|
Hospital Payment Systems: Will Payers Like The Future Better Than The Past?
Unsustainable health care cost growth has forced payers to reexamine goals for hospital payment systems. Employers want simplicity and transparency, with comparative performance data available in the public domain. Insurers favor simplicity but prefer to keep the analysis of comparative performance data and pricing private. Thirty-five pay-for-performance experiments have been devised in the private sector, to reward hospitals for higher quality and move toward more effective payment systems. Definitive results are not yet known, and caveats remain, but early signs are promising. We develop three scenarios for future hospital payment systems and identify policy actions to improve outcomes.
EVOLUTION IN HOSPITAL PAYMENT SYSTEMS is likely because of growing frustration with health care cost growth. Historically, payment systems have been changed only when cost growth appeared unstoppable. Hospitals still capture the largest share of the health care dollar, and hospital costs have been the major component of recent cost growth.1 At the same time, renewed focus on quality measurement and improvement and on medical-error reduction has heightened interest in paying for performance, rather than just reimbursing providers for services rendered. Some see the evolution toward a performance-based payment system as the key to long-run cost-growth reduction and quality improvement, just as the Medicare prospective payment system (PPS) was seen as a necessary innovation against continued hospital cost inflation in the 1980s. The paper describes what an ideal hospital payment system might look like, from the perspective of payers. We first explore how payment systems might differ from employers’ and insurers’ perspectives. We then describe the hospital-focused, performance-based payment experiments under way in the private sector, and how they compare and contrast to the Medicare pay-for-performance (P4P) experiments. Finally, we develop scenarios for how payment systems might evolve under various technical and policy circumstances. We identify impediments to an improved hospital payment system, and we discuss concerted actions that might increase its likelihood.
Although the 175 million private group insurance enrollees are far less likely than Medicare beneficiaries to be hospitalized, they are so numerous that group insurance payments still account for 30 percent of total hospital revenue—the same percentage as Medicare. Thus, the group market as a whole is important to hospitals. At the same time, payments to hospitals represent 34 percent of claims paid by employers, second only to the 35 percent paid for physician and clinical services.2 So hospital spending is important to employers as well. Given the contributions of hospital spending to overall health care cost growth, it is not surprising that employers, and their health plan agents, focus on hospitals when devising cost containment strategies.3 Employers are most interested in hospitals and the health care system today, to reduce costs and offer competitive health benefit packages. Some large employers and the most savvy employer coalitions, such as the Leapfrog Group, the Midwest Business Group on Health, and the National Business Coalition on Health, are beginning to focus on improving clinical quality, hoping to reduce costs. Some estimates of the savings potential are quite high—as much as 30 percent of total health system costs.4 Because increases in employers’ payments for health care have returned to unsustainably high rates, large employers (predominantly self-insured) and their agents, the larger insurers, feel compelled to "look under the hood" of hospital payment systems in more detail than ever. What do they see? Compared with most input suppliers, hospitals present a Byzantine array of pricing structures. They range from the infamous chargemaster or fee-for-service (FFS) price list to bundled payment systems such as diagnosis-related groups (DRGs), with various forms of "discounts off charges" and "per diems" somewhere in between.5 Faced with complexity and continued cost growth, employers understandably desire more simplicity and transparency. Absolute simplicity is impossible, but relative simplicity—reducing the number of prices to negotiate—is imaginable. For example, an overall "discount" percentage off of the charge master is analogous to a single conversion factor for a resource-based relative value scale (RBRVS) system of physician payment. In contrast to the RBRVS, however, employers have long been aware that the schedule of chargemaster prices is not competitive. Thus, while getting a larger discount than others is better than the alternative, even the largest discount is unsatisfactory if all underlying prices seem too high, or if the rationale behind the variation in pricing of hospital services is unclear. One might infer that demand for payment reform arises when trust in the efficiency and fairness of the existing payment system breaks down. The United States could be nearing that point, as more employers take action to reduce cost shifting aimed at them. The "best" hospital payment system, from an employer’s perspective, has a manageable number of negotiable prices, linked to aggregated sets of clinically appropriate services. The ideal system would also be transparent enough to enable the payer to know both which services are bundled within a particular price and how actual service use, quality, and prices compare across hospitals. Comparing quality in a publicly accessible way is also key