Several attempts have been made to integrate acute and

Transcription

1 MODELS OF GERIATRIC CARE, QUALITY IMPROVEMENT, AND PROGRAM DISSEMINATION Providing All-Inclusive Care for Frail Elderly Veterans: Evaluation of Three Models of Care Frances M. Weaver, PhD, ab Elaine C. Hickey, RN, MS, cd Susan L. Hughes, DSW, ef Vicky Parker, DBA, dg Dawn Fortunato, MGS, h Julia Rose, PhD, ij Steven Cohen, MD, h Laurence Robbins, MD, k Willie Orr, MD, l Beverly Priefer, PhD, APRN-BC, m1 Darryl Wieland, PhD, n andjudithbaskins,rn,bsn op Frail elderly veterans aged 55 and older who met state nursing home admission criteria were enrolled in one of three models of all-inclusive long-term care (AIC) at three Veterans Affairs (VA) medical centers (n 5 386). The models included: VA as sole care provider, VA-community partnership with a Program of All-inclusive Care for the Elderly (PACE), and VA as care manager with care provided by PACE. Healthcare use was monitored for 6 months before and 6 to 36 months after enrollment using VA, DataPACE, and Medicare files. Hospital and outpatient care did not differ before and after AIC enrollment. Only 53% of VA sole-provider patients used adult day health care (ADHC), whereas all other patients used ADHC. Nursing home days increased, but permanent institutionalization was low. Thirty percent of participants died; of those still enrolled in AIC, 92% remained in the community. VA successfully implemented three variations of AIC and was able to keep frail elderly veterans in the community. Further research on providing variations of AIC in general is warranted. J Am Geriatr Soc 56: , Key words: long-term care; veterans; PACE From the a Center for Management of Complex Chronic Care, Hines Veterans Affairs Hospital, Hines, Illinois; b Northwestern University, Chicago, Illinois; c Center for Health Quality, Outcomes and Economic Research, Bedford Veterans Affairs Hospital, Bedford, Massachusetts; d Boston University, Boston, Massachusetts; e Center for Research on Health and Aging, Chicago, Illinois; f School of Public Health, University of Illinois, Chicago, Illinois; g Center for Organization, Leadership and Management Research, Veterans Affairs Boston Health Care System, Boston, Massachusetts; h Veterans Affairs Medical Center, Dayton, Ohio; i Division of Geriatrics and Bioethics, Department of Medicine; j Aging and Cancer Program, Case Comprehensive Cancer Center, Case Western Reserve University, Cleveland, Ohio; k Geriatrics and Extended Care, Department of Veterans Affairs Eastern Colorado Health Care System, Denver, Colorado; l Medical Services, Total Longterm Care, Denver, Colorado; m William S. Middleton Veterans Affairs Hospital, Madison, Wisconsin; n Division of Geriatrics; o School of Medicine, University of South Carolina, Columbia, South Carolina; and p Palmetto Health Geriatric Services, Columbia, South Carolina. 1 Formerly: AIC Project, Eastern Colorado Health Care System, Denver, Colorado. Address correspondence to Frances M. Weaver, PhD, Center for Management of Complex Chronic Care (151H), VA Hospital, Hines, IL frances.weaver@va.gov DOI: /j x Several attempts have been made to integrate acute and long-term care delivery and financing at the state level to develop and provide alternatives to nursing home placement for elderly and disabled people. Because of the high cost of nursing home care, alternatives to placement that keep elderly people in the home or community setting are desired. The belief is that these alternative care strategies can save money while accommodating the preferences of the great majority of adults to receive care at home. 1 The Social Health Maintenance Organization (S/HMO) and the Program of All Inclusive Care for the Elderly (PACE) are two examples of more-integrated models of healthcare delivery and financing that can serve as alternatives to institutional placement. 2 5 S/HMO is limited to Medicare enrollees and offers limited community and nursing home services. In contrast, PACE has been successful in integrating delivery of acute and long-term care and Medicare and Medicaid financing in a way that enables participants to remain in the community as long as possible. 4,6 The PACE model is available in many states and has become more accessible to older adults through a provision of the Balanced Budget Act (BBA) of BBA 1997 established PACE as a Medicare provider and allowed states to enact legislation that enabled privatesector providers to establish PACE programs. The core features of PACE are targeting clients who are nursing home eligible but who choose to receive long-term care services in the community, integrating funding and provider financial risk through capitated Medicare and Medicaid reimbursement, delivering integrated services through adult day health care (ADHC) centers, and using multidisciplinary teams for case management. 4 Potential enrollees must be aged 55 and older and meet state nursing home eligibility criteria to enroll in PACE. Enrollees who do not have Medicaid are expected to pay for the Medicaid portion (approximately two thirds of the total rate) out of pocket, 4 although this accounts for relatively few enrollees. Nationally, the average PACE enrollee is approximately 80 years old and female and has three activity of daily living (ADL) impairments and 7.9 medical conditions. 4 Although PACE is available to adults covered by Medicare and Medicaid, the Department of Veterans Affairs (VA) has not covered it as a reimbursable service. The future of long-term care for veterans is of particular concern to the VA. Approximately 2.4 million veterans who were eligible to JAGS 56: , 2008 r 2007, Copyright the Authors Journal compilation r 2008, The American Geriatrics Society /08/$15.00

