Pay-For-Performance and Unintended Consequences: The International Language

Pay-For-Performance (or P4P) is an important tool in our effort to improve the quality of the nation’s health care, and control cost.  It may prove to be a critical and effective tool. However, we need to tread cautiously; P4P is no panacea.

Central to all health care reform proposals, across the political spectrum, is the observation that part of our problem is perverse economics.  The economics are warped in (at least) two ways.  First, since the post-World War II growth of employee provided insurance, and the 1965 introduction of Medicare, consumers have largely been removed from any direct financial impact of the provision of medical services.  With no consequence to consumption, demand grows without restraint.  This long-standing dynamic is starting to be reversed as employers are steadily shifting more of the financial cost of insurance to employees.

The second dynamic is equally pernicious.  In any city in the United States we have physicians who deliver excellent, evidence-based care; who are great communicators with outstanding bedside manner; who achieve outcomes beyond those of their peers.  We also have doctors in the same city (or same practice) who are insensitive, abrupt, below average clinicians with below average outcomes.   I believe, thankfully, the former category substantially outnumbers the latter.  However, here is where the economics are twisted:  the good, kind and efficient doctor will see the same sort of patient and perform the same types of procedure as the gruff doctor with poorer outcomes.  Both doctors will submit a bill for a unit of service to Medicare, Medicaid or private insurance, and both doctors are paid exactly the same thing.  A consequence of the payment system in America is the only way for a physician (or hospital) to do better financially, is to do a greater volume, pure and simple.  The sole incentive is to do more, not to do better.

Some studies show family physicians spend only 7-8 minutes with a patient in a typical visit.  It is not by their choice.  I know dozens, perhaps hundreds of family physicians.  As a group, they are wired to take time, engage in meaningful dialog, dig into complex interpersonal dynamics and try to understand how an illness impacts a patient as a human being.  This is why they chose their field.  Left to their own devices, most would happily linger with a patient for an hour or longer. 

The unfortunate economic dynamic prohibits this sort of interaction.   Family physicians are already among the poorest paid of physicians, earning less than half, or even a third of their specialist colleagues.  To pay office staff, cover overhead, keep the lights on and take home a reasonable wage, the typical doctor must see at least 30 patients a day.  I will do the math for you…over an 8-hour day, that works out to 16 minutes per patient.  Take from that time for documentation, phone calls, and sicker patients who demand more time, and you quickly get to 8 minutes per patient.

Enter P4P.  The concept is seductive in its simplicity.  Let’s stop paying doctors based on volume, and start paying them based on quality of patient outcomes and patient satisfaction.  On a purely intuitive level, this idea has almost universal appeal.  However, because something makes intuitive sense, does not necessarily mean it will work, especially when rolled out on a very large scale across a complex system.

I recently read two studies related to physician compensation as a tool to modify behavior.  Both studies had excellent and worthy goals.  Both implemented well-designed and rational compensation changes.  Otherwise, the two studies could not have been any more different, except for one other significant similarity… neither worked as expected.

The first study comes from an unexpected source:  a rural delivery system in a poor province of China.¹  The physicians in that community had two sources of income.  They billed for services in a traditional fee for service model – more volume, more income – and they were permitted to sell drugs at a substantial mark up.  The health care planners in the province were concerned that this payment mechanism was encouraging unnecessary service and over-prescription of medications.  The fix was to revise the compensation system.

The payment scheme was revised so that rural providers would now have three sources of revenue: 1) a base salary ($15/month); 2) a volume bonus ($0.06/outpatient); and 3) a quality bonus, based on defined metrics ($12.50/month).  The economic incentive to sell medications was eliminated.  Move the decimal point about three places to the right, and this payment system looks much like many US compensation models. 

Rational, thoughtful, well designed.  So what were the results?

The designers had a number of expectations. Spending would decrease on the village level.  Unnecessary care would be reduced, particularly to the young and healthy.  There would be a reduction in prescription of unnecessary drugs.  In fact, all these results were achieved, but with one significant unexpected consequence.  Sicker patients were more costly for the rural doctors to treat, and negatively impacted their quality metrics (and pay).  As a result, there was a sharp increase in the referral of the sicker patients to more expensive city clinics.  The net result after 5 years? No cost savings to the healthcare system.

From a small study in China to a massive study in the United States.²  Starting in 2005, ten physician practices were voluntarily enrolled in the Medicare Physician Group Practice Demonstration (PGPD).  These groups joined what was considered to be a precursor of Accountable Care Organizations (ACO).  Quality metrics were introduced, and cost saving monitored in the demonstration groups.  Any savings realized from more efficient, more effective care, would in part return to the physician practice.

From 2005 through 2009 there were 990,177 patients enrolled in PGPD practices.  In a quasi-experimental design, this group was compared to patients in non-PGPD practices, both during the trial, and for the four years preceding the trial.  Keep in mind, the practices that enrolled in this demonstration were best of class.  They would not have enrolled if they were not confident they had the ability to succeed. 

Like the Chinese study: good and noble goals, rational design.  The results?  Although not a total failure, the positive impact was limited to a relatively small subset of patients (see figure below, taken from the JAMA article in the footnotes).

For Medicare patients, over a five-year period, there were no statistically significant savings.  In fact, the statistical confidence interval was such it was possible the care was actually more expensive.

The group that did seem to benefit was the Medicare “dually eligible”, that is, those patients with Medicare, who were also eligible for some portion of Medicaid benefits.  This group is primarily comprised of the disabled, and the elderly poor.  The dually eligibles also tend to suffer from multiple chronic illnesses, and have a disproportionately high rate of mental health issues, so it is not necessarily surprising they benefited from the enhanced case management typically part of an ACO, “medical home” model.

Two studies, different sides of the world, different cultures, aligning incentives in a rational manner…both with disappointing results.

This is admittedly a very limited and selective review of the literature.  Although at the time health care reform was passed there was a dearth of evidence to prove P4P actually works, there has been subsequent demonstration of efficacy, at least in select populations. 

What lessons can we take away from this? I think there are three:

• Often we see the need to improve care, and the need to engage physicians and other providers in supporting change. This usually leads to a conversation around “aligned incentives”. The groupthink sometimes seems to be, “if we could design the perfect compensation system, solutions to all these problems will fall into place.” It is clear; compensation design is not a magic bullet. 

• General solutions, widely applied over a large population, are not likely to be effective for all patients. A team-based, case management intensive model may work exceptionally well for an elderly disabled patient with multiple medical conditions. It may not be the right model for a healthy young adult. 

• More important than a payment system of accountability, we need to develop a medical culture of accountability. Compensation may be a tool in shifting culture, but it is only a single tool. 

¹ Wang H, Zhang L, Yip W, Hsiao W. An experiment in payment reform for doctors in rural China reduced some unnecessary care but did not lower total costs. Health Affairs. 2011; 30(12): 2427-36.

² Colla C, Wennberg D, Meara E, et al. Spending Differences Associated With the Medicare Physician Group Practice Demonstration. JAMA. 2012; 308(10): 1015-1023.  Click for link to article