Fraudulently or not, overlooked practice of ‘upcoding’ costs Medicare billions

Among the health headlines this week came the news that private Medicare Advantage plans have overbilled the government by billions of dollars, according to a critical report from the Government Accountability Office.

The audit came on the heels of the Center for Public Integrity’s 2014 series “Medicare Advantage Money Grab,” which found these plans overstated patients’ health risks, resulting in nearly $70 billion in “improper” payments to health plans from 2008 through 2013 alone.

In the wake of these news stories on Medicare Advantage plans, we caught up with Michael Geruso, a professor of economics at the University of Texas, Austin, and one of the leading researchers on the practice known as “upcoding,” which he defines as manipulating patient diagnoses in order to game payment systems, fraudulently or not. In this working paper, Geruso and Timothy Layton of Harvard Medical School used large datasets to investigate physician upcoding among private Medicare Advantage insurers.

In an interview with the Center for Health Journalism, Geruso explained a key distinction that is important for health journalists to understand: Focusing on fraud only tells part of the story.  The overall practice of more aggressively documenting a patient’s health conditions is what’s driving the vast majority of the extra costs.

“It is really sexy to think about targeting fraud, but fraud compromises the minority of the differences in coding and it’s not the economically-relevant phenomenon,” he said. “If the reporting doesn’t understand that, we might be missing the forest for the trees.

Even when upcoding is not technically fraudulent, it doesn't help patients.

For example, say that doctors working with insurer A were incredibly meticulous in documenting a patient’s condition in the claims submitted to the insurer, while doctors working with insurer B provided exactly the same sets of visits, procedures, and tests, but simply paid less attention to recording diagnoses in the claims files. In Medicare, the public will be paying more for enrollees that choose insurer A, even though the patients had the same medical experience, Geruso said.

“And it certainly imposes a huge cost on the public financing of health care,” he said.

How risk adjustment works

To explain upcoding in Medicare managed care, Geruso offered a quick primer on how risk adjustment works in these health plans, which now enroll more than 17 million Americans.

Under Medicare Advantage, the government is essentially giving consumers who are eligible for Medicare a voucher to buy a private plan of their choosing. To keep these private insurance plans from cherry-picking the healthiest Medicare patients, the government offers risk-adjustment payments. These payments compensate insurers for factors that might drive up the cost of insurance, such as one’s age, gender or a previous diagnosis such as cancer, diabetes or Parkinson’s Disease.

“Risk adjustment is trying to make all enrollees appear equally attractive,” he said. “When risk adjustment works perfectly, the insurer’s net costs of enrolling all patient are equal.”

"It is really sexy to think about targeting fraud, but fraud compromises the minority of the differences in coding and it’s not the economically-relevant phenomenon. If the reporting doesn’t understand that, we might be missing the forest for the trees." — Michael Geruso, University of Texas, Austin

But the method isn’t working perfectly.

“The moment you tie a plan’s payment to the level of sickness of the consumer the plan enrolls, you put in incentives for plans to try to make people appear sicker than they are,” he said.

For example, if someone were on the margin of having diabetes, they could be coded as having diabetes or being pre-diabetic – a difference that could result in thousands of extra dollars a year for Medicare Advantage plans.

That leads to huge financial incentives for more aggressive coding.

The latest findings

Unlike traditional fee-for-service Medicare, private Medicare Advantage plans receive diagnosis-based subsidies based on patients’ risk scores, a measure of their current diagnosed health conditions. Higher risk scores result in higher government subsidies, Geruso said.

During the same period of time, researchers found a significant difference in these risk scores between traditional Medicare and Medicare Advantage patients. The researchers estimate risk scores were inflated between 6 and 16 percent across Medicare Advantage plans. The higher payments were especially pronounced when doctors or hospitals owned their own insurance company, a practice known as “vertical integration.”

While fraud is what often captures headlines, a large portion of upcoding is likely not legally fraudulent.

“It’s just a matter of having a strong financial incentive to make sure every code plausible is consistently put on a patient,” he said. “A lot of the cost difference is coding intensity.”

That extra coding intensity translates into big government expenditures — to the tune of more than $2 billion annually, according to Geruso’s estimate.

For example, the same person who might generate $10,000 in yearly medical bills on regular Medicare will generate $10,700 a year in a Medicare Advantage Plan – simply because of more intense use of medical billing codes.

Researchers also looked at which conditions might encourage more intense coding, some of which were surprising, Geruso said. They went into the study thinking there would be little difference between traditional Medicare and Medicare Advantage plans in the diagnosis of cancer since that’s not a fuzzy diagnosis. But cancer shows some of the largest differences. After some digging, researchers realized that’s because the cancer category includes benign tumors. While traditional Medicare doctors might not code for this if there’s no procedure involved, Medicare Advantage plans typically include the code since it bumps a patient into the lowest category of a cancer diagnosis.

Insurers encourage upcoding

Through conversations with doctors, physician groups and insurers, Geruso researched the various ways insurers encourage more intense coding.

The practice can occur at the level of the doctor’s office. Typically, front office staff take physicians’ diagnostic notes and put them into claim forms, which are then sent to insurers. Even though these staffers work for the physicians, insurers can train them to code more intensely, or to watch for certain codes. 

The insurer can also reject claims and send them back to doctors if they are not appropriately coded. There’s even software that can “scrape” the doctor’s notes for evidence of codes they might have missed.

Medicare pays these Advantage plans based on how sick their enrollees are, rather than fee-for-service.  So when these insurers deal with large physician groups, they can offer incentives for more intense coding.

Another strategy is to send home health agency staff to a patient’s home to ensure they receive a diagnosis in a calendar year if they haven’t had a chance to visit the doctor. (Of course, home visits for sick patients aren’t necessarily bad, Geruso acknowledged.)

What can be done?

The Centers for Medicare and Medicaid Services (CMS) acknowledges that Medicare Advantage is coded more intensely, so they lower the subsidies. But they’re not deflating subsides enough, Geruso said. One possibility is to deflate Medicare Advantage risk scores even further.

Eliminating risk adjustment all together, though, isn’t an option, Geruso said.

“If we didn’t have risk adjustment, insurers would find ways to completely avoid unhealthy Medicare populations,” he said. “Risk adjustment does an important job.”

For reporters looking to cover this story in their community, Geruso suggested interviewing physicians and doctor groups. Ask how much of their time and resources are devoted to coding. Compare the answers between physicians who primarily serve traditional Medicare patients versus Medicare Advantage patients.

When the researchers posted their working paper, they received a call from a doctor who wanted to vent about why she left her large group practice, which was focused so intently on coding that she felt it was a disservice to patients. That physician was part of a physician-owned insurance group.

He also cautioned reporters to be clear and purposeful in how they define upcoding in their articles. In many news reports, it’s used to refer to fraudulently altering codes. Others, such as Geruso and Layton, use the word to mean differences in coding intensity.

[Photo by Wonderlane via Flickr.]