What is prospective payment system

what is prospective payment system



To examine skilled nursing facilities (SNFs) “make-or-buy” decisions with respect to rehabilitation therapy service provision in the 1990s, both before and after implementation of Medicare's Prospective Payment System (PPS) for SNFs.

Data Sources

Longitudinal On-line Survey Certification and Reporting (OSCAR) data (1992–2001) on a sample of 10,241 freestanding urban SNFs.

Study Design

Principal Findings


The financial incentives associated with changes in reimbursement methodology have implications that extend beyond the boundaries of the health care industry segment directly affected. Unintended quality and access consequences need to be carefully monitored by the Medicare program.

Keywords: Transaction costs, skilled nursing facilities, rehabilitation services, prospective payment

In a typical skilled nursing facility (SNF), numerous parties exchange goods and services in the delivery of care. Resident care is rendered through a complex series of transactions among these parties, including, but not limited to: physicians, pharmacists, equipment distributors and manufacturers, rehabilitation therapists, laboratorians, and dieticians. Some of these transactions occur among the employees of the nursing home (intra-organizational exchange) while others involve contractual relationships with outside parties (inter-organizational exchange). In this article, we focus on one particular transaction: the provision of rehabilitation therapy services.

Transaction cost economic (TCE) theory provides a theoretical framework for identifying the organizational and environmental circumstances governing the choice of inter- or intra-organizational transactional form (Williamson 1975 ; Williamson 1991 ). It has been used (although not extensively) in health services research to explain hospital vertical integration (Mick and Conrad 1988 ; Dansky, Milliron, and Gamm 1996 ), diversification (Robinson 1994 ), service innovation (Stiles, Mick, and Wise 2001 ), and quality control (Stiles and Mick 1997 ). Transaction cost economic theory argues that contractual relationships among and within firms arise from efficiency-seeking behavior in a world of limited information and incomplete enforcement possibilities (Oster 1990 ). A fundamental principle of TCE is that organizations incur costs as a result of planning, implementing, and enforcing exchanges with other organizations. Firms tend to structure exchanges in ways that will minimize the cost of using the transactional apparatus (Oster 1990 ). With respect to external supplier exchange relationships, transaction costs can include costs associated with contract negotiations, monitoring adherence to contractual terms, providing financial incentives or penalties, and losses resulting from supplier noncompliance.

Increases in environmental uncertainty requiring adaptation between exchange parties increase market or externally driven transaction costs (Williamson 1975 ; Williamson 1991 ). Historically, SNFs operated in a stable, if not particularly munificent environment, posing few competitive threats. However, events conspired to increase environmental uncertainty for SNFs throughout the 1990s. The implementation of Medicare's Diagnostic Related Groups (DRGs) for hospital reimbursement began to have a major case-mix impact on SNFs at the onset of that decade (Cornelius et al. 1994 ). Compounding the DRG effect, the growth of managed care to almost 30 percent of the insured population by 1998 promoted reduced hospital length of stay, intensifying the medical care needs of patients discharged to SNFs. In addition, there was tremendous growth in viable institutionally based substitutes for SNF care at both the high end (hospital-based SNF units) and low end (assisted living) of the care continuum. Finally, concerns about nursing home quality resulted in a wave of new regulations that influenced virtually every aspect of nursing home operation (Harrington and Carillo 1999 ). Thus, SNFs exposed to increasing environmental uncertainty created by the escalating impact of these underlying forces throughout the 1990s may have experienced increased transaction costs in structuring relationships with outside suppliers.

In addition, the transaction cost implications of the 1998 change in the financing of SNF care may have promoted major alterations in the terms governing external exchange relationships. From its inception in 1964, the Medicare program reimbursed SNFs under a retrospective, reasonable cost based system. Outside contractors, paid on a fee for service basis, directly billed the Medicare program for services rendered onsite, allowing SNFs to avoid clinical and financial expenses associated with administering these services. According to 1998 data compiled by the Centers for Medicare and Medicaid Services (CMS), nearly 70 percent of nursing home facilities contracted with outside vendors for all physical or occupational therapy services.

