The Program of All-Inclusive Care for the Elderly (PACE):

A Description and Evaluation of Existing Sites and the Connecticut PACE Pilot Program

September 1998

Abstract

This report describes and evaluates the Program of All-Inclusive Care for the Elderly (PACE), both as it currently exists at sites across the USA, and in the form proposed by the Connecticut PACE Pilot Project. Particular attention is paid to financial and health care/life satisfaction outcomes associated with PACE, the various ways in which the Connecticut PACE Pilot Program will differ from the traditional PACE model, the likely pros and cons of these proposed departures, and the objectives of and criteria for evaluating the Connecticut PACE Pilot Program.

Acknowledgments

I would like to thank, without implicating, the following for their help in the preparation of this report: Yvonne Able, Abt Associates; Lawrence Branch, Duke University; John Capitman, Heller Graduate School for Advanced Studies in Social Welfare, Brandeis University; Robert Friedland, National Academy on Aging; Barbara Fletcher, Connecticut Department of Social Services; Bill Irwin, Alliance For Health Reform; Harriet Komisar, Institute For Health Care Research and Policy, Georgetown University; Steve Miller, HCFA; Cathy Shannon, On Lok; Christine Van Reenen, National PACE Association; and Joshua Wiener, Health Policy Center, The Urban Institute.

I INTRODUCTION

1. The Desire to Integrate Health Care Delivery Systems

A major concern in contemporary health policy circles is with the notion that the provision of acute and long-term health care services is both fragmented and un-coordinated, and that there is a pressing need to integrate these services into a single, seamless system of health care provision. The population that stands to be affected by integration initiatives comprises frail-elderly and non-elderly disabled persons who not only require long-term care, but are also substantial consumers of acute care. This population includes, of course, the so-called Adual eligibles@ C those who qualify for both Medicare and Medicaid benefits. Indeed, it has been argued that Aa good deal of the impetus for integrated care models springs from...the >dual eligible= population@ (Ashbough and Smith, 1997, p. 5). In the context of an aging population, the number of persons requiring both acute and long-term care is only likely to grow over time, an observation that adds to the sense of urgency surrounding the issue of integrated acute and long-term care.

In what sense is the current system of providing acute and long-term care fragmented and un-coordinated? It is possible to identify two interrelated dimensions of fragmentation, affecting the financing of services and the flow of services to beneficiaries respectively. The financing of acute and long-term care are profoundly fragmented, and there exists no explicit mechanism for co-ordinating the different sources of financing for the population identified above. Medicare is primarily responsible for financing acute care, under the jurisdiction of the federal government (specifically, the Health Care Financing Administration or HCFA). Meanwhile, Medicaid is largely responsible for financing long-term care under the auspices of state governments (albeit with HCFA oversight). To complicate matters further, the Social Services Block Grant, Department of Veteran Affairs, Older Americans Act programs, private insurance and out-of-pocket expenditures also play various roles in financing acute and long-term care (Wiener and Skaggs, 1995, p.2). Because the financing of acute and long-term care is fragmented, then so, too, is the delivery of care, with different agencies responsible for making decisions with respect to a patient=s eligibility for various types of care and the availability of the required care. This fractious, discontinuous decision- making with respect to acute and long-term care contrasts with the continuous, alternating needs for acute and long-term care amongst a population who suffers both acute and chronic conditions and for whom acute and long-term conditions often interact as part of a single experience of being unwell. As Wiener and Skaggs (1995, p.1) suggest, Ain an ideal world, the elderly and non-elderly population with disabilities would receive the services they need in a seamless system that does not sharply distinguish between acute and long-term care...Hospitals, physicians and other acute care provides would coordinate the care plan with nursing homes, home care and other long-term care providers, and vice versa...Unfortunately, this ideal world does not exist.@

It is widely held that the (again, interrelated) consequences of the fragmentation and lack of co-ordination identified above are higher costs of provision and reduced quality of care. For example, a patient may unnecessarily remain hospitalized at the expense of Medicare simply because more appropriate and cheaper long-term care services funded by Medicaid are not available (Wiener and Skaggs, 1996, p.47). The separate financing and administration of acute and long-term care seems to mean that, at best, each system is indifferent about efforts and initiatives that would save money for the other (Wiener and Skaggs, 1995, p. 3). At worst, the Medicare and Medicaid systems engage in Acost shifting@ C that is, explicit attempts to pass the costs of care off onto each other, as, for example, when Medicare attempts to reduce its costs by shifting patients into Medicaid financed long-term care programs (Ashbough and Smith, 1997, p. 5). This eagerness to shift costs may raise the total costs of providing long-term care. Because, for example, it is easier and quicker to discharge patients from hospitals into nursing homes, Medicare frequently does this. However, because of the cost of nursing home care and the fact that it is more difficult to move patients out of nursing homes than to prevent their entering one initially, this raises the costs to Medicaid of providing long-term care (Allen, 1997, p. 2). In this way, A...each part of the system is motivated to guard its resources jealously, shifting patients and their costs to the other part of the system rather than managing those costs@ (Saucier and Riley, 1995, p. 7 cited in Allen; 1997, p. 2).

The quality of care also suffers. First, beneficiaries are confronted by Apaperwork requirements and administrative rules [that] are voluminous@ (Bullen, 1997, p. 2) as they attempt to gain access to the services to which they are entitled. Second, the care they receive may be compromised by the fragmentation and lack of co-ordination between acute and long-term care provision. The following example (National Senior Citizens Law Center, 1998, p. 4) is instructive. A dually eligible individual in a Medicaid-funded nursing home bed requires hospitalization, which will be paid for by Medicare. Following hospitalization, he/she may be required to have Medicare pay for post-hospital nursing home care. Unfortunately, this may prevent him/her from returning to his/her original (Medicaid-funded) bed, because Medicare certified beds may be kept in separate wings of nursing homes. Furthermore, Medicare nursing home coverage is limited, and by the time it has expired, the patient=s original Medicaid bed may no longer be available. He/she will be forced to move to a different nursing home in order to continue receiving Medicaid-funded nursing home care. It has been suggested that his sort of situation A... >whipsaws= low-income elders between the two systems [Medicare and Medicaid] in a manner that is unacceptable for the individual, family members and policy-makers@ (Bullen, 1997, p. 2). Of course, if needs actually go un-met as a result of this >whipsawing=, then the health of the patient may begin to decline.

Set against the back-drop of these problems is the apparent promise of integrated acute and long-term care. Advocates of integration argue that it will both improve the quality of care and reduce the costs of providing care. The former benefit would stem largely from the ability of an integrated system to tailor treatment programs to the health needs of individuals C regardless of whether these needs are, at any particular point in time, for acute or long-term care C without the interruption and administrative difficulties that characterize current provision. (Stone and Katz, 1996, pp. 223-4). Cost-savings, meanwhile, are envisaged as arising from the ability of an integrated care system to substitute less expensive preventative and home and community-based care for more expensive hospital and nursing home care, coupled with the financial incentives to perform such substitutions that would exist under a system of capitated payments to providers (Stone and Katz, 1996, p. 225; Wiener and Skaggs, 1995, p. 4).

Not all researchers are convinced of the benefits of integration, however. There are concerns that the acute care environment will dominate in the event of integration, resulting in the over-medicalization of long-term care (Stone and Katz, 1996, p. 227: Wiener and Skaggs, 1995, pp. 4). Moreover, Ashbough and Smith (1997, pp. 3-6) are skeptical of the managed care entities that would take responsibility for administering integrated care, arguing that they would substantially increase overhead costs, and that their administrative techniques, honed in the acute care environment, are largely un-proven in the long-term care arena. Most importantly, there is concern as to whether or not the objective of quality improvement will be sacrificed to that of cost cutting in an integrated system (Stone and Katz, 1996, p. 226).

Cost reduction and quality enhancement need not be mutually exclusive. However, costs can be reduced by providing fewer and/or lower quality services. Hence the desire to use integrated service systems in order to, for example, limit the right of Medicare beneficiaries to disenroll from managed care entities (National Senior Citizens Law Center, 1998, p. 7), or to bypass regulatory requirements such as the Medicare process of certifying home health agencies (Stone and Katz, 1996, p. 225) raises concerns amongst advocates for the elderly and people with disabilities regarding the manner in which integrated care would achieve cost reductions. But, these reservations aside, it would seem that perception of the problems associated with fragmented care delivery and of the benefits of integration is sufficiently widespread to have created a broad-based interest in the integration of acute and long-term care in policy circles. Sufficient, in fact, for Stone and Katz (1996, p. 217) to describe integration as A...one of the buzzwords of the 1990s,@ even as the National Senior Citizens Law Center (1998, p.7) warns of the need A....to assure the program beneficiaries, not the state and federal fisc, are the primary focus of policy development.@

Of course, the pros and cons of integration notwithstanding, there remains the question of how integration can or should be effected. Financially, integration involves pooling resources from the various sources of funding for acute and long-term care. In terms of service delivery, however, the notion of integration is less clear cut. Wiener and Skaggs (1995, pp. 6-7) identify two different models of integrated service delivery, which may, in principle, have different implications for both cost and quantity/quality of care outcomes. The first of these is the Ahands-off@ model, in which the provision of acute and long-term care remain separate, but each patient=s access to and hence transition between the acute and long-term care systems is explicitly co-ordinated by a case manager. This model is associated with, for example, Social HMO=s (see, for example, Leutz et.al., 1994). The second is the Ageriatrics@ model. In this model, the notion of integration extends far beyond the co-ordination of otherwise separate acute and long-term care services. The objective is to eliminate the distinction between acute and long-term care and to replace it with a holistic vision of health care including multidisciplinary teams. This brings us to the Progam of All-Inclusive Care for the Elderly (PACE) with which this report is primarily concerned, since PACE and the On Lok system from which it derives are A...relatively pure examples of the [second] approach, unifying acute and long-term care intellectually, physically and fiscally@ (Wiener and Skaggs, 1995, p. 7).

2. Outline of the Report

The purpose of this report is to describe and evaluate the PACE model as it appears at the various sites across the US where it already exists and then to analyze the forthcoming Connecticut PACE Pilot Program in light of these findings. PACE is not intended to provide a comprehensive solution to the fragmentation and lack of co-ordination within and between the acute and long-term care systems. However, it is a model that impacts the Adual eligible@ population which, as was noted earlier, has provided much of the impetus for the current interest of policy makers in the integration of acute and long-term care.

In Section II, the PACE model is described, and the literature evaluating its performance to date is summarized, with particular attention paid to program eligibility criteria, the number and type of clients served, Program finances and health care outcomes. The section ends with a discussion of two specific PACE sites, On Lok in San Francisco, CA and Comprehensive Care Management (CCM) in the Bronx, NY. Section III turns to the Connecticut PACE Pilot Program, first describing the essential features of this Program as stipulated in Public Act 98-198, An Act Establishing a Pilot Program of All-Inclusive Care for the Elderly (PACE). The Connecticut PACE Pilot Program is then evaluated, with particular emphasis on the following key issues: how (if at all) will the Program differ from existing PACE sites and why; will the Connecticut PACE Pilot Program ultimately become a separate program for all dual eligibles, regardless of their need (or lack thereof) for long-term care; what are the objectives of the Program from a health -care perspective; and what criteria will be used to assess the demonstration and on what basis will it be judged a success or failure? Finally, Section IV of the report offers some conclusions.

