Paying for Care: In Depth

A stethoscope atop a heap of 100 dollar bills

Photo by Natalia Merzlyakova/Fotolia

As decades of research have shown, how providers are paid strongly influences their approach to clinical practice. In the early 1990s, the fee-for-service payment model dominated the health care landscape. But steadily rising health care costs, coupled with ongoing concerns about quality, stimulated alternative payment models intended to influence physician behavior in ways that lowered costs while improving patient outcomes.

We take an in-depth look at how some of these models work and explore their effects when implemented in various settings.

Transforming How Providers Are Paid

The evolution of payment systems over the last 25 years documents an ongoing effort to incentivize physicians to provide better care, provide lower cost care, or both.

Pay for Service

Since 1992, when Congress instituted the Medicare Fee Schedule for physician services, the dominant payment model in the Medicare program has been an approach called "fee for service." The fee-for-service payment model pays physicians separately for each exam, test, and treatment they provide. As a result, physicians have a strong incentive to supply as many services as possible. The more services they provide, the more they get paid, irrespective of whether the care provided is of high or low quality. At the same time, some important types of services, such as efforts to coordinate care, are not typically reimbursed under fee-for-service payment systems.

Ironically, ordering lots of tests and treatments does not ensure that patients receive the care they need. A 2003 RAND study, conducted in 12 communities nationwide, determined that adult patients were receiving recommended care only 55 percent of the time. Other studies found similar outcomes for the elderly and for children.

With no incentives to conserve, paying-per-service also contributed to other undesired effects that drove up costs, such as:

  • Overuse of services, some which were not necessary and
  • Duplication of services such as lab tests or imaging.

Hospitals and doctors might actually benefit as a result of poor care—for example, if a patient needed to be readmitted to the hospital due to poor care coordination. And expensive new technologies and treatments—like specialty drugs—were hitting the market, contributing to the growth in health spending.

Pay for Quality

In the fee-for-service model, the more services physicians provide, the more they get paid—irrespective of the quality of care.

In the late 1990s, private commercial health plans were creating a new approach to paying for care, based on the idea that financial incentives could be used to drive improvements in quality performance. One of the first models tested was pay-for-performance (P4P). This approach first measures provider performance on a set of quality of care measures—such as the percent of patients whose high blood pressure is under control. It then financially rewards providers who attain a high level of performance on those measures or improve their performance on them compared to the prior year.

Equity is an important concern in P4P schemes. Providers who have a disproportionate number of patients with low socioeconomic status tend to perform less well than other providers on the quality measures used in P4P approaches. This is because disadvantaged individuals tend to have poorer overall health and more chronic conditions. They may also see a doctor less frequently. If quality measures are used in the same way to assess care for lower SES groups as for the rest of the population, incentive payments may be redistributed away from providers who most need them to improve care. Case-mix adjustments and other modifications can be made to ensure that providers are not "penalized" for caring for disadvantaged patients.

Initial experiments showed that P4P had modest effects on quality of care, perhaps because the financial incentives in early efforts weren't sufficiently large to engage providers. The public and private sectors have continued to experiment with financial incentives as a way to drive improvement in care.

It remains unclear which P4P design elements are critical to motivate quality improvement. A recent systematic review found slight evidence that P4P programs could improve processes of care, but no clear evidence that they improve patient outcomes.

An additional concern has been that P4P does not focus on value in health care, where value is defined as both the costs of delivering care and patient outcomes. Recently, private health plans and Medicare have signaled a shift in how they pay providers—going beyond "paying for quality performance" to "paying for value."

Pay for Value

Value-based purchasing (VBP) provides financial incentives to health care providers who deliver "value" in health care, where value means both the outcomes of care and the costs of delivering it. The Affordable Care Act (ACA) called for the Centers for Medicare and Medicaid Services (CMS) to roll out VBP in all of the settings in which it delivers care—hospitals, home health, skilled nursing facilities, etc.

Largely based on work that RAND conducted in 2007 to help CMS design a VBP program for hospitals, CMS implemented a Hospital Value-Based Purchasing Program. Starting in 2012, Medicare began rating hospitals on their performance on clinical quality and patient experience measures and tying incentive payments to performance.

