Payers in population health contracts are using claims data to identify the physician responsible for a patient’s health and spending. It’s an inexact science that can aggravate providers before empowering them.
New care delivery and payment models all seem to have the same key elements: a defined population, panels of people for whom health care providers are accountable, and a way to tie measures of quality and cost to those responsible. With that foundation in place, organizations can reward the individual and group efforts of clinicians to improve population health metrics of care at a reasonable expense.
Unfortunately, it’s not that simple. There are no neatly defined panels of patients tied to specific providers. The dominant form of health plan, the preferred provider organization, generally permits people to seek care from a sizable group of providers and to change their minds at will. In addition, these people may change their addresses or switch insurance carriers, which can confound attempts to define a panel for any measurable length of time. But some organization must be responsible for these patients so that provider networks can “more effectively manage the population, and to ensure that everyone is getting the care that’s appropriate to their [respective] condition,” says Michael Simon, principal data scientist with analytics vendor Arcadia Healthcare Solutions.
To resolve that issue is a process called attribution. It uses mainly medical claims to identify all the providers that a patient sees and the costs applied to the patient’s illness and wellness, and determines who among a patient’s providers should be accountable for his or her condition and health care expenditures. Attribution is neither exact nor simple. It has many variables, and each tweak in its logic can have an impact on which patients are attributed to a particular physician, and on the final calculation of quality and cost-control scores that may affect such issues as financial rewards.
The details of attribution formulas usually are determined by health insurance plans that make value-based assessments part of their contracts, though providers can influence or even assume the task if they are sophisticated enough, says Nathan Gunn, president of the population health division of Valence Health, an analytics and consulting firm.
But no matter how attribution is set up, it has the potential to create physician frustration and resistance, especially if it affects the amount of supplemental payments for patient coordination and management, or if doctors are penalized based on results from other providers a patient may see.
Governing boards “should recognize that this is an art — it’s not a science — [and] that there needs to be flexibility in terms of interacting with physicians,” Gunn says. At minimum, trustees should insist on a way for doctors to air concerns that a measure is wrong, or that a person attributed to one of them is not really his or her patient.
Backward and Forward
The most accurate way to attribute patients to accountable physicians is to track and measure the behavior of a target population after the end of a reporting period, which is typically one year. Software loaded with attribution logic, usually employed by the payer originating a value-based contract, looks at every encounter with the health system and the costs incurred, and it retrospectively assigns a patient to the physician he or she appears to see most often.
Payers generally assume that a person is under the care of a certain primary care physician that he or she sees during a plan year, even with patients’ freedom to engage specialists for chronic or acute ills or go outside the contracted system, says Christopher Stanley, M.D., vice president for care management at Catholic Health Initiatives, Englewood, Colo. Physicians use that retrospective determination of attributed patients to inform their actions in the coming year. But, unless there are ways to update and inform doctors on patients’ movements, “[they] really don’t know what that patient is going to do over the next 12 months,” he says. The frustration for physicians is that they won’t know which patients they are responsible for until the end of the year. “By then, it’s too late to reach out to them.”
The alternative is to make the process prospective, using a similar attribution process as a basis for accountability going forward instead of looking back. Attributed patients are known to their providers, who can use the care and cost data that come in during the year to respond to problems immediately, says Dana Gelb Safran, senior vice president for performance measurement and improvement for Blue Cross Blue Shield of Massachusetts. “If you’re going to ask a provider to be accountable for a population, it’s good if they know who they’re accountable for, so that they can be actually managing care in a way that will deliver good quality and affordability results,” she says.
Health systems that use an attribution method from Arcadia Healthcare Solutions are going the prospective route, using claims supplemented by a host of other sources including electronic health records, Simon says. With pay-for-performance programs, “a provider attribution can ultimately mean the difference between success and really falling behind,” he says. When providers have no sense of who is going to be officially responsible for their patients nor a way to get at that determination, “that leaves a sense of powerlessness in the face of an incentive contract,” he says.
Nevertheless, the multiple pilot projects emanating from the Affordable Care Act, such as the Pioneer Accountable Care Organization model and the Medicare Shared Savings Program, mean momentum is behind retrospective attribution. The Centers for Medicare & Medicaid Services recently allowed some measure of patient recruitment into a particular ACO’s attributed population to provide some predictability, but the basis for deciding which Medicare beneficiaries are in or out of an ACO is still mainly tied to the previous year.