II PACE: A DESCRIPTION AND EVALUATION OF EXISTING SITES

1. What is PACE?

The chief purpose of this section is to describe the configuration of a typical PACE site. We begin by outlining the general aims and approach of the PACE model and by providing a brief history of the model to date. The organizational structure of a PACE site as it affects both the delivery and the financing of health care services is then discussed in more detail.

i) Basic Aims and Approach

The aim of PACE is to provide integrated acute and long-term care to a frail elderly population who are at high risk of institutionalization, so as to stabilize chronic conditions and assist basic functioning in a manner that is conducive to elders remaining in familiar surroundings and living as independently as possible (Irvin et.al., 1993, p. 21; Lee et.al., 1998, p. 65; Kunz and Shannon, 1996, p. 302). PACE is a voluntary program, and federal legislation requires that PACE participants be at least 55 years of age, although some states set the age limit higher at 60 or 65 (Irvin et.al., 1993, p. 21). In addition, seniors must live within the defined catchment area of a PACE site (Kunz and Shannon, 1996, p. 302; Eng et.al., 1997, p. 224), and must be state certified as eligible for admission to a nursing home (Irvin et.al., 1993, p. 21; Saucier, 1995, p. 26; Eng et.al., 1997, p. 224) in order to enroll in PACE. As such, PACE participants are not just elderly, but are also frail and tend to have multiple medical conditions and functional dependencies. According to Lee et.al. (1998, p. 65), Athe typical enrollee is age 80, has 7.8 medical conditions and is dependent in 2.7 activities of daily living.@

Once enrolled, a participant receives all of his/her health care from PACE staff or PACE contracted providers (Eng et.al., 1997, p. 224). This includes all Medicare and Medicaid acute and long-term care services as well as home and community-based services. The type of services that an enrollee might expect to receive from PACE staff or contractors include (but are not limited to) physician visits, prescription drugs, rehabilitation services, visits from personal care workers, hospitalization, and nursing home care (Lee et.al., 1998, p. 65). The idea is that in PACE, the formal distinction between acute and long-term care all but disappears (Kane et.al., 1992, p. 771). Instead, the Program aims to deliver a seamless stream of health care services, tailored to meet the specific needs of specific individuals (Kunz and Shannon, 1996, p. 302).

PACE sites receive capitated payments from both Medicare and Medicaid, and, in fact, most PACE participants are dually eligible. Some enrollees, however, are not Medicaid eligible, and these participants are personally liable for the Medicaid portion of their health care costs (Saucier, 1995, p. 26; Eng et.al., 1997, p. 224). Furthermore, some sites enroll participants who are Medicaid eligible but who do not qualify for Medicare, with Medicaid paying the full capitation rate for these individuals (Kane et.al., 1992, p. 772). PACE sites use their blended Medicare and Medicaid monies to provide services not allowable in the traditional, fee-for-service Medicare and Medicaid programs. Meanwhile, participants pay no further amounts (in the form of co-payments or deductibles, for example) for their health care once enrolled at a PACE site. A key principle of PACE is that individual sites (eventually) adopt full financial risk for the provision of health care services to their enrollees (Saucier, 1995, p. 26; Irvin et.al., 1993, p. 21). Any shortfall of revenues relative to health care expenditures must be made up by the site itself.

ii) The Historical Development of PACE to Date

The origins of PACE can be found in the On Lok (Cantonese for Apeaceful, happy abode@) Senior Health Services Program which began in the Chinatown, North Beach and Polk Gulch areas of San Francisco, California in 1972 (Zawadski and Eng, 1988, p. 75; Eng et.al., 1997, p. 224). Based on the British day hospital, On Lok originally sought to provide day health and social services to frail elders in a manner that would avoid their being institutionalized. Community adult day health centers C which remain central to the PACE model C were established in 1973, following which On Lok began providing home support services in 1975. In 1978, under the auspices of a HCFA-funded demonstration, it added primary medical services, thus effecting a then-unique integration of acute and long-term care (Eng et.al., 1997, p. 224).

In 1983, On Lok was granted Medicare and Medicaid waivers that enabled it to receive capitated payments from both programs in return for its assuming full risk for the provision of acute and long-term care to its enrollees (Eng et.al., 1997, p. 224; Kunz and Shannon, 1996, p. 302). In 1986, On Lok was granted indefinite extensions to these waivers. Meanwhile, the Omnibus Budget Reconciliation Act (OBRA) of the same year initiated the first generation of what thereafter became known as PACE demonstration sites, allowing On Lok to replicate itself in up to 10 new locations throughout the USA. This initiative was aided by large grants from the Robert Wood Johnson and John A. Hartford Foundations (Eng et.al., 1997, p. 224; Zowadski and Eng, 1988, p. 76; Irvin et.al., 1993, pp. 22-3). A second generation of demonstrations was initiated by the OBRA of 1990, with the result that since 1986, there have emerged 12 PACE sites (in addition to the original On Lok site) offering dual (i.e. Medicare and Medicaid) capitated, integrated acute and long-term care. As of April 1998, these were:

Centers For Elders Independence, Oakland, CA

Sutter Health=s Sutter Senior Care, Sacramento, CA

Total Long Term Care, Inc., Denver, CO

East Boston Neighborhood Health Center=s Elder Service Plan, East Boston, MA

Henry Ford=s Health System=s Center for Senior Independence, Detroit, MI

Beth Abraham Health Services= Comprehensive Care Management, Bronx, NY

Via Health=s Independent Living for Seniors, Inc., Rochester, NY

Sisters of Providence=s Providence ElderPlace, Portland, OR

Palmetto Richland Memorial Hospital=s Palmetto Senior Care, Columbia, SC

Bienvivir Senior Health Services, El Paso, TX

Community Care Organization=s Community Care for the Elderly, Milwaukee, WI

Elder Care of Dane County=s Elder Care Options, Madison, WI

Although these sites share the essential features of the PACE model in terms of their financing and methods of care delivery (see discussion below), there are some important differences between them. For example, it is typical for a PACE site to be based around a sponsoring organization, although this is not always the case, and there are considerable differences between the types of sponsor organizations found at the various PACE sites (Branch et.al., 1995, p. 350; Irvin et.al., 1993, p. 23). The Columbia, Rochester, and Portland sites, for example, are sponsored by a neighborhood health center, the Bronx site by a nursing home and the El Paso site (which is realtively new) has no sponsoring organization (Irvin et.al., 1993, p. 23). Furthermore, there is substantial variation in the size of catchment areas as between PACE sites. The On Lok site=s original catchment area was only 2.5 square miles (expanded to 10 square miles in 1989-90) and that in East Boston is only 5 square miles. Meanwhile, the catchment areas of other sites vary from 42 square miles (Bronx) to 1400 square miles (Columbia) (Branch et.al., 1995, p. 350).

The Balanced Budget Act of 1997 (P.L. 105-33) endowed PACE with permanent provider status under Medicare, and also allows states to treat PACE as a Medcaid benefit (National PACE Association, 1998; Lee et.al., 1998, p. 65). HCFA is currently working on regulations that will codify the operation of PACE sites under the auspices of the Medicare program. P.L. 105-33 allows the total number of PACE sites to expand to 40 within the first year of its enactment, and legislates the creation of an additional 20 sites per annum thereafter. Furthermore, these targets are cumulative C unused site authorizations carry over from year to year (National PACE Association, 1998). The legislation also authorizes a 4-year demonstration project that will facilitate the creation of up to 10 for-profit PACE sites (National PACE Association, 1998; Lee et.al., 1998, p. 65). From its beginnings in San Francisco over 23 years ago, PACE has thus evolved through several national demonstrations to the point where it has now been certified as a permanent feature of Medicare and (subject to the discretion of individual states) Medicaid, and has been federally mandated to expand the scale of its operations (as measured by the number of sites operating throughout the USA.

iii) The Organizational Structure of PACE sites: Delivery of Services

As indicated earlier, PACE sites provide enrollees with an expansive range of acute and long-term health care services. There are two outstanding features of the PACE model with regard to its organization of the delivery of services: the adult day health center (ADHC) and the multidisciplinary term.

ADHC=s providing Aadult day care@ is not a new idea. For example, as Chatterji et.al. (1998, p. 5) note, in Weissert et.al.=s (1998) survey of home and community based long-term care programs, some 12 of the 27 programs surveyed offer adult day health services to enrollees. However, what is unique about PACE=s use of the ADHC is that, rather than being optional, attendance is a requirement for Program participants. The typical PACE ADHC services approximately 120 Program participants with 60-80 staff (Eng et.al., 1997, pp. 225-6), and acts as the geographical Afocal point@ of the Program. A typical PACE ADHC includes a medical practice, rehabilitation center and offices for nurses and social support staff and offers a variety of activities and classes (Lee et.al., 1998, p. 66). In this way, it serves three important functions that are central to the PACE model of care: it allows PACE staff to monitor the physical and mental well-being of their enrollees (Wiener and Skaggs, 1995, p. 15; Kane et.al., 1992, p. 772); it serves as the primary setting for the delivery of health care to Program participants (Kane et.al., 1992, p. 772); and it provides a social focus for elders who might otherwise be housebound and/or socially isolated (Kane et.al., 1992, p. 772; Wiener and Skaggs, 1995, p. 15).

The second outstanding feature of the PACE model in terms of its service delivery structure is the multidisciplinary team. The composition of these teams, the frequency with which they meet and their internal dynamics are all site specific (Chatterji et.al., 1998, p.5). However, it is possible to make some generalizations. First, multidisciplinary teams are composed of all those PACE staff - professional and para-professional C who have personal contact with Program participants and may include a physician, nurse, social worker, nutritionist, physical, recreational and/or occupational therapists, personal aides and van drivers (Irvin et.al., 1993, p.24; Wiener and Skaggs, 1995, p.15; Lee et.al., 1998, p.66). Second, PACE staff devote considerable time resources to discussion of participant care within the context of the mutidisciplinary team. Kane et.al. (1992, p.775) estimate that formal team meetings occupy about 8 hours per week of a typical staff member=s time. Finally, critical to the functioning of the multidisciplinary team is that it operates as a team. Hence although the team includes health professionals (including a physician) as well as social workers, the former do not dominate as in the traditional medical model of service delivery (Irvin et.al. , 1993, p.24).