2 346 WEAVER ET AL. FEBRUARY 2008 VOL. 56, NO. 2 JAGS Table 1. Description of Three All-Inclusive Care Models Model Components Model I Model II Model III Program characteristics Reimbursement model Services provided VA is sole provider of all healthcare needs. Veteran used existing VA healthcare benefits (copayments applied as appropriate for those less than 70% service connected or those who were non-service connected, dependent on their VA eligibility category). VA provided all care: hospital, specialty, nursing home, outpatient doctor visits, laboratory tests and procedures (e.g., x-ray), hospice, home health care, adult day health care, homemaker/chore, supplies and medications, and transportation. VA partnered with a community provider (local PACE provider) to share care responsibilities. Goal was to use Medicare and Medicaid patient benefits to pay for care; if patient had neither or was 70% or greater serviced connected, VA paid $3,700/month. This included a $470 discount from usual monthly fee for PACE based on state negotiated rate for PACE (for the care provided by VA). Depending on Medicare and Medicaid benefits of patients, monthly billing rates to VA varied. Patient eligibility was reassessed monthly. VA provided hospital care, specialty care, medications and supplies, laboratory tests and procedures. PACE provided adult day center including primary care and activities, home care, social work, homemaker/chore services, and transportation. VA contracted with a community provider (local PACE provider) to provide all health care to veteran. The total monthly cost of PACE was $2,823. VA paid the full amount for a patient who had no insurance or who was at least 70% service-connected disabled. If veteran had Medicare insurance, Medicare covered $1,312. If veteran also had Medicaid (or converted to Medicaid), VA no longer paid for any care. Once veteran enrolled in PACE, VA no longer provided any health care. PACE provided all care: hospital, specialty care, nursing home, outpatient doctor visits, laboratory tests and procedures (e.g., x-ray), hospice, home health care, adult day health care, homemaker/chore, supplies and medications, and transportation. VA 5 Department of Veterans Affairs; PACE=Program of All-Inclusive Care for the Elderly. receive health care in VA medical facilities in 1999 were aged 65 and older. This represents 39.2% of the population that the VA served. 7 In fiscal year 2001, the VA spent approximately $3.1 billion on long-term care services, but only 8% of these dollars were for noninstitutional long-term care, despite the preferences of older adults. 8 The aging veteran population, specifically those aged 85 and older, is expected to grow for at least the next 10 years. Consequently, the dollars spent on long-term care also are expected to grow, particularly those spent for institutional care. 9 To reduce anticipated growth in institutional care costs, the VA is examining strategies to reduce institutional care use and increase cost-sharing for medical care services for frail elderly veterans. The projected aging of the veteran population, along with growing concerns about financial ability to support the needs of this aging veteran population, prompted Congress to pass Public Law Section 102 of the law required the VA to implement and test three models of all-inclusive care (AIC), modeled after the PACE program, to determine whether any of the three models would reduce hospital and nursing home use by frail elderly veterans while maintaining them in the community. Three AIC models were established in 2000 as part of a national demonstration: Model I (VA as sole provider of all health and supportive care), Model II (a VA community partnership model that shared care responsibilities), and Model III (VA as care manager model with all care contracted to a community provider). This article describes the characteristics and healthcare use by frail elderly veterans who used each of the three different models of AIC provided through this VA demonstration. METHODS Design This observational program evaluation of three pilot AIC programs examined healthcare use for participants in the 6 months before enrollment and at 6-month intervals after enrollment for the duration of the demonstration (up to 36 months for early enrollees). Subjects were enrolled between June 2001 and June Study Sites and Participants Three VA medical centers served as study sites. Each site provided a different model of AIC: VA as the sole provider of all care, a VA community partnership model, and a VA as care manager only model, wherein community agencies provided all care. The VA as sole provider model (Model I) was developed at a VA facility in Ohio that had most components of AIC already in place, including on-site ADHC, a nursing home care unit, and skilled home care services. This VA added a homemaker/home health aide program and identified and arranged for transportation of participants for ADHC and healthcare needs through a variety of mechanisms. The VA community partnership program (Model II) was implemented at a VA facility in Colorado that already had a well-established PACE program in the community. VA partnered with the local PACE program to share AIC responsibilities. The VA provided hospitalization; short-term nursing home for subacute rehabilitation; subspecialty consultation; and laboratory, imaging, and pharmacy services, and PACE assumed responsibility for primary care, ADHC, transportation, home health care, homemaker/chore, and other supportive care needs. The VA as care manager program (Model III) was implemented at a VA facility in South Carolina that also had a well-established PACE program functioning in the community. This VA contracted for PACE to provide all care, with the understanding that veterans would not return to the VA for any healthcare services as long as they were active AIC participants. More details on the characteristics of each model are provided in Table 1. Participants were veterans aged 55 and older who meet the state criteria for nursing home admission for the state

