The Employer-Led Health Care Revolution


Intel was rigorously managing its equipment suppliers but not its health care suppliers…….Intel decided it  could use its purchasing power in markets where it had operations to influence health care players—care providers, health plan administrators or insurers, and other employers—to rise above their competing self-interests and work together to redesign the local health care system….

The Employer-Led Health Care Revolution


The Employer-Led Health Care Revolution

In the years leading up to 2009, Intel tried a number of popular approaches to tame its soaring health care costs. To encourage employees and their families to be more involved in the purchase of their care and aware of its actual cost, the company implemented “consumer-driven health care” offerings such as higher-deductible plans with lower premiums, tax-advantaged accounts, and tiered-provider options. To save employees time and improve access, it opened primary care clinics at Intel work sites in Oregon, New Mexico, and Arizona. It offered wellness and fitness incentives, including optional annual health checks that would reduce premiums or deductibles, health coaches, and free on-site fitness classes.

While those programs generated improvements in employee awareness, engagement, and accountability, it had become clear by 2009 that they alone would not enable Intel to solve the problem, because they didn’t affect the root cause: the steadily rising cost of the care that employees and their families were receiving. Intel projected that expenditures for its 48,000 U.S. employees and their 80,000 dependents would hit $1 billion by 2012—triple the amount it spent in 2004. Intel’s leaders were torn: They wanted to protect the bottom line but were reluctant to shift more of the cost to employees, concerned that it would become harder to attract and retain top talent.

One of us (Patricia McDonald) suggested another option: Intel could use its purchasing power in markets where it had operations to influence health care players—care providers, health plan administrators or insurers, and other employers—to rise above their competing self-interests and work together to redesign the local health care system. Specifically, the company would use its deep expertise in supply chain management to improve quality, remove waste, and thereby reduce costs in both the clinical and administrative sides of local health care enterprises while putting the needs of their customers—patients—at the center of everything they did.

Intel was rigorously managing its equipment suppliers but not its health care suppliers.

Intel would urge the health systems to standardize work by adopting best-practice clinical processes and adapting them to their own situations. In this case, the source would be Virginia Mason Medical Center, a health system based in Seattle. It was one of several providers in the United States that employed a version of the famed Toyota Production System to make its processes “lean”—in other words, strip them of activities that did not add value and caused delays or waits in patient care. Intel would pay for the clinical processes and Virginia Mason’s expertise in installing them and would train people at the local health systems to use Intel’s version of TPS to adapt them. Finally, Intel would enlist its health plan administrator, Cigna, to contribute the claims data required to establish priorities and track progress.


The Toyota principles can be applied in operations involving expertise.

Intel’s pilot Healthcare Marketplace Collaborative (HMC) was launched in metropolitan Portland, Oregon. Over five years, it successfully implemented new clinical processes for treating six medical conditions and for screening patients for immunizations status and illnesses such as diabetes and high blood pressure. Although assessing the HMC’s full impact was not easy—and in a number of cases impossible given how the experiment was designed—the results that could be measured were significant: The HMC reduced the direct costs of treating three of the conditions by 24% to 49%—a tremendous accomplishment in an industry where slowing the rate of cost increases is considered a major achievement.

The HMC also emphasized evidence-based care (clinical decision making backed by validated research); eliminated unnecessary care; allowed patients to access care and return to work faster; generated high levels of patient satisfaction; and cut more than 10,000 hours of waste in business processes.  What’s more, it did all this within the confines of today’s fee-for-service reimbursement system, which is widely considered a major impediment to improving the U.S. health care system.

The need to accelerate the transformation of health care in the U.S. is urgent—for both patients and employers. We have seen some hopeful signs that the tide may be turning: Thanks to the Affordable Care Act, the proportion of adult Americans without health care coverage fell to 12.9% in 2014 from 18% in 2013. And the rate of increase in U.S. health care spending has recently slowed, although it’s hard to know whether that’s simply a by-product of the Great Recession. Still, the crisis is far from over.

