Medical Students Crunch Big Data To Spot Health Trends

Oct 30, 2015
Originally published on November 2, 2015 4:04 pm

Medicine, meet Big Data.

For generations, physicians have been trained in basic science and human anatomy to diagnose and treat the patient immediately in front of them.

But now, massive stores of data about what works for which patients are literally changing the way medicine is practiced.

"That's how we make decisions; we make them based on the truth and the evidence that are present in those data," says Marc Triola, an associate dean for educational informatics at New York University's medical school.

Figuring out how to access and interpret all that data is not a skill that most physicians learned in medical school — but that's changing.

"If you don't have these skills, you could really be at a disadvantage," says Triola, "in terms of the way you understand the quality and the efficiency of the care you're delivering."

That's why every first and second year student at the NYU School of Medicine is required to do what's called a 'health care by the numbers' project. Students are given access to an enormous database with more than 5 million anonymous records — information on every hospital patient in the state for the preceding two years. "Their age, their race and ethnicity, what zip code they came from," Triola lists, as well as their diagnosis, procedures and the bills paid on their behalf.

The project, funded in part by an effort by the American Medical Association to update what and how medical students are taught, also includes a companion database for roughly 50,000 outpatients. It's called the Lacidem Care Group. (Lacidem? That's "medical," backwards.) The database contains information from NYU's own faculty practices — scrubbed to ensure that neither the patients nor the doctors can be identified. Students use analytic tools provided by the project to "look at quality measures for things like heart failure, diabetes, smoking and high blood pressure," Triola says, "and drill down and look at the performance of the practice as a whole, and [the performance of] individual doctors."

Some med students — including Micah Timen, now in his second year — have taken to the assignment with relish. Timen likes numbers. A lot. A former accountant before applying to med school, he keeps a spreadsheet to track his study hours in preparation for a test. Regarding an upcoming test on the digestive system, for example, Timen says, "I know I have 18 hours and 40 minutes left to make sure I feel comfortable walking into my exam."

For his database project, Timen wanted to know if the cost to patients of hip replacement surgery around the state vary as much as the cost of a fast-food hamburger. Timen says they tried comparing the varying prices of a hip replacement using The Economist magazine's famous Big Mac Index, which measures purchasing power between currencies. "But when you call McDonald's, they don't give you prices over the phone," he says. So he tried Plan B: "Burger King gave it to me."

Using the "Whopper Index" instead, Timen found, not surprisingly, that the price of a giant burger sandwich is higher in New York City than, say, Albany. So, too was the amount of money patients paid for a hip replacement. But the margin was much wider for health care than for hamburgers, meaning patients are paying more in some places than simple geography would suggest. Timen says he'd like to explore why that might be, "but unfortunately med school is a little bit time consuming," so that may have to wait.

Meanwhile, these classes appeal not just to "data junkies," like Timen, but also to those who were not already steeped in crunching data.

"I really have no statistical background," says Justin Feit, also a second-year student. "I don't even know how to use Excel well."

So Feit was partnered with Jennifer Lynch, who already has a Ph.D. in physics. Lynch says that if medicine wasn't moving in the direction of including more interpretation of big data sets, "I don't know if I would have gone into medicine."

Together Feit and Lynch looked at the rates of cesarean births around the state, and found that the rates of C-section — like the cost of a hip replacement — varies widely. But their project will get more than just a grade. A faculty member at NYU is using it as part of a bigger research project headed for publication.

Triola says he hopes that will happen more and more.

"With literally millions of records, these in-class student projects often involved more patients than the published literature," he says. "It's incredible."

And the concept of having students learn to use health data is catching on quickly. Triola says NYU has offered its database and program to other medical schools, and seven are already incorporating it into their curricula.

Copyright 2015 Kaiser Health News. To see more, visit http://www.kaiserhealthnews.org/.