3 JAGS FEBRUARY 2008 VOL. 56, NO. 2 ALL-INCLUSIVE CARE FOR ELDERLY VETERANS 347 Table 2. Participant Demographics Demographics VA as Sole Provider (n 5 181) VA Community Partnership (n 5 102) VA as Care Manager (n 5 85) Male, % Race, % White Black Hispanic Other Marital status, % Married Widowed Divorced or separated Single Enrolled in Medicaid, % Enrolled in Medicare, % Mean age at enrollment (range) 75.5 ( ) 77.1 ( ) 76.2 ( ) Service-connected disability, % Not service connected Service connected o70% Service connected 70% Caregiver spouse, % Percentage of participants who were already enrolled in Medicaid or enrolled after entering the all-inclusive care program. VA 5 Department of Veterans Affairs. (required by PACE) in which they resided. A comparison of state criteria across the three sites revealed that the admission criteria were similar. Data Sources Multiple data sources were used to obtain healthcare use for study participants. The VA maintains national-level administrative databases for inpatient, outpatient, pharmacy, and long-term care services that it provides. The National Patient Care Database is an SAS-formatted database that includes inpatient admissions; diagnoses; procedures; patient demographics; and outpatient clinic visits, diagnoses, and procedures. These files also contain information on VAprovided home health care, ADHC, and nursing home care. Data on medication use are available through the VA s Decision Support System. To assess Medicare use, Medicare claims data were obtained on participants who enrolled before January 1, Medicare data were not available at the time of analysis for subsequent years. The hospital inpatient, nursing home, skilled home care, hospice, outpatient visit, and provider claim files were reviewed to identify all Medicare reimbursed health care used by study participants. In addition, the national PACE organization has created and maintains a database of PACE participants, referred to as DataPACE. It includes enrollment information, patient characteristics, and healthcare use including ADHC visits, home care, transportation, and other supportive services. DataPACE was used for veterans who participated in the demonstration in the VA-community partnership and VA as care manager programs (Models II and III) to supplement what was not available through VA or Medicare files. Study participants and their informal caregivers were interviewed at study enrollment. Cognitive and functional status was assessed using the Short Portable Mental Status Questionnaire 10 and the Katz ADL 11 scale. The initial assessment provided information regarding self-care deficiencies and the need for specialized services such as dialysis or oxygen therapy and a description of psychiatric, cognitive, or behavioral problems. Patient and caregiver satisfaction with AIC was also collected and will be presented in a separate article. Data Analyses Analyses were primarily descriptive, because the goal of the evaluation was to assess the effect of each program on patient outcomes rather than a comparison of programs. Frequencies, means, and standard deviations are presented, along with some comparisons pre- and postenrollment using t-tests and chi-square tests. The institutional review boards at each of the participating VA facilities and at the two facilities in which the program evaluators were based reviewed and approved the study. RESULTS A total of 368 veterans were enrolled across the AIC model programs between June 1, 2001, and June 30, (The Model I program recruited veterans and randomized them to receive AIC or customary care. This article reports only on veterans enrolled in AIC.) The demographic characteristics of enrollees according to site are provided in Table 2. Almost all participants were male (96%). Although the majority of AIC participants in Models I and II were white

4 348 WEAVER ET AL. FEBRUARY 2008 VOL. 56, NO. 2 JAGS Table 3. Baseline Functional and Cognitive Status Model I Model II Model III Status Variable % Cognitive status (errors on Short Portable Mental Status Questionnaire) Normal (0 2) Moderate impairment (3 6) Severe impairment (7 10) Katz index of ADLs Independent in all tasks Dependent in 1 2 ADLs (one function; bathing plus one other function) Dependent in 3 4 ADLs (bathing, dressing, plus one other function; bathing, dressing, going to the toilet, plus one other function) Dependent in 5 6 ADLs (bathing, dressing, going to the toilet, transferring, continence, and feeding) Dependent in at least two functions but not classified in above categories (e.g., continence and transfer) Received home care services at the time of AIC enrollment Skilled care needs at enrollment Medications Wound care Therapies Urinary catheter Nursing visits 41 time per week Behaviors and symptoms Wandering Verbal abusive Physical abusive Resisting care Hallucinations Note: A small number of participants had missing data on one or more of these assessments. ADLs 5 activities of daily living. (75% and 62%, respectively), the majority of participants in Model III were black (54%). More than half were married (less than half in Model II), and most were enrolled in Medicare. The percentage of veterans receiving Medicaid was low at the time of enrollment, although in the partnership model, 38% of AIC participants had enrolled in Medicaid by the end of the evaluation period. Veterans who were classified as service-connected disabled (veterans who have a medical or psychological condition or disability resulting from their military service can receive health care for that disability free from VA; no copayments are required) at any level varied from 20% in Model III to 26% in Model I and 36% in Model II. Slightly more