The multidisciplinary team serves a variety of functions, including the assessment and re-assessment of enrollee=s well-being, the design of plans of care, and the oversight and (in most cases) direct delivery of care (Kane et.al., 1992, p.772). Note that in this way, PACE integrates not only types of care (acute and long term), but also the planning and delivery of care (see Figure 1 below). In the PACE model, staff members who deliver most of the care that enrollees receive also jointly plan this care. Case management is central to the PACE model, then, but there is no case manager as such. Instead, the case management function is served by those directly involved with the delivery of services operating as the multidisciplinary team. The main perceived advantage of this approach is the potential for information sharing (Eng et.al., 1997, p.227; Lee et.al., 1998, p.66). For example, therapists and aides who are in a position to notice changes in a patient=s physical functioning can not only draw these changes to the attention of the team physician, but the information can also be rapidly incorporated into the subsequent planning of care through the discussions of the multidisciplinary team. The main disadvantage of the approach is the loss of the independence of the traditional case manager, who is not involved in service provision and, as such, is well placed to recommend and monitor care impartially (Zawadski and Eng, 1988, p.80).

Figure 1: The Role of the Multidisciplinary Team in the PACE Model of Integrated Acute
and Long-Term Health Care.

Source: Eng et.al. (1997, p.227).

Finally, although it is less generally recognized as such in the literature describing PACE, Kane et.al. (1992, p.772) and Branch et.al. (1995, p.354) argue that there is a third outstanding feature of the PACE model=s care delivery system: protective, congregate housing. The financing of this service is not included in the capitated payments received by PACE sites (see the discussion in section II.2.(vi) below), but housing is identified by Kane et.al. (1992, p.772) as being instrumental to the success of PACE sites in reducing the hospitalization rate and the length of hospital stays of their enrollees. In Branch et.al.=s (1995, p.354) survey of 8 PACE sites, more than one fifth of enrollees were found to be occupying PACE housing at three sites (On Lok, East Boston and Rochester) and more than 50% of enrollees were found to be occupying PACE housing at one site (the Bronx).

iv) The Organizational Structure of PACE Sites: Financing

As intimated earlier, Amature@ PACE sites are financed by capitated payments from Medicare and Medicaid, together with (in some cases) direct payments from non-Medicaid eligible enrollees, and assume full financial risk for the services they provide. Because funds are pooled and traditional payment restrictions do not apply, the pattern of PACE site expenditures can vary from those observed in the traditional fee-for-service Medicare and Medicaid programs. For example, PACE may C indeed, does C offer services not available under these programs, such as social services intervention or respite care (Lee et.al., 1998, pp.65-6). Or it may continue expenditures on a particular treatment beyond the time or dollar limitations that would ordinarily be imposed (Irvin et.al., 1993, p.24). According to Lee et.al. (1998, p.73), the typical PACE site directs 32% of its expenditures towards ADHC-based services, 22% towards in-home care, 21% towards plant and administration, 17% towards inpatient services and 8% towards other medical services.

Both the Medicare and the Medicaid capitated payments to PACE sites are designed so as to build in savings to each program, when compared to estimated expenditures for fee-for-service care. Medicare capitated payments to PACE sites are calculated on the basis of the same methodology that HCFA uses to calculate payments to Medicare HMOs C namely, calculation of the Adjusted Average Per Capita Cost (AAPCC) of care for the catchment area of the site. PACE sites are then paid 95% of the AAPCC, after the latter has been inflated by a special adjustment factor of 2.39 to reflect the frailty of PACE=s target population. There is substantial variation in the size of the Medicare capitated payment across PACE sites and this range also varies over time. Medicare capitated payments per person enrolled per month ranged from $500 to $1100 in 1990 (Kane et.al., 1992, p.773), from $689 to $1562 in 1994 (Eng et.al., 1997, p.224) and from $737 to $1623 in 1995 (Kunz and Shannon, 1996, p.302).

Calculation of the Medicaid capitation rate has been less straightforward and has produced even greater variation in capitation rates between sites (Kane et.al., 1992, p.773), largely because the precise methodology used to calculate the payment varies by state. Essentially, each state calculates its Medicaid capitation rate by calculating what it pays, on average, to care for a frail elderly population in a non-PACE setting. However, the choice of comparison group varies by state. For example, in Oregon, which has a strong tradition of home and community based care, PACE is deemed a substitute for assisted living and the capitation rate is based on the average costs of providing this form of care. Meanwhile, the Columbia, SC and Rochester, NY sites are viewed as alternatives to nursing home care, so that their Medicaid capitated payments are based on the calculation of average nursing home costs (Chatterji et.al., 1998, p.5). Further variation arises from the fact that PACE sites receive only a fraction of these estimated average costs of alternative care as their capitation rates C and these fractions vary between 85% and 95% of the average costs of alternative care depending on the state (Chatterji et.al., 1998, p.6). Not surprisingly, then, Medicaid capitation rates vary widely across sites, and this range of rates also varies over time. The Medicaid capitation rates received by PACE sites ranged from $1,486 to $4,465 per person per month in 1994/5 (Kunz and Shannon, 1996, p.302; Eng et.al., 1997, p.225) and from $1,486 to $4,006 per person per month in 1996 (National PACE Association, 1997, p.16). It follows from what has been said above that total (Medicare plus Medicaid) capitated payments also vary widely across sites, and do so in a fashion that is dominated by the variations in Medicaid capitation rates (Branch et.al., 1995, p.348). In the early 1990s, total capitated payments (per person, per month) to PACE sites varied from $4,830 in the Bronx to just $1, 885 in Denver and Portland (Kane et.al., 1992, p.774).

One final point that remains to be made is that according to some commentators (Eng et.al., 1997, p.225; Kunz and Shannon, 1996, pp.302-3), the way that PACE finances are organized has positive repercussions for the nature and quality of patient care at PACE sites. Hence because PACE takes responsibility for all (acute and long-term) care and is at full financial risk in so doing, it is unable to shift costs to another part of the health care system; it is the health care system for its enrollees. Hence there exist no incentives to withhold certain types of medical services in the belief that any subsequent financial consequences this may have will be borne by a different provider. At the same time, there exists an incentive to engage in aggressive preventative care, in order to reduce the costs associated with acute conditions. In sum, the all-inclusivity of PACE, coupled with its adoption of full financial risk for health care, means, it is argued, that PACE sites are unable to trade off the quality/quantity of care they provide for cost savings.

2. Evaluating the Performance of Existing PACE Sites

Based on the variety of studies that discuss outcomes associated with PACE, it is possible to evaluate the performance of existing PACE sites according to a number of different social, financial and health-related criteria.

i) Program Eligibility Criteria

As previously discussed, PACE enrollees must be at least 55 years of age, live within a site=s catchment area and be state-certified as eligible for admission to a nursing facility. However, PACE does not serve a random sample of this eligible population. Rather, its clientele is subject to two forms of selectivity. First, PACE is a voluntary program; eligible elders are not required to participate and may choose not to. Second, PACE sites are not required to accept all eligible elders who do wish to participate in the Program; they can and do screen applicants, admitting only those deemed appropriate for participation.

According to Branch et.al. (1995, pp.350-1), the process by which PACE sites screen applicants is relatively uniform across sites. Would-be participants are usually referred to the Program and initial screening (to check, for example, the individual=s financial status) is followed by a home visit. This permits construction of an applicant profile C including his/her living arrangements, current health status and a limited medical history C which is referred to the multidisciplinary team. If no objections are raised, the applicant visits the ADHC in order to meet staff and (at most sites) to allow the multidisciplinary team to conduct a more detailed evaluation. Following the site visit, staff make final decisions with respect to the (un)acceptability of the applicant.

Branch et.al. (1995, p.351) identify a number of characteristics that PACE sites self-report as influencing their decision as to whether or not to accept a particular PACE applicant. Sociability, transport safety and family support are identified as leading factors in favor of acceptance; cognitive insufficiency, disruptive behavior and substance abuse are leading factors making for denial. Mental disorders are identified as particularly problematical, posing a challenge to all PACE staff (from care providers to home helps and van drivers) and making a patient unsuitable for the atmosphere and environment of the ADHC (Branch et.al. , 1995, pp.353-4). The calculation of admission to referral ratios is suggestive of significant differences between sites as regards admission decisions and/or the quality of referrals that sites receive. In the survey of 8 PACE sites by Branch et.al. (1995, p.353), On Lok was found to have the lowest admission to referral ratio (0.11), stemming from an average rate of admission but a much higher than average referral rate. East Boston and Denver were also found to have low admission to referral ratios (0.12 - 0.16), but this time because of their lower than

average rates of admission. Portland=s admission to referral ratio of over 0.5 was the highest amongst the sites surveyed.

ii) Number and Type of Clients Served

Total enrollment in PACE has grown steadily over time, from 2004 persons in 1994, to 2700 persons in 1995, to 4053 persons in 1996 (Wiener and Skaggs, 1995, p.23; Eng et.al., 1997, p.228; National PACE Association, 1997, p.6). Despite this expansion C which the National PACE Association (1997, p.6) calculates as representing an annual average rate of growth of enrollment of 36% per annum since 1990 C these figures suggest that PACE is really quite small in terms of the total number of clients that it serves. At the 11 Amature@ PACE sites (those operating under both Medicare and Medicaid capitation) extant in December 1996, the total number of clients served varied from 135 (Oakland) to 550 (On Lok), the average number of clients per site being approximately 293. The small total number of clients served is certainly not due to a propensity on the part of PACE participants to disenroll from the Program. Less than 5% of all persons served during 1996 disenrolled for reasons other than death (National PACE Association, 1997, p.6). The PACE disenrollment rate for reasons other than death averages just 0.9 persons per month per site across all sites (Eng et.al., 1997, p.228). However, numerous commentators have suggested that PACE sites suffer enrollment problems C particularly during the early stages of their development. For example, Kane et.al. (1992, p.778) report that of the 4 sites that Amatured@ in 1990 (East Boston, Columbia, Portland and Milwaukee), none achieved the target enrollment of 120 specified by On Lok, with fewer than two-thirds of those who had participated in the pre-capitation programs at these sites Arolling over@ into the capitated programs. Branch et.al. (1995, p.350) report that these enrollment problems remained at the time of their site visits in 1993. A number of factors have been identified as contributing to enrollment difficulties at PACE sites. In the Branch et.al. (1995, p.351) survey, the three Abarriers to enrollment@ most commonly reported by PACE staff were the distaste of potential clients for regular visits to the ADHC, financial considerations (either an unwillingness to privately pay the Medicaid capitation rate or, in some cases, an unwillingness to apply for Medicaid), and the unwillingness of potential clients to commit themselves to site providers and thereby forego provider choice (especially their choice of physician). Kane et.al. (1992, p.778) also argue that PACE sites may face institutional opposition from case managers, who might otherwise refer potential clients to PACE sites, either because they are wary of recommending programs rather than treatments to their patients or because they have their own ideas about what will best serve a patient=s interests and/or are unwilling to give up patients to PACE sites. It is also possible that in their early stages of development, it is difficult for PACE sites to market their services because of ambiguity about what they provide and how it is financed. Hence immature sites may offer services that differ from those that will ultimately be offered under full capitation and are certainly trying to sell services for which the funding arrangements will ultimately change (from fee-for-service to capitation) (Kane et.al., 1992, p.778). This ambiguity may be off-putting to potential participants and may also help to explain the poor Aroll-over@ rates in the transition to full capitation that were reported earlier.