However, as of July 1, 1998, SNF reimbursement changed to case-mix adjusted payments under the Medicare Prospective Payment System (PPS) for the costs of all SNF care provided to Medicare recipients (Health Care Financing Administration 1998 ). Medicare beneficiaries served under the SNF benefit are now classified into one of 44 resource utilization groups (RUGs), and Medicare payments for services furnished from all sources are bundled into a single RUG payment received by the SNF. The SNF in turn reimburses outside service vendors out of this RUG payment. Prior to PPS, SNFs were not accountable for the consequences of poorly negotiated contracts with external suppliers, as these costs were passed on to the Medicare program. Under PPS, nursing homes receive a flat per diem charge per RUG regardless of the cost of services rendered. Poorly negotiated contracts with outside vendors could result in diminished profitability. Thus, under PPS, facilities face increased transaction costs associated with outsourcing. To reduce or eliminate externally or market-driven costs, the facility may opt to provide all, or most, of these services through in-house employees. Control over rehabilitation services would have high priority because to qualify for the SNF benefit, Medicare beneficiaries must have documented rehabilitation potential entailing the provision of a minimum number of therapy hours per week.

Using TCE theory as a conceptual framework, this study has two objectives. The first is to determine if transactional characteristics affected SNF “make-or-buy” decisions with respect to rehabilitation services in response to increased environmental uncertainty during the turbulent decade of the 1990s. Thus, this study extends the existing literature by demonstrating the applicability of the TCE framework to long-term care service. The second is to examine whether a specific environmental shock that increased transaction costs associated with outsourcing, the implementation of PPS for SNFs in 1998, was associated with an increased likelihood that facilities would choose to make rather than buy rehabilitation services. From a policy perspective, this study demonstrates how efforts directed at increasing efficiency in one health care sector can have unintended, far-reaching consequences for other market participants.

Theory and Hypotheses

Transactional Characteristics Affecting SNF “Make or Buy” Decisions

From a theoretical perspective, transaction costs increase under conditions of environmental uncertainty requiring adaptation between the exchange parties. Gilley and Rasheed (2000) hypothesized that many benefits offered by outsourcing activities are offset by environmental uncertainty, and the more uncertain the environment, the fewer benefits realized.

Although environmental uncertainty during the 1990s influenced the exchange relationships of all nursing homes, the impact on transaction costs is not uniform. For example, environmental uncertainty is more likely to increase externally or market-driven transaction costs when transactions between exchange partners occur frequently (Williamson 1975 ). Much of the volatility in the SNF industry during the 1990s was attributed to reforms in the public sector, particularly in the Medicare program. Thus, the frequency of transactions involving Medicare reimbursement is particularly relevant. There is considerable variation in the number of Medicare beneficiaries in the total patient census of individual skilled nursing facilities. Some admit a high volume of Medicare beneficiaries in an effort to achieve the economies of scale necessary for Medicare participation to be financially feasible (Dor 1989 ). In a study of the information services industry, Poppo and Zenger (1998) found that when the output volume of a component manufactured in-house is sufficient to produce economies of scale, firms are less likely to outsource. These considerations provide the rationale for our first hypothesis:

H 1. Facilities with a higher proportion of Medicare residents in total census will be more likely to attempt to reduce externally or market-driven transaction costs by bringing therapy services in-house.

Walker and Weber (1984) found that output (demand) uncertainty led automobile manufacturers to produce components in-house rather than purchase from outside suppliers. They suggest the “make” response takes precedence as managers attempt to avoid transaction costs associated with renegotiating contracts with suppliers in order to accommodate fluctuations in demand. In the SNF setting, demand uncertainty with the potential to increase transaction costs associated with outside therapy contracts would be manifested in wide variations in the number of Medicare referrals admitted from one year to the next.

H 2. Facilities with greater fluctuation in the annual number of Medicare admissions will be more likely to attempt to reduce externally or market-driven transaction costs by bringing therapy services in-house.

Exchange relationships may also differ in terms of the underlying complexity of the transaction (Williamson 1975 ). Complex transactions have many components or sequential decision points, and resist standardized approaches. The greater the complexity of the transaction, the greater the need for contractual control to ensure that tasks are appropriately carried out. In the nursing facility context, transactional complexity varies with the care requirements of the residents.

H 3. Facilities characterized by greater case-mix complexity will be more likely to attempt to reduce externally or market-driven transaction costs by bringing therapy services in-house.