than half had a spouse as their primary caregiver in Models I and III, whereas only 42% of Model II enrollees had a spouse caregiver. The cognitive and functional status of veterans at enrollment also varied according to model (Table 3). Model III enrolled the greatest number of severely cognitively impaired (19% had a score of 7 or higher). This model also enrolled more veterans who were functionally impaired (26% were dependent in five to six ADLs, compared with 10% in the other models). Approximately 15% of the Model II subjects had seven or more errors on the Short Portable Mental Status Questionnaire, and 41% were dependent in one to two ADLs. Model I participants were moderately dependent on ADLs, with 39% having one to two impairments and 27% having three to four impairments, and the majority had normal cognitive functioning (73%). The evaluators examined disease severity indicators to more fully describe and compare enrollees in each of the programs. Factors such as the need for skilled nursing care and the presence of inappropriate or disruptive behaviors are also indicators of risk for nursing home placement and of the amount and type of support services required. AIC enrollees from all three sites required a wide range of skilled care services at baseline. The VA sole provider group had the greatest need for medication management (28%), the partnership model subjects required frequent nursing care (22%), and the care management model had the highest percentage of participants with behavioral impairment. Model III had the greatest percentage of enrollees who were verbally abusive (21%), and in Models II and III, approximately one-quarter of enrollees exhibited wandering behavior at baseline. Healthcare use by enrollees was examined first by comparing healthcare use in the 6 months before enrollment to use that occurred during the first 6 months after AIC enrollment (Table 4). There were no statistically significant changes in inpatient hospital or outpatient clinic

5 JAGS FEBRUARY 2008 VOL. 56, NO. 2 ALL-INCLUSIVE CARE FOR ELDERLY VETERANS 349 Table 4. Healthcare Services Use 6 Months Before and 6 Months After All-Inclusive Care Enrollment (n 5 191) Model I: VA as Sole Provider of Care (n 5 86) Model II: VA Community Partnership (n 5 51) Model III: VA as Care Manager (n 5 54) Types of Use Pretest Posttest Pretest Posttest Pretest Posttest Inpatient use, % Hospital admissions/patient Total inpatient days/patient Nursing home use, % Nursing home admissions/patient Nursing home days/patient Outpatient use, % Home care use, % Adult day healthcare use, % Po.02,.001. VA 5 Department of Veterans Affairs. use in the 6 months before and after AIC enrollment at any of the demonstration sites. Hospital days increased slightly for Model I but stayed the same for Models II and III, despite the fact that hospital admissions increased in Model III. In contrast, nursing home admissions and days increased in all three models. The percentage of patients who used nursing home care more than doubled in the Model III site after AIC enrollment, although nursing home days per patient did not change, suggesting that much of the nursing home use was for short-term rehabilitation or respite needs. Although almost half of all patients used home care services in Model I before and after AIC enrollment, the number of patients who used home care services increased significantly in both of the other models. Finally, after AIC enrollment, more than half of the Model I subjects used ADHC, whereas 100% of patients in Models II and III used ADHC at their respective PACE partner sites. Healthcare use for all AIC enrollees was also examined in the 12 months after AIC enrollment at each of the model sites (Table 5). Included in Table 5 is any use of VA-provided care, care paid for by the VA provided by community providers (i.e., fee-basis care), and any care provided through the PACE program (for Models II and III). Medicare data were not available for the entire study period, so Medicarereimbursed utilization was not included in Table 5. For approximately one-quarter of the total sample, at least 12 months of Medicare data were available. There was no use of Medicare services in Model III and limited use of inpatient and durable medical equipment (DME) in Model II, but a large number of Model I veterans used Medicare services for emergency department, hospital, and outpatient care, with limited use of Medicare home care and DME. Given this information, it is likely that healthcare use was undercounted in Model I in general across the study period because Medicare data were not available for the entire study period. Utilization data are displayed in Table 5. Approximately half of the AIC participants in Model I used inpatient care during the 12 months after enrollment, averaging 8.55 days of care (median 5 0 days). Similarly, 54% of the Model I participants used the VA s ADHC program, averaging 14 visits over 12 months (or approximately 1 visit per month), whereas 26% had any nursing home use, averaging 11 days of care. Almost all participants in Model I used outpatient care, averaging 23.5 visits in 12 months, and 42% received home health care (mean visits per month for users). The pattern of care used was somewhat different for Models II and III. In both models, 100% of participants used ADHC provided by PACE. Somewhat greater frequency of ADHC visits occurred in the Model III program, with an average of 13.8 visits per month, versus 10 visits per month in Model II. The vast majority of participants in both models also received home health care, with a median of 6 visits per participant per year. Approximately 35% to 40% of participants used inpatient or nursing home care after enrollment. Outpatient clinic use was similar for the Model II and III participants. None of these numbers have been