The brief sketch of the PACE clientele in section II.1.(i) above has already revealed that the typical PACE participant is elderly (aged 80) and very frail, having 7.8 medical conditions and being dependent in 2.7 activities of daily living (Lee et.al., 1998, p.65). In fact, approximately one-third of PACE enrollees are 85 or older (National PACE Association, 1998b). More than half are incontinent and about 50% suffer from dementia, making the latter the fourth most common medical diagnosis amongst PACE enrollees, after diseases of the eye (which afflict 60% of enrollees), hypertension (which afflicts 57% of enrollees) and arthritis (which afflicts 54% of enrollees) (National PACE Association, 1998b; National PACE Association, 1997, p.8). Some 39% of PACE enrollees live alone in the community C 14% without any informal support from family and/or friends (Eng et.al., 1997, p.229). In addition, compared to applicants who ultimately decide not to enroll in PACE, those who do decide to enroll are more likely to be female and to have low levels of educational attainment, less likely to own a home, more likely to have had experience with adult day care before enrollment, more likely to have been in a hospital or nursing home immediately prior to enrollment, more likely to self-report that they are in good or excellent health and satisfied with life and health care arrangements, and more likely to be IADL impaired (Kidder, 1996, p.2; Irvin et.al., 1998, p.2).

One of the distinguishing features of PACE as compared to other long-term care programs is that the majority of enrollees are either African American, Hispanic or Asian (Wiener and Skaggs, 1995, p.23). Only 46% of the total population of PACE enrollees are white: 26% are African American, 15% Hispanic and 12% Asian (with 2% other) (National PACE Association, 1997, p.7). The racial/ethnic composition of the PACE clientele does vary dramatically between sites. Some sites (for example, East Boston, Portland and Rochester) serve a predominantly white clientele, others (for example, Detroit, Oakland and Columbia) serve a predominantly African American population, whilst at some sites, enrollees are predominantly Asian or Hispanic (San Francisco and El Paso respectively) (National PACE Association, 1997, p.7). The Bronx and Denver sites are examples of racially mixed sites, serving populations of enrollees that are 29% white, 27% African American, 43% Hispanic and 1% Asian, and 52% white, 27% African American, 20% Hispanic and 1% other respectively (National PACE Association, 1997, p.7). As Branch et.al. (1995, p.353) remark, newer PACE sites have not replicated On Lok=s original, predominantly Asian catchment area. This suggests that any successes accredited to the PACE model (regarding health care outcomes, cost effectiveness etc.) are not likely due to the specifics of the racial/ethnic composition of Program participants.

One final noteworthy characteristic of PACE participants is that at all sites the vast majority of enrollees are Medicaid eligible. Although the proportion of private-pay participants (those eligible only for Medicare, and who must pay the Medicaid capitated rate out-of-pocket) varies between sites, it is typically less than 10% (Branch et.al., 1995, p.353). This is not surprising. Non-Medicaid eligible PACE participants must contribute co-payments that vary between $5.00 and $3,000.00 depending on the site and the participant=s income and wealth (Irvin et.al., 1998, p.36). The average monthly co-payment in 1994 was $2,361.00 (Eng et.al., 1997, p.231). Although this is less than the average monthly cost of a nursing home bed, the latter may often appear preferable to a frail elder=s family because it includes room and board, whilst many middle-income elders prefer to simply go without formal long-term care rather than pay for either PACE or nursing home care (Eng et.al., 1997, p.231).

iii) Service Utilization

Given the structure and objectives of PACE regarding the integration and management of acute and long-term care in a community setting with emphasis on preventative care, we should observe a APACE effect@ in the pattern of health service utilization that is characteristic of PACE participants, as compared to that of other frail elders. In fact, it seems that this is the case. However, it is important to note that the service utilization comparisons that follow lack a true experimental design (Chatterji et.al., 1998, p.48). When comparing PACE participants with other frail elders, we are contemplating a non-random group that has chosen to participate in PACE and has been approved by PACE staff. Hence we cannot be sure whether the APACE effect@ reported below results from the intrinsic nature of the Program itself or from characteristics of PACE participants, by virtue of the sample selection bias that is necessarily involved in comparisons of PACE participants with other groups of frail elders.

The rate of hospital utilization by PACE enrollees is consistently below that of the general Medicare population, despite the fact that PACE participants are, on average, more frail than the general Medicare population C a fact that Irvin et.al. (1993, p.27) deem Aremarkable@. In 1992, PACE enrollees averaged 2,777 hospital days per thousand members per annum, compared to 2,811 days per thousand persons per annum for the general Medicare population in 1990 (Irvin et.al., 1993, p.27). In 1995, meanwhile, PACE participants averaged 2,399 hospital days per thousand persons per annum, compared with 2,448 days per thousand persons per annum for the general Medicare population (Eng et.al., 1997, p.229). The average length of stay in hospital is also lower for PACE participants than for the general Medicare population C 5.4 days per admission in 1992 compared to 9.0 days per admission for all Medicare beneficiaries in 1990 (Irvin et.al., 1993, p.27) and 4.9 days per admission in 1995 compared to 7.5 days per admission for the general Medicare population in 1994 (Eng et.al., 1997, p.229). Finally, despite the fact that all PACE participants are state certified as being nursing home eligible, the rate of nursing home utilization amongst PACE participants is very low. According to Kunz and Shannon (1996, p.303) and Eng et.al. (1997, p.229), only 5-6% of PACE capitation days are nursing home days. At the end of 1996, fewer than 7% of Program participants resided in a nursing home (National PACE Association, 1998b).

The importance of the descriptive statistics on service utilization reported above is confirmed by Kidder (1996) and Chatterji et.al. (1998), who find that, relative to comparison groups of similarly frail elders, PACE has statistically significant impacts on hospital and nursing home utilization. PACE enrollees are less likely to be admitted to a hospital (20% admitted compared to 37% for the comparison group) and are less likely to be admitted to a nursing home (recording a 9% admittance rate compared to 20% for the comparison group) (Kidder, 1996, p.3; Chatterji et.al., 1998, p.47).

PACE has also been found to reduce the utilization of other health care services. For example, in 1994, PACE enrollees made only 1 visit every 4 months to a specialist, compared to 1.05 visits per month to a primary care physician C a low utilization rate of specialists relative to primary care physicians, according to Eng et.al. (1997, p.229). The same authors argue that PACE reduces prescription drug use, favorably comparing an average number of medications per client at PACE sites of 5.53 with Beers et.al. (1992) finding of an average 7.2 medications per patient for elderly nursing home residents.

PACE participants are, not surprisingly (given the design of the PACE model), significantly more likely than other frail elders to use ADHC services (Chatterji et.al., 1998, p.47) and to use significantly more of these services (Kidder, 1996, p.3). According to Branch et.al. (1995, pp.354-5), the percentage of PACE enrollees who use the ADHC is typically close to 100% and no less than 85% at any site. On average, PACE enrollees attend the ADHC 11 days per month (National PACE Association, 1997, p.11). They visit a PACE primary care physician an average of 1.05 times per month and receive an average of 6.15 nursing visits per month in the ADHC (Eng et.al., 1997, p.229). All of this has been interpreted as A... evidence for the success of primary care in PACE@ (Eng et.al., 1997, p.229). Sharma (1998) provides some rather more substantial evidence in support of this latter claim, in a study of the East Boston PACE site. The author uses preventable hospitalizations (PH) as an indicator of success in the provision of primary care, arguing that although no causal link has been proven, there is support for the hypothesis that a link exists between inadequate primary care and the need for hospital admittance, at least for some medical conditions (Sharma, 1998, p.2). Adjusting for age and gender differences between groups, the PH rate for elders at the East Boston PACE site is found to be 18.46%, which compares favorably with the 34.19% PH rate calculated for the author=s comparison group of similarly frail, non-PACE enrolled elders. According to Sharma (1998, p.3), Athese results strongly indicate that PACE is an effective intervention at prevention and primary care, when measured in terms of preventable hospitalization rates@.

However, it should be noted that there is variation in health service utilization between PACE sites. At the time of their evaluation, Kane et. al. (1992, pp.774-5) reported that some PACE sites were having difficulty in encouraging use of the ADHC C particularly the Bronx site, at which, prior to full capitation, fewer than 50% of enrollees used the ADHC. In Branch et.al. (1995, p.355), the Bronx site was found to have increased the proportion of its clients utilizing the ADHC to 85%. However, it continues to be the site at which there is the lowest level of per-participant, daily attendance at the ADHC C just 7.0 days per enrollee per month, compared to a Program-wide average of 11 days per month (National PACE Association, 1997, p.11).

Furthermore, other service utilization rates C and in particular, hospital and nursing home utilization rates C vary between sites (Saucier, 1995, p.27; Branch et.al., 1995, pp.354-6; Eng et.al., 1997, p.229). For example, hospital utilization rates ranged from 480 days per thousand enrollees per annum to 5040 days per thousand enrollees per annum across sites in 1995 (Eng et.al., 1997, p.229). Meanwhile, the percentage of clients using a nursing home ranged from 1% (the Bronx) to 22% (Denver) in 1992 (Branch et.al., 1995, p.355). According to Kane et.al. (1992, p.775), there are also variations in the degree to which clinical activities are integrated into the other activities of the ADHC. Their evaluation suggests that at some sites clinical and other activities are little more than located in the same building, whereas at other sites clinical activities are part of the

daily routine, enabling physicians to observe patients engaging in everyday activities and enabling nurses and even social workers to assist in the monitoring of each participant=s physical well-being.

iv) Health and Quality of Life Outcomes

According to the National PACE Association (1998b), enrollment in PACE is associated with lower mortality rates, improved health status and better quality of life. In fact, independent research corroborates these claims. Hence across all PACE sites in 1993, there were, on average, 138 deaths per 1,000 persons per annum, compared to some 186 deaths per thousand persons per annum amongst those residing in nursing homes (Eng et.al., 1997, p.229). Furthermore, both Kidder (1996, p.3) and Chatterji et.al. (1998, p.47) report a positive impact of PACE enrollment on life expectancy, although only in the latter study is this found to be statistically significant.