A final characteristic distinguishing transactions reflects how well the transaction itself is understood by all the parties involved. Theoretically, even if frequent or complex, transaction costs may not increase if the transaction is well understood by the buyer (Williamson 1975 ). On the other hand, if the transaction entails the application of knowledge and skills beyond the buyer's expertise, terms must be more formally and explicitly articulated to ensure successful fulfillment of the contract, increasing transaction costs. Under these circumstances, it may be more efficient to bring the transaction in-house. However, if the firm possesses the expertise to monitor compliance with contractual terms, outsourcing may be the preferred option. In the nursing facility context,

transaction-specific uncertainty is mitigated by the expertise reflected in the qualifications of professional staff.

H 4. Facilities characterized by higher levels of professional staffing will be less likely to bring therapy services in-house.

The Impact of PPS for SNFs on the Provision of Rehabilitation Services

By increasing environmental uncertainty, the implementation of PPS increased transaction costs in SNF/rehabilitation vendor relationships. For nursing homes to maximize reimbursement by qualifying the resident for the highest possible RUG payment category, rehabilitation therapy services must be ordered and provided promptly after admission. Thus, under PPS, among the most compelling externally driven transaction costs are the losses incurred if suppliers fail to provide appropriate services in a timely and cost-effective manner. This may provide the impetus to move from the provision of services through external contract to the hiring of in-house staff.

H 5. In response to PPS implementation, facilities will attempt to reduce externally or market-driven transaction costs by bringing therapy services in-house.

Finally, although not specified by TCE theory, there are organizational characteristics that could increase the propensity of a facility to exercise greater control over services provided by rehabilitation therapists in response to PPS implementation. For-profit facilities presumably are the most market-oriented providers and, all other things being equal, may be quicker to recognize changes in financial incentives that signal the need to reconsider the “make-or-buy” decision with respect to therapy services. On the other hand, perhaps because of the availability of centralized contracting capability and other facilitating administrative resources, facilities operated by chains may be more effective at negotiating with outside contractors (Zinn et al. 1999 ). These considerations lead to the following exploratory hypotheses:

H 6. In response to PPS implementation, for-profit facilities will be more likely to increase control over therapy services by bringing services in house.

H 7. In response to PPS implementation, chain-affiliated facilities will be less likely to increase control over therapy services by bringing services in house.


Data Sources and Sample

The primary source of data for this analysis is the On-line Survey Certification and Reporting (OSCAR) system, which we compiled into a longitudinal data file spanning from 1991 to 2001. The OSCAR data contain organizational and aggregated resident data routinely collected as part of the annual licensure and certification process. The longitudinal OSCAR data contained 167,600 surveys from 18,613 hospital-based and freestanding nursing facilities in urban and rural counties nationwide. The Area Resource File (ARF) provided market(county)-level data that were matched to the facility surveys in OSCAR.

We restricted our analysis to a subsample of urban, freestanding nursing facilities surveyed between 1992 and 2001 (CMS overwrites OSCAR data so earlier years are not available), excluding facilities that changed either from or to hospital-based status over that time period (N=335, about 2 percent of total facilities). Thus, study findings are generalizable to this population only. We further excluded facilities represented by only one survey in the data file in order to allow information obtained from a previous survey to predict the outcome (the manner in which the SNF provides therapy services) at the current survey. Accordingly, each observation in the final dataset links the current outcome with information from the previous survey. We further excluded counties where no Medicare residents were reported in any facilities in a given year. Altogether, ten counties (27 facilities) were dropped from the final analysis. Following these rules, the final dataset contains 68,114 outcome surveys from 10,241 freestanding nursing facilities located in 815 urban counties. Of these surveys, 67,524 with complete data were available for multivariate analysis. Relatively few facilities exited the Medicare program over the study period (approximately 1 percent per year). Additional analysis controlling for exiting facility did not change any results reported here.

Variable Specification

The dependent variable is the arrangement the facility has to provide rehabilitation services at the time of the current survey. This is classified into four categories: “No PT/OT,”“All contract,”“Mixed,” and “All staff.” These categories are based on the availability of physical or occupational therapy (PT/OT) and whether some of or all of these services are provided through contract with outside venders. If no PT/OT full-time equivalents (FTEs) are recorded, a facility is classified as “No PT/OT.” A facility is considered “All contract” if all PT/OT FTEs are reported under a contract basis, and “Mixed” if some PT/OT FTEs are reported to be staff and others under contract. A facility is classified as “All staff” if all PT/OT FTEs are hired employees. Our data indicate that the “mixed” model is one in which supervision over the quality and quantity of therapy is based in-house, but some services are provided by an outside contractor. Thus, a facility that is either all staff or mixed staff has increased internal control over the provision of therapy services, relative to an all contract arrangement. We treat these as categorical not ordinal options.