The Healthcare Marketplace Collaborative model has the potential to be a game changer. One of us (Lindsay Martin) led a two-year research project by the Institute for Healthcare Improvement to identify initiatives in which employers or unions, health plans, and care providers joined forces to redesign the local health care system to achieve the triple aim of improving the health of the local population, reducing the per capita cost of care, and enhancing the patient experience. Of the dozen efforts that IHI studied, the HMC model stood out in terms of its results and its potential to be replicated.

We believe that other large employers can and should follow Intel’s example. As large purchasers of health services and experts in quality improvement and supplier management, corporations are uniquely positioned to drive transformation of health care in the United States.

The Birth of the Portland Collaborative

In 2007, Pat McDonald was the manager of what was then Intel’s highest-performing chip factory in the world: Fab 20, in the Portland suburb of Hillsboro. While exploring how her plant might apply a lean approach to solve problems in complex engineering processes, Pat attended a conference whose speakers included another coauthor of this article, Robert Mecklenburg, of Virginia Mason.

A leader in applying lean techniques in health care, Virginia Mason had persuaded Starbucks, Aetna, and (later) other major employers to collaborate in a successful effort in Seattle to improve the processes for treating a number of conditions. The health system had done this in response to a threat by Aetna to exclude Virginia Mason from its provider network because the prices it was charging for some specialties were higher than competitors’. McDonald toured Virginia Mason’s facilities and was amazed at what she saw. For instance, the flow of work in the pediatric clinic was so efficient that the waiting room was empty.

In 2009, McDonald was invited by Richard Taylor, Intel’s senior vice president and director of HR, to join a committee charged with figuring out how to bring the company’s health care costs under control. “Why not solve the health care problem the same way we would solve manufacturing problems?” she said. She pointed out that Intel rigorously managed its equipment suppliers, monitoring their safety, quality, and costs, but not its health care suppliers, because like most large employers outside health care, it felt it lacked the expertise.

McDonald insisted that wasn’t true. She told the committee about the Seattle effort and suggested that Intel create a health care collaborative in metropolitan Portland, where its health plan covered nearly 18,000 employees and their nearly 21,000 dependents. The committee agreed, and the Healthcare Marketplace Collaborative was born. McDonald was called on to lead the initiative. She enlisted two key champions: one from HR (Taylor) and one from manufacturing (Steve Megli, a vice president of the Technology Manufacturing Group, who was interested in applying the lean approach to health care).

Some care delivery organizations aren’t waiting to be told how to improve the system—they’re already doing it. Here’s what’s working.

Mecklenburg, the leader of the Seattle initiative, was brought on as a key adviser. Previously Virginia Mason’s chief of medicine, he had recently become the medical director of its new Center for Health Care Solutions, a unit formed to encourage employers to use a collaborative model with providers and health plans to drive improvements in health care.

Mecklenburg strongly believed that for the approach to spread throughout the country, employers—not health care systems or insurers—should be the driving force. He felt that on the whole, neither providers nor health plans were consistently acting in the best interests of employers or their workers. Providers’ services were too costly and their quality variable. Health plans weren’t reimbursing providers on the basis of quality and were willing to pay for unnecessary visits, procedures, and medicines. Without strong pressure, they would not make enough effort to provide the highest-quality, lowest-cost care possible. Although Mecklenburg was hopeful about many elements of the legislation that would become the Affordable Care Act, he feared that it would not generate relief fast enough. He concluded that only employers, with their purchasing power, were in a position to accelerate the pace of change. McDonald’s invitation to join the HMC experiment presented an opportunity to put his ideas to the test.

Let’s take an in-depth look at the Healthcare Marketplace Collaborative model and the elements that were critical to its success.

1: Make explicit what each player is bringing to the effort

Intel initially invited Cigna; Providence Health & Services, a multistate health care system; and Tuality Healthcare, a small local system with two community hospitals, to join the collaborative. On Providence’s recommendation, two state agencies, Oregon’s Public Employees’ Benefit Board and the Oregon Educators Benefit Board, were asked to participate in 2010. Each organization that joined the collaborative brought skills and capabilities that the others lacked and information that players in conventional health care systems rarely share for the betterment of the overall market. It was important that each group’s unique value be recognized so that all team members would feel they were equals and would be motivated to fully engage. Making explicit what each player brought to the table helped forge a strong partnership, which was necessary to overcome the natural challenges and conflicts over priorities.