At the same time, there are statistically significant differences between the self-reported health-care status and quality of/satisfaction with life of PACE enrollees and other frail elders. For example, PACE enrollees are more likely to consider themselves as currently being in good health and to expect good health in the future (Kidder, 1996, p.3; Chatterji et.al., 1998, p.47). In addition, they are more likely to find life satisfying, attend social programs, and report having both some discretion with respect to how they spend they time, and at least some confidence when facing life=s problems (Chatterji et.al., 1998, p.47). It is important to note that Chatterji et.al. (1998, p.48) report that some of these effects C particularly those associated with life satisfaction and health status - appear to be relatively short lived. Furthermore, both Kidder (1996, p.5) and Chatterji et.al. (1998, p.48) find that the beneficial outcomes of PACE enrollment on health outcomes and life satisfaction are most noticeable amongst the frailest elders C those who are most dependent and suffer the most severely limiting conditions. This leads Chatterji et.al. (1998, p.48) to conclude that A... the PACE approach is particularly suited to addressing the needs of the frailest members of the nursing home eligible population@.

v) Ethnicity Issues

As discussed above, PACE sites have succeeded in enrolling participants from a variety of racial/ethnic backgrounds. Indeed, PACE is the only long-term health care program that has a majority of non-white participants. It is important in this context that studies of PACE participants that distinguish between the racial/ethnic backgrounds of enrollees have found important differences between the health care wishes, and the ways that these wishes are expressed, of different racial/ethnic groups. Hence in Hornung et.al.=s (1998) study of 1193 PACE participants, 91% of whites were found to express their own health care wishes, compared to 85% of Hispanic, 83% of Asian and 67% of African American patients surveyed. Alternative decision makers C a child, spouse, or in some cases (particularly amongst black men), neither C were identified for fully 33% of African American patients and 15% of Hispanics and Asians, but only 8% of whites. In addition, Eleazer et.al. (1996) find that amongst PACE participants, African Americans are significantly more likely than other patients to select aggressive health care interventions, whilst white patients were significantly more likely than other patients to express their health care wishes in the form of a written instrument, such as a living will or Durable Power of Attorney. It is not at all obvious that these outcomes are connected with enrollment in PACE per se. However, they may have an important bearing on the success with which PACE sites C which vary considerably with respect to the racial/ethnic composition of their enrollments C are able to function, owing to the general emphasis within the PACE model on discussing health care wishes with patients and obtaining advance health care directives (the latter often in written form) (Eng et.al., 1997, p.228).

vi) Program Finances

One of the main aims of PACE C and certainly one of the things that makes it attractive to state and local governments C is to achieve greater efficiency in caring for elders through the integration of acute and long-term care, which, in principle, facilitates the substituting of cheaper long-term and primary care services for more expensive acute care services, thus reducing the overall costs of delivering care. At the present point in time, it is difficult to say to what extent PACE is successful in realizing this goal of cutting the costs of delivering health care and hence reducing per capita state and federal health care expenditures. As noted earlier, the method of calculating both the Medicare and Medicaid capitation rates paid to PACE sites builds in savings to both programs C at least compared to what it is estimated that the programs would otherwise have had to pay to provide PACE enrollees with traditional, fee-for-service care. Hence PACE sites receive 95% of the Medicare AAPCC and 85-95% of what state Medicaid officials estimate to be the average cost of caring for a comparably frail group of elders C apparently yielding 5% and 5-15% savings in total expenditures to the Medicare and Medicaid programs, respectively. Some studies suggest that these estimates of savings due to PACE are, in fact, too low. The National PACE Association (1998b) cites a study by DataChron Health Systems, Inc. that estimates a 12% saving to Medicare due to PACE. Meanwhile, an oft-cited study by Gruenberg et.al. (1993) argues that PACE sites are under-compensated by Medicare. Recall that HCFA adjusts the AAPCC by a single frailty factor of 2.39 when calculating the PACE capitation rate. Gruenberg et.al. (1993) argue that the frailty of the PACE population is such that this adjuster should actually be between 2.42 and 3.60, so that PACE, in effect, reduces Medicare expenditures by between 14% and 39% of the estimated cost of fee-for-service care.

It has also been argued that PACE creates savings in capital as well as current expenditures. Hence Eng et.al. (1997, p.230) suggest that the development of a PACE site serving 250-300 members requires approximately $1.5 million in capital expenditures C i.e., approximately $5,000 per potential PACE enrollee. This, they argue, is substantially less than the capital cost per bed of a nursing home C a likely alternative source of care for the frail elderly population served by PACE.

Perhaps inevitably, given the paucity of studies that exist at this point in time, there is substantial debate surrounding the cost effectiveness of PACE. Wiener (1996a, p.49; 1996b, p.9) argues that such savings that are in evidence may stem simply from the fact that PACE is a capitated program, and not from any efficiencies arising from the integration of acute and long-term care. Given the current state of research, it is simply not possible to distinguish between the contributions of these two potential sources of savings at the present point in time. Others (Wiener, 1996b, p.9; Irvin et.al., 1993, p.58) express concern about the small size of PACE sites. According to Irvin et.al. (1993, p.58), PACE sites enroll only 2% of the PACE eligible population in their catchment areas and it is impossible to know whether or not their cost effectiveness could be replicated with substantially increased enrollments.

Some researchers are concerned that PACE may not be cost effective at all C that the savings reported above are, in fact, more apparent than real. One factor motivating this concern is that savings are calculated relative to the estimated cost of providing PACE-eligible elders with traditional, fee-for-service care. As we have already seen, there is substantial debate surrounding the frailty adjustment in these calculations, which may well lead to the estimated savings due to PACE being understated. However, a different concern is the possibility that, whatever their exact value, the estimated expenses of providing fee-for-service care to the PACE eligible population differ from C specifically, are higher than C the actual expenses that would, in fact, be incurred as a result of providing fee-for-service care to the PACE enrolled population. The reason for this is straightforward. If Medicare and Medicaid payments to PACE sites reflect the expense of caring for the average PACE-eligible elder, but PACE sites Abeat the average@ C i.e., enroll a non-random sample of the eligible population that is, on average, healthier and therefore less costly to care for than the eligible population C then PACE sites will be systematically overpaid for the care they deliver. Rather than resulting in savings to Medicare and Medicaid, PACE would, in this scenario, cost these programs more than alternative forms of care. Note that the potential for this problem to arise is clearly present C those who enroll in PACE are a non-random sample of elders who partly self-select into PACE and are partly selected by the Program itself through the decisions of the multidisciplinary team. Note also that the problem is not necessarily indicative of any unscrupulous decision making on the part of PACE staff. Their enrollment decisions may be based entirely on non-financial considerations and in any case, as was noted above, PACE enrollees are partly self-selecting C they have to decide to enroll in the Program themselves. Nevertheless, the possibility arises that there will be an adverse (from the point of view of Medicare and Medicaid) selection of frail elders into the PACE program.

Not surprisingly, the prevalence of this problem in reality is a subject of heated debate. Some authors have vigorously denied its existence. Clauser et.al. (1996) argue that the screening of clients by PACE staff is vital to ensuring the suitability of the individual for the Program and that the average PACE enrollee is, in any case, as frail as his/her unenrolled peers, having 1.78-3.42 ADL dependencies in 1993, compared to 3.2 dependencies for the average Medicaid eligible nursing home patient. However, as noted above, the fact that PACE staff make enrollment decisions on the basis of medical/social criteria and not financial criteria (i.e., they are not deliberately trying to Acherry-pick@ the cheapest clients to care for) is besides the point, because the medical/social criteria they use to govern enrollment decisions may, however unintentionally, correlate with the cost of care, so that the most costly patients to care for are not admitted. Recall then from section II.2.(ii) that Branch et.al. (1995, p.351) report that cognitive insufficiency, disruptive behavior and substance abuse are the three most common factors that PACE staff report for denying an enrollment. If any or all of these are positively correlated with the cost of care C and this suggestion is purely hypothetical, because Branch et.al. (1995, p.351) do not investigate the possibility C then PACE staff will be unwittingly selecting out (i.e., denying enrollment to) some of the highest cost-of-care patients in their catchment areas; patients who, nevertheless, will feature in the calculation of the capitation rates that accrue as revenues to PACE sites.

Furthermore, a variety of authors present evidence suggesting that, contrary to the claims of Clauser et.al. (1996), PACE enrollees are, on average, less frail than the general population of elders from which they are drawn. Eng et.al. (1997, p.229) suggest that the average PACE enrollee is dependent in 2.7 ADL=s C below the Medicare nursing home average. This is confirmed by data from On Lok (1993) cited by Wiener (1996a, p.48) and Wiener and Skaggs (1996, p.24), suggesting that PACE enrollees are less frail than nursing home residents in all ADL=s. In 1985, nursing home patients were more dependent than PACE participants in bathing (91% versus 82%), dressing (78% versus 67%), toileting (63% versus 51%), transferring (63% versus 48%) and eating (40% versus 26%). Meanwhile, Irvin et.al. (1998, pp.2, 36) argue that health status plays a statistically significant role in influencing the willingness of individuals to enroll in PACE. Hence they find that whilst PACE sites attract patients requiring the greatest IADL assistance, death within three months of interview, increasing numbers of impairments requiring homemaker services and prior Medicare reimbursements C all indicators of substantial frailty C are negatively associated with the likelihood that an individual will enroll in PACE. Hence the authors conclude that Athe role played by health status suggests that those with the lowest levels of health and incurring the largest Medicare expenditures are the least likely to enroll [in PACE]@ (Irvin et.al., 1998, p.2). Finally, a rather different problem, reported by Kane et.al. (1992, p.776), is that once a person has been deemed eligible for PACE, they remain eligible indefinitely C their eligibility is not subsequently re-assessed. This is significant because elders= health status tends to fluctuate over time, so that a person deemed eligible for nursing home care today may not be deemed so in the near future. Kane et.al. (1992, p.776) also refer to the Anumerous studies@ showing that eligibility for, and actual use of, nursing homes are not well correlated. In other words, it is possible that at any point in time, PACE is caring for elders who either would not be eligible for, or would not enter a nursing home, were they not enrolled in PACE. The implication is that the estimated cost of caring for these individuals C used in the calculation of the health care savings due to PACE C is greater than the actual cost that Medicare and/or Medicaid would incur were they not enrolled in PACE.

Two final aspects of the financing of PACE sites remain to be discussed. First, a number of important issues surround the initial financing of a PACE site. Saucier (1995, p.27) emphasizes the importance of development grants during the start-up phase of a site, if it is to remain financially viable. On a similar note, both Kane et.al. (1992, p.778) and Branch et.al. (1995, p.350) remark on the important role of the sponsor organization in determining the financial viability of a site C especially prior to full capitation. Hence Kane et.al. (1992, p.778) report that sponsors of successful sites provide substantial formal and informal assistance, including free or subsidized administrative work, supplies and assistance with site construction/renovation. Meanwhile, Branch et.al. (1995, p.350) refer to the typical practice of sponsor organizations underwriting losses during the early development of a site.

A different but, nevertheless, related theme raised by Saucier (1995, p.27) concerns the arrangements for risk sharing during the pre-capitation phase of a PACE site=s development. He argues that because losses are shared by Medicare and Medicaid in proportion to their capitation rates, forcing Medicaid to absorb more losses (because of its proportionately larger capitation payments) despite the fact that these losses typically originate from higher than expected hospital utilization (which would normally be Medicare financed) and because there is no provision for profit sharing, state Medicaid agencies have the incentive to set their Medicaid capitation rates as low as possible. This problem could clearly have an adverse effect on PACE site finances, particularly during the start-up phase of the site when it is most financially vulnerable.

