On the Front Lines of Health Care Fraud
For faculty and students in the Master of Health Care Delivery Science program, explaining and rooting out fraud is an important part of reducing health care costs.
Plenty of people disagree sharply about how the United States should overhaul its health care system.
Just look at the short, unhappy life of the Affordable Care Act, which will cover tens of millions of previously uninsured Americans. After a year of bitter political infighting, it passed by only the slightest of margins. And this week, almost two years before much of the law takes effect, the Supreme Court will hear three days of oral arguments on the question of its constitutionality. The law might not see its third birthday.
Yet despite the endless bickering, almost everyone agrees with one thing: health care fraud, especially in Medicare and Medicaid, is pervasive and must be stopped. The arrest of a Texas doctor a few weeks ago gives a sense of the scale of the problem. Dr. Jacques Roy and six colleagues were charged with perpetrating a scheme—signing up homeless people for home health care they never received—that cheated the federal government out of nearly $375 million. Broadly speaking, the FBI estimates that fraud, waste, and abuse account for 3 to 10 percent of health care spending, or somewhere between $70 and $240 billion per year. Rooting out the bad actors in a health care system so large and bureaucratic has not been easy, but it is essential to the overarching goal of reigning in health care costs.
For faculty and students in Dartmouth’s Master of Health Care Delivery Science (MHCDS) program, a new degree that draws on the expertise of faculty from the Tuck School of Business and the Dartmouth Institute for Health Policy and Clinical Practice, the challenge is to not only understand the underlying causes of the fraud, but to design better processes that can prevent it from happening in the first place.
Robert G. Hansen, the Norman W. Martin 1925 Professor of Business Administration, is one of the faculty members who teaches the Health Care Economics class in the MHCDS program. He also teaches an elective in Tuck’s MBA program on the economics of the credit crisis, and he says there are important parallels to draw between health care fraud and the conditions that facilitated the 2008 real estate crash.
“You’ve got a huge government bureaucracy that’s paying fixed prices to doctors and hospitals,” he explains about Medicare and Medicaid. The fixed prices are set administratively, not by the prevailing market conditions, which can result in some perverse incentives. “When prices are too low, hospitals and doctors have an incentive to not do those services,” he says. “When they’re too high, the incentive is to do as many as they can, even if the service is not necessary.” In effect, millions of health care decisions are made based on artificial values.
This sort of administrative tinkering played a role in the credit crisis, Hansen says, because the price for different kinds of risks on banks’ balance sheets were set by banking regulations. The regulations classified different kinds of securities as being less risky, and therefore needing less bank capital. Within the same risk class, some securities would offer higher yields, “and the banks just loaded up on them,” says Hansen.
Fixed prices in health care pose another problem: they circumvent the laws of supply and demand. “Normally, as doctors do more of these services, the price of the service would come down,” Hansen says. “But if the price is fixed by someone in Washington, it doesn’t come down. In any other industry, excess capacity would lead to lower prices.” Such a system not only keeps the prices higher than they otherwise would be, it also provides an incentive to make fraudulent claims.
Economic principles don’t tell the entire story, however. Another key cause of Medicare and Medicaid fraud is the naïve and antiquated procedure for reimbursing claims. Called “pay and chase,” the system automatically pays every claim within 30 days of being submitted; only afterward is the claim investigated and “chased” if it appears to be fraudulent. Designed in the 1960s, this system was built like a house, says Edward Baker, the director of Vermont’s Medicaid Fraud Control Unit. “You put up the frame, put in the windows and doors, and the last thing you do is put the locks on. Fraud enforcement is an afterthought.”
Bret Anderson D’05, MHCDS’13 hopes to change that. He works as a health care consultant for Booz Allen Hamilton (BAH) and last year won the top prize at the company’s annual Ideas Festival. The festival is a chance for junior consultants to pitch a game-changing idea that BAH can implement for its clients. The winners receive funding from the firm to research and build out the idea. Anderson’s winning proposal was to develop a rapid-learning algorithm that assigns a risk score to Medicare claims so that fraudulent claims can be detected before they are paid. “I ripped a page out of the credit card industry’s playbook,” Anderson says. “With credit cards, if the system suspects fraud, it doesn’t allow the transaction to go through. I thought, Why doesn’t Medicare follow that lead?”