Intel brought a huge customer base—its employees and their families constituted a significant proportion of the health care customers in the Portland area. It also contributed its deep expertise in system engineering, improvement methodology, and supplier management. The state agencies—which together provide health care coverage and other benefits for 270,000 active and retired employees of state agencies and universities, school districts, and community colleges and their dependents—extended the endeavor’s reach in the region and added another employer perspective. Perhaps more important, their involvement showed the greater community that the work was being done to benefit everyone, not just Intel employees.


Providence and Tuality each brought a unique perspective and, because of their differing sizes, were critical in demonstrating that standard, lean processes for health care could be created and applied in a range of environments. Having both providers as part of the initiative also broadened access for employees and accommodated their preferences for big or small institutions.

Insurers or administrators.

Cigna played a limited but crucial role: It provided the claims data essential for identifying which conditions should be priorities, establishing baselines for improvement efforts, and tracking progress—something that insurers and administrators generally don’t do. Cigna’s involvement ensured that patients’ privacy rights would not be violated. It should be noted that while the information shed light on the costs and utilization of health care in the region, it was raw data. Intel had to hire a third-party vendor to analyze the data and turn it into actionable information.

Physician leader.

Keenly aware that it had limited knowledge of health care, Intel felt it needed a hands-on adviser to help guide the initiative, so it asked Mecklenburg to serve as physician leader. He was an expert in applying lean techniques to health care, and as a clinician, he would be effective in explaining the approach to the care providers and getting them to engage.

The role of physician leader is critical in an employer-led collaborative. Candidates may come from inside or outside a participating health system. They must have a track record in leading change and be advocates of collaborative decision making. And they must be driven by dissatisfaction with the current state of medicine and the belief that care provider organizations today are wasting employers’ and taxpayers’ money and are not doing their level best for patients.

Among all the players in a collaborative, the anchor should be the employer. In many cases it will be a large corporation, but state governments, pressured by their tight budgets, are also ideal. (The ongoing Bree Collaborative in Washington State is a good example.) The founding employer can invite other like-minded employers, providers, and health plans to come to the table. Upon seeing the results, additional employers may seek to join the effort—or begin to purchase health care differently.

Such an endeavor must be organized as a serious enterprise, not treated as an “extracurricular activity” that employees have to squeeze into their already-full workloads. It requires significant personnel resources. Management and operating teams are needed to bring the stakeholders together and establish a healthy operating rhythm.

2: Establish a shared aim

The purchase of health care services for employees is often a game in which each player—employer, payer, or health care provider—tries to use its market power to secure the best deal for itself in annual negotiations. Payers (insurers or third-party administrators) attempt to wrangle discounts from providers. Providers strive to raise prices independent of cost or quality. Payers and employers try to shift actuarial risk—the risk that the assumptions built into models for pricing insurance policies may turn out to be wrong—to providers. And to offset rising costs, employers force workers to pay more for premiums and shoulder higher deductibles and copays.

Making matters worse, little information is shared among key local players. Health plans and providers often shut employers out of the discussion of costs. Health plans may be contractually prohibited from disclosing negotiated rates with providers. And although progress in getting health systems to share learning has been made via government and nonprofit initiatives, providers within a given market rarely share data on outcomes. The grand result is a system in which it is rare for all the local players to work together in a transparent way to improve the quality and cost of health care and enhance the patient experience.

To break this dynamic, the HMC’s members agreed to focus on an aim that would be in the interests of all the stakeholders, including patients: providing the right care in the right place at the right time and the right cost for Intel employees and families and all other Portland-area health care users. They would strive to eliminate waste, achieve zero defects, and, where possible, focus on keeping people well, reducing the need for reactive care.

Only large corporations, with their vast purchasing power, can accelerate the transformation of health care.