The second aspect of Program financing mentioned above is more controversial and concerns the role that supportive housing plays in the delivery of PACE services. As discussed in section II.1.(iii) above, supportive housing arrangements have been identified by some researchers as an integral feature of the PACE model. The proportion of Program enrollees who live in PACE housing varies widely between sites, from less than 1% at 5 sites to between 20% and 50% at 4 others and a high of 89% at the Portland site (National PACE Association, 1997, p.14). The concern exists that because these housing services are not included in the Medicaid and Medicare capitation rates, but are instead separately financed (Kane et.al., 1992, p.772), they compound the difficulties associated with developing an accurate appreciation of the cost effectiveness of the PACE model.

vii) Lessons For the Future

What, if anything, can be learned from the PACE evaluations surveyed above that might usefully inform future PACE sites C such as those planned in Connecticut C and the state officials involved in their development? Six general issues can be identified.

Site Development C Eng et.al. (1997, p.230) emphasize the considerable resources that must be mobilized when developing a PACE site, the typical site to date having spent $1.5 million and operated for a 3-5 year start-up period before achieving Amaturity@ (dual capitation). Kane et.al. (1992, pp.778-80) emphasize the vital importance of there being sufficient funding during this start-up phase to ensure the financial viability of a new site C some of which may necessarily emanate from external, grant giving agencies.

Staff Development C According to Kane et.al. (1992, p.780) PACE sites have typically experienced high turnover amongst physicians and other ADHC staff C but the development of a stable multidisciplinary team with genuine abilities to function in a multidisciplinary environment is critical to the success of a PACE site. Eng et.al. (1997, p.231) suggest that recruitment and retention of physicians and health care professionals has been a particular challenge for PACE sites C in 1993, 83% of PACE physicians were internists. The authors also bemoan the lack of geriatrics-trained physicians, although they believe that this may improve in the future as a result of ongoing efforts to increase the emphasis on geriatric care in medical schools. This is, perhaps, just as well, if Kane et.al. (1992, p.780) are right to identify geriatric training or a strong background in

community health, combined with an ability to work in a multidisciplinary setting, as the critical attributes required of PACE health professionals.

Enrollment C as discussed earlier, PACE enrollment has been lower than expected, at least during the developmental stages of many PACE sites and sites have found it difficult to retain enrollees when they make the transition to full capitation. Kane et.al. (1992, p.780) and Branch et.al. (1995, p.350) stress the importance of appropriate marketing strategies to counteract the problem that, at least initially, a PACE site is trying to promote a service that is new and in flux. Branch et.al. (1995, p.350) suggest that in-person contact, frequent follow up with prospective clients, and group presentations to referral agencies have proven more successful than mass media forms of advertising. Attracting middle-income, private-pay patients remains a problem because of the magnitude of co-payments and will likely remain so unless PACE is able to structure financing arrangements in a fashion that is attractive to this group (Eng et.al., 1997, p.231).

Housing C the importance of assisted housing to the functioning of at least some of the existing PACE sites was discussed earlier, when it was also remarked that housing services are not included in the Medicaid and Medicare capitation rates. As such, Kane et.al. (1992, p.780) argue that the development of housing facilities requires often complicated arrangements to access other sources of financing. The successful mobilization of such alternative financing for housing facilities may, therefore, be important in the development of at least some PACE sites.

Ethnicity and Care Giving C Eng et.al. (1997, p.228) note the emphasis that the PACE model places on talking to patients and obtaining advance directives in the course of delivering care. Section II.2.(v) above suggested that preliminary studies have found important differences between the way different racial/ethnic groups report their care preferences and their willingness to issue advance directives in the form of more formal, often legal, written instruments such as living wills. This suggests that Aone size does not fit all@ as regards the way that frail elders from different racial/ethnic groups should be cared for in a PACE environment and that PACE sites may need to pay attention to the ethnic composition of their enrollments if they are to maximize the well being of those for whom they are caring.

Is PACE a Panacea? C the simple answer to this question is no. This is not a criticism of PACE because the simple fact is that it is not designed to be a universal model of care for the elderly and/or those with disabilities. It is obvious from research to date (see section II.2.(iv) above) that, in terms of both health outcomes and quality of/satisfaction with life, those enrolled in PACE do very well indeed. There can really be little doubt that the Program benefits its enrollees. It should be noted, however, that these enrollees are a small percentage of the total population who require long-term care. Hence according to Don Sherwood of HCFA A[PACE is] not for everybody. Only 5% of those over 65 are frail enough to be eligible and many don=t want to change doctors or go to a day health center. But for those who want comprehensive care, it=s very good@ (New York Times, February 14, 1994, p.A11). This has prompted some researchers to worry less about the experiences of those who do enroll in PACE (which seem to be very positive) than the fates of those elders C the vast majority C who do not. For example, in reference to integrated acute and long-term care strategies in the US and UK, Leutz (1997, cited in Ashbaugh and Smith, 1998, p.3) suggests that whilst Afull integration may be very effective and even efficient for a few, it is important that while we are trying to identify who those few are, what they need, and how to provide and pay for their care, we should ensure that the many also get what they need@ (emphasis in original). These Amany@ are those with Amild and moderate impairments as well as stable medical and functional conditions that are unlikely to become unstable and need urgent attention, ... [who] may receive short-term or long-term specialized services but whose care is usually but not always routine@ (Leutz, 1997, cited in Ashbaugh and Smith, 1998, p.3). Of course, to the extent that federal and state governments are overpaying for PACE services (see discussion in section II.2.(vi) above), there is a second equity issue here. Specifically, we should be concerned not only with the disproportional allocation of time and planning resources towards those who benefit from PACE, but also the disproportional allocation of the health system=s financial resources towards PACE enrollees. It should be noted that this second concern is very much dependent on the extent to which PACE sites are, in fact, overpaid, something that is very hard to convincingly establish or refute given the current state of research. However, it may be that there are important lessons for federal and state government agencies as regards the financing of managed care and, in particular, the calculation of capitation rates, arising from experience with the operation of PACE (see also Branch, 1996, p.9).

3. Two Examples

The purpose of this section is to provide more detailed insights into the workings of two Amature@ PACE sites. The sites that have been chosen are On Lok Senior Health Services in San Francisco, California, and the Comprehensive Care Management (CCM) Program in the Bronx, New

York. One of these sites C On Lok C is the oldest PACE site, whilst the other only recently (during 1995) adopted full financial risk for its operations.

i) On Lok Senior Health Services, San Francisco, CA

As has been mentioned before, On Lok is the original PACE site. Traditionally, it has operated in a relatively small catchment area and its enrollees are predominantly Asian. In the 1994-5 programming year C the program=s tenth under its system of continuous waivers C On Lok was operating with 4 ADHC=s and 3 housing facilities (Zimmerman et.al., 1998, p.1). Data from the National PACE Association (1997, pp.11-14) suggest that in 1996, each On Lok enrollee attended the ADHC an average of 11.1 times per month. He/she experienced, on average, just over 2 encounter days per month with a physician or nurse practitioner for primary care, 10.3 encounter days per month with a nurse, 2.5 encounter days per month with a social worker, and approximately 7 encounter days per month with a physical or occupational therapist. He/she also averaged approximately 0.2 outpatient specialist consultations per month, 5 prescriptions filled per month and 36.6 hours per month of personal care in the home and home help. On Lok ended the 1994-5 programming year with a $750,000 surplus (Zimmerman et.al., 1998, p.1)

One of the main issues for the On Lok site in recent years has been uncertainty concerning the state determined Medicaid capitation rate, the calculation of which has been described as Aperennially clouded and elusive@ (Zimmerman et.al., 1998, pp.1-2). Ostensibly, the rate is set at 85% of the average cost of nursing home care. Since 1994, however, rates have been set based on realized program costs from the previous year, extracted by state officials from site rate proposals. Zimmerman et.al. (1998, p.2) report that On Lok is discussing with the state of California the possibility of defining a more predictable rate setting methodology C although they do also note that state officials believe that the current methodology is, On Lok=s objections notwithstanding, actually quite transparent.

On Lok=s main objective at the present point in time is expansion. In 1995, the program set itself the short-term goal of increasing enrollment from 428 to 530 by the middle of 1997 and the longer term goal of expanding its service area to cover the whole of San Francisco (Zimmerman et.al., 1998, pp.7-8). To this end, On Lok acquired in 1995 a new site in a predominantly Hispanic area of San Francisco. In a break with the past, during which it eschewed formal marketing and relied on word of mouth recommendations and high visibility within its relatively small catchment area, On Lok began planning formal marketing campaigns aimed, in particular, at middle-income (i.e., non-Medicaid eligible) and non-Asian elders (Zimmerman et.al., 1998, pp.4-5). On Lok has also begun to adjust its internal organization in response to actual and expected program expansion. A new staffing model was implemented in 1996 which involved most staff reporting to their ADHC manager rather than to a cross-center discipline supervisor. This was designed to resolve reported conflicts of allegiance amongst staff, to centers on the one hand and to their disciplines on the other. Plans were also laid to complement this move towards ADHC-based management by transferring budget responsibilities to individual ADHC=s (Zimmerman et.al., 1998, p.6). Whether or not On Lok=s long-term plans for city-wide expansion are realized will no doubt depend in some part on the

substantial competition for enrollees that it faces from California=s many other public and private managed care interests (Zimmerman et.al., 1998, p.3).

On Lok In Action: A Case Study C the following case study is reproduced in full from Zawadski and Eng (1988, p.80).

An 88-year-old woman, who lived alone and had Alzheimer-type dementia and a previous right hip fracture, was found on the floor of her apartment one morning by On Lok staff. She had fallen as she had arisen from her bed and subsequently was unable to get up or to summon help. She was admitted to the hospital by On Lok=s staff physician after X-rays revealed a displaced left hip fracture. Orthopedic consultation was obtained on admission. The following day, under spinal anesthesia, the patient underwent open reduction and placement of a hip prosthesis. Postoperatively, there were no complications. Physical therapy began the first postoperative day.

The patient was discharged from the hospital on the first postoperative day to On Lok=s respite care unit. For the next 2 weeks, she attended the day health center daily, where On Lok=s physical therapist supervised a treatment plan of progressive ambulation with the patient using a walker. The clinic nursing staff monitored the surgical wound for infection, and dressing changes were done daily. At the end of each day, she returned from the day health center to reside in the respite care unit, where she received personal care from home health aides in the evening and night. Personal care included helping her to learn to use a bedside commode safely at night. At the end of the 2 weeks, the patient was ambulatory with a walker, without hip pain, and her surgical wound was healed.

