Health Affairs, 22, no. 5 (2003): 97-109 doi: 10.1377/hlthaff.22.5.97 © 2003 by Project HOPE	New Online   Obama's Success & Next Steps   3 Health Policy Options for President Obama   A Hybrid Proposal   Bipartisan Compromise

This Article Abstract Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when Comments are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via ISI Web of Science (9) Citing Articles via Google Scholar Google Scholar Articles by Lave, J. R. Search for Related Content PubMed PubMed Citation Articles by Lave, J. R. Related Collections Hospitals Medicare Mental Health/Substance Abuse

Medicare Reform

Developing A Medicare Prospective Payment System For Inpatient Psychiatric Care

PROLOGUE: On some issues—disparities, for example, or quality—the policy challenges in mental health are not essentially different from those in the general medical sector. On others, such as parity or the role of the states in providing services, mental health is in a unique position. At times, though, the gravitational fields of the general and the particular intersect. Long-deferred efforts to develop a prospective payment system for inpatient psychiatric care in Medicare represent such a hybrid case.

As the following discussion by Judith Lave demonstrates, diagnosis-based payments for psychiatric inpatients do an inadequate job of capturing variations in treatment modalities and costs. A DRG system like that used for most Medicare hospital payments would whipsaw psychiatric facilities between unpredictable extremes of over- and underpayment—and invite systematic exploitation and abuse. Lave details the reasons why variations in psychiatric treatment are even more pronounced and changeable than those observed in medical care by John Wennberg and his colleagues. But the logical alternatives to diagnostic approaches tend to be data-intensive and add to administrative burdens as they increase in their statistical power to predict actual variations in care and cost.

Rapid changes in treatment norms and the organization and financing of mental health services have compounded the difficulties of creating a new system: fast-growing use of psychotropic medications, drastically reduced inpatient lengths-of-stay, and the rapid rise of managed behavioral health care, to name a few. Similar care is now provided at different types of sites with different costs, which is both a motive for creating a new payment system and an obstacle to setting the baseline for it.

Lave is a professor of health economics, chair of the Department of Health Policy and Management, and codirector of the Center for Research on Health Care at the University of Pittsburgh. Her doctorate in economics is from Harvard University, and she has served for several years on the Medicare Payment Advisory Commission.

Abstract

 
Under the Balanced Budget Refinement Act (BBRA) of 1999, the secretary of health and human services was mandated to implement a prospective payment system (PPS) for psychiatric inpatient facilities that were exempt from the Medicare inpatient PPS. This paper reviews the reason for the initial "distinct-part" exemption, describes research that has been conducted to inform the development of a psychiatric inpatient PPS, and examines some of the issues that must be addressed as a PPS is designed. In addition, some changes in the overall inpatient psychiatric hospital sector are discussed.

Medicare now pays for inpatient psychiatric care under two sets of rules depending on the setting: (1) It pays for care in scatter beds under the Medicare hospital inpatient prospective payment system (PPS); and (2) it pays for care in units and psychiatric hospitals under TEFRA (the Tax Equity and Fiscal Responsibility Act of 1982) rules. However, the Balanced Budget Refinement Act (BBRA) of 1999 mandated that the secretary of health and human services (HHS) develop a per diem PPS for inpatient hospital services provided by psychiatric hospitals and units and that its patient classification system reflect differences in patient use and costs among such hospitals; this PPS was to be implemented by 1 October 2002.² Although the implementation deadline was missed, the Centers for Medicare and Medicaid Services (CMS) is developing regulations for a per diem PPS.

In this paper I focus on the way Medicare pays for inpatient psychiatric care and some of the issues that must be addressed as a PPS is developed. I begin with a brief discussion of the recent major changes in inpatient psychiatric care. I then discuss briefly the history of Medicare payment for psychiatric inpatient services and the reasons why psychiatric facilities were exempted from the inpatient PPS. I discuss some issues surrounding the development of a PPS for psychiatry, and I describe the main thrusts of the current research in this area. Finally, I lay out some of the decisions that the CMS will have to make as it moves ahead.³

Recent Changes In Inpatient Psychiatric Care

Top Recent Changes In Inpatient... Medicare Payment For Psychiatric... Developing A PPS For... Current Research On Developing... Moving Ahead With Prospective... Editor's Notes NOTES

 
The inpatient psychiatric care sector has undergone remarkable changes, with expansion in the late 1980s and early 1990s and contraction in the late 1990s. The number of psychiatric hospitals increased from 579 in 1988 to 702 in 1994 and then decreased to 483 in 2002. Units followed a similar pattern: They increased from 969 in 1988 to 1,490 in 1998 and then decreased to 1,410 in 2002.⁴ The number of psychiatric admissions to hospitals increased, but lengths-of-stay fell. For example, average length-of-stay in hospitals reporting to the National Association of Psychiatric Health Systems fell from 25.6 days in 1990 to 10 days in 2000.⁵ These changes were stimulated by many factors, including changes in types of treatments available (new drug therapies, partial hospitalization, residential facilities) and in treatment philosophy (for example, treatment in the least restrictive setting); the extent of insurance coverage; and the growth of managed care.