When Intel approached Tuality and Providence about joining the collaborative, each asked—reacting to past experiences with employers—“Does Intel want a separate system for its employees alone?” Intel realized that would undercut the goal of establishing a shared aim and responded with a resounding no. Creating special care for Intel employees would not be beneficial for the health care providers: It could result in parallel clinical and business work processes, which would be inefficient and, in health care, could lead to an increase in “adverse events or errors”—patient injuries due to medical interventions rather than to underlying medical conditions. Moreover, a separate system might limit employees’ choices and would not be in the best interests of the greater community.

3: Don’t reinvent the wheel

Rather than develop new protocols from scratch, the HMC’s two health systems accepted Intel’s proposal to start out by acquiring proven clinical content and work processes—or, to use lean lingo, “value streams”—and quality metrics from Virginia Mason, whose lean clinical processes were evidence based and focused on the patient, addressing convenience, rapid access, cost, patients’ lifestyles, and family considerations as well as quality of care. All this appealed to Intel, which was accustomed to finding the best in the industry and buying it. Intel opted to cover the full price of the value streams, because it believed they would benefit its employees and accelerate the standardization of care in the health systems. Intel also provided training for employees at the health systems in using its proprietary Rapid Integrated Lean version of the Toyota Production System to adapt the Virginia Mason processes to fit their own contexts.

Employer-led collaboratives can draw on a number of sources of processes and expertise. For example, Bellin Health Care Systems, in Wisconsin works directly with employers to drive down costs; ThedaCare, which is also in Wisconsin, offers courses and direct coaching on lean health systems and improvement methodologies; and Salt Lake City–based Intermountain Healthcare, well known for standardizing care, teaches courses in health care quality improvement.

4: Make it flexible

No two health care providers are exactly the same in terms of size, structure, and operations. In some instances, a provider may already have an effective method of treating a targeted condition. In others, internal or structural issues (physical space, available resources, state regulations) may make it difficult, if not impossible, to simply install a clinical process without changes.

Recognizing this, the collaborative agreed at the outset that Providence and Tuality would each decide whether or how to adopt each of the new clinical processes. For example, some of them called for tasks to be performed by an advanced registered nurse practitioner who had earned at least a master’s degree. Because that role did not exist in either organization at the time, Tuality and Providence had to adjust the value streams to have other personnel perform those tasks. In the end, Tuality chose to adopt some form of all the value streams. Providence adopted four but decided that its programs for upper respiratory illness, diabetes, and screening were robust and would be kept; it was still committed, however, to achieving HMC goals for all three.

Proposed changes to any of the Virginia Mason value streams were vetted at meetings of Intel’s global medical director, Donald C. Fisher, MD; Providence’s and Tuality’s medical directors; and the physician leader. The directors evaluated all the changes by testing and monitoring the results and then reconvened to reflect on how each health system could improve.

5: Prioritize on the basis of impact and difficulty

Intel combed through Cigna’s claims data and chose which medical conditions to focus on initially—those whose improvement would most benefit its employees, their dependents, and the company. About two years into the effort, the medical directors at Providence and Tuality selected additional conditions. The group used four criteria to establish priorities:

How can health care professionals ensure that the quality of their service matches their knowledge and aspirations? As a number of hospitals and clinics have discovered, learning how to improve the work you do while you actually do it can deliver extraordinary savings in lives and dollars.

Expenditures and impact on patients.

Intel focused on types of care on which it spent a lot of money and treated the most patients—both Intel’s employees and the community at large. Although the treatment costs for certain conditions are relatively low, they typically result in greater total expenditures because they occur so frequently. Therefore, team members considered frequency as well as cost in setting priorities.

Level of complication and risk.

Intel chose to start with less complicated and less risky conditions to make it easier for Providence and Tuality to put the new clinical processes in place. For example, because of their complexity and the intense emotions that patients and their families typically experience, the collaborative’s steering committee opted not to tackle pregnancy or cancer while in learning mode. In addition, Intel knew from experience that tackling the most complicated challenges first—something teams are often tempted to do because the potential payoff is the highest—increases the likelihood that a program will bog down or fail.