for employers trying to reestablish employee trust, which diminished in the managed care era when providers were perceived to have been excluded from networks for cost reasons alone. Chargemasters across hospitals include thousands of prices; they are neither identical nor easily ranked. Thus, comparisons would be difficult even if the chargemasters were more publicly available. (California has made its charge-masters publicly available with little observed effect, other than providing comical examples for researchers.)6 The transaction costs of comparing thousands of prices could easily outweigh potential savings. Per diems and DRGs are both attempts to reach some acceptable trade-off with better incentives than the FFS charge- master but with less risk to the hospital than simple capitation per insured member. DRGs are really risk-adjusted per diems and therefore safer but not more popular among hospitals, for they are often accused of adjusting too slowly to new technologies. Although each payment system in use reflects the relative market power of the hospital and the payer in question, the inability to control costs has pushed large, sophisticated employers to search for ways to reward effective providers relative to ineffective ones. P4P schemes have arisen to help them identify each type. Small employers are still primarily focused on premiums; quality remains a second-order priority for them.7 Four accurate perceptions explain the growing interest in hospital P4P initiatives among large employers: Quality varies across hospitals, current payment systems make it difficult to reward providers of higher-quality or more-cost-effective care, and insurers’ incentives to encourage services with longer-run health or cost savings payoff are weak, given that workers stay with employers longer than with health plans.8 In addition, after the managed care backlash, private insurers are hesitant to alienate their provider base by highlighting differential performance.9 Hence, more of the impetus for P4P might come from employers, which will likely work through their insurers.10 Most private-sector P4P experiments link relatively small bonus payments to performance on specific process measures, which have been associated with better outcomes for costly and relatively common conditions. These first-generation P4P experiments will evolve as knowledge is gained and as the science of measurement, information tracking, and the underlying clinical evidence base progresses.
Health plans are classic middlemen—agents of employers and their workers—with specialized knowledge of claims processing and risk pooling and the ability to negotiate with providers via their market leverage. Through their experiences with managed care in the past twenty years, many insurers have also acquired unique knowledge of health service provision and how to use financial incentives to affect providers’ behavior.11 They now use both knowledge bases in their role as purchasing agent for employers and workers. Some insurers are focused on enhancing and preserving their ability to add value to employers’ and workers’ efforts to secure health services and to be compensated for that value added.12 Thus, they are increasingly interested in data that describe providers’ and enrollees’ actions, costs, and health outcomes, so that they can document their added value to a skeptical marketplace. The primary source of divergence between insurers’ and employers’ interest is around transparency in pricing: If all prices for all hospitals were posted publicly, along with the typical services bundled into the aggregate payment forms (for example, per diems and DRGs), then a large part of the delivery system component of insurers’ specialized knowledge and negotiating power would be in the public domain. If this occurred on a wide scale, insurers would have to find other ways to add value in addition to claims processing, administrative support, and steering enrollees to certain providers. They could earn a decent rate of return at this, but their scope of services and potential profits would shrink. Like employers, insurers would like to see data on quality in the public domain. Given their historical tensions with providers, insurers alone could not generate such proprietary data on quality (performance), because they would encounter resistance from providers. Insurers’ added value—the piece they would like to keep private—is their analysis of these performance data in combination with price data. This is where insurers diverge from employers. Insurers bring value to employers by negotiating lower prices than their competitors. They would resist price transparency, and uniform pricing, because it would limit their ability to negotiate lower prices. The irony, of course, is that the demand for information, which might lead to an infrastructure for information production, sows the seeds of insurers’ future inability to sell specialized access to delivery system knowledge. The National Committee for Quality Assurance (NCQA), which was created to measure health plan performance, also provides a kind of competitor to health plans selling delivery system expertise by branching out to measure the quality of providers and report upon them as it has in recent years. Multi-employer databases, such as the one maintained by Medstat, are another information tool that competes with insurers’ expertise and data for employer business. How long health insurers maintain a role of organizing the delivery system for employers will depend in large part on the health information system we build and on access to it.