adjusted to take into account survival, so although an individual may have received an average total of home care visits in a year, these visits may have occurred close together in a short span of time. At the end of the evaluation, approximately 30% of veterans enrolled in AIC had died (28%, 28%, and 34% in Models I, II, and III, respectively). Slightly fewer than half of all deaths occurred in a nursing home (Table 5). The number of participants who disenrolled varied from 12% in Model I to 31% and 34% Models II and III, respectively. The most common reason for disenrolling in Model I was refusal of AIC care, whereas the most common reason for Models II and III was that the participant moved to a nursing home. In Model I, four patients disenrolled and went to a nursing home, two moved outside the catchment area, and two preferred care from their own physician. Ten patients in Model II disenrolled to go to a nursing home, and six moved outside the catchment area. In Model III, nine patients left PACE to move to a nursing home, and six left, because they were dissatisfied with PACE. Near the end of the AIC demonstration, the VA indicated that it might not be able to provide continued financial support of AIC for veterans. Veterans who were not dually enrolled in Medicare and Medicaid were more likely to leave the program, often with

6 350 WEAVER ET AL. FEBRUARY 2008 VOL. 56, NO. 2 JAGS Table 5. Healthcare Use by All-Inclusive Care Enrollees at Each Demonstration Site 12 Months After Enrollment and Status at End of Evaluation Period Healthcare Use and Study Status Model I (n 5 158) Model II (n 5 102) Model III (n 5 85) Healthcare use Patients with inpatient admissions, % Inpatient admissions/patient, mean SD Total inpatient days/patient, mean SD Patients with nursing home admissions, % Total nursing home admissions, mean SD Nursing home days/patient, mean SD Patients with outpatient clinic visits, % Outpatient clinic visit/patient, mean SD Patients with home care visits, % Number of home care visits/patient, mean SD Patients with adult day healthcare use, % Adult day healthcare visits/patient, days, mean SD Status at end of the evaluation period, n (%) Deaths 52 (28) 28 (28) 29 (34) Died in nursing home 24 (46) 14 (50) 16 (55) Died elsewhere 28 (54) 14 (50) 13 (45) Disenrolled 16 (9) 23 (22) 19 (22) Survivors 113 (62) 51 (50) 37 (66) Residing in nursing home 7 (6) 8 (15) 1 (3) Residing in community setting 106 (94) 43 (84) 36 (97) SD 5 standard deviation. the assistance of VA staff, because they could not afford to pay out of pocket for PACE care. Of those who survived and remained in AIC to the end of the demonstration, 92% were residing in the community. DISCUSSION The VA developed and evaluated three models of providing AIC to frail elderly veterans who met the criteria for nursing home placement. Baseline data indicate that there is variability in the characteristics of veterans who are appropriate for nursing home care. The variability in race reflects the geographic locations of the three sites. Differences in cognitive and physical functioning across sites also illustrates heterogeneity that exists across individuals who are considered eligible for nursing home placement. Model III had the most participants with serious cognitive impairments, whereas Model II had the most participants with two or fewer ADL impairments. The patterns of healthcare services used reflects different approaches taken by the demonstration sites to provide AIC. Model I appears to have used a home-based approach to care in which approximately half of the subjects received home care services, ADHC, or both. Further analysis according to site suggests that participants with cognitive impairments were more likely to use AHDC, whereas those with physical impairments used home care. All Model I participants used VA outpatient clinics. They also retained their primary care providers in most cases. In constrast, Models II and III were more similar to traditional PACE. All participants in Models II and III used PACE ADHC. Although Model I offered AIC using a model of care that did not rely on ADHC as the primary care management mechanism, it provided care using one funding stream (the VA) to cover any medical care and a single care-management program using elder care coordinators. Other similar programs have been described in the literature. The Wisconsin Partnership Program (WPP) provides a similar variant on PACE in which integrated funding from Medicare and Medicaid is channeled into a single program and a single care management model, 12 although similar to Model I, individuals enrolled in WPP maintain their primary care providers and do not rely on ADHC for care management. Similar to the PACE and VA AIC models, WPP s goals include reducing use of institutional acute and long-term care, although while WPP did not affect hospitalizations, an earlier study of PACE found a substantial reduction in acute hospital use. 13 The extent to which these programs compare with traditional PACE with respect to efficiency of managing care for frail elderly people requires further study. One of the primary goals of the demonstration was to delay or prevent permanent institutionalization. Over the 3-year period, there was a high death rate (approximately 30%), although only half of the deaths occurred in nursing homes. Excluding deaths and disenrolled subjects, most participants resided in the community until death or study end. Given that all participants were nursing-home eligible, these data suggest that each AIC model tested was able to maintain the majority of patients in the community. A recent article reported that disenrollment from PACE programs is low (approximately 8%) and is more likely for