Ease of standardization.

Intel wanted processes that could be standardized easily across multiple care-delivery systems. So it initially chose value streams that Virginia Mason had already developed and successfully implemented in several health care sites.

Benefit to the health systems.

Although Intel set the initial priorities, it recognized that all the stakeholders needed to benefit from implementation of the value stream. Certain “production” costs for the two health systems would be reduced by eliminating unnecessary procedures (such as MRI scans), optimizing staff (for example, using clinical professionals other than physicians to diagnose and treat uncomplicated conditions), and reengineering administrative processes. Revenue would grow as a result of increasing patient throughput—for example, by offering patients with uncomplicated conditions rapid access to treatment and by providing patients with complicated conditions rapid access to specialists (whose schedules were no longer filled by patients with uncomplicated conditions). The benefits of increased volume and reduced costs would more than offset any reduction in revenue associated with eliminating unnecessary care. Indeed, the health care providers would most likely increase their market share as better outcomes and lower costs translated into a stronger financial position.

The goal was to actually reducecosts, not just slow the rate of increase.

Uncomplicated back pain was selected as the first value stream to improve because it was high on Intel’s list in terms of frequency and total cost; Virginia Mason had used this lean process to treat thousands of back patients since 2005 and had solid experience standardizing the clinical process at multiple sites; and Providence and Tuality treated a high number of patients with the condition.

6: Choose simple metrics and goals

U.S. health care providers measure more than 100 indicators of the quality of their clinical processes—such as the rate of ventilator-associated pneumonia, the percentage of patients whose prophylactic antibiotics are discontinued within 24 hours of surgery, and so on—which they report to public- and private-sector bodies. Although they are all valid quality indicators, few are useful in an employer-driven initiative like the HMC. The collaborative needed a set of simple, standard metrics to measure progress; it chose five that had been adopted by the Seattle collaborative and addressed the aim of better, faster, and more-affordable care. And it set audacious goals for each.

Better care.

The HMC used two metrics for this goal. One gauged medical quality: whether or not patients received evidence-based care. The other tracked patient satisfaction: the proportion of patients who responded “probably” or “definitely” to the survey question “Based on today’s visit, would you refer a friend to our medical clinic?” The goal for both measures was 100%.

Faster care.

The HMC choose two metrics here: same-day access to care and return to function (how many days before patients could resume their normal daily routines). The goal was that 85% of patients who called Monday through Friday could get an appointment with an appropriate provider within one business day of their call. For the return-to-function metric, the team members from Intel, the medical directors from the two health systems, and the physician leader determined the targets for each value stream. The goal was for 90% of patients to meet or beat the target.

More-affordable care.

The first metric for affordable care was the total cost to employer and patient of treating a condition—in other words, the fees paid to providers. (Historically, this has been hard to calculate in the United States because most providers charge for each item or service, not for the full treatment of the condition.) The collaborative compared costs from when the need for care arose and when the problem was resolved using both the new approach and the one typically used in the health system. The goal was to actually reduce costs (no numerical target was set), not just slow the rate of increase.

Although not calculated in dollar figures, the return-to-function metric also was considered in gauging progress on more-affordable care. The cost of lost productivity often is much greater than the cost of care—and is something that health systems in the United States typically don’t consider.

There were inevitably differences of opinion about how to implement value streams, and using common measures allowed the health systems to test approaches in parallel and easily compare the results, accelerating decisions. For example, Virginia Mason’s process called for a physician and a physical therapist to jointly evaluate whether a patient whose lower back pain seems to be uncomplicated might have something more serious. Specialist physicians at both Providence and Tuality believed that the therapist alone would suffice. So the two systems tested the approach for three months—Tuality with a physician and Providence without. Recovery rates and satisfaction levels were essentially the same, so both health systems decided that physician involvement was unnecessary.

