The Volokh Conspiracy - Paying Health Care Providers: How to Make Agency Problems Worse:

Over the weekend, the New York Times had a nice short piece about how the way we compensate health care providers can create problematic incentives. By and large, Medicare pays providers on a fee-for-service basis -- tied to patient encounters and services provided, rather than the results of those interactions. For hospitals, this takes the form of a flat payment for each hospitalization, with the specific amount determined by the discharge diagnosis. (Physicians are paid through a separate system, which creates its own difficulties).

This payment structure for hospitals has advantages compared to the cost-based reimbursement system it replaced, but it is far from perfect. The article focuses on one specific manifestation of the problem -- hospital readmissions within a short time (<30 days) of discharge. Unless the readmission is within a very short time frame, or otherwise indicates "gaming" of the reimbursement system, the hospital will be paid for both admissions.

Readmissions may reflect poor quality care, but they need not. However, eliminating readmissions that result from poor quality care results in a better outcome for the patient, lower spending on health care for the payor, and a reduction in income for the provider:

Millions of patients each year leave the hospital only to return within weeks or months for lack of proper follow-up care. One in five Medicare patients, for example, returns to the hospital within 30 days. Over all, readmissions cost the federal government an estimated $17 billion a year.

But even when hospitals find ways to greatly reduce the return trips, saving money for Medicare and other insurers, their efforts go unrewarded. In fact, because insurers typically pay hospitals to treat patients — not to keep them away by keeping them healthy — hospitals can actually lose money by providing better care. Empty beds mean lost revenue.

The article gives two concrete examples of institutions that lowered their readmission rate, but suffered substantial reductions in income as a result. One has discontinued the program, and another is only continuing the program out of a sense of "moral obligation" but notes that it is "getting killed" by it.

The problem is not unique to this setting; as I observe in a forthcoming piece, our fragmented health care looks the way that it does because our "encounter-based, primarily fee-for-service payment system has a distinct tendency to reward unbundling and inefficiency. Even under the best of circumstances, the current payment system does not create systematic incentives to deliver efficient high quality care."

The root cause of this failure is simple; as former Assistant Secretary of Health and Human Services Dr. Philip Lee once noted, providers "get paid for what we do, not what we accomplish." The failure to tie compensation to variables that correlate closely with patients' needs and desires means that providers rarely have an economic incentive to invest in quality or prevent error. (Of course, they have other incentives to address these issues; as I like to remind my law students when we cover medical malpractice, "no one gets through medical school with the goal of being below-average or providing low-quality care.") But, the underlying problem remains -- and at an institutional level, it is hard to create a "business case for quality" or a sense of urgency in addressing such problems when improving quality makes the provider financially worse off.

Regardless of one's views on how the American health care financing and delivery systems should be structured, it is hard to justify punishing providers financially for doing a better job for their patients. As I noted in a long-ago article with Charlie Silver on using compensation to align the incentives of providers and patients, "No rational system of payment rewards an agent for a behavior that makes a principal worse off."

Finally, the specific problem noted in the New York Times article is not new; a 1984 article in the New England Journal of Medicine notes that the readmission rate during 1974-1977 was 22% -- roughly the same as it was in 2004.

Going forward, we should do more to use payment incentives, as well as other strategies, to align the interests of providers and patients -- a goal that is easier stated than accomplished. This article gives some sense of the complexities, even for a problem as long-standing and seemingly straightforward as readmissions.

As the conclusion to my forthcoming article notes,

In health care, we get what we pay for -- and what we pay for is the provision of specific services --virtually irrespective of whether they are provided efficiently, or even needed. Because payment is conditioned on the laying of hands (or eyes) upon a patient, time spent coordinating care doesn't create a billing opportunity. When we don't pay for something, it generally doesn't get done. Similarly, providing integrated care doesn't pay better than fragmented care -- and in some instances, it pays worse. The results are entirely predictable -- and until the incentives created by the payment system are modified, we will continue to get what we've already got: a fragmented non-system for delivering care of highly variable quality at high cost. In our health care delivery non-system, coordination/integration is the dog that doesn't bark -- because under our current payment system, no one has any interest in actually buying the dog.