Narrowing Down the Provider
Another variable in the attribution process is the type of practitioner ultimately deemed the main influence on care services and costs. Chronic-illness care calls for a primary care provider who has a big say in the treatment plans of patients, even if they more often go to a cardiologist or pulmonologist. But specialists sometimes call the shots, especially in an episode of care requiring a procedure or other intensive intervention [see The Care Team’s Quarterback, Page 11].
Clinically integrated networks already are running with educated decisions on the formulation of their attributed population. “Really, truly understanding patient behavior, and starting with primary care” is essential in the attribution process, says Daniel Edelstein, director of Northwest Alliance ACO, a Minneapolis-area network bringing together five clinics operated by Allina Health and four clinics operated by HealthPartners.
“All the models I have with both commercial and the government payers are based on where a patient seeks primary care services,” says Edelstein, who hails from the Allina side of the alliance. “We’re not attributing based on where they’re getting inpatient care, or even specialty care.”
One premise is that primary care physicians, while not providing specialized care, are determining the need, authorizing care and sending patients to a certain person or place. “A lot of decisions made by the primary care providers clearly will influence both health scores and outcomes, and, of course, affordability,” Edelstein says. “The model can get quite complicated, but we keep it simple: The primary care provider that you get attributed to is the one that you see the most.”
Making the decision on the proper primary care physician involves looking for certain patient activities, says Safran of Blue Cross. Insurers first look for “well visits” to the doctor. “Typically a patient does not schedule a well visit with somebody other than their primary care doctor,” she says. That covers a lot of the population. Next, look for urgent care visits, the sources of prescriptions, and then other kinds of PCP visits, such as for a blood draw. “You’re using your claims data to tell you, based on the number and type of encounters, who appears to be the doctor for this patient, and then what system that doctor is in,” she says. “You take all the patients assigned to doctors in that system and that becomes the system’s population.”
Attribution in Action
As population health-oriented approaches become more prevalent, the impact of attribution becomes increasingly clear. “If we’re really trying to move toward accountable care and managing populations, it’s all about what population is being served, and attribution clearly defines that,” Edelstein says. “Off that population served, and off the value-based arrangements that are contractually made with the plans, we can quite easily measure total cost of care and put a real accountability on affordability.”
To understand how a population seeks care, how that activity is categorized and what should be done to improve it, says Stanley, it first takes an understanding of how each patient is connected to a care practitioner, whether in a primary care physician’s office, the emergency department, or with a specialist replacing a joint — and that’s where attribution comes in.
With the data derived from attributing patients to their responsible doctors, Northwest Alliance ACO and its physicians are better equipped to identify opportunities to improve care quality and affordability, Edelstein says. “If we, for example, can say that [attributed patients] used the emergency department five percent more than the rest of the metro, what are the unique nuances of our practice that may lead to that? Do we have a good triage process? Do we have good access where patients can get in? That would ideally reduce unnecessary emergency department utilization.
“It clearly has people asking the right questions,” he adds. For example, the higher-than-usual use of inpatient mental health services was a concern for the ACO. “So, what are the driving causes of this? Physician practices? Gaps in the care continuum? Lack of integration?” In the past, advocates for services such as integration of primary and behavioral care had few champions and little momentum, but the accountability for cost data by attributed provider has made primary care physicians more engaged, Edelstein notes. As a result, the ACO has added “a ton of resources” to mental health, including bringing behavioral health inside primary care clinics.
The onus on primary care physicians also provides momentum for team-based care, including the use of physician assistants and nurse practitioners, which might not happen otherwise, says Stanley. Doctors responsible for a range of activities, such as immunizations, have a stake in delegating routine services to others who can work in parallel, supporting the role of physician as coordinator. Optimal use of mid-level practitioners goes hand in hand with how care is paid for, he says.
Getting Past the Disruption
Like other aspects of health care reform, the application of attribution methods is essential, disruptive and, at times, seemingly unfair. The phase-in likely will spark pushback even as it helps to push the organization forward. The challenges start with the attribution formula and proceed into how doctors are being asked to think, says Stanley.
“As a provider group, if you’re given 20,000 individuals [to manage], population health is much, much different than physicians have ever practiced and really think about practicing,” he says. After focusing for years on “the 25 patients who are coming to see me in my office today,” and developing a relationship with them, the notion of attributing an individual life to a doctor can seem artificial.