Exhibit 1  describes the more common hospital performance measurement initiatives—not P4P programs, but a list of measures used in P4P. As of June 2005 we could identify thirty-six hospital P4P programs (Exhibit 2) via three methods: Internet searches on the names of health plans and calls/e-mail messages to those plans; searching the Internet using the terms "pay-for-performance" or "quality" combined with "hospital"; and follow-up of programs noted in prior publications, case studies, or the Leapfrog Compendium.13
Other than the CMS Premier demonstration program, the hospital P4P experiments are run either by commercial plans (which can be stratified into larger national plans versus smaller regional plans) or by employer-payer coalitions.14 Most of the programs include process and structure measures, and some also include clinical outcome measures for specific conditions.15 Most include some measures from the Joint Commission on Accreditation of Healthcare Organizations (JCAHO), Centers for Medicare and Medicaid Services (CMS), Leapfrog, National Quality Forum (NQF), Agency for Healthcare Research and Quality (AHRQ), or Institute for Healthcare Improvement (IHI).16 The private and employer P4P programs range in their coverage from 3 percent to 26 percent of a state’s insured population. Although hospitals can gain a wide range of incremental revenue from successful performance under these programs—from less than 1 percent to 15 percent—to date, most hospitals do not face large financial risks under P4P programs.17 Some hospitals also reward their chief executive officers with bonuses if the hospital performs well on P4P measures, and some are planning rewards for lower-level directors.18 Public versus private programs. Private P4P programs have similarities to as well as differences from public efforts. Whereas CMS incentives are budget-neutral, private payers’ programs are not. The CMS program is the only P4P program with direct financial penalties: 2 percent lower DRG payment for clinical conditions in hospitals scoring below the tenth decile in year three of the CMS Premier demonstration. Although a few plans are creating tiered networks, it is usually a response to hospitals’ overall costs rather than to data on quality performance. For example, the California Public Employees’ Retirement System (CalPERS) is implementing a narrow hospital network by eliminating twenty-eight so-called high-cost hospitals from its Blue Shield health maintenance organization (HMO) plan.19 For the CMS P4P program, the DRG is the base payment from which hospital incentives are calculated. Private P4P programs for hospitals usually pay bonuses as an incentive above the agreed-upon reimbursement rate. Some of these private payers, such as Blue Cross Blue Shield of Michigan, use the DRG base rate in this incentive calculation; others, such as the Independence Blue Cross Hospital QIPS in Pennsylvania, might increase their per diem rates. There is wide variation even within private P4P plans in contracting around incentives by location. Thus, we expect experimentation with payment incentives to continue for some time. A more rational reimbursement system, which rewards quality of care rather than simply doing more to patients, is the short-run goal of paying for performance. The longer-run goal is also to make the health care system more efficient. It has become clear that under existing reimbursement structures, current market forces are insufficient to ensure either higher-quality or more-cost-effective care.20 Limitations of existing hospital P4P programs. Because P4P is in its infancy, several limitations need to be addressed before it is more widely implemented. These include defining and unifying measures across the vast number of reporting initiatives, risk adjustment for clinical outcome measures, resource burdens on smaller versus larger hospitals, and the need for data on the effectiveness of P4P in improving care processes and outcomes. Measure definition and risk adjustment. Validity of the measures that indicate high- quality care or good performance is paramount. Inadequate information systems, as well as imperfect algorithms and data to control for patient-level risk factors, severely limit the ability to risk-adjust clinical outcome measures. This is a major barrier to more widespread implementation of P4P and to convincing some physicians, who manage complex patients on a daily basis, of the value of these measures. As we move forward with hospital-based P4P, we must consider how to encourage different insurers to use the same measures and risk adjusters so that comparisons can be made. In the absence of coordination, which could be hampered by antitrust law or lack of leadership, this standardization might occur slowly, if at all. Lack of data on P4P’s effect on provider performance. Except for the CMS and Inter-mountain Health Care, few other P4P programs have publicly documented results concerning their program’s impact on clinical quality.21 Some programs have documented their reduction in expenses and shift in use by enrollees from less- to more-favored hospitals (such as the Blue Shield of California Tiering Program) or in uptake of reporting measures by hospitals (such as the Memphis Business Group on Health). The Piedmont Clinic Performance Improvement Plan in Georgia has found a 2 percent improvement in patient satisfaction and a 20 percent increase in use of computerized physician order entry (CPOE). The CMS has enough market share to require and make public those hospital performance results that Congress will permit. Intermountain Health Care is an integrated health system that is also an insurer, and its quality improvement efforts could be driven more by a provider-led culture of pursuing excellence than by financial incentives alone.
The apparent discrepancy in public disclosure of results by organization type raises larger questions about the most powerful motivating factor behind performance improvement.22 It now appears that most plan-driven P4P programs are not forthcoming about the details of the program parameters, whereas the public-sector and employer-driven programs are more willing to share structure information (Exhibit 2 Special cases. Most existing P4P programs are geared toward medium and large hospitals, so smaller and rural hospitals present a special case for which different measures or subsets of measures could be appropriate. Smaller rural hospitals might have smaller populations, fewer numbers of certain complicated procedures, and fewer resources to devote to data collection. P4P measures are being tailored for these smaller hospitals—for example, by Leapfrog and the CMS. Academic medical centers and teaching hospitals also could suffer under P4P systems that include efficiency measures. Physicians at these institutions spend time teaching medical students and residents; thus, their efficiency appears lower under conventional measures. Payment reform will need to consider how such teaching responsibilities skew performance on efficiency measures. Lack of information technology (IT) infrastructure. If hospital administrative data are going to be used to make hospital-to-hospital comparisons, increased investments in hospital medical records staffing and auditing will be necessary. The cost of such an endeavor, if coordinated across hospitals, could be shared among all payers in proportion to the size of their enrollee populations. Alternatively, a dedicated separate public funding stream for IT investment, similar to that used to create the interstate highway system, could be allocated. The validity of P4P measures, especially the clinical outcome measures requiring risk adjustment, will hinge on the availability and quality of a future electronic medical record. We must avoid penalizing hospitals in low-income areas that are the only service providers there. Using absolute level of performance as opposed to improvement on measures as the criteria for receipt of incentives will disproportionately punish the hospitals with fewer resources and poorer performance at baseline. Underresourced hospitals might not attempt to meet P4P goals if they function too far below the absolute targets that programs reward. Such institutions might need more funding to bring them "up to speed." Rewarding improvement rather than simply attaining an absolute target will also help encourage such hospitals.
