The annual debate on how to spend Medicare dollars is here again. We have many modern analytical tools to look at health care spending, but it may be worth going back to some basic insights of microeconomics. About 140 years ago, a group of economists showed that rational decision making meant looking at marginal cost and marginal benefit. What Jevons, Menger, and Walras demonstrated was that decisions to consume or produce goods and services are optimized when marginal costs equal marginal benefits and, in particular, that price was the language for doing this calculation. Take a look at their formula:

Marginal Benefit = Price = Marginal Cost

Virtually all modern economists subscribe to this maxim. What does this mean for the Medicare program? Medicare faces the challenge of having to look at equity as well as efficiency. Nonetheless, Medicare still has to make purchasing decisions, which means it has to build a set of policies that somehow approximate this fundamental formula.

There is no pocket calculator to do this. Classical economics suggests that Medicare, being a consumer rather than a producer of care, should “simply” assess the marginal benefit of the various services and procedures and pay accordingly. But the modern-day equivalents of marginal benefit such as quality-adjusted life-years (QALYs) don’t have the analytic heft to drive Medicare’s almost $400 billion of purchases. Medicare, though a consumer, does an interesting thing. It goes over to the producer side of the equation — the marginal cost side — and sees what it can find there. We are all familiar with the result: cost reports and various utilization formulas that serve as the marginal cost proxies to calculate price and, by extension, marginal benefit to beneficiaries and the public.

There is a deep problem here.

Payment problems. In setting payment levels through tools such as relative value unit (RVU) calculations, Medicare looks at what economists call “average cost” rather marginal cost or ideally marginal benefit. In other words, it cost $X for an appendectomy. This is an acceptable approximation for well-defined treatments such as surgery or medications. It is far more problematic for diagnostic procedures.

Diagnosis can range from the physician walking into the room and making the diagnosis in 10 seconds (Arthur Conan Doyle’s Sherlock Holmes, if you will) to subtle findings undetectable except by modern imaging and testing. Diagnostic work is almost never a stand-alone pill or surgery where the marginal cost tends to equal the average cost. Diagnosis comes from a mix of man, machine, and time. Diagnosis is inherently an integrative process. Implicitly basing diagnostic payments on individual test cost calculation methodologies leads to massive problems.

The recent March 2007 MedPAC report to Congress describes this tension. MedPAC (the Medicare Payment Advisory Commission) notes great difficulties paying primary care physicians who are fundamentally diagnosticians and decries the increase in modern imaging costs. These are precisely the areas where Medicare payment methodologies are most broken. Diagnostic costs should be looked at in the context of costs for the entire episode of care. An average cost of an office visit or test does not equal the marginal cost or benefit of that visit or test in discovering what is wrong with the patient.

Why is the cost of diagnosis, both low-tech and high-tech, increasingly of concern? First, we acknowledge the fundamental disconnect of patients’ not bearing the cost of their purchase decisions. But a major part of the tension in paying for diagnostic services is that as diagnostic choices expand well beyond the office visit and simple blood tests and x-rays, Medicare needs to update its payment methods as well.

When the Medicare program began in 1965, advanced diagnosis was not an issue. There was almost no ultrasound, there was no CT, there was no MRI, and endoscopies were rigid (think metal tube). DNA microarrays were undreamed-of.

For Medicare to be a smart shopper and drive providers towards the smartest patient care in this era of incredible diagnostic insight into the human body, it has to use not just the average cost — the price of each test — as a proxy for benefit but to look at the total diagnostic evaluation and the total episode of care. In other words, the program must look at the marginal benefit for these tests, whether they include the time of smart, highly trained clinicians or CPU cycles of incredibly engineered machines. Defining episodes of care to calculate costs is not simple. But it can be done, as the June 2006 MedPAC report to Congress suggests.

In 1965, the diagnosis of appendicitis was made by vivisection, by a surgeon cutting into the abdomen and looking. In 2007, we diagnose the presence or absence of appendicitis with a 30-second CT scan. Medicare should stop looking at the year-over-year growth in imaging costs without also formally looking at the stunning benefits of the new diagnostic procedure. Even in 2007, marginal economics is the way to go.