As is well known, managed care spread in the 1980s in response to rapidly rising health care costs. Separate institutions emerged to manage the delivery of behavioral health care services. Today, most behavioral health care in the United States is managed by specialized managed behavioral health care organizations (MBHOs) under "carve-out" contracts.

There has been much research on the effects of managed care.⁶ Most researchers have found that carving out behavioral health care was associated with sizable decreases in overall spending on behavioral health care services. MBHOs achieved these savings by lowering payment rates (usually, negotiated per diem rates), controlling admissions, and aggressively managing lengths-of-stay. Although the growth of managed care has affected the overall health care system, it has arguably had stronger effects on the behavioral health care sector.⁷

As a consequence of these changes, providers of inpatient psychiatric services are under stress. This stress will be exacerbated by the likely cutbacks in the Medicaid program resulting from current state budget problems.⁸ At the same time, Medicare, which accounts for approximately 30 percent of the revenues of psychiatric institutions, is proposing to implement a new payment system while "the psychiatric inpatient delivery system...is already fragile and beset by problems," according to the American Psychiatric Association.⁹

Medicare Payment For Psychiatric Inpatient Services: Historical Context

Top Recent Changes In Inpatient... Medicare Payment For Psychiatric... Developing A PPS For... Current Research On Developing... Moving Ahead With Prospective... Editor's Notes NOTES

 
Medicare’s enabling legislation mandated that hospitals be reimbursed for the reasonable costs incurred in providing services to beneficiaries. Between 1965 and 1982 program costs increased rapidly, and cost-based payment was considered to be a contributing factor. The first major changes in reimbursement method were implemented under TEFRA in 1982. TEFRA established both limits on hospital costs per discharge and target rates for cost increases. The second major change took place in 1983 with the passage of the inpatient PPS. Under the PPS the case is the unit of payment, and all cases were originally classified into one of 486 diagnosis-related groups (DRGs), of which fifteen were psychiatric DRGs (nine were for mental disorders and six for substance abuse disorders).

As the PPS was being designed, there was intense pressure to exempt certain types of providers from the new system. Many in the psychiatric community argued that psychiatric hospitals and units should be made exempt because the new system was unlikely to pay these facilities fairly for the care they provided. Their arguments were based primarily on three interrelated factors.¹⁰ First, different types of facilities had developed to serve the diverse needs of the mentally ill.¹¹ These needs included clinical needs (defined in terms of diagnosis, level of impairment, and treatment requirements, including treatment failure), social needs (defined in terms of the patient’s ability to cope with daily living), and legal needs (in terms of the patient’s and society’s need to be protected from harm). Diagnosis (the basis of DRGs) did not adequately capture patients’ treatment needs. Second, almost no data from psychiatric hospitals had been used in developing DRGs. Third, psychiatric DRGs accounted for much less of the variation in costs and lengths-of-stay for psychiatric cases than for medical-surgical cases. Furthermore, the coefficients of variation for the costs or lengths-of-stay for almost all psychiatric DRGs were higher than for most medical and surgical DRGs.¹²

Psychiatric hospitals and units that met certain requirements were exempted from the PPS. Patients treated in scatter beds or in units that did not seek or were not granted an exemption were paid under PPS rules.¹³ Medicare pays for care provided in exempt facilities under TEFRA rules. It establishes both a baseline cost and a target cost per discharge (determined by raising baseline costs, or prior target amounts, by an update factor set by legislation) for each hospital. Initially, exempt providers were paid their actual costs per discharge as long as it did not exceed the target amount. They also received bonus payments if operating costs were less than the target and some relief payments if operating costs exceeded the target. The structure of the TEFRA system was modified slightly over time. The most important changes were made under the Balanced Budget Act (BBA) of 1997, when the method for establishing the baseline costs for new providers was changed; a payment cap set equal to 75 percent of the target rate was instituted; and a variable update factor, one that varied with the difference between the provider’s actual costs and the target amount, was implemented.¹⁴ The cap was eliminated 1 October 2002.

