By Aaron Krol 

January 22, 2014 | It’s only January, but it's already been a bad news year for the Oregon healthcare system. If you’ve been following the news out of the Pacific Northwest, you’ve probably heard that Cover Oregon, the state’s online health exchange established under the Affordable Care Act (ACA), failed to sign up a single applicant for any health insurance plan for more than three months after it went online. It’s grown so discouraging that Oregon is one of only two states, along with Maryland, to consider switching from its own health exchange to the federal site at Healthcare.gov. You may also have heard about a recently-published study that looked at Medicaid recipients in Oregon, and found that they visited the emergency department a full 40% more often than their uninsured peers – a possible omen of serious new burdens for hospitals nationwide, as the ACA enters millions of Americans into public health plans.

Here are two figures you might not have heard: at least 160,000 Oregonians have already gained coverage as a result of the ACA. And over the past year, according to preliminary reports, the state has seen a 9% reduction in emergency department use by Medicaid recipients.

These are just a few of the seeming contradictions that will emerge as the country’s patchwork health care system is disrupted by the patchwork interventions of the ACA. But Oregon will hold a special place in policy debates, thanks to a government that has been eager to experiment with its public health plans, and willing to gather copious data on the results. After years of unconventional policy initiatives, and with the efforts of devoted data crunchers, the state has become an important case study in both what happens when you expand Medicaid in a vacuum – and how that vacuum might be filled.

 

A Controlled Study 

The national news media isn’t often drawn to the methodology of controlled experiments, but a scientific journal with a flair for timing and stage presence can sometimes carry it off. Earlier this month, Science all but dared pundits not to give a nod to the scientific method when it published a controlled trial of Medicaid expansion just one day after the first round of ACA Medicaid coverage kicked in.

The paper was the third to emerge from the Oregon Health Study, which has been ongoing for over five years and may be the best available proxy for national outcomes of expanding Medicaid. Oregon isn’t the most obvious candidate to stand in for a whole nation’s worth of new Medicaid customers: its eligible population is on the small side, for one thing, and overwhelmingly white. But thanks to a peculiar experiment that took place in 2008, that population includes a large cohort that could almost have been purpose-made for research.

In 2008, the state’s Department of Human Services found it had enough money to expand its Medicaid program, the Oregon Health Plan, to 30,000 extra residents. The eligible population, however, was estimated at twenty times that number, and policymakers could see no clear reason to select some of the uninsured for coverage and not others. With no better options, the state opened a lottery for health insurance, receiving around 90,000 valid applications, most of which would have to be rejected.

Although this program was only intended as a fair way to distribute a limited amount of funding, it had an unanticipated consequence. “What this gives us is essentially a randomized trial of public insurance,” says Sarah Taubman, a health policy researcher at the National Bureau of Economic Research (NBER) and the Harvard School of Public Health. In fact, it’s the only program in American history to randomly distribute public insurance, providing both a test group of new Medicaid recipients, and a control group of similar applicants who were eligible for Medicaid but didn’t receive it. “And so we’ve used that to study the effects of Medicaid on a whole range of outcomes,” adds Taubman.

Taubman joined the Oregon Health Study in its early days, after being recruited by principal investigators Amy Finkelstein, an MIT economist, and Kate Baicker, a health policy expert who knew Taubman from Harvard and the NBER. She’s been a co-author on all three papers, whose results have ranged from almost tautological – Medicaid recipients were less likely to have medical debt or catastrophic health costs – to gently encouraging for the efficacy of public insurance plans: recipients in the study used both primary care and prescription drugs more often than the uninsured, and were less likely to be diagnosed with depression.

Some findings have added interesting wrinkles to our understanding of how Medicaid changes the lives of its recipients. Self-reported health, for example, seems to diverge pretty widely from any easy measures of physical wellbeing. Researchers measured the blood sugar, blood pressure and cholesterol levels of thousands of subjects, and found no significant difference between those with Medicaid and those without. Nevertheless, subjects in the Medicaid cohort were around 25% more likely to report their own health as “good,” “very good,” or “excellent.” As Taubman says, “Health is very complex and multidimensional. [We] measure a handful of important markers of health, but not necessarily ones that cause the most symptoms. And so when people answer a question about their health, they may well be thinking about lots of different, other kinds of conditions… [It] may be that individuals feel much better about their health because they feel they have access to health care that they’ll be able to afford.”