Anderson’s proposal garnered a six-figure investment from BAH. As the project manager, Anderson has been using the funds to hire in-house advanced analytics programmers, buy Medicare data sets, and go on marketing trips and site visits. Like the consumer profiles credit cards build for their customers, Anderson’s system is creating physician profiles for every type of doctor that accepts Medicare. The profiles establish norms for services and claim frequency. If a claim falls outside those norms, it’s flagged for further review.
Amazingly, Medicare doesn’t have a system like this in place already. Instead, it uses rule-based analysis to flag potential fraud. For example, Medicare will set a threshold for claims from a certain provider; if the provider meets or exceeds the threshold in a given period of time, Medicare will investigate. But the organized crime syndicates that perpetrate Medicare fraud quickly learn the thresholds and are careful to stay just below them. By contrast, “the advanced learning algorithm will identify behavior when it nears the threshold,” Anderson explains, “and knows that it’s highly correlated with fraud.”
For the moment, Anderson is building his system for general Medicare claims and Accountable Care Organizations, but it’s just a matter of altering the algorithm so that it can analyze claims for Medicaid and private health insurance companies. For Baker, who leads a team of eight people who investigate and prosecute Medicaid fraud in Vermont, a more modern fraud detection process would be a welcome addition. Currently, Baker says, Vermont is losing $30 to $100 million per year to Medicaid fraud; his team, which is working flat out, is recovering just a fraction of that. Without an improved method of fraud detection, and the resources to prosecute it, the losses are likely to get worse. That’s because, under the Affordable Care Act, about 16 million more people will be Medicaid eligible in 2014. About Anderson’s system, Baker says, “We need more ideas like that, and certainly technology is going to be part of the solution to reducing the amount of fraud in the system. Here in my office, we’re on the front lines of fraud detection and some days I feel like we’re winning. Other days, I feel like we’re barely keeping pace with our expanding caseload.”
Anderson has been in the health care consulting business ever since he graduated from Dartmouth, and he found that the MHCDS program aligned nicely with his career. “The focus on health care delivery is exactly the type of thing I’ve been interested in,” he says, “finding the best bang for the buck in our health care spending.” In the program, Anderson joined the Vermont single-payer learning group, which consists of doctors, administrators, and health care professionals who are designing the benefits package for Vermont’s new health insurance plan. That experience is helping Anderson in his job at BAH, and he’s also formed fruitful working relationships with other participants, especially the students from United Healthcare. “These folks are light years ahead of Medicare in terms of reducing fraud, waste, and abuse,” he says. “Having those perspectives in class has been a huge asset.”
In fact, these sorts of perspectives are not only useful for designing a better fraud detection system, but for developing solutions that get at the core of the problem: the inefficiency of a payment model that disburses fees for individual services. Such a system encourages providers to pile on the procedures and tests, because they will get reimbursed for everything (sometimes at a rate well above relevant cost.) “Payment reform is one of the big issues we’re trying to deal with in our program,” Hansen says. (The topic is woven into the syllabi of classes such as Health Economics and Policy; Population Health; Finance Essentials; and Strategy for Health Care Organizations.) “Right now, we’re in a fee-for-service world, but that’s shifting slowly to a bundled payment model.”
Under the bundled system, a provider will be reimbursed for the cost of, say, caring for a diabetic patient for an entire year. The good thing about this is that it will reduce the volume of claims and give doctors an incentive to offer patients only as much treatment as they “need”—ordering test after test, service after service, will only eat into the provider’s bottom line.
But the flip side of that is the possibility that bundled payments make it easier to commit fraud, since a provider can just submit one bill for a year of services that it never rendered. Or, the bundled system could provide an incentive to give less care than is necessary, making the patient suffer. “If we can measure what the health care providers are actually doing, in terms of quality of care, then the bundled payments will probably work better,” Hansen says.
These efforts—earlier fraud detection, more efficient and accountable payment models—could represent the best type of health care reform: lower costs and better care. Not many people can disagree with that.
—By Kirk Kardashian, March 2012