7 JAGS FEBRUARY 2008 VOL. 56, NO. 2 ALL-INCLUSIVE CARE FOR ELDERLY VETERANS 351 private-pay patients, those with hospital admissions, and those with increasing nursing home lengths of stay up to approximately 80 to 90 nursing home days, when PACE disenrollment begins to decline. 14 The uncertainty as to whether support of AIC would continue beyond the end of the demonstration prompted VA staff to work with veterans who could not afford to remain in PACE in Models II and III to return to the VA for their care. Model I disenrollment rates were more similar to traditional PACE rates. The veterans who enrolled in AIC differed from the traditional PACE participant. The veteran population is almost all male, and the majority are married and living with a spouse, whereas the PACE population is predominately female, is not married, and approximately one-third live alone. 15 The PACE population also has a greater number of enrollees who are severely cognitively impaired. Despite the averaged findings across sites provided here, there was significant variation across the VA AIC and PACE programs in patient characteristics. 3 If frailty is defined as physical and cognitive impairment, Model III participants appeared to be more similar to the traditional PACE population. For the VA to contract with PACE in other areas, PACE would need to consider whether any changes in their ADHC activities might be needed to meet the needs of a predominately male population. The finding that one-third to one-half of AIC enrollees had hospital use was somewhat unexpected, because PACE tries to keep its patients out of the hospital. Almost all AIC enrollees in this study were male. Male sex was found to be a risk factor for hospitalization in a national study of acute hospital use by PACE enrollees. 13 Because the majority of PACE enrollees are traditionally female, this sex effect is not seen unless use is examined according to sex. Over the last 20 or more years, VA has conducted several evaluations of geriatric community-based interventions. A common goal of these interventions was to reduce use and cost of acute care and institutional care services through use of community-based programs such as ADHC, home health care, homemaker/chore, and assisted living. Veterans targeted for these programs were typically disabled, homebound, elderly, frail, nursing home eligible, terminally ill, or some combination of these. For the most part, these interventions did not reduce costs but in many instances improved satisfaction with care. A multisite randomized trial comparing the VA s home-based primary care (HBPC) with customary care (including Medicare home health care) demonstrated greater satisfaction with care and lower caregiver burden in HBPC accompanied by 12% higher total per person costs at 12 months. 16 An earlier study of ADHC, considered to be an alternative to nursing home care at the time, increased the cost of VA care 15% over customary care, with no differences in patient or caregiver health outcomes. 17,18 The most recent study, a demonstration of the provision of assisted living to veterans, also suggests that costs were higher in the treatment group than the customary care group. 19 Results of community care interventions in the private section have fared similarly. An early review of home and community care programs as alternatives to institutionalization showed that increases in overall healthcare use and cost usually accompanied these services. 20 An early evaluation of PACE questioned its slow enrollment and suggested that there may be niche-marketing of participants, limiting its generalizability. 21 Despite this, PACE was able to demonstrate sufficient success to be designated as a permanent Medicare program in Growth in the PACE program remains slow because of financing constraints (95% of the PACE population are dually eligible), the requirement to give up one s primary provider at entry, and the reliance on use of ADHC. A national PACE study found lower short-term acute care use in PACE than for fee-for-service Medicare beneficiaries, 13 although this was a retrospective database study. Three HMOs undertook an innovative program to provide primary care in long-stay nursing home patients. This program was compared with care provided to traditional fee-for-service residents at the same nursing homes. The HMO that provided the most primary care visits per month (2 visits) experienced fewer emergency department visits and fewer hospitalizations per resident than fee-forservice residents (1.1 visits), 23 although this population was already institutionalized. Recent interventions such as the Medicaid consumerdriven Cash and Counseling model for personal care services and home- and community-based services also have found costs to be higher in the intervention group. The authors commented that costs were higher, because those in the traditional Medicaid system did not receive the services to which they were entitled. 24 These data suggest that participants in new programs tend to have unmet care needs. The meeting of these needs may result in more utilization of services in the short term, which longer stays in the community in the long term may offset. Further long-term evaluation would be needed to test this assumption. It is difficult to comment on the costs of providing AIC for frail elderly people because of the limited cost data that were available and the differences in the patient samples across the three models. Furthermore, this was a demonstration of the feasibility of implementing AIC in VA. It is also important to mention that, because the models were designed to have different characteristics, not all VA facilities could implement each model. For example, if a VA facility sought to provide all care (Model I), that facility would have to already provide ADHC and home care on site, because there likely would not be new funds available to create these services. Model II would be most