It wasn’t until this month’s Science paper, however, that the Oregon Health Study released findings capable of really shaking preconceived notions about Medicaid. A widely-held, if far from universal, belief had held that Medicaid recipients should make fewer demands on emergency departments, because they would have readier access to preventive care in other settings. Instead, the Oregon Health Study’s analysis showed a substantial swing the other way, an increase of around 40% in ED use within their Medicaid group. This was more in line with the views of those health economists who thought basic market theories should apply to the emergency department: lower the cost of visits, and demand will naturally rise.

“Part of why I think emergency department use is such an interesting outcome to look at is because there really were active theories pointing both directions,” says Taubman. And with access to the full hospital ED records, gathered in cooperation with Oregon’s Office of Health Policy and Research, the study could not only settle this longstanding policy question, but also drill into how Medicaid recipients were using the ED differently than their uninsured peers.

Working with those records, the researchers sorted visits into five categories: “emergent, non-preventable” visits, or unforeseeable crisis situations; “emergent, preventable” visits, of the kind that might decrease with better access to preventive care; “primary care treatable,” which need immediate care but could be dealt with at any clinic; and “non-emergent,” using the emergency room for basic clinical services. As it turns out, Medicaid recipients logged somewhat more visits in every category, but all the statistically significant increase fell under the last two. Essentially, it seems that public insurance encouraged patients to get at least some routine care at the emergency department, increasing non-emergency workloads. This is a dangerous trend, as EDs are more expensive than other care settings, don’t establish long-term relationships with patients, and need to be kept clear for emergencies.

It’s important to remember, says Taubman, that “this is happening in a context of increased access to primary care, and increased use of primary care.” That is, the Medicaid population wasn’t flocking to the emergency room for all its basic care needs; they were taking advantage of preventive services and still increasing their visits to the emergency department. Public plan doomsayers rushed to publicize the results. After all, with a huge sample 25,000 ED visits and a p-value of less than .001, the findings could hardly be coincidence, and such a large change in use surely meant the burden of these Medicaid patients on the health care system had to be tremendous.

 

A Dose of Perspective 

Well, with certain caveats. Traveling to the emergency room is not exactly an everyday event. In the period of the study, only around 35% of the control group made any visits to the ED at all. Many of that group made multiple visits, however, resulting in an average of around .67 visits per year among the uninsured. The 40% boost is on top of that figure – meaning that the average Medicaid user made excess demands on the emergency room to the tune of an extra .27 visits a year. Unsurprisingly, this change was too small to show up in subjects' self-reported ED use in patient surveys.

Of course, the point of measuring ED use isn’t to gauge the impact on patients – who receive more or less the same care at the ED that they would elsewhere – but to weigh the cost to hospitals and public health plans. This is a complex factor to estimate, as costs vary widely between different procedures and points of care, but the authors make the effort to be thorough. “We’ve done some back-of-the-envelope calculations,” says Taubman, “which suggest that in the control population, emergency department use would be around 10-15% of their costs.” 

If that estimate is correct, then the increase in ED visits would have racked up an extra 4-6% in costs for Medicaid users over the period of the study, as compared to the uninsured. The Science paper even tries to pin a rough dollar value on that increase: $120 annually. That’s far from negligible – even in Oregon’s relatively small expansion of 30,000 new Medicaid plans, it amounts to over $3.5 million a year – but it’s also a small fraction of overall healthcare costs.

Another crucial piece of context is the timeframe of the research. The Oregon Health Study is justly proud of its scientific rigor, but the rare chance to perform a controlled study of Medicaid has its limitations. “Depending on the data source, we have data for the first year to two years,” says Taubman. “Looking beyond that period of about two years, the Medicaid program was expanded again, and a large number of our controls were able to obtain insurance, which means we don’t really have the random assignment beyond that time period.” If the Oregon Health Study had followed their Medicaid group for more than two years, they would have lost a consistent and reliable control group for comparison.

For the ED use study, the researchers were able to acquire 18 months of relevant data, spanning from March of 2008, when the first new Oregon Health Plans took effect, to September of 2009. So this study is a gold-standard measure of health behaviors in the first year and a half of Medicaid insurance, but can’t comment on whether these behaviors hold over time. It’s possible that the low-income individuals who qualify for Medicaid will keep using the ED for its convenience, finding it hard to schedule primary care appointments around their work demands. It’s also possible that their ED use will taper off as they transition to more routine care, or that greater access to prescriptions and preventive care will slowly reduce their preventable ED visits.

“The long-term outcomes are clearly a very, very interesting question,” says Taubman, “and we certainly would be very excited if we could figure out a way to continue to get after them in such a clear way.” But even with the state’s cooperation, it seems unlikely that a strong control population can be found after 2009. For now, she adds, “One thing that our study really highlights is the value of doing randomized evaluation for these kinds of public policy questions,” which can at least separate the short-term impacts of Medicaid coverage from wider trends that affect all patients. “Our hope is that by providing estimates, we can allow policymakers and providers to make better decisions.”