	[David Hyman, May 11, 2009 at 11:55pm] Trackbacks Paying Health Care Providers: How to Make Agency Problems Worse: Over the weekend, the New York Times had a nice short piece about how the way we compensate health care providers can create problematic incentives. By and large, Medicare pays providers on a fee-for-service basis -- tied to patient encounters and services provided, rather than the results of those interactions. For hospitals, this takes the form of a flat payment for each hospitalization, with the specific amount determined by the discharge diagnosis. (Physicians are paid through a separate system, which creates its own difficulties). This payment structure for hospitals has advantages compared to the cost-based reimbursement system it replaced, but it is far from perfect. The article focuses on one specific manifestation of the problem -- hospital readmissions within a short time (<30 days) of discharge. Unless the readmission is within a very short time frame, or otherwise indicates "gaming" of the reimbursement system, the hospital will be paid for both admissions. Readmissions may reflect poor quality care, but they need not. However, eliminating readmissions that result from poor quality care results in a better outcome for the patient, lower spending on health care for the payor, and a reduction in income for the provider: Millions of patients each year leave the hospital only to return within weeks or months for lack of proper follow-up care. One in five Medicare patients, for example, returns to the hospital within 30 days. Over all, readmissions cost the federal government an estimated $17 billion a year. But even when hospitals find ways to greatly reduce the return trips, saving money for Medicare and other insurers, their efforts go unrewarded. In fact, because insurers typically pay hospitals to treat patients — not to keep them away by keeping them healthy — hospitals can actually lose money by providing better care. Empty beds mean lost revenue. The article gives two concrete examples of institutions that lowered their readmission rate, but suffered substantial reductions in income as a result. One has discontinued the program, and another is only continuing the program out of a sense of "moral obligation" but notes that it is "getting killed" by it. The problem is not unique to this setting; as I observe in a forthcoming piece, our fragmented health care looks the way that it does because our "encounter-based, primarily fee-for-service payment system has a distinct tendency to reward unbundling and inefficiency. Even under the best of circumstances, the current payment system does not create systematic incentives to deliver efficient high quality care." The root cause of this failure is simple; as former Assistant Secretary of Health and Human Services Dr. Philip Lee once noted, providers "get paid for what we do, not what we accomplish." The failure to tie compensation to variables that correlate closely with patients' needs and desires means that providers rarely have an economic incentive to invest in quality or prevent error. (Of course, they have other incentives to address these issues; as I like to remind my law students when we cover medical malpractice, "no one gets through medical school with the goal of being below-average or providing low-quality care.") But, the underlying problem remains -- and at an institutional level, it is hard to create a "business case for quality" or a sense of urgency in addressing such problems when improving quality makes the provider financially worse off. Regardless of one's views on how the American health care financing and delivery systems should be structured, it is hard to justify punishing providers financially for doing a better job for their patients. As I noted in a long-ago article with Charlie Silver on using compensation to align the incentives of providers and patients, "No rational system of payment rewards an agent for a behavior that makes a principal worse off." Finally, the specific problem noted in the New York Times article is not new; a 1984 article in the New England Journal of Medicine notes that the readmission rate during 1974-1977 was 22% -- roughly the same as it was in 2004. Going forward, we should do more to use payment incentives, as well as other strategies, to align the interests of providers and patients -- a goal that is easier stated than accomplished. This article gives some sense of the complexities, even for a problem as long-standing and seemingly straightforward as readmissions. As the conclusion to my forthcoming article notes, In health care, we get what we pay for -- and what we pay for is the provision of specific services --virtually irrespective of whether they are provided efficiently, or even needed. Because payment is conditioned on the laying of hands (or eyes) upon a patient, time spent coordinating care doesn't create a billing opportunity. When we don't pay for something, it generally doesn't get done. Similarly, providing integrated care doesn't pay better than fragmented care -- and in some instances, it pays worse. The results are entirely predictable -- and until the incentives created by the payment system are modified, we will continue to get what we've already got: a fragmented non-system for delivering care of highly variable quality at high cost. In our health care delivery non-system, coordination/integration is the dog that doesn't bark -- because under our current payment system, no one has any interest in actually buying the dog.