likely to succeed if the VA and a community provider, such as PACE, already had a relationship and could negotiate a shared arrangement of services. Model III could work if a community provider would agree to assume the entire responsibility for the veteran s care, although many community providers such as PACE have limited catchment areas, which would restrict which veterans could receive AIC, raising questions about equal access to service for veterans within the VA hospital s catchment area, which is usually much larger than PACE. It was learned that the VA can establish and operate AIC programs. The relationships that were established in Models II and III suggest that there are advantages to developing partnerships with and between agencies to meet the diverse needs of a frail elderly population. Model I attempted to provide the full range of services within the VA, without partners. Although they met their goal to provide AIC, it required creation of a homemaker program on site

8 352 WEAVER ET AL. FEBRUARY 2008 VOL. 56, NO. 2 JAGS and the piecing together of a variety of strategies to provide transportation. Without continued operation of these models at these locations over a longer period of time, it was not possible to clearly understand the functions, operating characteristics, patient characteristics, and outcomes that would be needed to determine the success of these programs in providing a cost-effective alternative to institutionalization. There are a number of limitations to this evaluation. There were incomplete data on Medicare use. As a result, it is likely that healthcare use was undercounted, particularly for Model I. It is likely that the sample sizes available for comparisons between the pre- and postenrollment periods were underpowered and might mask any true differences. Regardless of what kind of program a veteran receives, in an emergency situation, an ambulance transports the veteran to the nearest non-va medical facility. This explains why a veteran enrolled in the Model I program could be expected to have experienced Medicare-reimbursed emergency department and hospital use. If Model I patients were not transferred to the VA and instead were discharged from the non-va setting, this is likely to have precipitated the ordering of Medicare home care and DME, which would also have been undercounted. Furthermore, the AIC program was designed as a demonstration, and it was not possible to evaluate the demonstration using a controlled clinical trial or a comparison of the three AIC programs. This limits the ability to compare or generalize findings to what might be expected if one or more of these AIC models were implemented at other VA facilities. Nonetheless, this evaluation confirmed that the VA can provide AIC using any of the three models of care and that the majority of nursing homefeligible frail elderly veterans who received AIC remained in the community. Given the waiting times for older veterans to be seen in VA clinics and the imbalance in VA spending for institutional versus home care, it is not surprising that this study found an overall increase in utilization of home care and ADHC. When older adults who have unmet needs for care are provided care, they appear to use it. Whether the increased use of home- and community-based services results in longer maintenance in the community is an important one, but one that could not be addressed in this study without customary care control groups. The most obvious limitation to this evaluation was the inability to assess the effect of AIC provision on patient outcomes. The low disenrollment rate due to dissatisfaction with services provided suggests that patients and their families were satisfied with the care provided in each AIC model. Perhaps as important as patient outcomes is the strong likelihood that nursing home capacity will not grow to meet future demand, making it necessary to find alternative means of caring for frail elderly persons. In fiscal year 2003, VA spent $2.3 billion on nursing home care for veterans, but only one-third of this was for long stays (90 days). 25 Unless veterans requiring long-term care have a VA service-connected disability benefit of 70% or greater, it is likely that they will apply to Medicaid to pay for nursing home care. Figures for 2005 indicate that the average expenditure for a nursing home resident was $52,000 a year. 26 The cost for provision of AIC, including hospital and nursing home care, to a veteran in Model III for 1 year was substantially less ( $33,900). If it is assumed that outcomes are at least comparable for AIC and nursing home care, this finding suggests that it will not be more expensive to invest in AIC care and could, in fact, produce cost savings. Ongoing investigation into the economics of providing alternative models of long-term care will be critical in view of the rapid growth in the population aged 80 and older, many of whom will need long-term care. Although legislation exists to continue the VA AIC program, no additional special program funds were made available to continue support of the Model II and III programs beyond the demonstration. Nevertheless, the participating VA sites, the community PACE partners, and the patients and caregivers who enrolled and received AIC care considered the demonstration to be successful. 27 PACE has continued to serve only a small number of potentially eligible frail elderly people, and some have perceived it to be a niche program. 