She was discharged from respite care back to her home, where increased hours of home care were arranged for her each morning and evening. She continued to attend the day health center 5 days a week for physical therapy. Although she was ambulatory with a walker, she was unable to negotiate the three flights of stairs to her walk-up apartment. She was accommodated by being carried up and down the stairs by two drivers until she could navigate the stairs herself.

ii) Comprehensive Care Management (CCM), Bronx, NY

The CCM program was developed in 1985 by the Beth Abraham Health Services (BAHS) agency, only subsequently evolving into a PACE site (Lee et.al., 1998, p.65). By the end of 1995, it had 331 enrollees (a drop of 78 from the previous year) serviced by 2 ADHC=s and access to BAHS=s three HUD 202 housing facilities (Zimmerman et.al., 1998, p.49). Data from the National PACE Association (1997, pp.11-14) show that in 1996 each CCM enrollee attended an ADHC an average of 7.0 days per month (the lowest of the 11 sites surveyed in National PACE Association, 1997) and experienced approximately 1 encounter day per month with a physician or nurse practitioner, fewer than 1.7 encounter days per month with a nurse, 0.4 encounter days per month with a social worker and fewer than 2 encounter days per month with a physical or occupational therapist. The average Bronx enrollee also experienced over 0.7 outpatient specialist consultations per month (by far the highest of the 11 PACE sites surveyed in National PACE Association, 1997), almost 6 prescriptions filled per month and 85.2 hours of home care and home making per month. In 1995, CCM generated a surplus of $1.7 million and by the end of the same year was holding $7 million in its risk reserves (Zimmerman et.al., 1998, p.49).

The growing CCM risk reserve has created a debate between CCM and New York state Medicaid officials in recent years, with regard to the appropriate capitation rate for the site. CCM itself proposed a rate reduction of 9% for 1995 and actually experienced a rate reduction of 11% (Zimmerman et.al., 1998, p.50). Despite this, the site still generated a surplus of $1.7 million, as reported above. Relations with the state of New York have not been helped by CCM=s claims that it has suffered from the hostile reaction to the state=s initiatives to cajole Medicaid beneficiaries into HMOs (Zimmerman et.al., 1998, p.50). There is also concern that after years of support, the state is now wont to believe that a panel of approximately 350 participants is too small to indicate that CCM demonstrates long-term viability in the state of New York (Zimmerman et.al., 1998, p.52).

In light of this latter concern and the necessity of competing with other managed care initiatives, CCM has identified Adramatic= program growth as a priority (Zimmerman et.al., 1998, p.51). To this end, following state approval of its desire to expand beyond the Bronx catchment area, CCM opened a new ADHC in Westchester, NY (with a capacity of 100 participants) in 1996, BAHS began the pursuit of a 71-unit HUD 202 housing facility in White Plains, and plans were laid for a third ADHC in the Bronx and a new ADHC in Manhattan (Zimmerman et.al., 1998, pp.51-2).

There have also been important developments in and changes to the organization of CCM in recent years. Unlike most other PACE sites, CCM has increased its primary care staff by recruiting physicians rather than nurse practitioners as it has grown C a development that Zimmerman et.al. (1998, p.53) attribute to the site=s not having encountered the same difficulties in recruiting physicians as other sites and the fact that the unionization of nurses in New York and the consequent increases in their salaries has eroded much of the financial incentive to hire nurse practitioners that existed prior to their unionization. Other changes include contracting with a mail order pharmaceutical company to provide prescription drugs C which is believed to make the tracking of participant=s drug consumption easier C and the adoption of a Acluster care@ home service model, whereby one home worker provides aid to a number of enrollees, all of whom live in the same area (Zimmerman et.al., 1998, p.53). There have also been changes designed to improve the efficiency of transportation C such as including the transportation co-ordinator in multidisciplinary team meetings C and to improve the efficiency of the multidisciplinary team meetings themselves. Previously, nursing staff were required to summarize cases before these meetings, whereas now, all disciplines are responsible for reviewing case details on a prior basis. This is believed to reduce the amount of time spent during multidisciplinary team meetings on the summation of cases and increase the amount of time devoted to their discussion (Zimmerman et.al., 1998, p.54).

CCM in Action: A Case Study C the following case study is reproduced in full from Lee et.al. (1998, pp.66-73).

Mrs. L.I., aged 89, lived in a unique senior citizens= complex, in which a staff of personal care workers provided 24-hour care for her and two other clients in a large apartment. Mrs. L.I. had chosen this housing arrangement instead of a nursing home, when her daughter had moved away and she did not want to go with her. This illustrates one of the possibilities that sites can develop within the PACE model.

Mrs. L.I. had a history of lacunar infarcts and hypertension, diabetes, mild dementia, was legally blind due to diabetic retinopathy, and had a severe hearing deficit. She had hypercholesterolemia, polyneuropathy and lumbosacral plexopathy with paraparesis.

For more than 3 years, she received assistance in all her activities of daily living. A nurse would see her daily to administer her insulin. With the help of a van provided by the PACE program, she came to the day health center for activities twice a week until the last few months of her life.

During her last year, she had developed progressive renal failure, and renal ultrasound showed no sign of obstruction. ACE inhibitor therapy had been stopped, but this did not improve her renal function. Her medical team felt that Mrs. L.I. had a relatively aggressive form of diabetic nephropathy with declining renal function and fluid retention.

Three months prior to her death, she was hospitalized for an initial episode of congestive heart failure. Evaluation by cardiac enzymes and ECG ruled out a myocardial infarction. Ecocardiography showed diastolic dysfunction, and she was discharged to care at home.

She declined renal dialysis after a discussion with her physician and health care proxy, during which she weighed the risks and benefits of the intervention and asked appropriate questions. She received increasingly frequent visits at home from her nurse and social worker to help manage her deteriorating medical condition.

During her last few weeks, she developed respiratory distress secondary to congestive heart failure. This was treated with oxygen and frequent adjustments of increasing doses of furosemide and then metolazone. She subsequently required IM furosemide and liquid morphine for comfort measures, which were provided by home care nursing. Her long-time personal care worker provided substantial psychosocial support through the last few days of her life, and she died in her home.

III PACE IN CONNECTICUT

The purpose of this section is to describe and evaluate the nascent Connecticut PACE Pilot Program. As will become clear, many of the details of this Program have still to be finalized. However, its intended design and objectives have already begun to emerge from the legislation that establishes the program and from discussion within the Department of Social Services (DSS), the state agency responsible for the Program=s administration.

1. The Connecticut PACE Pilot Program: A Preliminary Overview

Public Act 98-198, an Act Establishing a Pilot Program of All-Inclusive Care for the Elderly, passed in June 1998, mandates the establishment of a PACE Pilot Program in Connecticut beginning no later than July 1, 1998. The Act, which identifies the DSS as the state agency responsible for administering the Program, comprises just four sections, one of which is an amendment of Section 113 of Public Act 97-2, the Act that required the DSS to apply for a 1115 research and demonstration waiver. This waiver application (DSS, 1997) sought the establishment in Connecticut of a program called Connecticut LINC (Lifelong Integrated Networks of Care), the aims of which were to provide integrated preventative, acute and long-term care, of the sort currently available through either Medicare or Medicaid, through integrated service networks to a dual eligible population aged 65 and over. The integrated service networks were envisaged as taking a variety of forms, including (but not limited to) coalitions based around community organizations, provider groups, and HMO=s (DSS, 1997, pp. 1-2.) The principle objectives and target clientele of the Connecticut LINC were obviously very similar to those of PACE and it would appear from Public Act 98-198 that the Connecticut PACE Pilot Program has been conceived as a replacement for the aborted Connecticut LINC project.

Section 2 of Public Act 98-198 C which is an amended version of section 113 of Public Act 97-2 C stipulates the purposes of the Connecticut PACE Pilot Program as being the provision of comprehensive health care and care management services to a population of elderly and disabled Medicaid beneficiaries who may also be Medicare eligible. It also stipulates that Athe program shall be designed to reduce costs and increase efficiency in the operation of the Medicaid program and to improve the co-ordination of health care benefits with the Medicare program@ (Public Act 98-198, Section 2.(a)). The Commissioner of Social Services is instructed to contract with integrated service networks for the provision and management of health care to and for participating Medicaid beneficiaries. These integrated service networks must first successfully complete a feasibility study with the aid of a PACE Technical Center and are explicitly instructed by the Act to emphasize primary and community-based health care services and to avoid utilizing institutional (i.e., hospital and nursing home) care. Section 2 of the Act also prohibits integrated service networks from demanding co-payments from Medicaid-eligible enrollees and from offering lower levels of optional care than those offered by the traditional fee-for-service Medicaid program. Finally, it is stipulated that those eligible will be allowed to choose between receiving care from one of the integrated service networks and from the traditional fee-for-service Medicaid program.

The third section of the Act requires the Commissioner of Social Services to report to the Connecticut General Assembly on the progress of the PACE Pilot Program. A first report is required by January 1, 1999 and a final report by January 1, 2000. Regarding the content of these reports, Public Act 98-198 Section 3 stipulates only that the final report A...shall analyze the cost-effectiveness of the Pilot Program and shall include data on (1) the number of individuals served by the program, (2) the number and type of PACE program services offered under the program, and (3) the monthly cost per individual under the program.

It should be clear from the preceding description of Public Act 98-198 that many of the details of the Connecticut PACE Pilot Program have yet to be finalized. These specifics will only become clear once site proposals C which are currently being prepared by Hebrew Home and Hospital of West Hartford, and Masonic Home and Hospital of Wallingford C are received by the DSS. This will not likely occur until HCFA succeeds in finalizing its regulations for the operation of PACE at the national level. Even then, the proposals must undergo DSS scrutiny and be approved before the exact nature of the PACE model at its Connecticut sites becomes apparent. One thing that is clear, however, is that, at present, PACE is only a Pilot Program in Connecticut. Although, as mentioned in Section II.1.(iii), it is no longer necessary to obtain federal waivers in order to offer PACE Medicare and Medicaid beneficiaries, Public Act 98-198 clearly indicates that, in Connecticut, the PACE model is being tested as a method of providing integrated acute and long term care. Furthermore, despite the fact that there is, as yet, no operational PACE site in Connecticut, it is already possible to make some evaluative comments on the way that the Connecticut PACE Pilot Program is evolving. This is the purpose of the following section.

2. The Connecticut PACE Pilot Program: A Preliminary, Comparative Evaluation

In what follows, features of the nascent Connecticut PACE Pilot Program gleaned from Public Act 98-198 and information provided by the DSS are analyzed in the context of what is known about the structure and operation of the PACE model at its exiting sites, as discussed in section II of this report. Preliminary findings raise important questions C many of which, it should be said, are well understood by the DSS C about the APACE-ness@ of the emerging Connecticut PACE sites, the objectives of the Pilot Program, and on what basis the Connecticut PACE Pilot Program will be deemed fit to make the transition C if, indeed, it ever does C from a Pilot Program to a permanent provider of Medicare and Medicaid services in Connecticut.

i) How, If At All, Will Connecticut PACE Sites Differ From Other, Currently Existing, PACE Sites and Why?

According to DSS (1998a), there are six principle ways in which Connecticut PACE sites are likely to differ from those currently in existence elsewhere. First, the Connecticut sites will seek to expand consumer choice and self-direction beyond the levels that are typical of existing PACE sites. This will principally involve allowing Connecticut PACE enrollees to use their own (i.e., non-PACE staff) physicians and make less regular use of ADHC=s. Both of these are, in and of themselves, important departures from the PACE model as it currently exists and as such they are discussed in greater detail below.

Second, the Connecticut PACE Pilot Program will develop risk adjusted rates in order to calculate the capitation payments due to PACE sites. Recall from section II.1.(iv) above that there is considerable variation between states in the Medicare and Medicaid capitation rates that form the principle revenues of PACE sites. This variation is usually due to differences in the characteristics of those who reside within the various PACE site catchment areas and (in the case of the Medicaid rate) differences in the comparison group used in the calculation of the capitation rate. Usually, however, PACE sites receive only one capitation payment from each program (Medicare and Medicaid). But in Connecticut, multiple capitation rates will be calculated, designed to reflect Athe different clinical and functional risks of the participants@ (DSS, 1998b, p.37). Initially, the state will begin by calculating two rates C one for Aheavy-care@ clients and one for Alight-care@ clients (all of whom are, of course, nevertheless nursing home eligible) and sites will be remunerated accordingly, depending upon the composition of their enrollments. However, because Connecticut plans ultimately to expand the PACE-eligible population to include those who are not nursing home eligible (see discussion below), it is expected that more rates will be developed in the future to adjust for the less frail conditions of non-nursing home eligible participants (DSS, 1998b, p.14). At the time of writing, Connecticut=s risk-adjusted capitation rates are still in the process of being calculated. The motivation for developing these multiple, risk adjusted rates is to reduce any financial incentive that sites perceive in avoiding the enrollment of the frailest elders. This motivation is explicitly identified in DSS (1998b, p.38). Some of the research discussed in section II.2.(vi) above certainly identifies adverse selection as being a problem characteristic of existing PACE sites, although there is little or no suggestion in this literature that the problem comes about because of sites deliberately pursuing financial incentives associated with the enrollment of less frail elders. However, since adverse selection, whatever its motivation, does threaten the cost- effectiveness of PACE from the point of view of the public sector, Connecticut=s system of risk adjusted capitation rates may succeed in reforming PACE finances in a manner that addresses current criticisms of PACE financing, as identified in sections II.2.(vi) and II.2.(vii).

A third and major difference between the Connecticut PACE Pilot Program and existing PACE sites is that the former will seek to expand the PACE-eligible population well beyond those who have traditionally qualified for PACE (see also DSS, 1998b, pp.25-6). The initial target population during the first two years of the Connecticut PACE Pilot Program will be quite traditional, confined as it will be to dually eligible persons aged 55 and over, who are state certified as being nursing home eligible, and whose income is no greater than 300% of the Supplemental Security Income (SSI) level. However, modifications to the eligible population are planned for year 3 of the Program. First, the program will be opened to those who qualify for Medicare or Medicaid, but not both. Second, and more importantly, the Program will seek to expand the PACE eligible population beyond its traditional confines. This expansion will occur along two dimensions. First, the Program will be opened to all Aat risk@ elders, not just to those who have been state certified as nursing home eligible. Connecticut=s PACE sites may therefore become the first to provide services to frail but currently non-nursing home eligible elders who are considered to be at risk of future institutionalization C many of whom are, in Connecticut currently eligible for the state funded portion of the CHCP. Second, the Program will be opened to those whose income exceeds 300% of the SSI. Although not formally excluded from enrollment at existing PACE sites, this latter group is under-represented amongst the PACE-enrolled population because of the prohibitively high co-payments demanded of them (see discussion in section II.2.(ii)). The innovation that will see the Connecticut PACE Pilot Program attempt to expand its enrollment of this population is the introduction of a premium scale, that will enable those with incomes exceeding 300% of the SSI to receive some (but not full) state assistance towards the Medicaid costs of their PACE enrollment. These enhancements to the definition of the PACE-eligible population are explicitly facilitated by Public Act 98-198 Section 1.(a), which permits the definition of an Aeligible individual to be amended by waivers granted by the US Department of Health and Human Services@. Because it deviates from the definition of the PACE eligible population in Public Law 105-33 (the Balanced Budget Act of 1997, which grants PACE permanent provider status at the federal level), Connecticut is required to apply for federal waivers before proceeding to allow sites to actually enroll members of the expanded population described above. This federal waiver application is already in process. Whilst just 200 enrollees will be targeted during the initial phases of the Connecticut PACE Pilot Program=s operation, it is hoped to expand enrollment to approximately 2000 once the Program has been opened to a broader population after its second year of operation (DSS, 1998b, p.26).

If the waiver application is successful, the result will obviously be that the Connecticut PACE Pilot Program will cater to a population that is broader than that at any existing PACE site and will include elders who are less frail than any who are currently enrolled at existing PACE sites. One possible motivation for this development is that, following as it does in the wake of the failed Connecticut LINC project, the Connecticut PACE Pilot Program is essentially a replacement for Connecticut LINC rather than a Apure@ PACE project in its own right. It is therefore designed to create a distinct health care system for dual eligibles rather than one that caters to the traditional PACE population of the frailest elders. This possibility and its implications are discussed in greater detail in the section that follows. Another possibility is that Connecticut is alive to the problems that existing PACE sites have had with enrollment. As indicated in Section II.2.(ii), existing PACE sites remain small, the average number of enrollees per site in 1996 being just 293. And, as the case studies in Section II.3 suggest, sites such as On Lok and CCM are very much alive to the need to increase the size of their enrollments C in the face of what appears to be growing state ambivalence regarding the long-term viability of the PACE model in the case of CCM. The proposed expansion of the PACE eligible population in Connecticut may therefore rejuvenate the PACE model if the Connecticut PACE sites, by appealing to a broader audience, are able to avoid the enrollment problems that have bedeviled existing sites. On the other hand, and whatever the precise motivation for seeking to increase the size of the PACE-eligible population may be, the Connecticut PACE Pilot Program must be aware of the problems that may arise as a result of its attempting to expand PACE enrollment beyond the traditional PACE target clientele. As reported in Section II.2.(iv) both Kidder (1996, p. 5) and Chatterji et.al. (1998, p. 48) report that the beneficial effects of PACE on health outcomes and life satisfaction are most noticeable amongst the frailest nursing-home eligible elders. The question arises, then, as to whether or not the PACE model will actually work C i.e., better the health outcomes and life satisfaction C for the less frail elders to whom the Connecticut PACE Pilot Program is seeking to open PACE enrollment. Furthermore, as discussed in Section II.2.(vi), even authors who accept the cost-effectiveness of PACE have suggested that this cost-effectiveness may be related to the small size of PACE enrollments. Hence, Irvin et.al. (1993, p. 58) argue that it is, at present, impossible to know whether or not the financial successes of PACE that these authors report can be replicated with substantially larger enrollments.

Perhaps the most dramatic departures of the Connecticut PACE Pilot Program from the traditional PACE model of care delivery that are reported in DSS (1998a) are those indicated earlier: its intent to de-emphasize the role of the ADHC and to experiment with the use of community (as opposed to PACE staff) physicians. Hence DSS (1998a) refers to Aminimizing@ the ADHC requirement, which may involve requiring enrollees to make only one ADHC visit per month to consult with their physician, or possibly to abandon ADHC attendance as a PACE requirement. Furthermore, the Connecticut PACE Pilot Program will test relaxing the requirement that PACE enrollees receive (almost) all of their health care from PACE staff in favor of allowing enrollees to keep their own physicians upon enrollment. Specifically, community practitioners will be allowed to arrange with Connecticut PACE sites to continue in their capacity as a Connecticut PACE Pilot Program enrollee=s primary care physician (DSS, 1998b, p.21). Given that Branch et.al. (1995, p.351), as discussed in Section II.2.(ii), identify a distaste for regular ADHC visits and an unwillingness to forego provider choice as two of the three most important barriers to enrollment amongst the PACE-eligible population, these developments may help Connecticut PACE sites to avoid the enrollment problems that existing sites have suffered. However, it must also be noted that changing the ADHC requirement and the relationship of physicians to the rest of the multidisciplinary team involve quite fundamental revisions to the traditional organization of a PACE site C revisions that may threaten the very functioning of the PACE model itself. Recall from Section II.1.(iii) that two of the outstanding features of the PACE model of care delivery were identified as being the requirement that enrollees attend an ADHC, and the use of a multidisciplinary team C which includes staff primary care physicians C to plan, supervise and deliver care. The ADHC was described as being the geographical Afocal point@ of the PACE model, allowing PACE staff to continuously monitor enrollees= health status, serving as the main setting for care delivery and also providing a social focus for program participants (Kane et.al., 1992, p. 772; Wiener and Skaggs, 1995, p. 15). Meanwhile, the multidisciplinary team C in which staff physicians participate as equals with other medical and non-medical PACE staff members C was described as the primary instrument for designing, overseeing and delivering care to enrollees (Kane et.al., 1992, p. 772), the main perceived advantage of this team organization being the capacity for information sharing amongst team members that it creates (Eng et.al., 1997, p. 327; Lee et.al., 1998, p. 66). The question that arises, then, is whether or not Connecticut PACE sites will be able to successfully function as PACE sites if they minimize the ADHC requirement and allow PACE enrollees to use non-staff physicians? The answer would seem to depend on how well the Connecticut sites are able to integrate non-staff physicians within their multidisciplinary teams and the extent to which they are able to find an alternative (to the ADHC) mechanism for monitoring their enrollees= health. Hence the DSS fully expects the multidisciplinary team to be a central feature of the proposals it receives and for team members with the exception of an enrollee=s primary care physician to be PACE staff. The challenge, then, is to integrate the non-PACE physician into the multidisciplinary team C a challenge which the DSS expects sites to take up in their proposals. Meanwhile, Aminimizing@ the ADHC requirement may not mean abolishing it altogether C although the possibility of requiring Connecticut PACE enrollees to make just one site visit per month would still mean that Connecticut enrollees would be making far fewer ADHC visits per month than enrollees at any currently existing site. Hence recall from section II.2.(iii) that, at present, PACE enrollees attend their ADHC=s an average of 11 times per month, and although there is substantial variation in attendance rates between sites, the lowest current attendance rate at any site (recorded at CCM in the Bronx) is 7 days per enrollee per month (National PACE Association, 1997, p.11). One possibility would be to enhance the role of the home health aid within the PACE model. Indeed, the further development of assisted living within the PACE model is the sixth and final Connecticut specific feature of the Connecticut PACE Pilot Program listed in DSS (1998a). This feature appears to be motivated precisely by recognition of the need for Connecticut PACE sites to fulfill the enrollee-monitoring function of the ADHC by means of some alternative mechanism. Exactly how the functions of the ADHC are served if Connecticut sites do,