So far we have explained the logic of the preferences of employers and insurers and described the range of current hospital P4P experiments. Here we speculate about how hospital payment systems might evolve over time, given these historical preferences and existing P4P experiments. We develop three potential paths for future hospital payment systems, outline the factors affecting the likelihood that each will come to fruition, and identify some key policy choices that could improve or retard their likelihood. We start with systemwide performance-based payment, the least likely to occur but perhaps the most instructive to contemplate initially. Hospital P4P experiments to date have focused on measures of specific dimensions of care for particular conditions, or on processes of care within specific episodes of illness. In developing our scenarios, we allow for the possibility that P4P might evolve to focus more on outcomes and health status. Scenario 1: Utopia—exclusively performance-based payment. In such a system, the percentage of payment contingent on process or outcomes performance would be high enough that consistently low-performing hospitals could not survive. Initially low-performing hospitals would be given some time to "catch up," by rewarding improvement in the early years of the transition to Utopia. The system would not be simple but would pay substantial bonuses for behavior consistent with best practices and evidence-based medicine in most clinical areas, and for efficient resource use. "Bonus" is really a euphemism; good performers would earn the efficient cost of provision, and mediocre-to-poor performers would not be paid enough to cover costs. Base prices in Utopia would be either capitation per enrollee or, more likely, DRGs, for these base payment systems provide the strongest incentives to achieve better specific outcomes with lower resource use and more measured technology acquisition over time. DRGs pose less financial risk to the hospital than capitation because of the potential for unequal case-mix distributions associated with capitation and the hospital’s inability to control who gets admitted. For these reasons, DRGs are likely to dominate capitation in Utopia in the long run. Large differentials in hospital efficiency—measured as target outcomes per dollar—within DRGs could not persist in the long run, since each payer would adjust the DRG base rate over time to reflect the evolving state of efficiency in the best local or regional hospitals. Competition among payers would prevent this "base rate plus bonus payment" from being driven too low. In effect, the payment system would become one in which hospitals implicitly "bid" by DRG, since payers will contract with all or most hospitals and will direct patients to hospitals with the most attractive price-performance combination for each specific condition. In hospital payment Utopia, from a payer’s perspective, local providers’ market power would be effectively negated by organized buyers’ monopsony power, and all DRG-type prices would be driven to an efficient hospital’s marginal cost. The road to Utopia is long and winding. It would require coordinated buying power to counter local providers’ market power (present in most communities) and an information infrastructure that supports widespread performance and utilization measurement as well as comparative data collection and dissemination across a variety of clinical conditions and processes. Although employer groups’ potentially countervailing market share as a whole approaches Medicare’s 30 percent, it is diffused over many employers and competing health plans. When hospitals’ market power exceeds buyers’ market power, incentive-rich payment systems and transparency can be resisted, for they would force costly process innovation and financial risk. The most powerful payer in the market today is FFS Medicare, and there is no substitute for its leadership and catalytic role. The strongest market force for payment system reform would be for Medicare and a critical mass of private payers to agree on a payment system. This partnership is needed because only Medicare can compel compliance with information requirements and is likely to finance sufficient clinical research, while private-sector payers have a comparative advantage experimenting with payment incentives of varying types and sizes. Compelling, credible, and reproducible examples of quality and efficiency gains as a result of payment experiments would be the best catalysts to overcome the antimeasurement lobbies that will oppose change. If Medicare and private payers ultimately fail to agree, it is hard to imagine much progress toward either a performance-based payment system or a more efficient delivery system. Other barriers to Utopia include the reality that P4P quality measures are now limited to a few specific actions or outcomes associated with far less than 25 percent of hospital admissions and spending.23 Process-oriented quality improvement activities could spread to unmeasured domains in the hospital, or not. The likelihood of spread could be related to the size of the bonus or the percentage of payment "at risk" for performance; it also could depend more on medical professionals’ underlying commitment to pursuing excellence if they come to think that they are underperforming compared with their peers. None of this is known, and that is why experiments are necessary. Perhaps the largest barrier to Utopia is the difficulty in sharing the gains from increased hospital efficiency and higher quality with