 

The Changing Public Health Environment 

Those decisions have only grown more urgent since the 2008 Medicaid lottery. A key question that follows this research is how similar 10,000 Medicaid recipients in Portland* will be to the millions of Medicaid enrollees nationally who are entering the system this month. It “may be very different,” says Taubman, “when there’s a linear expansion like we were studying, where 10,000 individuals gained Medicaid, compared to millions of newly-covered individuals. There may be system-wide effects that we didn’t pick up.”

That could be a cause for optimism – if states take systematic approaches to moving their Medicaid populations toward primary care – or for real foreboding, if the much larger influx of enrollees pushes hospitals beyond capacity. Oregon officials have been aggressive about falling into the first category.

In early 2011, one year after the period of the Oregon Health Study came to a close, the Oregon legislature passed HB 3650, the “Health System Transformation” law, with sweeping consequences for Medicaid administration in the state. At that time, “We were in a scenario where we were having rising costs, year after year after year,” says Alissa Robbins, a spokesperson for the Oregon Health Authority (OHA) Transformation Center. “Every state in the country has experienced it, the rising cost of health care. We also knew there was the possibility of more people coming onto Medicaid on January 1, 2014… We needed to do things differently.”

The Health System Transformation changes both how public healthcare is administered in Oregon, and how Medicaid funding is distributed. Regional collectives of clinics, hospitals, and other points of care have been merged into Coordinated Care Organizations (CCOs), which manage state Medicaid funds, make executive decisions, and report to the OHA during periodic reviews. The first CCO was formed in August of 2012, and Oregon now has sixteen of these organizations, covering the entire state and over 95% of its Medicaid recipients. The hope is that local medical groups will have more flexibility to innovate and be more responsive to their patient populations than a statewide authority could. To nudge that innovation forward, CCOs have unprecedented freedom in how they choose to spend their Medicaid funds.

There’s also an element of competition in the new system. Part of the Oregon Health Plan fund has been split off into a “quality pool,” which will pay out bonuses to CCOs that meet statewide benchmarks in certain key metrics, mainly based on patient outcomes for Medicaid users. Those metrics range from enrollment in primary care, to cancer screenings, to hospital readmissions; they also include a few miscellaneous objectives like the adoption of electronic health records. Meanwhile, the larger funding pool for reimbursement of Medicaid services will steadily decrease in the coming years (as measured per patient), forcing CCOs to grow more efficient each year or come up short.

The benchmarks by which CCOs qualify for quality pool funding allow the OHA to set priorities for improving healthcare across the state, without trying to regulate the activities of every clinic and hospital. Looking at the metrics chosen, you can identify both local health dilemmas – Oregonians are less likely than any other Americans to vaccinate their children, so children age 2 and under getting their recommended vaccinations makes the list – and concerns that are troubling health systems across the country. Wherever possible, the benchmarks are set at either the 75^th or 90^th percentile nationally for Medicaid services. (Although for a few, like screenings for substance abuse, national standards aren’t available.) These are compared to baseline levels as measured in 2011, before the CCOs were formed.

CCOs aren’t expected to meet all the benchmarks right away, but they need to make measurable progress in a metric for it to contribute to their funding – at least 10% of the difference between their 2011 baselines and the statewide benchmarks, a goal the state borrowed from a successful program in Minnesota. The more metrics see improvement year over year, the more funding a CCO receives from the quality pool.

Unsurprisingly, one of the metrics in the incentive plan is ED visits. “It’s a huge priority,” says Robbins. “It’s a major cost driver. We know that the work that we’re doing is reflected in that ED visit use.” The OHA set a target of 44.4 ED visits per 1000 Medicaid recipients, which looks ambitious when viewed next to the Oregon Health Study’s recent paper. The new benchmark is less than half the figure logged by Medicaid users in that study – and almost a third lower than the ED use of the uninsured.

But CCOs seem game to try. Robbins lists several local initiatives taken to keep unnecessary ED visits in check. Some CCOs are hiring community health workers who try to identify chronic health problems in the home; Robbins cites the case of a child who was repeatedly admitted to the emergency room for asthma, before one of these workers found that a household vacuum cleaner with an improper filter was causing the attacks. She also names a project to place community workers in the emergency departments themselves. These workers are on hand when physicians feel an ED visit could have been better dealt with in primary care, to connect patients with primary care facilities on their way out.