7: Use one improvement methodology

Getting all the members of a collaborative to agree to use the same improvement methodology is essential. The good news is that virtually any quality improvement approach can be applied, including varieties of the Toyota Production System, Six Sigma, and the Model for Improvement (created by Associates in Process Improvement and used by the Institute for Healthcare Improvement). Intel’s Rapid Integrated Lean, or RIL, approach had many benefits, especially its pace: It was designed to deliver exceptional results in three weeks. It does this by focusing individuals, teams, or managers on standardizing work that is under their control and by concentrating them on problems where they can make the most impact. Intel loaned the HMC several of its lean experts and trained 48 people at Providence and Tuality in the technique. In exchange, Intel required that both providers share results with everyone in the collaborative.

In using RIL to tailor and implement the Virginia Mason processes, a core HMC operating team and specialist clinicians from the two health systems mapped out each value stream on a wall for everyone to see. People from Providence and Tuality provided input on how the workflow would fit into each of the organizations. They highlighted differences in practice and explained why each was needed. The medical directors then agreed on which changes to test.

As the work progressed, the core operating team tested pieces of the value stream (including proposed variations), observed them in action, gleaned insights, and tried again—continually learning and sharing best practices. All changes to the existing workflow were tested in small experiments and then gradually scaled up—an approach that helped get buy-in from people in the health systems. Every week the core operating team met and used Post-it notes to document what had been accomplished and what hadn’t, what could be done differently, and what barriers had been encountered. It typically took 10 to 14 weeks to design and test a clinical value stream and get it ready to be implemented more broadly. Value streams for administrative processes took three to five weeks.

In addition to serving as a tracking mechanism, the visual representation of the work was a team-building exercise. Everyone learned how to work together and how efforts were progressing across all players in the system.

8: Fix the business side

Administrative costs consume an enormous chunk of the money that goes to health care—more than 25% of U.S. hospitals’ expenditures, according to a 2014analysis by David Himmelstein and colleagues published in Health Affairs. So any serious effort to make health care more affordable has to tackle not only the clinical side but also the business side. In addition, administrative processes such as booking appointments and billing have a significant impact on the patient experience.

The HMC used RIL to remove waste and non-value-adding activities from Providence’s and Tuality’s business operations, including billing, inventory management, checking in patients, processes for getting patients into rooms to be treated, and cleaning. As of March 2014, a total of 48 business processes had been improved at the two systems, generating an estimated $2.6 million in annual savings. Since then, both have continued their improvement efforts.

Understanding the Challenges

Naturally, the Healthcare Marketplace Collaborative encountered challenges. As the anchor of the initiative, Intel brought the group together, implemented RIL, and drove players to abide by the process. Sometimes people at the other member organizations resented being told what to do, so Intel had to be careful not to push too hard.

Persuading clinicians in the health care systems to accept standard clinical processes wasn’t always easy. At Intel, if a manufacturing process was deemed an improvement or a best practice, it was documented and then implemented at all the company’s factories. There were specific methods and procedures for putting new process steps in place. This was not the case at the health systems. Not all clinicians, even within a specialty, practiced the same way; standard protocols often hadn’t been instituted. Some medical leaders of clinics and individual care providers didn’t like being told what was best for their patients. The collaborative had to find a balance between allowing highly trained clinicians to use their judgment in unique cases and creating standard approaches for treating the vast majority of patients. In addition, the health systems differed in their willingness to make changes that would reduce office visits with a physician and, as a result, their reimbursements.

The medical directors on the HMC team played an instrumental role in overcoming such resistance. Launching value-stream efforts at Providence and Tuality clinics and then demonstrating the results also proved effective in getting people at other sites on board. Ultimately, embedding protocols proven to provide the best care in an electronic-medical-record system will help clinicians stick with them. Virginia Mason already does this. For example, its clinicians cannot order an MRI for lower back pain unless they identify a medical indication for it in the EMR system or phone a designated expert.

Persuading patients to abide by the new process also requires a concerted effort. For instance, the health systems couldn’t force patients with uncomplicated back pain to go directly to a clinic and not see their primary care physicians first. Intel and the two state agencies understood this and used a number of means to publicize the streamlined services and encourage people to use them. Still, it takes time to change mindsets.