But it’s important to accept the imperfections of attribution and move on to what it can achieve for the organization, Stanley advises. “In our approach, we’re trying to lead by saying attribution and similar types of methodologies should not ever be used to blame a physician, blame a patient or negatively look at how care is being provided, but rather should be a mechanism by which we can actually improve the health and well-being of the communities that we serve,” he says. “This type of tool, as long as you understand the limitations and how it will help support your tactics, can be a very, very powerful tool to help improve the health of the population.”
With physicians who feel powerless, Stanley says the key is to turn the questions back to the protesting physicians and say, “Here are our shared goals, here’s the patient-focused approach and our methodology to reach that, and here’s how we’d like to take ownership and accountability for it.”
Boards of clinically integrated networks typically include participation from affiliated practices as well as the hospital or hospitals involved, and the particular type of attribution formula can be worked out in the appropriate committees of the board, says Gunn of Arcadia Healthcare Solutions. “When things are developed in the clinical committees and the financial committee, you have to have key leaders from your affiliate community physicians brought in. It will overcome a lot of mistrust,” he says. “You make sure everybody arrives at the least-worse answer together, so there is alignment around it.
“At the highest level, attribution is an issue for which there’s no technological answer,” Gunn says. “You have to give the docs a tool so they can raise their hands and say, ‘The patient’s not mine.’ Other than that, the issue is just picking your poison — and understanding that whichever methodology you’ve selected, there will be an upside and a downside.”
As leaders who already are pursuing value-based contracts have found, trying to fit population health aims into a business landscape still dependent on fees for service requires inventiveness on a large scale. Attribution is one such invention for what comes down to a long transition period toward full capitation, after which there is no need for attribution to nail down responsibility for patients, says Gunn. Once incentives are fully aligned under full risk at a fundamental level, “you don’t have to do all these workarounds, which is what attribution is: workarounds in a fee-for-service world.”
John Morrissey is a contributing writer to Trustee.
The Care Team’s Quarterback
When patients are seriously chronically ill, or need anything from a knee replacement to a new kidney, various specialists can perform the bulk of care services over a period of time. Yet, the total costs of a patient’s care usually are attributed to his or her primary care physician. What gives?
The analogy of the football quarterback often comes up when describing the position of primary care physicians in team care models. The outcome on the gridiron may be largely the result of plays by specialists such as running backs or wide receivers, but the quarterback initiated the handoff or pass, enabled various team members to do what they do best and, ultimately, absorbed the credit or blame. Similarly, the primary care physician directs, coordinates and has a vested interest in patient outcomes.
At Northwest Alliance ACO, the primary care focus of attribution has helped to develop teamwork with specialists, says Daniel Edelstein, the network’s director. “If a particular patient has a complex, specialty-type of condition — let’s say, diagnosed with cancer — there’s no question that a lot of the care will be actually guided by the oncologist. And the primary care physician’s decisions will likely not be the drivers of health care spend.” But those oncologists “are clearly in a close working relationship with the primary care provider.”
Meetings often are scheduled between primary care physicians and selected specialty groups around achieving health care’s Triple Aim of optimal quality delivered with maximum efficiency and high patient satisfaction with the care experience, Edelstein says. Specialists have a vested interest in maintaining good relations, he adds, because they depend on primary care physicians for referrals.
When care is concentrated in the hands of a specialist for a period of time, “it would seem that the responsibility has shifted,” says Michael Simon, principal data scientist of Arcadia Healthcare Solutions, “but you could also argue that your primary care physician has a vested interest in continuing to monitor all of the [patient’s] conditions, especially if [that patient has] chronic conditions related to the surgery.” Or the surgery could be unrelated, but the primary care physician still would want to make sure the patient is getting proper treatment for other ills.
For the most part, the performance of other specialists burnish the attributed doctor’s quality metrics, says Dana Gelb Safran, senior vice president for performance measurement and improvement for Blue Cross Blue Shield of Massachusetts. A specialist only can be helping by doing tests that are important to the care of a network’s patient population, which accrues to the primary care physician’s advantage. For a diabetes patient who sees an endocrinologist for relevant screenings, to cite one example, that physician’s contributions will help to keep the patient’s diabetes under control and contribute to higher quality of life while preventing unnecessary hospital expense, Safran says.
With primary care measures being predominant, she says, “It’s possible, in theory, that patients are seeing other providers that could hurt [the attributed doctor] on quality measures. I’m just not sure how.” — J.M.
For more about attribution, including how it chages physicians' mindsets, read "The Engine that Powers Population Health".