TEFRA payment policy also contributed to the changes in the overall psychiatric sector noted earlier. The original method for setting payment rates for new providers encouraged new providers to enter the system and probably contributed to the expansion observed between 1988 and 1997.¹⁵ The changes in payment method mandated by the BBA reduced those incentives considerably. In addition, since TEFRA essentially set a payment per discharge, exempt providers had incentives to reduce lengths-of-stay. Between 1986 and 2000 the length-of-stay of patients treated in scatter beds (under PPS rules) decreased by 39 percent, while those of patients treated in units, nongovernment psychiatric hospitals, and public psychiatric hospitals fell by 36 percent, 45 percent, and 0 percent, respectively. In calendar year 2000 there were 645,146 Medicare psychiatric discharges, of which 33 percent were from general hospitals, 50 percent from units, and 17 percent from psychiatric hospitals (5 percent were from public hospitals). This meant that more than two-thirds of psychiatric discharges were paid for under TEFRA.¹⁶

Developing A PPS For Psychiatry

Top Recent Changes In Inpatient... Medicare Payment For Psychiatric... Developing A PPS For... Current Research On Developing... Moving Ahead With Prospective... Editor's Notes NOTES

 
There has been extensive research related to the development of PPS for psychiatric inpatient care. This research can be classified into three interrelated areas: patient classification systems, response to payment incentives, and system design. I first discuss the older research and then look at some recent research directly related to the development of a per diem system.

Patient classification systems. It is generally agreed that if patient classification is to be used as the basis of a PPS, the groups (1) should be medically meaningful; (2) should be homogeneous with respect to resource requirements; and (3) should be based on criteria that cannot be gamed.¹⁷

Under any fixed-price payment system, hospitals are exposed to two types of risks: random and systematic.¹⁸ Random risk occurs when a hospital admits more (or less) profitable cases within a category as a result of "luck of the draw." Systematic risk occurs when a hospital treats patients within a category who are systematically more (or less) profitable either because they traditionally attract more expensive patients (because of special expertise, geographic location, or traditional referral patterns) or because they engage in practices that actively encourage or discourage the admission of certain types of patients. The less homogeneous the patients are who are assigned to a particular group, the higher the risk.

There has been extensive research on the development of psychiatric classification systems for payment purposes.¹⁹ Researchers first worked on classification systems in which the unit of payment was the case. Many researchers restricted their potential classifying variables to those found in the Medicare claims data (the Uniform Hospital Discharge Data Summary)—the same variables that were used to construct DRGs.²⁰ Some researchers, such as Susan Horn, collected additional information from the medical record, to construct indicators of patient severity.²¹ A few researchers, such as Marie Ashcraft and colleagues, collected additional patient-level information on patients’ symptoms and severity.²²

In a review of the various classification systems that had been developed by 1987, Connie Horgan and Steve Jencks found that the majority of the systems accounted for less than 8 percent of the variation in resource use or length-of-stay.²³ Horn’s proposed severity-of-illness system had the most explanatory power, but the most powerful variable was one called "response to treat," which was not usable in a payment system. In 1999 the Medicare Payment Advisory Commission (MedPAC) summarized the state of this research at that time as follows:

Since the Congress implemented TEFRA, researchers have explored the potential of several classification systems for psychiatric patients. Work in that area has reaffirmed the inadequacy of DRGs alone to account for resource variation across psychiatric patients and has resulted in more comprehensive diagnoses-based designs that incorporate additional patient characteristics...Classification research also has revealed the difficulties of predicting resources use of both acute care and chronic care patients within a single design.

While designs that predict resource use during inpatient stays have potential for acute care patients, outlier mechanisms or systems that measure per diem resources are necessary to classify patients with extremely long stays. Indeed, resource use and practice patterns vary greatly between facilities that predominantly treat one or other of these patient types, and a psychiatric case-mix classification system may be possible; however, much work remains.²⁴

Analysts believe that there are two primary reasons why developing patient classification systems for psychiatry is difficult. First, most researchers have developed such systems using diagnosis as the primary classifying variable. However, the diagnosis of psychiatric disorders is complex, and, as noted above, treatment strategies do not uniquely line up with diagnoses. The cost of treating the patient will depend in part on the reason for hospitalization and presentation of symptoms. This type of information is not available in administrative data.

Second, it is difficult to ascertain the cost of treating an individual patient. Under the PPS a relative weight is assigned to each DRG, and this weight indicates how much more a hospital will receive for taking care of a patient classified into a particular DRG relative to a patient classified into another DRG. These weights in turn are based on hospital charges. For example, Medicare pays hospitals about two times more for patients in DRG 111 (major cardiovascular procedures without cardiac catheterization) than for patients in DRG 89 (simple pneumonia) because, across hospitals and patients, hospitals charge about four times more for taking care of the former than they do for the latter, even though the length-of-stay for DRG 89 is about 35 percent longer. They charge more because these patients spend more time in the ICU, use the operating suite, and receive more ancillary services. However, if hospitals charged the same amount per day for each patient, the relative weights assigned to these two DRGs would almost be identical.

