Over the past few years, policymakers, practitioners, and scholars have proposed several reforms to the health care system. These have included, among others, a systems focus, an engineering focus, and a collaborative focus. What these reports share in common is the recognized need for changes in the payment system to accompany changes in the delivery system — in particular, the need for greater clinical integration. However, the order in which the changes are promoted will have long-lasting consequences on the ultimate shape of the U.S. health care system. We argue that changes to the payment system, and the process through which clinical integration emerges, must drive changes in the organization of the delivery of care, and not the other way around.

The consensus among the reports cited is that bundled payments to organized provider organizations are the key initiative to pursue. Reinforcing that point, three prominent integrated delivery networks — Intermountain, Kaiser, and Mayo — have proposed a series of action steps and pay-for-performance approaches that center principally on large medical groups and integrated delivery networks (IDNs). Most recently, Stuart Guterman and colleagues have similarly proposed large medical groups, hospital systems, and IDNs as the provider hubs for receiving bundled payments. These proposals are consonant with earlier and current calls for “Accountable Care Organizations” of hospitals or local physicians, or both. Consistent with these recommendations, the Centers for Medicare and Medicaid Services (CMS) has launched a four-state Acute Care Episode (ACE) demonstration that bundles physician and hospital payments to physician-hospital organizations (PHOs).

We believe that the commonality among these approaches — their reliance on IDNs, PHOs, and forced hospital-physician relationships (HPRs) as a prerequisite to payment reform — is also their major weakness. To be sure, the two biggest buckets of health care expenditures (nearly 50 percent) are hospitals and physicians, so it makes sense to concentrate on these two sets of providers to achieve cost-containment and quality objectives. However, the organizational models developed to date by these two sets of providers have been weighed in the balance of academic research and been found seriously wanting, perhaps because they have risen mainly to optimize gain in a currently flawed payment system. As one of us argued seven years ago, these models have been preoccupied with the structure of health care delivery (for example, vertical or virtual integration) rather than the process of health care delivery (for example, coordination, teamwork, learning). As such, making payment reform contingent on this type of delivery system “reform” seems inconsistent with the evidence to date.

Most IDNs have not demonstrated an ability to improve quality, increase access to care, or reduce costs — what some scholars have referred to as the “iron triangle” or “triple aim” of health care. These lackluster results hold for hospital systems and hospital-physician alliances, such as PHOs. PHOs, in particular, have been quite suspect in their ability to deliver value. They have been one of the most prevalent types of IDNs and yet have failed to develop any significant clinical or financial integration. The available literature suggests that the IDN provider models have been primarily developed to leverage managed care payers for higher reimbursement — which raises health care costs.

It Is Important Not To Prejudge The Optimal Organizational Form For Health Care Delivery

Of note, industrialized countries, and those on the pathway of full economic development, do not have the large integrated structures of care that have emerged in many parts of the U.S., and yet they consistently produce better results, partially because they are focused on implementing the right processes of care and have let care delivery organizations emerge naturally as a result of the payment incentives.

What we do know is that there are processes of care that can deliver better value. A review of the Medicare Participating Heart Bypass Center Demonstration of the 1990s, which was the first federal experiment with bundled payment, and the federally sponsored investigation of the “Winners and Losers under the Medicare Program” provide valuable lessons on the ingredients needed for clinical integration. They show that realizing gains under an episode-of-care system involves the following process elements: a clinical nurse specialist in charge of each patient; cost-accounting systems to compute direct costs incurred by specific physicians; outside consultants to review cost and utilization profiles with individual clinicians; clinician leaders to actively encourage peers to adopt more cost-effective practices based on these profiles; investment in physician relations programs; supply-chain management practices (for example, product standardization, generic substitution, vendor bargaining with physicians at the table, and so on); experimentation with different ways to share savings with physicians (for example, bonuses, payments in kind such as more time in the operating room); improved billing and collection; and decomposing patient lengths-of-stay into clinically meaningful components to engage relevant physicians. None of these processes makes the existence of a formal organization an absolute necessity. They do, however, need the right financial incentives to flourish.

So, given all this, where do the true innovations in health care delivery need to occur? We believe it is a mistake to automatically assume that IDNs and PHOs will be the vanguard of this movement. Vertically integrated structures are notorious for being inflexible, bureaucratic, internally divisive, and slow to change — the opposite of what we are trying to achieve. It is also unlikely that an IDN that is good at producing an efficient episode of care in one condition will be equally adept at doing so across all other conditions. Providers, however organized, will need to be nimble to render efficient and effective care across multiple types of patient conditions by coordinating multiple sets of providers. Innovation may occur in community hospitals as well as IDNs. It may occur in small and medium-size practices that become virtually integrated. It may occur in organizational designs that are yet to emerge. Prejudging the optimal form of an organization that can best respond to a new form of payment before naturally observing what such organizations could be seems the height of arrogance. A new payment model will create incentives around which new organizations will emerge. Let us test the boundaries of these organizations in pilots before deciding what those boundaries should be.