CMS is now seeking to implement value-based payment across all of its payment systems. In 2015, then Health and Human Services (HHS) Secretary Sylvia M. Burwell announced measurable goals and a timeline to move the Medicare program, and the health care system at large, toward paying providers based on the quality, rather than the quantity of care.

Specifically, in 2016, 85% of Medicare fee-for-service payments were to be tied to quality or value through programs such as the Hospital Value Based Purchasing and the Hospital Readmissions Reduction Programs; the target increases to 90% by 2018.

In 2016, 30% of Medicare payments were to be tied to quality or value through alternative payments models such as accountable care organizations (ACOs) or bundled payment arrangements. CMS has already hit the 30% target. The goal increases to 50% by 2018. This is the first time in Medicare's history that HHS has set explicit goals for alternative payment models and value-based payments.

Measuring Success in Value-Based Purchasing Programs

Value-based purchasing (VBP) models are still a work in progress. In a recent study, RAND experts summarized experience with value-based purchasing in both the public and private sector. Their assessment is based on the published literature and on consultation with experts.

Recommendations emerging from their assessment are grouped into two broad categories:

  • Design and implementation: What do successful VBP programs look like? How should VBP programs be implemented to achieve the desired results?
  • Monitoring and evaluation: What kind of information do we need to understand where VBP works and under what conditions?

For example, recommendations include setting performance targets, engaging providers in designing and implementing VBP programs, using a broad set of evaluation measures that emphasize patient outcomes, augmenting quantitative assessment of programs in the public sector with qualitative evaluations to capture private-sector experience, and evaluating new models of VBP as they emerge.

The study also recommends that HHS develop a research agenda to fill important gaps in our current knowledge about VBP.

In 2012, Medicare began rating hospitals on their performance in clinical quality and patient experience measures and tying incentive payments to performance.

Shifting Payment Structures to Influence Provider Behavior

Paying for quality or paying for value entails shifting how physicians are paid in order to modify their behavior. However, these incentive systems are largely built on top of a fee-for-service system. More fundamental shifts in organizing, delivering, and paying for care have been proposed to drive significant improvements in cost and quality. The ACA fostered experimentation with several alternative models, including:

  • Accountable care organizations
  • Bundled payments
  • Medical homes

Accountable Care Organizations (ACOs)

An ACO is designed to integrate different components of care—primary care, specialists, hospitals, home health care, etc.— and ensure that all of the parts “work well together.” The arrangements of ACOs vary:

  • In some arrangements with payers, ACOs that save money while also meeting quality targets keep a portion of the savings.
  • Under other arrangements, the ACO is also responsible for some of the incurred costs.
  • In order to qualify for bonus payments, all ACOs must meet specific quality benchmarks, including focusing on prevention and managing patients with chronic diseases.

ACOs are proliferating across the country, not only for Medicare beneficiaries but also for patients with private insurance. By December, 2017, there will be an estimated 780 ACOs nation-wide, covering 23 million lives.

Pilots will help build the experience base needed to assess how well ACOs achieve their twin goals of quality improvement and cost containment. For example, Blue Shield of California gave $20 million to 18 provider organizations to develop health information technology infrastructure and clinical delivery systems for supporting ACO implementation. A RAND team evaluated the pilot, and study results are forthcoming.

Bundled Payment

The Affordable Care Act fostered experimentation with alternative models of care delivery and payment.

A medical team works together to provide unified care

Photo by Africa Studio/Fotolia

Bundled payment provides a single payment for “bundles” of related services during an episode of care. For example, a heart operation or a hip replacement would be paid with one payment, rather than paying the hospital, physicians, and other medical providers for each unit of service they provide. The expectation is that this approach will encourage providers to work together to provide care more efficiently and to eliminate duplicative and ineffective treatment. The bundled payment approach is supposed to be better for the consumer and better for the bottom line.