This is a problem for psychiatric cases. Many services that psychiatric patients receive are not billed for separately. Thus, although staff time devoted to individual patients may vary, all patients treated in a given hospital unit will receive the same room charge and would appear to be equally costly. Hence, a case-mix system must be developed to account for variation in cost across patients.

Response to financial incentives. Any change in the method of payment will change providers’ financial incentives. All payment systems have built-in financial incentives that reward or penalize certain behavior. For instance, under a case-mix-based system, all hospitals that treat cases of a given type will receive approximately the same payment.²⁵ This should encourage providers to increase the efficiency with which they deliver care—that is, to lower the cost of care without impairing quality. However, providers can respond to these incentives differently: They may discharge patients sooner than appropriate, transfer patients to other settings, cut back unnecessarily on services, or engage in practices to encourage the admission of less costly patients within a group. In addition, if the payment is higher than the cost per case, there will be incentives to increase admissions. Similarly, if hospitals are paid per day, there will be incentives to decrease the daily intensity of resources used. Furthermore, if the per diem payment is higher than marginal cost, there are incentives to increase length-of-stay.

The results of research on responses to financial incentives suggest that providers’ responses might be stronger for psychiatric than for general medical and surgical services.²⁶ Several explanations have been offered for this differential response: (1) There is more waste in average psychiatric inpatient care and thus more room for improvement; (2) there is a wider range of acceptable treatment practices, and providers respond to financial incentives to provide the most appropriate care given the financial incentives; or (3) psychiatric cases are more amenable to undertreatment because quality is more difficult to measure.

PPS design. There also has been extensive research on the design of prospective payment systems in general and those for psychiatric hospitals in particular.²⁷ Given the problems inherent in designing adequate patient classification systems and the observed strong behavioral response, many researchers have focused on methods to mitigate the strong incentives inherent in a strict case-based system. The case-based system is a "high-powered" payment method in that providers face strong incentives to control the cost of care; these incentives would be mitigated under lower-powered payment systems, in which providers received additional payments for providing more services.

In this respect, analysts have proposed that hospitals be paid differently for outliers (cases that have notably higher or lowers costs); be paid blended payments (a fixed amount per case and an additional amount each day of care to cover some variable costs); or be paid per day.²⁸ For example, Richard Frank and I have argued that Medicare should consider a payment system based on a series of prospectively set per diem rates.²⁹ We anticipated three blocks of rates: The first would be a relatively high per diem reimbursement rate that would cover the costs associated with prevailing clinical practice that assesses and stabilizes condition and develops a treatment plan during the initial part of the stay. Subsequent days would be reimbursed at a reduced rate for a prescribed number of days, followed by a further reduction in the per diem rate on the assumption that the per diem use of resources decreases with the day of stay. We recommended that the per diem rate decline with the day of stay to mitigate incentives to increase length-of-stay.

Current Research On Developing A Medicare Per Diem PPS

Top Recent Changes In Inpatient... Medicare Payment For Psychiatric... Developing A PPS For... Current Research On Developing... Moving Ahead With Prospective... Editor's Notes NOTES

 
Four ongoing research projects should provide information for developing a per diem PPS. Three address the two reasons why it has been difficult to develop a classification system for psychiatric cases, as discussed above. The fourth proposes developing a per diem PPS using available administrative data.

Collecting patient information. The CMS has funded a study led by Fries to develop a survey instrument to collect patient information that could be used to classify psychiatric patients for payment purposes.³⁰ Following an extensive review of the literature and input from experts, Fries and his colleagues developed a draft instrument to collect information such as legal status, depressive symptoms, psychotic symptoms, danger to self and others, and activities of daily living (ADL) and instrumental activities of daily living (IADL) limitations. Research plans call for this information to be collected in a number of sites and to be linked with Medicare claims data, to determine whether researchers can identify patient groups that account for variation in average Medicare daily charges or daily costs (where the per diem costs would be determined by multiplying covered hospital charges by the Medicare cost-to-charge ratio and dividing by the covered length-of-stay).

