This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

Click here to view a Webcast of an interview
with Robert Berenson and Steven Lieberman
(available December 10, 2003).

Read related papers by
Steven Lieberman et al. • Robert A. Berenson •
David E. Wennberg and John E. Wennberg

P E R S P E C T I V E : 
V A R I A T I O N S
W E B  E X C L U S I V E

10 December 2003

Addressing Variations:
Is There Hope For The Future?

Demonstration projects hold promise as Medicare uses its size
and purchasing power to affect patterns of care for all Americans.

by David E. Wennberg and John E. Wennberg



ABSTRACT:

The papers by Robert Berenson and by Steven Lieberman and colleagues show that variations remain a true challenge for those trying to improve the delivery of health care. Recent clarifications in the understanding of unwarranted variations allow us to address variations in a more logical and manageable fashion. In this Perspective we describe key challenges in addressing variations in the context of these recent clarifications. The Centers for Medicare and Medicaid Services (CMS) needs to move forward on information-sharing interventions and use demonstrations to pursue innovative strategies to improve the delivery of care through its purchasing power.

Hundreds of articles on geographic variations have been published in the past forty years; these have variously described the phenomenon, ascribed root causes, and prescribed solutions. Yet, as evidenced by the accompanying papers by Robert Berenson and by Steven Lieberman and colleagues, variations remain a true challenge for those trying to improve the delivery of health care. While the pessimists among us might wonder why things always seem to be the same, much has been learned during the past four decades that can guide structural changes to address the underlying drivers of variations.

Categorizing Unwarranted Variation

Two recent “discoveries” in the science of variations allow health care providers, policymakers, purchasers, and consumers to address variations in a logical and manageable fashion. The first is a clearer definition of what unwarranted variation is: care that is not consistent with a patient’s preference or related to a patient’s underlying illness. This “discovery” allows a more fruitful debate about appropriate versus inappropriate care. The second is the categorization of unwarranted variations into three categories of care: effective, preference-sensitive, and supply-sensitive.¹

Recognizing these categories gives tremendous insight by elucidating the “underlying problems” and pointing toward possible interventions. Thus, the question of whether to intervene in unwarranted variations at the patient, geographic, or provider level turns on whether one is interested in reducing variation in effective, preference-sensitive, or supply-sensitive care.

Effective care. For effective care, the leverage points are with patients and providers. Unlike preference- or supply-sensitive care, there are no “conflicts” between patients, providers, or health care systems that drive unwarranted variations in effective care. As the Institute of Medicine (IOM) and others have illuminated, it is a problem not of “bad doctors” but rather of “nonsystems” of care. Within the Medicare program, interventions with the patient will likely be limited to education about the relevant effective care opportunities and minor support to allow the enrollee to implement this knowledge.² These types of interventions are those that disease management companies attempt and, as suggested by Lieberman and colleagues, should be targeted at specific patients regardless of where they live.

At the provider level, the Centers for Medicare and Medicaid Services (CMS) could play several key roles to reduce variation in effective care. At a minimum, the CMS should use the tremendous wealth of data it collects as a part of paying claims to create data feedback reports for providers. These reports would inform them about their performance in effective-care opportunities for their panel of patients.³ Second, the CMS has the ability to create a systems approach to dealing with effective-care opportunities for beneficiaries.⁴ Third, as the largest purchaser of health care in the United States, the CMS has tremendous opportunities to influence providers’ behavior through pay-for-performance reimbursement or selective contracting; the risks and benefits of some of these strategies are well outlined in Berenson’s paper.

Preference-sensitive care. For preference-sensitive care, the leverage points are patients and, to some extent, providers. For patients, a general educational approach with more sophisticated targeting of information and decision-support tools for those at most risk would begin to address the knowledge imbalance that makes many medical decisions in this category of care non–preference based. Several decision aids are available that greatly improve patient-centered decision making.⁵ The critical issue related to these interventions is not the tools, but rather the implementation of the tools in real time.

Under ideal circumstances, the perfect setting for interventions to reduce variation in preference-sensitive care would be the provider’s office; the provider is the key diagnostician and prescriber of preference-sensitive care. Unfortunately, the provider is also the key driver of unwarranted variation in preference-sensitive care. Therefore, in current models of reimbursement, the most relevant provider for preference-sensitive interventions is likely to be the primary care physician, who receives neither financial reward nor professional status from performing the procedures most relevant to preference-sensitive interventions.⁶ However, systematic provider interventions such as selective contracting or incentive-based network tiers could be ideal mechanisms for decreasing unwarranted variation in the delivery of preference-sensitive care.

Supply-sensitive care. For supply-sensitive care, the most effective leverage point almost by definition is the health system.⁷ Health systems, whether they be individual physicians, hospitals, or integrated delivery systems, are the key decisionmakers (if decisionmakers exist at all) when it comes to recruiting new providers, building new beds, or adding new services. There are, of course, interventions that will affect people’s use of these services. For example, much of the disease management literature supports the finding that patients with moderate-to-severe illnesses will benefit from the coordinated care that these vendors provide. These benefits could include reduced use of the emergency room or hospital.⁸ Lieberman and colleagues argue that these types of interventions will be more effective than the politically charged, operationally challenged, regionally based interventions. However, unless the interventions at the patient level result in a reduction of capacity at the system level, the association between supply of hospital beds and hospitalizations for chronic illness documented in the Dartmouth Atlas of Health Care predict a reciprocal increase in hospitalization rates among the unmanaged population. Why? Because the dynamics of the system work on all members, not just those with chronic illness: The population-based variation patterns of use of supply-sensitive services are experienced by the sick and the well in nearly equal ratios.⁹