Transcript

ARI SHAPIRO, HOST:

Big Data is becoming a big deal in medical schools. For a long time, doctors have been trained to care for patients based on the medical knowledge they've acquired. Today, doctors are expected to treat people, in part, based on how those patients compare to millions of other patients with similar conditions. This is where the ability to understand and use huge amounts of data comes in. Julie Rovner visited a medical school in New York and sent this report.

UNIDENTIFIED MAN: The portal vein has what next to it, then?

UNIDENTIFIED WOMAN: The arterial and the bile duct.

JULIE ROVNER, BYLINE: This classroom at New York University on Manhattan's East Side is bustling with medical students doing what medical students do - learning basic science. Micah Timen is a Cleveland native and second-year student at NYU. He's planning to become an ear, nose and throat doctor.

MICAH TIMEN: We're in the histology lab, which is basically where you look at slides, but on a computer. So right now we're doing it about hepatocytes, or liver cells.

ROVNER: NYU medical students are using computers for something else these days, too. Every first and second-year student is required to do what's called a health-care-by-the-numbers project. Students are given access to an enormous database with more than 5 million anonymous records - information on every hospital patient in the state for the past two years.

MARC TRIOLA: What zip code they came from, the diagnosis they were treated for and what procedures they had, the bill that they received or their insurance company received at the end of the visit.

ROVNER: Marc Triola is an associate dean at NYU medical school who helps run the program. He says that learning to use and manipulate data will make students better doctors.

TRIOLA: Medical school has been designed to teach students how to take care of that patient in front of them, but not necessarily to think on that meta level about how they and the health care system they're working in are doing across the 2,000 patients that each provider has or 20 or 30,000 patients that our practice has.

ROVNER: Triola says that because the database includes cost information, students get a head start learning not just about how well medical care works, but about its cost effectiveness.

TRIOLA: Everything is pushing us at using those data in ways that we can improve the efficiency and the value of the care that we deliver. And physicians should play a big part in leading that initiative. They should be skilled in this.

ROVNER: Cost was the focus of the project Micah Timen did with his partner. Their question - does the cost to patients of hip replacement surgery around the state vary as much as the cost of a fast food hamburger? Timen says they tried comparing hip replacement costs using The Economist magazine's famous Big Mac Index, which measures purchasing power between currencies.

TIMEN: But when you call McDonald's, they don't give you the prices over the phone. So I don't know how The Economist got it necessarily, but Burger King gave it to me.

ROVNER: So Timen used what he called a Whopper index instead. He found, not surprisingly, that the price of a giant burger sandwich is higher in New York City than, say, Albany. So, too, was the amount patients paid for their hip replacements. But the margin was much wider for healthcare than for hamburgers, meaning patients are paying more in some places than simple geography would suggest. Now, Timen is not necessarily the norm when it comes to med students doing data analysis. He's a former accountant who keeps a spreadsheet to track every hour he studies. Right now, it's tracking his review for a test on the human gut.

TIMEN: So I know I have 18 hours and 40 minutes left to make sure that I feel comfortable walking into my exam.

ROVNER: Some of his classmates, like fellow second-year Justin Feit, are not so comfortable with big data.

JUSTIN FEIT: I really have no statistical background. I have a clinical research background, but nothing - I don't even know how to use Excel well, so...

ROVNER: So Feit was partnered with a student who already has a Ph.D. Together, they looked at the rate of cesarean births around the state and, like the cost of hip replacement, found that they varied widely. But their project will get more than just a grade. A faculty member at NYU is using it as part of a bigger research project headed for publication. Associate Dean Marc Triola says he hopes that will happen more and more.

TRIOLA: With literally millions of records, these in-class student projects often involved more patients than the published literature. It's incredible.

ROVNER: And the concept of having students learn to use health data is catching on quickly. Triola says NYU is offering its database and program to other medical schools. Seven are already incorporating it into their curriculum. For NPR News, I'm Julie Rovner. Transcript provided by NPR, Copyright NPR.