Feasibility of per diem payment. The CMS also has funded Health Economics Research (HER) to work on the feasibility of developing a per diem PPS for psychiatric facilities excluded from the inpatient PPS. HER has developed a method for collecting data on staff time devoted to monitoring or taking care of individual patients over a twenty-four-hour period. In a pilot study, the group collected data on daily staff time spent in patient care–related activities for all staff (registered nurses, mental health specialists, caseworkers, and therapists) who interact with hospitalized patients.³¹ They developed a per diem patient resource intensity ("daily cost") indicator by costing the time allocation and linking it to individual patients. In addition, they gathered limited additional patient-level information such as Diagnostic and Statistical Manual, Fourth Revision (DSM-4) multiaxial assessment diagnoses, prior hospitalization, suicidality/combativeness, medication compliance, legal status, and global assessment of functioning.

HER has conducted some preliminary analyses to determine how "daily cost" varies with specific patient and hospital characteristics. They find considerable variation in these costs across patients; however, psychiatric diagnosis accounts for only 2 percent of it. They conducted multivariate analyses of the factors influencing the daily costs of 200 patients (patient days). This model accounted for about 30 percent of the variation in daily costs. However, the variables that accounted for most of the additional variation were day of the week (daily costs were significantly lower on Saturday and Sunday) and type of unit (geriatric unit versus other)—variables unlikely to be used in a payment system. The coefficients of the additional variables were statistically significant but quite small.

Estimating daily costs. John Hirdes and colleagues have completed a major project that both collected additional information on patients and developed estimates of daily costs.³² Specifically, thirty-four psychiatric facilities in three Canadian provinces volunteered to participate in the study by implementing the Residential Assessment Instrument–Mental Health (RAI-MH) and by participating in a staff time measurement study in one or more of their units.³³ Time was costed using wage rate information. The team obtained RAI-MH data and daily cost estimates on 1,998 patients and used decision-tree analyses to derive a patient classification system. The resulting System for Classification of In-Patient Psychiatry (SCIPP) contains forty-seven groups with cost weights ranging from 0.26 to 2.17.

Hirdes and colleagues found that the SCIPP accounts for about 29 percent of the variation in daily costs. Diagnosis alone, they found, accounts for only 2.4 percent of this variation. They also found a relationship with the day of stay—three of the most costly groups comprised patients who had been in the facility for up to four days, while most of the least costly groups comprised patients who had already stayed in the hospital more than ninety days.

Setting prospective payment rates. The fourth research project is being conducted by the Health Economics and Outcomes Research Institute (THEORI) of the Greater New York Hospital Association in cooperation with the American Psychiatric Association (APA).³⁴ THEORI has developed a payment methodology for setting prospective payment rates using standard administrative data. Specifically, THEORI specifies a regression model in which cost per day (defined as Medicare-covered charges times the Medicare cost-to-charge ratio for the psychiatric in-patient facility, divided by covered days) is the dependent variable; independent variables include hospital characteristics (area wage index, disproportionate-share indicator, interns and residents per bed); patient factors (age, sex, psychiatric DRG, comorbidity indicator); and length-of-stay variables. This model accounts for about 20 percent of the variation in per diem costs.³⁵ THEORI then uses the coefficients from the estimated regression equation to set per diem payment rates.

The results from the THEORI analysis imply that the per diem payments would vary with patient diagnosis (the coefficients on the DRGs and comorbidity indicators are small but significant) and with the day of the stay. In addition, there would be a redistribution of Medicare payments across types of facilities—relative to current law, payments to units would decrease and payments to government hospitals would increase.³⁶

The THEORI research indicates that administrative data could be used to develop a per diem PPS that would account for some of the differences in the types of patients treated by exempt facilities. HHS Secretary Tommy Thompson, in reviewing this work, stated "an inpatient psychiatric PPS, which relies on currently available administrative data, appears feasible."³⁷ According to Thompson, the CMS is performing further analyses of these data.

The basic question is how much better a per diem system would be if the patient classification system were based on patient characteristics not captured on the discharge record and if the weights were determined in part through staff time studies. It is possible to get some insight into this question by examining the research discussed above. First, although the results are preliminary, both HER and the Hirdes study found that diagnoses account for only 2–3 percent of the variation in daily costs, even when the costs are determined by careful costing methods. Thus, it probably does not make sense to invest in costing studies without also investing in gathering better patient information. Second, it unlikely that the collection of more patient-level information will lead to major improvements in a case-mix classification unless it is joined with costing studies that distribute the routine daily costs across individual patients. (However, we will know more about this once Fries and colleagues have completed their research.)