The concept of bundled payments is appealing, but it has proven difficult to implement in practice:

  • A RAND systematic review (PDF) of bundled payment interventions, commissioned by the Agency for Healthcare Research and Quality (AHRQ), found weak but consistent evidence that bundled payment programs have been effective in containing costs without major effects on quality. However, there was insufficient evidence to identify design or contextual factors of bundled payment systems that were related to program effectiveness.
  • RAND reported the results of the first evaluation of bundled payment implementation in the private sector—evaluating the PROMETHEUS Payment model. None of the pilot projects had a bundled payment system in place after three years of implementation, nor had they executed contracts between payers and providers. One reason for the lengthy delays may be that the payment model is extremely complex and is built on already-complicated fee-for-service payment models.
  • To determine whether bundled payment could be an effective payment model for California, the Integrated Healthcare Association convened stakeholders (health plans, hospitals, ambulatory surgery centers, physician organization, and vendors) to develop consensus on how to implement bundled payment for orthopedic procedures. The participants were enthusiastic, but the program failed to meet its goals, due to recruitment challenges, regulatory uncertainty, administrative burden and concerns about financial risk. A few contracts were signed, but not enough to actually test how bundled payment affected quality and costs.

As RAND experts have observed, it's not unusual for a demonstration to fall short of its original objectives. Learning from such cases is part of the innovation process. This is especially worthwhile for bundled payment, which has many potential benefits for patients, providers, and payers.

Medicare Experiments with Bundled Payment

Medicare has decades of experience with bundled payments. The Prospective Payment System of 1983, in which Medicare reimbursed hospitals a predetermined, fixed amount for a bundle of inpatient care based on the nature of care being provided, is credited with dramatically reducing Medicare spending.

Medicare's new Bundled Payments for Care Improvement (BPCI) initiative, phased in between 2013 and 2015, was intended to improve coordination of care and reduce costs by bundling inpatient and post-hospital care across providers. Hip fracture and joint replacement seemed to be ideal candidates for the National Pilot Program on Payment Bundling because they had strong potential for cost savings but posed less financial risk for providers than other bundles. An implemented bundled payment for surgery included the surgery, preparation for the procedure, and immediate post-operative care.

Initially, Medicare expressed concerns that the valuations for post-operative care and the global surgical packages might not be accurate. However, as part of MACRA, Congress mandated that CMS develop and implement a process to gather the necessary data to appropriately value surgical care.

Medical Homes

Group of doctors and nurses having a meeting

Photo by skynesher/iStock

Medical homes are primary care practices designed to provide comprehensive, personalized, team-based care, using patient registries, electronic health records, and other advanced capabilities. Medical homes are intended to improve quality while also containing costs. Health plans provide larger reimbursements for practices that achieve formal recognition as a medical home.

RAND has been evaluating implementation of medical homes in many settings to better understand their effect on quality, utilization, and cost. For example, we evaluated the Southeastern Pennsylvania Chronic Care Initiative, one of the earliest and largest multipayer medical home pilots conducted in the United States. Results suggest that the pilot produced modest improvement at best in quality and patient experience, but there was little evidence of effects on costs.

However, a recent study of a three-year medical home intervention in Northeast Pennsylvania showed improvements in quality, increased primary care visits, and lower use of emergency, hospital, and specialty care. Interviews with practice leaders revealed that both initial and ongoing costs of transforming a practice into a medical home ranged widely. The highest costs involved adding new staff to help better coordinate patients’ care; ongoing costs were larger than the one-time cost of transformation.

Practices receive subsidies to make the shift to a medical home, but this study suggests that transformation could be especially challenging for small and independent practices, which have higher per physician costs. The variability of costs across the practices suggests that one size of subsidy does not fit all practices. In addition, since ongoing costs are much larger than one-time costs, subsidies to promote transformation may need to have a longer time horizon.

No two medical home pilots are alike. Pilots differ both in nature (the context in which the medical home implemented), and nurture (the way in which the medical home is designed to operate). Given such differences, researchers suggest there is more to be gained by comparing the pilots rather than by lumping them together, including both "negative" and "positive" results.