Putting It All Together

So, how do we put this all together? The pivot point for the CMS and other U.S. payers is Berenson’s proposition that “traditional Medicare” needs to move from being a payer of claims to being a purchaser of health care. By virtue of its size and national scope, the CMS is the critical actor in the value-based purchasing arena, in which efficiency is now a key component of the “next act.” But, as Berenson rightly recognizes, political realities will make a paradigm shift very difficult if attempted on a national scale. Are there other opportunities? We argue that the current demonstration legislation is the perfect vehicle for experiments in “paradigm shiftlets,” a safe but important environment to test these admittedly unproven interventions. In fact, one of the current demonstration projects supports such a shift toward purchasing care. The Physician Group Practice Demonstration (PGPD) is a novel, provider-focused project that allows for physician group practices to share in the benefit of a coordinated care model. This demonstration aims to improve the quality and efficiency of care for the population of fee-for-service (FFS) Medicare patients who are “loyal” to the demonstration physician groups. The reward (if there is one) derives from a retrospective evaluation of the actuarial costs of such patients compared with a concurrent control group.¹⁰ If costs in the population of loyal patients are lower than in the comparison group, the practice group receives a proportion of the savings.

Although this model represents a tremendous step toward purchasing care rather than paying claims, it has two problems. First, the payment (assuming there is one) comes post hoc, so the fundamental payment model is still one of getting paid for doing more. Second, because hospitals are not included in the demonstration, there is a major conflict from a payment/ cost standpoint between the two key actors in use of supply-sensitive services. This conflict will be most acute when the group practice is the dominant if not the sole user of a given hospital. And in situations where hospital resources are shared with nonparticipating physicians, small-area variation studies predict an increase in hospitalization rates and per capita costs among the population loyal to nonparticipating physicians, resulting in no net overall reduction in per capita spending. This result would be particularly ironic if the nonparticipating population were serving as the concurrent control.

The current “déjà-vu” crisis in health care—rising costs, increasing numbers of uninsured people, and serious questions about quality—has created a flurry of stopgap and more comprehensive proposals to fix the “problem.” The tremendous size of the Medicare program tends to create resignation leading to inertia and thus to continuation of the status quo. On the other hand, some of us see opportunities in smaller interventions leading to larger changes.¹¹ In choosing the route of demonstration projects, we might be able to avoid the complete gridlock that resulted from the last proposed grand health care experiment of the 1990s and avoid the political traps that stop many changes from happening before they are even fully designed. Given the wealth of data resulting as a byproduct of the Medicare program, the CMS needs to quickly move forward on information-sharing interventions. We also encourage the CMS to use the demonstration route to pursue innovative strategies to improve the delivery of care through its purchasing power. Seen through the lens of unwarranted variations, the opportunities for such interventions abound. Can we go forward?

This work was supported in part by a grant (PO1 AG19783) from the National Institute on Aging.

NOTES

1. At the micro level, these are necessarily fuzzy categories with some overlap between types of care across categories. For example, cardiac revascularization can be both effective care (for those with left main disease) and preference-sensitive care (for those with stable angina and one vessel disease).
2. Within the commercial insurance world, other interesting experiments with incentives, flexible spending accounts, and so on are also being done or contemplated in an attempt to influence patients’ behavior in the use of effective-care services.
3. These same reports could be used to inform patients about the relative performance of providers, guiding provider selection through information as opposed to contracting.
4. The CMS’s Quality Improvement Organizations (QIOs) are attempting to create quality improvement systems. However, their interventions are underfunded and often indirect.
5. Ironically, many of these decision aids have been more rigorously evaluated than the underlying treatment options under consideration.
6. The CMS could, for example, allow providers to be paid for “informed medical decision making” under the evaluation and management counseling services codes.
7. Whether the system is defined as a region or a provider group will depend, in part, on the political options and the intervention contemplated.
8. Within the commercial insurance markets there are other interventions, such as increased copayments or health spending accounts that hold some promise for reducing use of supply-sensitive services at the individual level. However, without addressing the system, these will tend to “shift the dollars around.”
9. It is important to note that the level of service use does vary by illness level. For example, Medicare patients with an acute myocardial infarction (AMI) receive more tests, imaging studies, specialist visits, and hospitalizations than those in the Medicare Current Beneficiary Survey (MCBS) (a random sample of beneficiaries). However, across regional levels of spending, the MCBS (the group with the “best” health) living in the highest-spending region received 52 percent more services than those living in the lowest-spending regions, while AMI patients (a much “sicker” population) in high-spending regions received 58 percent more. See E.S. Fisher et al., “The Implications of Regional Variations in Medicare Spending, Part 1: The Content, Quality, and Accessibility of Care,” Annals of Internal Medicine (18 February 2003): 273–287; and E.S. Fisher et al., “The Implications of Regional Variations in Medicare Spending, Part 2: Health Outcomes and Satisfaction with Care,” Annals of Internal Medicine (18 February 2003): 288–298.
10. The comparison groups are unassigned Medicare enrollees residing within counties from which at least 1 percent of the “loyal” enrollees reside.
11.Small interventions in Medicare are, on an absolute scale, large; thus, we argue, they can give “stable” estimates of the impact of such interventions on a complex “organization” such as the U.S. health care system.

David Wennberg, davidwennberg{at}healthdialog.com, is director of the Center for Outcomes Research and Evaluation, Maine Medical Center, in Portland. John Wennberg directs the Center for the Evaluative Clinical Sciences at Dartmouth Medical School in Hanover, New Hampshire.

Read related papers by Steven Lieberman et al. • Robert A. Berenson • David E. Wennberg and John E. Wennberg.

©2003 Project HOPE–The People-to-People Health Foundation, Inc.