Sometimes the health systems lacked sufficient staff to implement a new value stream smoothly. For instance, staff turnover and difficulty lining up surgeons and radiologists explain why same-day access for the breast problems value stream was so low. Viewing those poor results as an improvement opportunity, Tuality decided to dedicate another surgeon to the breast problems value stream.

Another big challenge was getting data. In a number of cases, the health systems lacked the resources to track metrics; in others, it was difficult to track particular kinds of patients, such as unemployed Medicare patients. Cigna provided the cost data, but there was a frustrating lag of several months because the insurer had to wait for claims to be submitted, approved, and paid. In other industries, gathering cost data takes minutes or even seconds. (These slow data turns—the norm in the health care industry—are a major barrier to rapid improvements in clinical processes.) For some value streams, the HMC couldn’t get meaningful cost data because it had such information only for Intel employees and the number treated by the new value stream was too small to draw statistically valid conclusions. Because of the challenges in getting useful data, it’s wise to bring a data analysis engineer on board early in the development stage of a value stream.

Employers should take the lead in securing better health for local populations and lowering costs.

Finally, agreeing on a uniform set of medical classification codes was difficult. More than 16,000 codes for diagnoses and procedures are used for billing and reimbursement in the United States. Back pain, for example, has many possible codes, and preferences may differ by provider. The medical directors ultimately decided to use the codes in Virginia Mason’s handbooks to conduct cost analyses. But differences between the codes used for existing clinical processes and those used for the new ones made it difficult, if not impossible, to compare results. For the other metrics, the HMC did not use control groups to measure the degree of improvement. Instead, it created extremely ambitious goals and tried, where it could, to use them to gauge progress.

The Healthcare Marketplace Collaborative ended in June 2014, after the improvement process had been established at both health systems and Intel was no longer needed to drive the effort. While results of the HMC experiment were hardly perfect, and should be viewed as coming from a challenging work in progress rather than from an ideal end state, they proved that an employer can engage all the players in a market to accelerate health care reform. HMC’s focus on patient-centered care produced solid cost savings and, more important, behavioral changes grounded in evidence-based medicine, measures, results, and patient satisfaction.

Intel is now applying elements of the HMC approach to purchasing health care services in Oregon and New Mexico and plans to do so elsewhere as well.

The Bree Collaborative, in Washington State, in which Bob Mecklenburg is involved, is taking the Portland approach to the next level. It incorporates goals and metrics like those used in Portland in employers’ contracts with participating health systems, but it involves more employers and health systems, is tackling more-complicated conditions, and has created or is creating a standard bundle of all the services that patients with a given condition require through recovery. Each employer negotiates a fixed price for each bundle, which also includes a warranty against hospital readmissions for avoidable complications. The initiative has already created bundles for joint replacements and lumbar fusion and expects to complete one for coronary artery bypass grafting in the summer of 2015.

We encourage employers to use their purchasing power to drive the transformation of health care in their regions. That means taking the lead in securing better health for local populations, improving their experiences of care, and lowering costs for employees and companies. Employers should choose plans and providers on the basis of their willingness to join them in the effort. Ultimately, companies in such initiatives should report the results of health systems’ efforts to make care better, faster, and more affordable and encourage employees to use the information to select providers.

We urge care providers to join collaboratives in their regions and develop business models in which better outcomes and lower costs translate into improved financial performance. Providers should build new skills to standardize clinical and business processes and strive to integrate all aspects of care, including behavioral or mental health and social services. They should learn to be transparent about prices and outcomes to help employees choose the best providers and to help employers teach their employees that more care is not necessarily better care.

Health plans should develop business models that allow them to succeed in an environment in which health care costs and premiums are decreasing. They should identify high-performing providers of care or those on a positive trajectory. They should make information on providers’ prices and outcomes rapidly available to employers, employees, and their families in a useful form. They should lend their power and insight to efforts to change the payment system so that it is no longer an obstacle to improving outcomes and lowering costs.

In virtually all regions, at least some employers, providers, and health plans will be able to transcend narrow self-interest and cooperate to develop new business models that result in the best health outcomes for individuals at lowest cost. We enthusiastically invite them to join us on the journey.