News organizations have reported further CCO innovations, from the exotic – paying pregnant women to quit smoking – to the mundane but practical, like opening emergency room express wings or new walk-in clinics that can provide extra points of care for Medicaid users without primary care physicians.

“We’re also doing what are called learning collaboratives,” says Robbins. “Membership of the CCOs – sometimes their medical directors, sometimes others – have an opportunity to talk about specific subjects and share specific ideas and projects.” This formal framework for collaboration tries to make sure success stories don’t stay isolated in a single CCO’s jurisdiction, without stifling local initiatives through a top-down approach.

 

Data-Driven Creativity 

CCOs can be given this level of freedom because, ultimately, their funding depends on measurable results. The OHA recently released its first Health System Transformation progress report, covering the first six months of CCO activity, through June of 2013. The data is preliminary – it’s largely based on administrative claims, which won’t be finalized until a full year’s worth of records come in – but strongly encouraging.

This first report shows ED use by Medicaid patients falling 9% annually over the 2011 baseline, while ED costs fell 18%. The average CCO logged 55.2 ED visits per 1000 patients, moving almost a third of the distance to the major benchmark set by the OHA. Other metrics the OHA report highlights are a fall in hospitalizations for chronic care conditions – by nearly 30% in the cases of congestive heart failure and pulmonary disease – and a 36% uptick in primary care home enrollments.

“It was very exciting to have this report,” says Robbins, adding that “we need a full year of data” before any of these changes can be called a trend. “What we can say is that things are headed in the right direction.” The OHA will release an updated report in February, and a finalized report covering all of 2013 in July. Lori Coyner, the Director of Quality and Accountability at the OHA, also told Clinical Informatics News that her office has deeper claims data than is released in these reports, and should eventually be able to do many of the same analyses as were performed in the Oregon Health Study – seeing, for instance, how many ED visits were treated in outpatient care, and how many resulted in hospital admissions, a good proxy for necessary use.

For now, the first report is a hopeful sign just when change is most needed. CCOs are taking on a new patient population this month, much larger than most analysts had predicted. On January 1, 160,000 Oregonians received Oregon Health Plans under the ACA Medicaid expansion. Even as Cover Oregon has failed to deliver private plans to state residents, the OHA has waged a highly successful campaign to enroll eligible individuals in Medicaid. “We had names of people identified in advance through our SNAP [Supplemental Nutrition Assistance] Program,” says Robbins. “We knew what people would definitely qualify, and those people got a letter in the mail… That really helped enrollment be easier and more efficient.” That's a victory for expanding health insurance, but potentially a huge danger for CCOs, which have to keep up their progress while absorbing thousands of new patients, most of whom have no established primary care homes.

As part of its preparations, the state is doubling down on its local innovation drive. In addition to the quality pool funding, the Oregon legislature has earmarked $40 million in “transformation funds,” to award to CCOs that want to invest in projects for long-term healthcare savings. That includes some expected measures, like implementing CCO-wide electronic health record sharing, but also some more creative ones, like creating a mobile unit for behavioral health problems, or merging addiction services with primary care centers so physicians who notice signs of addiction can refer patients immediately. At least one CCO, AllCare Health Plan of Southern Oregon, plans to use its transformation funds to reenact the Health System Transformation in miniature, offering incentive dollars to providers who meet locally-set benchmarks.

It's hard to guess how successful any of these plans will be in advance. Thankfully, Oregon doesn't have to. The OHA will keep collecting data quarter after quarter, seeing which CCOs are making progress and which metrics are moving statewide. It's a program of data-driven creativity, a hallmark of a state government that wants to try untested approaches but also wants to make them testable.

One way or another, the status quo will have to be shaken up to accommodate Medicaid expansion – an endeavor that the Oregon Health Study has shown carries real meaning for those affected, even in the short term, but also comes at a significant cost. “I think it helps give us a sense of what some of the costs and benefits of Medicaid expansion may be,” says Taubman. “So we see increases in health care use. We see real improvements in financial outcomes and financial wellbeing. We see reductions in depression, and we can reject large improvements or large declines in physical health.”

Oregon wants to see its uninsured residents gain access to those benefits. As CCOs continue to experiment, the state is hoping that constant innovation and carefully measured feedback can bend back some of the costs.

 

 

*The Oregon Health Study paper on ED visits looked only at residents of Portland and its suburbs, rather than all 30,000 new Medicaid enrollees statewide. This allowed the researchers nearly complete access to the subjects’ ED records, which were almost entirely accounted for by twelve hospitals in the Portland area.