Finally, we can get a preliminary assessment of the value added in conducting a large research project that gathers better information by carefully reviewing the Hirdes findings. In that study, mean daily costs were $42.24; the range across the SCIPPs was $26.62 to $91.55. However, it is unlikely that any patient covered by Medicare would fall into the lowest SCIPP categories. First, many patients assigned to the lowest-weighted SCIPP categories had been in the facility for more than ninety days and are unlikely to be representative of Medicare-covered inpatients. Second, the costs studied were only a subset of the costs of caring for psychiatric inpatients—they do not include food, drugs, linen, housekeeping, and other ancillary services; overhead; and so forth. The smaller the proportion of costs that directly vary with patient characteristics, the less important it is to measure those patients and costs accurately.

I am pessimistic that current research on case-mix classification systems and improved methods to allocate staff time across patients will have important payoffs in terms of improving the overall payment system. Given that the payment rates would be based in part on special time and motion studies, there would be no easy way of updating them. Although the CMS should continue with its current research agenda, it will need to assess whether any improvements over using available administrative data are worth the costs associated with gathering the additional data. As an aside, this concern is not relevant to Canada, where the RAI-MH will be used for all patients and will be used initially for planning and quality-related purposes.

Moving Ahead With Prospective Payment

Top Recent Changes In Inpatient... Medicare Payment For Psychiatric... Developing A PPS For... Current Research On Developing... Moving Ahead With Prospective... Editor's Notes NOTES

 
It is possible to move forward in developing a per diem psychiatric PPS using administrative data. Obviously, many decisions need to be made; however, some decisions, such as how to adjust for geographic location and teaching status and how to phase in the new system, are similar to those made in designing other prospective payment systems.³⁸ However, some new, purely technical issues need to be resolved, such as the best way to analyze administrative data to determine how to vary the per diem payment rate with the day of the stay. The CMS will need to address specific problems; I consider four of them here.

First, the work by THEORI indicates that a per diem PPS will lead to a reallocation of Medicare payments across different types of psychiatric facilities. To what extent should this redistribution be mitigated? For example, should the CMS pay a higher per diem rates for patients treated in units? Other things equal, per diem costs are about $122 higher in units. Costs could be higher for a number of reasons, including the following: (1) Hospitals allocated a higher proportion of overhead costs to units; (2) units are more inefficient in their use of resources; (3) patients need more intense care; and (4) the treatment process is more intense.³⁹ An administrative pricing system should not adjust for the first two factors, should adjust for the second, and should adjust for the third only if there are measurable differences in the quality of care provided. More research is needed to understand the difference in costs.⁴⁰

Second, some hospitals treat psychiatric patients in both units and scatter beds. Whereas under TEFRA the basis of payment for the two groups of patients was similar (the case), under the new system Medicare will pay for care in units and in scatter beds on the basis of the day and the case, respectively. This could provide an opportunity for gaming the system.⁴¹ The CMS will need to monitor the distribution of patients across units and scatter beds within general hospitals and to support research to examine decision making with respect to treatment location.

Third, although the per diem PPS will be designed to mitigate incentives to increase length-of-stay, such incentives could still exist. If length-of-stay increases, it is likely that Congress will respond by lowering per diem payment rates. This response could have serious consequences for facilities that controlled their lengths-of-stay, which in turn could respond by reducing the quality of care. Therefore, the CMS will have to closely monitor length-of-stay and to examine the pros and cons of introducing utilization management.

Fourth, the impact of a per diem PPS will depend not only on the structure of the system but also on the level of payment. Under current law, the base payment rate will be set to make overall payments budget-neutral. Neither MedPAC nor the CMS has published data on the current financial margins of exempt facilities. Therefore, it is not possible to assess the full financial implications of these changes without more information.

Medicare is likely to change the way that it pays for inpatient psychiatric care at a time when the industry is undergoing much stress. Moving away from the TEFRA-based system that it pays for inpatient has several advantages. The new system should be slightly more sensitive to the characteristics of the types of patients treated in the different facilities, and payment rates should be more equitable across hospitals. Psychiatric facilities could see some advantage to Medicare’s paying for services on the same basis that most other payers use (per day, not per case).

However, the new payment system should be phased in, to minimize disruption and to give hospitals time to adjust. In addition, the new system should be supported with research that leads to a better understanding of the factors that influence and are associated with the distribution of patients across facilities, and the types of treatments used within hospitals. Finally, hospitals’ responses to prospective payment should be carefully monitored so that policymakers can assess alternative approaches to responding to their behavior. Under an administrative price-setting system, it is difficult to be sensitive to the peculiar conditions facing individual hospitals—those conditions that in a competitive market would lead to higher or lower prices. In this sense, the prospective payment systems implemented for the Medicare program are very different from the prospectively set rates negotiated between a managed care organization and providers.