the physicians who make most care decisions inside and outside the hospital. Medicare is statutorily prohibited from allowing physicians to benefit financially from care-reducing clinical decisions, even if they are perfectly evidence-based. This issue deserves serious attention in any Congress committed to making Medicare more efficient. Private-sector P4P initiatives now focus on either hospitals or physicians but not both.24 This partly reflects the infancy of P4P programs and the reality of the fragmented U.S. delivery system, but it also suggests that systemwide efficiency-increasing P4P payment systems will have to await future generations of experiments devised by public and private payers and the providers who serve them. If for no other reason, Medicare and the private sector should coordinate the next generation of P4P experiments, for measured impacts are likely to be far greater where the incentives in place across payers are complementary in a given locale. Scenario 2: slow progress. Although sufficient evidence on P4P initiatives is not yet widely available, there are enough promising signs of improved quality and providers’ responsiveness to incentives that expansion of measurement sets and conditions is likely to continue, at least among payers and providers in markets where measured hospital quality is important for marketing purposes.25 Such pressures on hospitals, however, are not sufficiently strong in the majority of U.S. local markets. Thus, the continued importance of local reputation, coupled with providers’ concern about imperfectly adjusted comparative quality measures, makes P4P unlikely to become a strong force in the majority of local markets soon. Still, P4P might spur small efficiencies and quality improvement here and there. Scenario 3: status quo—rejection of publicly reported comparative performance data. Although this is less likely than Scenario 2, inertia remains a powerful ally of many stakeholder interests. Employers are fairly desperate for savings and are increasingly tempted to retreat from transforming the delivery system to promised financial safety behind a wall of defined contributions, high-deductible health plans, and more consumer responsibility and risk. Health plans have a strong interest in pleasing their clients who are demanding cost growth reduction, but negotiating discounts off charges on the basis of market share could be more profitable than placing provider relations at risk by driving wholesale delivery system change. Finally, even if all payers adopted current P4P strategies, strong countervailing forces exist. People are not generally supportive of closing inefficient or mediocre hospitals in their own communities. They might reject the short-run implications of a far more efficient health care system and demand protection for the locally favored. The entire movement to P4P is predicated on the belief that there is much ineffective service use of high cost and varying quality, as well as waste in the current system, and that better incentives and purchasing techniques could deliver higher-quality care to more people for less money than we now spend. This belief is strongly held by some researchers and observers of the health care system, but it is not widely shared by the general public. Furthermore, there is a consensus that although excess use should be reduced, in the long run, cost growth is driven by technology, so focusing more efforts on technology assessment might be wiser than trying to wring inappropriate use out of some providers.26 In addition, Americans have never tolerated cost containment techniques when their own access to care appears to be at risk. Convincing citizens that some of the care that they now want is superfluous at best will require a concerted educational effort. IT WILL TAKE POLITICAL, ECONOMIC, AND health stakeholder leadership to create an environment in which quality improvement and performance-based payment systems can take hold and flourish. Performance-based payment could serve the interests of employers, insurers, and quality-focused stakeholders, but major collective effort would be required to create it. P4P systems might well be required if we are going to make health care affordable for all Americans, but this is not yet a consensus goal. Unless a stronger political will for equity develops, we can expect modestly improving quality for some patients over time, while more citizens become disenfranchised by health care cost growth that rises faster than incomes. This is the choice we face in deciding whether to improve the effectiveness of the health care system, and hospital payment systems in particular. If we do, we could then afford to implement a moral vision, should one someday capture our collective imagination.
Len Nichols (nichols{at}newamerica.net) directs the Health Policy Program at the New America Foundation in Washington, D.C. Ann O’Malley is a senior health researcher at the Center for Studying Health System Change in Washington, D.C. An earlier version of this paper was prepared for the Federation of American Hospitals conference, "The Future of Hospital Payment Systems," Washington, D.C., 15 July 2005. Ann O’Malley’s time on this project was funded in part by the Robert Wood Johnson Foundation and by the National Cancer Institute (Grant no. K07 CA 91848). The authors acknowledge the research assistance of Jeremy Ershow; however, they alone are responsible for the content.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||
Top
What Do Employers Want?
What Health Plans Want