The Medicare Access & CHIP Reauthorization Act (MACRA) replaced the Sustainable Growth Rate (SGR) formula, a calculation intended to ensure that the yearly increase in the expense per Medicare beneficiary did not exceed the growth in gross domestic product (GDP). According to the SGR schedule, if physician spending in a given year exceeded a target based on overall economic growth, Medicare fees were to be decreased accordingly. But Congress regularly overrode cuts in reimbursement rates.

Under MACRA, starting in 2019, physicians can choose one of two payment tracks:

  1. The Merit-Based Incentive Payment System (MIPS) would reimburse physicians on a fee-for-service basis with adjustments for performance. MIPS included bonuses or penalties depending on the quality of care physicians delivered, their meaningful use of electronic health records, and improvement in clinical practice.
  2. The second track includes physicians who opted to participate in an alternative payment model (APM). Exactly what is considered an "alternative" model will likely change over time. But its definition will significantly affect physician payment.

An APM is intended to change provider behavior. So a vital piece of information in defining and assessing any APM is how physicians respond to the APM's features. To explore that issue, a RAND team used a simulation model called PADSIM to estimate how sensitive physician behavior is to APM design. The overall purpose of PADSIM is to determine how providers respond to changes in "payment policy," and to understand the relationship between the payments they receive and the services they provide.

PADSIM's analysis shows that MACRA does change provider behavior, and Medicare payments will be lower under MACRA than without it. However, MACRA's largest effect might be on hospital revenue: physicians may respond to payment incentives by, for example, working to avoid hospital readmissions or reducing use of hospital care. The model also underscores how strongly provider participation in APMs depends on APM design and the kind of incentives offered. Design and incentives will drive physician's participation.

MACRA is definitely a work in progress, with multiple challenges to its success. For example, it may not be implemented as designed. The definition of APMs may change. Well-designed APMs might not be available, leaving physicians without the option of following the second payment track. And evidence of how APMs affect care is just emerging.

How Can Alternative Payment Models Become More Effective?

A joint study by RAND and the American Medical Association provided valuable details about how a number of alternate payment models affect physicians and physician practices. In case studies of 34 physician practices in six diverse states, researchers examined effects of bundled payments, shared savings, pay-for-performance, capitation, retainer-based (concierge) practices, ACOs, and medical homes.

They found that physician practices are indeed changing in response to the new models, and they are investing in data management capabilities to track their performance. But practices say adapting to the goals of the new payment strategies is difficult when necessary data are not available, or when payment models are in conflict because they use different performance measures tied to payment. Physicians also said that alternative models had not substantially changed how they delivered patient care.

Physicians in leadership positions were optimistic about alternative payment models; most physicians not in leadership roles expressed some apprehension.

The study team concluded that changing the payment system probably isn’t enough to improve patient care. Medical practices will need support and guidance if the new models are to succeed. They will need resources to collect and analyze data. And conflicts among performance measures used to determine payment will need to be resolved.

What’s Next for Alternative Payment Models?

Changing the payment system probably isn't enough to improve patient care.

The success of alternative payment models depends on how physician practices and their patients change care delivery. It will be essential to seize every opportunity to improve how these models work, how they are implemented, and how physician practices respond. There is widespread experimentation with incentives to change provider behavior, but poor understanding about how best to design and use incentives to improve health system performance.

Particularly challenging is that little evidence exists on which type and mix of incentives are effective and in which context. Prior research has been handicapped by the lack of systematic collection of information on incentive design features and implementation contexts. A current effort is designed to fill this gap.

The goal of the RAND Center of Excellence on Health System Performance is to identify, classify, track and compare health systems in today's health care markets and to characterize the attributes of high performing health systems – those that are able to more nimbly translate new research evidence into clinical practice and thereby improve quality, reduce costs, and achieve better outcomes for patients.

Particularly relevant for successful implementation of alternative payment models will be a Center catalog of financial and non-financial incentives that are used to drive health system and provider behavior. The study team will identify the type, mix, and strength of incentives applied externally (by payers) and internally (by health systems) to assess the alignment between external and internal incentives, and to examine the relationship between the mix, strength, and alignment of incentives, the adoption of new research evidence, and health system performance.