A facility is determined to have a high proportion of Medicare residents (H 1) if the proportion of Medicare residents in the total resident census exceeds 12 percent (the upper quartile), as indicated by a dummy variable (1,0). We dichotomized this variable (and others as well) rather than used it in its continuous form because the effect may not be linear. Also, a dichotomous variable facilitates interpretation of its effect as an odds ratio on a multinomial outcome.

We used the annual fluctuation in Medicare admissions to create an indicator variable for demand uncertainty (H 2). Specifically, we calculated the absolute change in the number of Medicare residents between the previous and current surveys. The SNFs in the upper quartile of the distribution of all absolute changes (i.e. more than five residents) were categorized as high-fluctuation facilities. On average, this amounts to approximately 63 percent of the previous year's total Medicare volume.

We selected three measures of case-mix complexity (H 3). The first identifies facilities providing a high volume of rehabilitation services, defined as either (a) having 35 or more rehabilitation residents and 30 percent or more residents receiving rehabilitation services, or (b) having 20 or more rehabilitation residents and 50 percent or more residents receiving rehabilitation services (Berg, Intrator, and Lemon 2001 ). These definitions account for both the absolute number of residents receiving services and the impact on the facility. The second measure indicates whether a facility has residents receiving intravenous (IV) therapy, and the third whether there are any residents receiving tracheotomy care. Each of these measures reflects a different dimension of resident case mix (e.g. rehabilitation versus subacute care).

To measure the availability of professional expertise (H 4), we included a dummy variable indicating whether a physician-extender (nurse practitioner or physician assistant) is available in a facility. To test the PPS effect (H 5), we included a variable indicating whether the predicted outcome occurred after PPS implementation (July 1, 1998). Dummy variables indicated whether a facility is run for-profit (H6), or is part of a chain (H 7).

A number of control variables that could influence the manner in which SNFs provide therapy services were also included. To control for calendar time trend, we included a variable that measured the number of years from January 1, 1992, up to the current survey date. Since prior commitment to a staffing arrangement may constrain the ability to change, we included three dummy variables indicating whether the facility was “All staff,”“Mixed,” or “No PT/OT,” at the time of the previous survey, with “All contract” as the omitted category.

Large facilities may be able to achieve economies of scale with respect to in-house rehabilitation service provision, so we included the total number of beds (centered at 118, the aggregate mean, with steps of 10 beds). Higher occupancy could signify a more traditional long-term-stay resident population with less need for rehabilitation services, so facility occupancy rate (centered at 88 percent, with steps of 5 percent) was also included. Facilities with more private pay residents are in a more favorable financial position, allowing them to hire more in-house staff or absorb external transaction costs associated with contracting. Therefore, the percent of private pay residents (centered at 26 percent, with steps of 5 percent) was included. In addition, a dummy variable was created to indicate whether a facility first became a Medicare/Medicaid provider after 1995. These providers had to accept the fixed federal rate without having their own experience taken into account, placing them at a disadvantage. Finally, to control for market supply and demand conditions, we included the annual unemployment rate (centered at 4 percent) and the proportion of population aged 75 and older (centered at 5 percent) in 1990.

Statistical Methods

Given the multinomial form of the dependent variable, we used a longitudinal multinomial logistic regression model (Hosmer and Lemeshow 1989 ) to predict the probability of the current outcome in each of the four categories, with “All contract” as the reference category.

Because we are modeling multiple observations for the same facility over the survey time period, the conventional assumption of independence between observations may not hold. If the correlation between observations within-facility is not adjusted for properly, the standard errors for parameter estimates tend to be downward biased (Goldstein 1995 ). To correct for this, we apply the Huber/White (Huber 1967 ; White 1980 ) robust variance estimates procedure available in STATA. which estimates standard errors adjusted for clustering within facility.


Descriptive statistics for the independent and control variables based on a cross-section of facilities operating in 1998 are provided in Table 1. Grand descriptive statistics and those from other years are available on request, as is the correlation matrix for covariates included in the model. We did not see any indications of even moderate collinearity between the covariates that might compromise our model results.

Source: www.ncbi.nlm.nih.gov

Category: Bank

Similar articles: