Second-Generation Pay for Performance: A New Paradigm
Background
In the past few years, increased attention has been paid to health care quality. Leading observers, including the Institute of Medicine (IOM), have noted a lack of consistent, high quality care in the United States, and studies have shown that both widespread regional care variations1,2 and medical error rates are more frequent and harmful than previously believed.3,4,5 Further, clinical best practices frequently are not applied. A recent study showed that only 55 percent of patients receive appropriate care for common diseases.6
At the same time, health care costs have continually escalated, outpacing inflation by 8.5 percent and 9.3 percent in 2001 and 2002 respectively.7 In 2002, health care spending reached $1.6 trillion totaling nearly 15 percent of GDP. The combination of an aging baby boomer population, continued technological advances, tax subsidies for employer-provided health insurance, and limited direct consumer exposure to spending, provides a positive feedback loop for ongoing cost increases. Perhaps not surprisingly then, health care payers have begun to increase their focus on so-called pay for performance (P4P) initiatives.
The fact that health care requires P4P initiatives contrasts starkly with other industries, where providers of a higher quality product or service are directly rewarded with higher volume, higher price, or both. However, in health care, caregivers are often paid the same amount regardless of outcome, or may actually be paid a greater amount when patients fare worse (e.g., when complications occur). This peculiarity is a function of several unique differences when comparing health care to classic consumer-seller markets, including:
- Consumers and payers are often unable to readily identify or distinguish quality differences and lack key comparative information to use in decision making;
- Consumers do not directly face the true price of care; and
- Due to historical and convenience factors, most care is provided locally without the benefit of national or international competition.
Other forces conspire to further distort the health care market incentives. For example, a lack of data standardization, an underinvestment in information technology, and weak cooperation among and between provider and payer organizations have combined to make cross-organizational information sharing and measurement critical aspects of high quality patient care and population-based quality improvement exceedingly difficult. Also, frequent consumer turnover within payer populations (employer or plan) makes long-term quality investments less rewarding since future benefits often accrue to another payer. A lack of payer market concentration reduces payer leverage and limits sustained provider responses to individual P4P programs.
Although hospitals and physicians are generally committed to improving care by both mission and professional ideals, they must generate a return from their business. Yet providers are routinely frustrated as their efficiency and quality improvement efforts accrue largely to payers. Even worse, quality improvements may produce direct economic losses (e.g., when improvements yield fewer patient visits or admissions under fee-forservice or case-rate reimbursement). In fact, a recent study evaluating the business case for quality concluded that current payment mechanisms allow, and even reward, defective care because they do not reward future benefit.8 In response to these observations, health care quality and policy leaders recently called for P4P to become a national priority.9
Market Response to Date
As of July 2003, more than 20 million mostly urban beneficiaries were participating in approximately 37 different P4P programs, originating from 31 separate sponsoring entities with a collective 2004 budget of more than $200 million.10 Continued growth is expected with the number of P4P programs projected to more than double by 2006.11 Most simply, the goals of purchaser- driven P4P programs can be characterized as striving to stimulate and reward greater quality and efficiency in the delivery of care. The most common forms of incentives applied to date include:12, 13
Public disclosure publishing provider-quality information to stimulate provider-driven quality improvement efforts and consumer demand for higher quality services. This has been the most commonly applied method, but with mixed overall results and few examples of consumer-originated change. In New York, the State Health Departments release of risk-adjusted coronary artery bypass graft surgery mortality rates correlated with a 41 percent reduction in deaths during the programs first four years. However, evidence indicates that consumers ignored the data when selecting providers.14 In a rare example of consumer-originated change, PacifiCare in California released physician group quality information just prior to open enrollment in 1999; 30,000 employees selected the higher quality physician groups resulting in $18 million in additional capitation for those groups.
Payment differentials bonuses, withholds, or other payment incentives for providers with improved or more commonly, high performance. This approach has become more common in the past year or two and also includes shared savings programs where providers share in the benefit from quality improvements. This form of P4P is particularly important given the capacity constraints that many providers face, which make rewards based upon increased patient volume less attractive. It is important to note that the form of this differential whether based upon relative rankings (i.e., rewards top performers, but may not encourage ongoing improvement or attainment of target performance levels), absolute performance (i.e., rewards top performers, but may be unattainable for certain providers or within certain markets), or change from baseline (i.e., moves the mean performance, but may not reward existing high performers) will produce substantially different impact within the health care market.
As an example, the Center for Medicare and Medicaid Services has launched a pilot program to pay performance bonuses to hospitals that submit quality information. Hospitals in the top 10 percent or 20 percent of a given clinical area (e.g., acute myocardial infarction) will receive a 2 percent or 1 percent payment differential, respectively, for that specific condition.15 More recently, this carrot was augmented with a stick. Hospitals outside of the pilot that do not voluntarily report selected performance measures to their quality improvement organization will be ineligible for a 0.4% payment increase scheduled for fiscal 2005.16 More than a decade ago, Intermountain Health Care in Utah negotiated an equal split of savings between its providers and health plan derived from clinical guideline implementation. In communityacquired pneumonia (CAP) alone, the average cost per case fell by nearly 50 percent, from $2,752 in 1994 to $1,424 in 1995. Unfortunately, the savings realized within their self-insured patients was more than offset by lost revenue from applying the CAP guideline to patients with other payers.
Cost differentials incenting consumers to select providers by instituting higher premiums or out-of-pocket costs when selecting low-quality providers. While less common, this method may hold considerable hope for truly changing consumer behavior through consumers pocketbooks. In 2001, General Motors lowered premium contributions for employees choosing high quality, high efficiency health plans (as defined by GE) resulting in employee migration to designated plans, producing an estimated savings of $5 million.13
Reduced administrative oversight and overhead reducing or eliminating administrative (e.g., precertification requirements) burdens on high-quality providers. As with cost differentials, this method has been less commonly applied than public disclosure or payment differentials.
Risk sharing and capitation Although these methods once held out great hope for health care system transformation, they have failed to take root outside of limited markets and have even regressed, particularly after the catastrophic failure of the attempted Clinton health care reform initiative in 1993. Although critics note that, theoretically, capitation rewards lower utilization, it could promote appropriate care and prevention that typical fee-for-service arrangements often lack if combined with low patient turnover.
Other, even less frequent models, include: quality grants, care planning reimbursement, technical assistance for quality improvement, and practice sanctions.
Lessons Learned and Opportunities for Second- Generation P4P Programs
The collective P4P experience to date provides valuable insight into both limitations and opportunities for P4P programs. While cautiously optimistic, physicians have voiced concerns, most frequently questioning measure validity, attribution accuracy (i.e., the difficulty of correctly assigning outcome measures to individual physicians in a group or shared practice model), data collection and reporting burdens, need for and appropriateness of severity adjustment, potential liability, and ultimately, the relative cost and benefit of participation itself (i.e., is there enough reward potential to offset the participation costs?).17 Second-generation P4P programs will build upon our collective experience to drive widespread implementation of successful P4P initiatives. Below, key items are highlighted for your program design consideration.
Use Agreed-Upon, Actionable, Accurately Attributed Performance Measures. Measuring quality to improve performance requires that payers and providers agree upon metrics. Effective measures must be evidence-based, broadly understood, clinically accepted, stable over their useful lifetime, and reasonably amenable to practical improvement efforts. Measures must also enable providers to care for sicker patients without penalty, establishing a need for appropriate risk adjustment. Numerous metrics now exist across multiple dimensions of performance for a limited set of clinical conditions. Organizations such as the Leapfrog Group and the National Quality Forum are building consensus around a broad range of performance metrics within employer and multi-stakeholder groups respectively. While difficult and political, these efforts should be continued and those measures applied accurately to organization and individual performance to match accountability and control with reward.
Use Both Patient- and Population-Level Clinical and Efficiency Measures, with a Special Emphasis on The Complex Patient. Second-generation P4P programs will utilize both patient- and population-level clinical and efficiency measures. Detailed, patient-specific clinical metrics are important and hold wide appeal for clinicians who generally think of specific patients and actions. However, it is not sufficient to target measures for a limited number of chronic diseases. This requirement has two implications. First, special attention should be paid to developing and implementing metrics for the complex patient. From a clinical standpoint, 40 percent of patients with a single chronic disease have more than one disease.18 From an efficiency standpoint, beneficiaries with three or more chronic conditions account for nearly 90 percent of total Medicare spending,19 with similar patterns in commercial populations. Second, buyers need to make decisions for large populations of consumers, some with specific illnesses and others who may become ill in the future. As a result, even perfect information on a limited number of chronic diseases is not fully actionable for payers. Infrastructure measures such as Leapfrog criteria, participation requirements for regional clinical data exchange, or adherence to clinical performance improvement processes can help address this problem. In addition, other broad measures such as inpatient complication rates are now readily available. Complications impact more than 25 percent of all inpatients and on average cost more than $5,000.20 For hospitals, a clinically tested, peer-reviewed complication measure provides important insight into organizationlevel clinical performance.
Maximize Use of Readily Available, Routinely Collected Data. Second-generation P4P programs will rely heavily upon readily available, routinely collected data to ensure that data collection burdens do not limit provider participation. Particular emphasis should be placed upon data captured electronically through standard care processes. Technology to integrate and normalize data from disparate systems is now readily available and proven effective. This burden also can be reduced by coordinating P4P programs, particularly within local markets. By contrast, today nearly all payer-based programs define their own measures. As P4P programs expand, providers exposed to multiple custom programs would face overlapping or conflicting data collection and analysis requirements that would frustrate and limit their ability to care for patients and improve clinical performance. Manual data collection should be limited only to those critical data elements used across P4P programs that simply cannot be captured otherwise.
Work with Technology Suppliers to Integrate P4P Activities Directly within Care Management and Revenue Cycle Management Applications. In order to truly move into the mainstream, second-generation P4P programs must work seamlessly with providers routine care and revenue cycle management activities. To accomplish this, payers and providers must work closely with technology suppliers to integrate P4P data collection and analysis activities directly into their care management applications. Similar integration should occur within revenue cycle management applications as P4P becomes a larger component of the providers revenue stream. Technology is now widely available to accomplish this goal through composite application development and Web services enablement without replacing existing systems. Further, P4P performance measures should not only be collected at the population level for reporting and payment purposes, but must also be embedded directly into the realtime provider workflow. If the metrics meet the validity and usability tests identified above, they should serve as microsentinel events to drive patient care to new levels of quality and efficiency.
Increase Minimum Patient Participation and Payout Thresholds. In general, a significant minority of a providers revenue stream (by patient volume) must be impacted in order to garner appropriate attention. Of 37 P4P programs recently reviewed, participating enrollee populations ranged from 1 percent to 40 percent, with a mean and median of 11.7 percent and 7 percent respectively.10 For providers, these numbers are diluted both by each payers proportion of that providers practice and the proportion of patients within the disease areas targeted by the program. Most incentive programs place less than 5 percent of provider compensation at risk across multiple diseases and several metrics per disease, substantially diluting the financial leverage of any given metric. Finally, most differential payments to physicians are made at the group or Tax ID level rather than to individual physicians directly. Second-generation P4P programs will target a broader component of a providers patient mix through collaboration with other payers, government, or as a simple requirement that P4P metrics be measured on all patients. Impacting more than 10 percent of a providers practice will move P4P from the experimental into the mainstream.
Eliminate Offsetting Incentives. Most providers find it difficult and often ethically inappropriate to provide different types of care to patients with the same clinical problem but different payers. For example, if the practice change desired by a P4P program results in a bonus payment of 2 percent, but that very same change produces an offsetting loss of 8 percent from other payers, it may be untenable to implement. Secondgeneration P4P programs will acknowledge the presence of these conflicting incentives and integrate solutions explicitly into the programs. This could be accomplished by altering the payment mechanism broadly (e.g., a switch from fee-for-service to capitation) or by narrowly establishing payment and savings targets that acknowledge and reward providers for quality and efficiency improvements within acceptable fraud and abuse limitations.
Be Bold and Creative in Future Program Design. Secondgeneration P4P programs can benefit from other creative design components. Opportunities that show potential promise include: working with insurers and lawmakers to reduce malpractice premiums and liability for those that meet agreed-upon quality processes and metrics, challenging providers to offer warranties (i.e., the promise of a particular measure of success, targeted to patient and/or procedure) or guarantees (e.g., IVF, weight-loss),21 and including consumer-level metrics (e.g., participation in and achievement of smoking cessation or chronic disease compliance goals) in addition to provider measures.
Summary
In the past few years, concerns over health care quality and costs have produced numerous pay for performance initiatives. Although some have shown isolated promise, numerous barriers remain, including the need to: use agreed-upon, actionable, accurately attributed performance measures; use both patientand population-level clinical and efficiency measures (with a special emphasis on the complex patient); maximize use of readily available, routinely collected data; work with technology suppliers to integrate P4P activities directly within care management and revenue cycle management applications; increase minimum patient participation and payout thresholds; and eliminate offsetting incentives. Now is the time for a second generation of bold and creative P4P programs that address these barriers and truly drive high quality, efficient care. We all have much to gain from improved health care delivery: personally, professionally, and socially.
Endnotes
1 Wennberg, J., and the Dartmouth Medical School et al. The Dartmouth Atlas of Health Care in the United States 1999 - The Quality of Medical Care in the United States: A Report on the Medicare Program.
2 Chassin, M.R., et al. Variations in the use of medical and surgical services by the Medicare population. New England Journal of Medicine. 1986; 314:285-90.
3 Leape, L. L., Error in medicine. The Journal of the American Medical Association. Dec. 21, 1994; 272 (23):1851-7.
4 Bates, D.W., Cullen, D. J., Laird, N., Petersen, L. A., Small, S. D., Servi, D., Laffel G., Sweitzer, B. J., Shea, B.F., Hallisey, R., et al. Incidence of adverse drug events and potential adverse drug events. Implications for prevention. ADE Prevention Study Group. The Journal of the American Medical Association. 1995 Jul 5; 274 (1):29-34.
5 Leape, L. L., Brennan, T. A., Laird, N., et al. The nature of adverse events in hospitalized patients: Results of the Harvard Medical Practice Study II. New England Journal of Medicine. 1991; 324:377-84.
6 McGlynn, E. A., et al. The quality of health care delivered to adults in the United States. New England Journal of Medicine. Vol. 348, No. 26, June 26, 2003, 2653-45.
7 Healthcare Spending Reaches $1.6 Trillion in 2002. CMS News. Thursday, January 8, 2004. www.cms.hhs.gov/media/press/release.asp?Counter=935
8 Leatherman, S., et al. The business case for quality: case studies and analysis. Health Affairs. 2003; 22 (2):17-30.
9 Berwick, D., et al. Open letter: Paying for performance: Medicare should lead. Health Affairs. 22 (6), 8-10, 2003.
10 Rosenthal, M. B., et al. Paying for quality: Providers incentives for quality improvement. Health Affairs. 2004; Vol 23, Issue 2, 127-41.
11 Baker, G., et al. Pay for Performance Incentive Programs in Healthcare: Market Dynamics and Business Process. A Research Report, 2003.
12 National Health Care Purchasing Institute. Provider Incentive Models for Improving Quality of Care, March 2002.
13 MEDPAC Report to Congress: Variation and Innovation in Medicare. Chapter 7: Using Incentives to Improve the Quality of Care in Medicare, June 2003.
14 Mehrotra, et al. Employers efforts to measure and improve hospital quality: Determinants of success. Health Affairs. Vol 22, Issue 2, 60-71, March/April 2003.
15 Chassin, M., Achieving and sustaining improved quality: Lessons from New York State and cardiac surgery. Health Affairs. Vol 21, Issue 4, 40-51, July/August 2002.
16 Press Release: HHS to Launch Medicare Demonstration to Promote High Quality Care in Hospitals, Thursday, July 10, 2003. www.hhs.gov/news/press/2003pres/20030710.html
17 Press Release: CMS Announces Guidelines for Reporting Hospital Quality Data. Wednesday, January 28, 2004. www.cms.hhs.gov/media/press/release.asp?Counter=955
18 Milbank Memorial Fund. Value Purchasers in Healthcare: Seven Case Studies. New York: Milbank Memorial Fund, 2001.
19 Doherty, R. B., How performance measures can actually help internists. ACP Observer, November 2003. www.acponline.org/journals/news/nov03/washington.htm
20 Multiple chronic conditions reference.
21 MEDPAC Report to Congress: Medicare Payment Policy. March 2004.
22 Cohen, J. and Krauss, N. Spending and service use among people with the fifteen most costly medical conditions, 1997. Health Affairs. Vol 22, Issue 2, 129-138, March/April 2003.
23 Brailer, D. J., et al. Comorbidity-adjusted complication risk: A new outcome quality measure. Medical Care 34 (5): 490-505, 1996.
24 MacStravic, S., Price for performance. HealthLeaders, March 5, 2004. www.healthleaders.com/news/feature52903.html?contentid=52903&CE_Session= 5f88acbc0ccb3a189e642d08ab3ac1d2