Editor's Notes

Top Recent Changes In Inpatient... Medicare Payment For Psychiatric... Developing A PPS For... Current Research On Developing... Moving Ahead With Prospective... Editor's Notes NOTES

 
The author thanks Philip Cotterill, Richard Frank, Karen Heller, Sally Kaplan, and Christopher Vaz for their comments on an earlier draft of this paper.

NOTES

Top Recent Changes In Inpatient... Medicare Payment For Psychiatric... Developing A PPS For... Current Research On Developing... Moving Ahead With Prospective... Editor's Notes NOTES

 

1. Many health insurance plans impose limits on the number of covered days (both annual and lifetime). Medicare has a 190-day lifetime limit for care provided in freestanding psychiatric hospitals.
2. The statutory language is found in Section 124 of the BBRA, P.L. 106-113.
3. See T.G.Thompson, "Report to Congress: Prospective Payment System for Inpatient Services in Psychiatric Hospitals and Exempt Units" (Washington: HHS, 2002).
4. The term "units" refers to psychiatric units that are exempt from the PPS. For both units and psychiatric hospitals, the 1988 data come from the 1988 HCFA annual statistics and were provided by Philip Cotterill of the CMS; the 1994, 1998, and 2002 data are from the Provider of Services File and were provided by Sally Kaplan of MedPAC.
5. National Association of Psychiatric Health Systems, Trends in Behavioral Healthcare Systems: A Benchmarking Report, The NAPHS 2001 Annual Survey Report (Washington: NAPHS, 2002).
6. For a general review of this literature, see R.G.Frank and J.R. Lave, "Economics," in Managed Mental Health Services, 2d ed., ed. S. Feldman (Springfield, Ill.: Charles C. Thomas, 2003), 146–165.
7. See, for example, T.M.Wicker and D. Lessler, "Effects of Utilization Management on Patterns of Hospital Care among Privately Insured Adults," Medical Care 36, no. 11 (1998): 1545–1554.[CrossRef][ISI][Medline]
8. National Conference of State Legislatures, 2003 State Health Care Priorities Survey Report, Health Tracking Series (Washington: NCSL, 2003).
9. K.S.Heller and C. Vaz, "Inpatient Psychiatric Prospective Payment System: A Model Classification and Payment Methodology" (Paper prepared for the American Psychiatric Association Committee on Reimbursement for Psychiatric Care in partnership with the Health Economics and Outcomes Research Institute [THEORI], Greater New York Hospital Association, 6 December 2001).
10. Statement of the APA to the House Ways and Means Subcommittee on Health, 15 February 1983.
11. See H.H.Goldman, C.A. Taube, and S.F. Jencks, "The Organization of the Psychiatric Inpatient Services System," Medical Care 25, no. 9, Supplement (1987): S6–S21.
12. See, for example, C.A.Taube, E.S. Lee, and R. Forthoffer, "Diagnosis Related Groups for Mental Disorders, Alcoholism, and Drug Abuse: Evaluation and Alternatives," Hospital and Community Psychiatry 55, no. 5 (1984): 452–455; and R.G.Frank and J.R. Lave, "The Psychiatric DRGs: Are They Different?" Medical Care 23, no. 11 (1985): 1148–1155.[CrossRef][ISI][Medline]
13. Cancer hospitals, long-term care hospitals, children’s hospitals, and rehabilitation hospitals and units were also initially exempted from the PPS.
14. The change in setting baseline costs was implemented because originally new providers’ baseline costs were set at their first year’s costs. This gave them an incentive to inflate their costs in their first year to establish high target amounts, setting the stage for receiving bonus payments in subsequent years.
15. MedPAC, Medicare Payment Policy: Report to the Congress (Washington: MedPAC, March 1999), 74.
16. Thompson, "Report to Congress," 12.
17. MedPAC,Medicare Payment Policy: Report to the Congress (Washington: MedPAC, March 2002), chap. 1; and M.Hornbrook, "Hospital Case-Mix: Its Definition, Measurement, and Use, Part 1: The Conceptual Framework," Medical Care Review 39, no. 1 (1982): 1–4.[Medline]
18. Frank and Lave, "The Psychiatric DRGs"; and R.P. Ellis and T.G. McGuire, "Insurance Principles and the Design of Prospective Payment Systems," Journal of Health Economics 7, no. 3 (1988): 215–238.[Medline]
19. For a review of this literature, see C.Horgan and S. Jencks, "Research on Psychiatric Classification and Payment Systems," Medical Care 25, no. 9, Supplement (1987): S22–S36.
20. M.P.Freiman et al., A Study of Patient Classification Systems for Prospective Rate-Setting for Medicare Patients in General Hospital Psychiatric Units and Psychiatric Hospitals, Final Report, NIMH Pub. no. 278-84-0011 (Silver Spring, Md.: Macro Systems Inc., 1985); andJ.T.English et al., "Diagnosis-Related Groups and General Hospital Psychiatry: The APA Study," American Journal of Psychiatry 142, no. 3 (1986): 131–139.
