evaluation of Medicaid Waiver Managed Long-Term Care Diversion Programs: Final Report. Tampa, FL: Florida Policy Exchange Center on Aging.
• Schmidt, F. L., (1996). Statistical significance testing and cumulative knowledge in psychology: Implications for training of researchers. Psychological Methods, 1 115-129.
• Shafer, W. D., (1993). Interpreting significance and nonsignificance. Journal of Experimental Education, 61 383-387.
• United HealthCare Services, Inc. (2004). Welcome to Evercare: Where our name is commitment. Retrieved March 8, 2005, from http://www.evercareonline.com.
• Weissert, W., Lesnick, T., Musliner, M., & Foley, K., (1997). Cost savings from home and community-based services: Arizona's capitated Medicaid long-term care program. Journal of Health Politics, Policy and Law, 22 1329-1357.[Medline]
• Weissert, W. G., & Hedrick, S. C., (1994). Lessons learned from research on effects of community-based long-term care. Journal of the American Geriatrics Society, 42 348-353.[Medline]
• Weissert, W. G., Hirth, R. A., Chernew, M. E., Diwan, S., & Kim, J., (2003). Case management: Effects of improved risk and value information. The Gerontologist, 43 797-805.[Abstract/Free Full Text]
• Wiener, J., Brown, D., Gage, B., Khatutsky, G., Moore, A., & Osber, D., (2004). Home and community-based services: A synthesis of the literature (Report No. RTI Project 07147.019.003). Waltham, MA: RTI International.
• Ziliak, S., & McCloskey, D., (2004). Size matters: The standard error of regressions in the American Economic Review. Journal of Socioeconomics, 33 527-546.

This Article Abstract Full Text (PDF) Services Download to citation manager PubMed PubMed Citation