4 Its ability to partner successfully with VA to enroll and follow veterans was seen as a way to expand the PACE program and identify new groups of participants. The lessons learned from this demonstration can be used to consider how to construct other VA community partnerships for care provision in the future. Policy makers and researchers should continue to identify and test other variations of prepaid AIC for frail elderly people, including veterans, in the future. ACKNOWLEDGMENTS The authors would like to thank Elaine Czarnowski, Laurie Todd, Katherine Skinner, Kristin Koelling, Scott Miskevics, and Raquel Hampton and our data collectors, John Palmer, Margaret McKaben, Kim Lopez, Katherine Abbott, Tanetta Anderson, and Melissa Cappaert for their efforts on this project. Finally, we would like to remember Marcia Goodwin-Beck, who had the vision to carry this project forward. This views and opinions in the paper are those of the authors and do not necessarily reflect those of Veterans Health Affairs. Conflict of Interest: Funded by the Geriatrics and Extended Care Strategic Healthcare Group, Veterans Health Administration, Washington, DC. Willie Orr: Vice-President Medical Services, Total Longterm CareFPACE provider in Denver. The editor in chief has determined that the authors have no conflict of interest related to this manuscript. Author Contributions: Frances M. Weaver, Elaine C. Hickey, Susan L. Hughes, and Vicky Parker: study design, acquisition of data, analysis and interpretation of data, and preparation of manuscript. Dawn Fortunato and Beverly Priefer: acquisition of subjects, interpretation of data, and preparation of manuscript. Julia Rose and Laurence Robbins: study concept, acquisition of subjects, interpretation of data, and preparation of manuscript. Steven Cohen: acquisition of subjects and manuscript preparation. Willie Orr, Darryl Wieland, and Judith Baskins: study concept, acquisition of subjects, and preparation of manuscript. Sponsor s Role: None. REFERENCES 1. Wiener JM, Stevenson DG. State policy on long-term care for the elderly. Health Aff (Millwood) 1998;17:

9 JAGS FEBRUARY 2008 VOL. 56, NO. 2 ALL-INCLUSIVE CARE FOR ELDERLY VETERANS Bodenheimer T. Long-term care for frail elderly peoplefthe On Lok Model. N Engl J Med 1999;341: Branch LG, Coulam RF, Zimmerman YA. The PACE evaluation: Initial findings. Gerontologist 1985;35: Gross DL, Temkin-Greener H, Kunitz S et al. The growing pains of integrated health care for the elderly: Lessons from the expansion of PACE. Milbank Q 2004;82: Harrington C, Newcomer RJ. Social health maintenance organizations service use and costs, Health Care Financ Rev 1991;12: Eng C, Pedulla J, Eleazer GP et al. Program of All-inclusive Care for the Elderly (PACE): An innovative model of integrated geriatric care and financing. J Am Geriatr Soc 1997;45: VIReC. Research Findings form the VA Medicare Data Merge Initiative: Veterans Enrollment, Access and Use of Medicare and VA Health Services (XVA ). Report to the Under Secretary for Health, Department of Veterans Affairs, 2003 [on-line]. Available at SourcesName/VA-MedicareData/USHreport.pdf Accessed January 17, GAO. Availability of Medicaid home and community services for elderly individuals varies considerably. Long term care, GAO , Report to the Chairman, Special Committee on Aging, U.S. Senate, September Kinosian B, Stallard E, Wieland D. Projected use of long-term care services by enrolled Veterans. Gerontologist 2007;47: Pfeiffer E. A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. J Am Geriatr Soc 1975;23: Katz S, Ford AB, Moskowitz RW et al. Studies of illness in the aged. The index of ADL: A standardized measure of biological and psycho-social function. JAMA 1963;185: Kane RL, Homyak P, Bershadsky B et al. The effects of a variant of the program for all-inclusive care of the elderly on hospital utilization and outcomes. J Am Geriatr Soc 2006;54: Wieland D, Lamb VL, Sutton SR et al. Hospitalization in the Program for All- Inclusive Care for the Elderly (PACE): Rates, concomitants, and predictors. J Am Geriatr Soc 2000;48: Temkin-Greener H, Bajorska A, Mukamel DB et al. Disenrollment from an acute/long-term managed care program (PACE). Med Care 2006;44: Wieland D, Lamb V, Wang H et al. Participants in the program of all-inclusive care for the elderly (PACE) demonstration: Developing disease-impairmentdisability profiles. Gerontologist 2000;40: Hughes SL, Weaver FM, Giobbie-Hurder A et al. The effectiveness of team managed home care: A randomized multicenter trial. JAMA 2000;284: Rothman ML, Hedrick SC, Bulcroft K et al. Effects of VA adult day health care on health outcomes and satisfaction with care. Med Care 1993;31:S38 S Rothman ML, Diehr P, Hedrick SC et al. Effects of contract adult day health care on health outcomes and satisfaction with care. Med Care 1993;31:S75 S Hedrick S, Guihan M, Chapko M et al. Evaluation of the Assisted Living Pilot Program, Final Report, for Geriatrics and Extended Care Strategic Health Care Group. Department of Veterans Affairs, Washington, DC: Weissert WG, Cready CM, Pawelak JE. Home and community care: Three decades of findings. Milbank Q 1988;66: Branch LG, Coulam RF, Zimmerman YA. The PACE evaluation: Initial findings. Gerontologist 1995;35: Gross DL, Temkin-Greener H, Kunitz S et al. The growing pains of integrated health care for the elderly: Lessons from the expansion of PACE. Milbank Q 2006;82: Reuben DB, Schnelle JF, Buchanan JL et al. Primary care of long-stay nursing home residents: Approaches of three health maintenance organizations. J Am Geriatr Soc 1999;47: Dale SB, Brown RS. How does cash and counseling affect costs? Health Serv Res 2007;42(1 Part 2): U.S. Government Accountability Office. VA Long-Term Care: Oversight of Nursing Home Program Impeded by Data Gaps. GAO-05-65, Washington, DC, November 10, Johnson KA. The looming problem of long-term care and Medicaid spending. The Heritage Foundation, October 8, 2005 [on-line]. Available at Accessed June 8, Weaver FM, Hickey E, Parker V et al Evaluation of three pilot programs of all inclusive long term care. Final Report, MRR , Office of Geriatrics and Extended Care, Department of Veterans Affairs, Washington, DC, April 2005.