21. S.D.Horn et al, "Psychiatric Severity of Illness, a Case-Mix Study," Medical Care 27, no. 1 (1989): 69–84.[CrossRef][ISI][Medline]
22. M.L.Ashcraft et al., "A Psychiatric Patient Classification System: An Alternative to Diagnosis Related Groups," Medical Care 27, no. 5 (1989): 543–557.[Medline]
23. Horganand Jencks, "Research."
24. MedPAC, Medicare Payment Policy (March 1999), 77.
25. The payment will vary with factors such as area wage rates or hospital teaching status.
26. See, for example, J.R.Lave and R.G. Frank, "The Effect of the Structure of Hospital Payment on Length of Stay," Health Services Research 25, no. 2 (1990): 327–347.[Medline]
27. MedPAC, Medicare Payment Policy (March 2002), chap. 1; and Horgan and Jencks, "Research."
28. M.P.Freiman, J.B. Mitchell, and M.L. Rosenbach, "Simulating Policy Options for Psychiatric Care in General Hospitals under Medicare’s PPS," Archives of General Psychiatry 45, no. 11 (1988): 1032–1036.;[Abstract]B.E.Fries et al., "A Comprehensive Payment Model for Short- and Long-Stay Psychiatric Patients," Health Care Financing Review 15, no. 2 (1993): 31–50;[Medline]R.P.Ellis and T.G. McGuire, "Hospital Response to Prospective Payment: Cost Sharing and Supply," Journal of Health Economics 5, no. 2 (1996): 129–151; and R.G.Frank and J.R. Lave, "Per Case Prospective Payment for Psychiatric Inpatients: An Assessment and an Alternative," Journal of Health Politics, Policy and Law (Spring 1986): 83–96.
29. Frank and Lave, "Per Case Prospective Payment for Psychiatric Inpatients."
30. Thompson, "Report to Congress."
31. J.Cromwell, "Report on the Feasibility of a Per Diem Prospective Payment System for Psychiatric Facilities Excluded from PPS," Draft Report (Waltham, Mass.: Health Economics Research, September 2001).
32. J.P.Hirdes et al., "A System for Classification of In-Patient Psychiatry (SCIPP): A New Case-Mix Methodology for Mental Health" (Paper prepared for the Psychiatric Working Group, Joint Policy and Planning Committee, University of Waterloo, March 2002).
33. The RAI-MH is an instrument similar to the Minimum Data Set (MDS) that is used to collect information on nursing home residents but that has been adapted for patients with psychiatric disorders.
34. The results discussed here are those found in Heller and Vaz, "Inpatient Psychiatric Prospective Payment System." After the CMS submitted its Report to Congress, THEORI continued to update its initial regression model to include facility type as an independent variable, among other changes. Christopher Vaz, THEORI, personal communication, 24 February 2003.
35. This R² is not comparable to those reported for the other studies. Both the dependent and independent variables are quite different.
36. There are a number of reasons for this redistribution. Other things equal, relative to a case-based system (such as TEFRA), a per case system will redistribute resources to hospitals with longer lengths-of-stay. In 2000 the average covered length-of-stay for Medicare patients was 7.0 days in scatter beds, 11.2 days in psychiatric units, 11.4 days in nongovernment psychiatric hospitals, and 17.0 days in government-owned psychiatric hospitals. In addition, other things equal, as one moves from a cost-based system to a prospective payment system, costs will be redistributed toward facilities with the lower costs. THEORI estimated a per diem cost regression that included facility type as well as the payment variables. Relative to freestanding private hospitals, other things equal, the per diem costs of units were higher by about $122 and those of freestanding government hospitals were lower by $7.
37. Thompson, "Report to Congress," 39.
38. MedPAC, Medicare Payment Policy (March 2002).
39. When units and other facilities were exempted from the PPS, hospitals had financial incentives to allocate costs to cost centers that were covered under TEFRA cost-based payment systems.
40. Ancillary charges represent about 15 percent of total charges of patients discharged from psychiatric hospitals and 25 percent of total charges of patients discharged from units. Exploring these differences could lead to increased understanding of the reasons for the differential costs. It might be possible to determine whether patients discharged from units have more medical burden (which could account for higher costs) by examining the past year’s Medicare costs, psychiatric and nonpsychiatric, separately.
41. A hospital would game the system if it admitted a patient to the setting with the expected highest payment rate rather than the setting that was most suitable for the patient.

                      What's this?