Policy Forum
May 2012

Assignment, Attribution, and Accountability: New Responsibilities and Relationships in Accountable Care Organizations

Harold S. Luft, PhD
Virtual Mentor. 2012;14(5):407-410. doi: 10.1001/virtualmentor.2012.14.5.pfor1-1205.


The Patient Protection and Affordable Care Act of 2010 (ACA) includes important features that are likely to change how medicine is delivered in the United States. The Supreme Court will rule this summer on various aspects of the act’s constitutionality, especially the individual mandate provisions, but it is unlikely to strike down the entire act. In section 3022 of the act, the Secretary of Health and Human Services is authorized to recognize arrangements between Medicare and collections of physicians and other providers as accountable care organizations, or ACOs [1]. If the costs to Medicare for people “assigned” to an ACO grow at a slower rate than costs for patient populations with similar patterns of medical use, and if the ACO meets certain quality targets, the implicit savings will be shared by Medicare and the ACO. These savings can be allocated by the ACO to reward its participating clinicians, build infrastructure to facilitate care, or pay for services not ordinarily covered by Medicare.

Several features of the Medicare version of ACOs and similar arrangements developed for the privately insured market (PACOs) are important. The first is that patients in ACOs and PACOs keep their traditional coverage, e.g., Medicare, rather than enrolling in an HMO-style health plan that significantly limits their choice of providers. (I use the term “provider” to include both physicians and other clinicians, such as nurse practitioners, who may be the patient’s usual source of care, and entities such as clinics and hospitals.) Indeed, some patients may be in an ACO or PACO without even knowing it. This brings us to the second feature—patients are attributed (ACA uses the term “assigned”) to an ACO based on their patterns of primary care use. The next section will explore this in more detail. The third feature of ACOs and PACOs is that they are held accountable for all services received by the patients attributed to them, even those received outside the ACO.

One might ask, “why would clinicians want to form an ACO or PACO and be held accountable for the quality and costs of care that they do not provide for patients who are not formally enrolled and have no financial incentives for receiving care within the organization?” Indeed, initial response to the concept was mixed, but it seems to be gaining traction, especially after CMS revised its initial regulations [2].

Perhaps the short answer is that many believe the current payment and incentive system leads to so much wasteful care and so many missed opportunities for quality improvement; the incentives and flexibility of the ACO/PACO model may encourage them to “do the right thing” more frequently and save money in the process. The lack of control inherent in the ACO/PACO structure (compared to an HMO) makes cost saving more difficult, but has the advantage of engaging providers and patients not willing to be in HMOs. More importantly, it requires and supports a different relationship between clinicians and patients.

Attribution versus Assignment

The “assignment” terminology in the ACA does not reflect an attempt to assign patients to providers, but instead reflects the origins of the accountable care organization concept in what were essentially epidemiological studies [3]. The term “attribution” better fits what is actually being done. Although the details for attribution to Medicare ACOs and various PACOs differ, the notion is that a person can be attributed to a specific primary care provider (PCP) based on a list of the providers from whom he or she receives care. Readily available insurance claims data allows one to “crunch the numbers” to do this attribution without ever asking the patient. Attribution is intended to be a feasible, rather than an ideal, methodology.

Even in a Medicare population with a large number of visits per person per year, it is difficult to attribute patients to individual PCPs [4]. Problems arise, for example, when a patient has an equal number of visits to two PCPs, or when so many PCPs are seen that no one accounts for a majority of the visits. Attribution typically focuses only on visits to PCPs, so problems arise when, because of a dominant chronic condition, a patient largely has his or her care managed largely by a specialist. The attribution challenge is exacerbated with younger adult populations because many have no visits to a PCP in any one year. Does this mean they do not have someone to whom they could go, or who should be monitoring their care?  Such concerns are mitigated substantially if one merely needs to attribute people to a set of PCPs affiliated with one ACO/PACO; the problem of “ties” usually disappears if the attribution is to “any PCP within the ACO/PACO.”

The key aspect of attribution, however, is that it reflects a relationship between patients and their provider organizations that is fundamentally different from patient-HMO relationships. An HMO takes responsibility for an enrolled population, meaning that the premiums it receives each month allow it to know exactly for whom it is responsible. The HMO contract with its enrollees, moreover, generally says it has no financial responsibility for medical care obtained outside its system, except in emergency situations or via explicit referrals. In contrast, ACOs and PACOs do not have monthly enrollment lists, and their attributed patients have comparable coverage for care from non-ACO/PACO providers. The challenge for them is to be so attractive to their patients that patients don’t want to seek care elsewhere.


HMOs typically bear full financial risk for their enrollees; in an ACO/PACO part of that risk will be borne by Medicare or private insurers. This is a necessity given the highly skewed nature of health care costs, in which a small number of patients account for a large share of overall expenditures. Risk sharing does not, however, allow ACO/PACOs to ignore the costs of the care received by their patients. The insurer can capture information on all the patient’s care, regardless of the providers’ affiliation, and the ACO/PACO shares in savings only if overall patient costs are lower than those for comparable patient populations. If it meets such targets, however, the ACO/PACO will receive lump-sum payments from the insurer (Medicare) not tied to the services of any particular provider. This flexible pot of money allows the ACO/PACO to focus on developing standard processes for the efficient management of the problems its patients face.

Efficiency in this context does not mean shaving a 12-minute primary care visit to 10 minutes and ending the session with two prescriptions and a referral to a specialist. On the contrary, it may mean spending 20 minutes to thoroughly understand the patient’s problem and working through treatment options, perhaps with a phone call 3 days later to see how the patient is doing. Even if the insurer does not pay for the extended visit or the time to call the patient, the ACO/PACO could compensate for that time with the savings achieved.

Balancing the incentives to reduce expenditures are quality metrics. Initially, these may focus on the standard preventive screening and process measures, but they will rapidly move toward clinical and patient-reported outcomes. The latter are not the oft-maligned “generic patient satisfaction” measures but specific patient assessments of their functional status, understanding of their condition, and experience of care—that is, measures patients care about.

Physicians and other professionals deliver medical care, but organizations create the infrastructure to ensure high quality. Quality care may begin with the face-to-face encounter, but it requires the ability to transfer information efficiently among all the clinicians involved, to delegate mundane tasks so the most skilled clinicians can attend to clinical cues, to know when a patient hasn’t come in when he or she should. Large medical groups already provide much of this. ACO/PACOs seeking to include providers such as independent or small group practices will need to create such infrastructures.

ACO/PACOs and the Patient-Physician Relationship

It is too soon to know how ACO/PACOs will function, but the logic behind them is quite different from that of a standard insurance plan or an HMO. Insurers are typically passive payors of claims after events have occurred. They focus primarily on tweaking benefit packages to create patient-focused incentives to reduce expenditures. HMOs (and managed care plans) sometimes act as if they “own” the patient—at least for a time—and exercise the right to say they will not cover certain services even if the physician thinks they may be needed. They typically also have more data about what is and is not done for their enrollees.

Most physicians in independent practice know just what they do for their patients, but have little or no information on their patients’ care from other providers. Because fees do not adequately compensate for time spent with patients, financial pressures discourage the development of close and trusting connections between patients and physicians. Well designed and effectively implemented ACOs should help those who deliver primary care become trusted elicitors of informed patient preferences and knowledgeable coordinators of care. It will take a few years, however, to know if they successfully seize this opportunity.


  1. Centers for Medicare and Medicaid Services. Medicare program; Medicare Shared Savings Program; Accountable Care Organizations. Final rule. Fed Regist. 2011;76(212):67802-67990.
  2. See, for example, Galewitz P, Gold J. HHS release final regulations for ACOs. Kaiser Health Newshttp://www.kaiserhealthnews.org/stories/2011/october/20/accountable-care-organization-rules-regulations.aspx. Accessed April 17, 2012.

  3. Luft HS. From small area variations to accountable care organizations: how health services research can inform policy. Annu Rev Public Health. 2012;33:377-392. http://www.annualreviews.org/doi/abs/10.1146/annurev-publhealth-031811-124701. Accessed April 17, 2012.

  4. Pham HH, Schrag D, O’Malley AS, Wu B, Bach PB. Care patterns in Medicare and their impliations for pay for performance. N Engl J Med. 2007;356(11):1130-1139.


Virtual Mentor. 2012;14(5): 407-410.



The viewpoints expressed in this article are those of the author(s) and do not necessarily reflect the views and policies of the AMA.