Big Data Peeps At Your Medical Records To Find Drug Problems : Shots - Health News It's been tough to identify the problems that only turn up after medicines are on the market. An experimental project is now combing through data to get earlier, more accurate warnings.

Big Data Peeps At Your Medical Records To Find Drug Problems

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript


It's MORNING EDITION from NPR News. I'm Steve Inskeep.


And I'm Renee Montagne. Drugs don't go on the market until pharmaceutical companies convince the Food and Drug Administration that they're reasonably safe and effective. Typically a potential medicine can only be tested on a few thousand people.

INSKEEP: Once it starts being sold many more people will likely take that drug and there is a chance that problems will occur. The question is how can the FDA rapidly spot the side effects. Today In Your Health, NPR's Nell Greenfieldboyce reports that the FDA has recently been trying something new and there's a good chance that it involves your medical records.

NELL GREENFIELDBOYCE, BYLINE: Here's the way the FDA has traditionally been alerted to possible dangers that start showing up once a drug is on the market. A doctor or pharmacist sends in a report that says, hey, my patient took this drug and later had a stroke or developed a strange rash. The reporting is completely voluntary and ad hoc.

JANET WOODCOCK: We get about a million reports a year that way. But those are random, OK, they're whatever people choose to send us.

GREENFIELDBOYCE: Janet Woodcock heads the division of the FDA that evaluates drug safety. Figuring out from those reports if a drug is really to blame isn't easy and what's more if the side effect is something common, there might not even be any reports because doctors might not connect their patient's symptom to the drug.

WOODCOCK: We need a rapid way to find out what's happening with drugs, especially safety of drugs, after they're approved and on the market.

GREENFIELDBOYCE: Increasingly to meet that need the FDA calls up folks in a brick office building near Fenway Park in Boston. Take the elevator up and there is Richard Platt, a mild-mannered man in a bow tie. He's a professor at Harvard Medical School's Department of population medicine and he works at a research institute founded by Harvard Pilgrim Health Care, a not-for-profit health plan. Platt is amused that I want a tour.

RICHARD PLATT: It looks like an office. There are - but, I'm happy to walk you around and show you the layout.

GREENFIELDBOYCE: These cubicles are the operations center for mini-Sentinel, a pilot program to mine huge databases of medical records, for signs that drugs may be linked to problems. Don't let the mini in the name fool you. The government has invested over $100 million into this project. Its aim is to transform how our nation monitors the safety of drugs on the market.

GREENFIELDBOYCE: And you guys gotten a lot of media attention?


GREENFIELDBOYCE: I mean, like have you ever had a reporter wandering around here asking you questions about this?


GREENFIELDBOYCE: The questions that do come in, come from the FDA . If federal officials have any concerns about a medicine, they can use mini-sentinels to quickly investigate. Mini-sentinel work's with 18 data partners, almost all health plans and insurance companies. That means it can do studies using the health records of nearly 180 million Americans.

GREENFIELDBOYCE: What are the chances that somebody listening to this NPR program has their data searched when you all do a query?

PLATT: If you have insurance through a private health plan, the chances are pretty good.

GREENFIELDBOYCE: Almost all of it is billing records. So, not what your doctor scribbled in your chart but rather the diagnostic and procedure codes on your medical bill. Platt says, don't worry about your privacy, the data partnership has been carefully set up to preserve it. He says his own records have probably been part of a study.

PLATT: I have no way of knowing that. We built the system so the analysts here would have no way to say, let's see if Doctor's Platt's data is in there. The system is built, so, that's an impossible question to ask.

GREENFIELDBOYCE: It's taken five years for Platt's team to build mini-sentinel up from scratch and make it a routine part of the FDA's work.

PLATT: We're now doing hundreds of queries a year.

GREENFIELDBOYCE: For example they ran a search when the FDA got some troubling reports about a blood pressure drug called Olmesartan and intestinal problems.

PLATT: They had noticed that patients who'd received this drug were having these complications they thought might be related to the drug.

GREENFIELDBOYCE: At first mini-sentinel found no connection. Then the FDA asked for a second search, this time focusing on people who took the drug for long periods. Platt says they found a link and the FDA added a warning to the drug's label.

PLATT: It was possible to provide an answer that otherwise would, frankly, just be unavailable to them.

GREENFIELDBOYCE: Mini-sentinel was an experiment to see what was possible. It's contract ends in September and the FDA will be deciding what to do next. Almost everyone agrees that the ability to sift through huge amounts of patient data is the way of the future. But not everyone believes we really know how to do it yet. Thomas Moore works for a nonprofit called the Institute for Safe Medication Practices.

THOMAS MOORE: I think it's a good and important step that the FDA is moving in this direction. The problem is, I think they've underestimated how far they have to go.

GREENFIELDBOYCE: He says billing data was never meant to be used this way. He questions how well it can reveal side effects from drugs.

MOORE: The biggest danger is that people will get a false reassurance about safety.

GREENFIELDBOYCE: He points to a controversy over a new blood thinning medicine. Reports had come in about serious bleeding episodes.

MOORE: The FDA published an article in a leading medical journal, basically discounting all that, saying that using mini-sentinel they had seen no unusual risks for this drug.

GREENFIELDBOYCE: The agency is now taking another, more nuanced, look at this drug, after critics questioned how well the FDA had designed the original mini-sentinel study. People at the FDA are aware that for this kind of research with big data the way you set up the question can affect the answer. Janet Woodcock says it's a real issue.

WOODCOCK: Things may change depending on which claims data set you use or how you run your definitions, how you set up your parameters and so forth. And there's no doubt these studies are vulnerable to all of these changes.

GREENFIELDBOYCE: She says that's why the FDA is also involved in another effort to explore all of these potential effects. This effort is led by The Reagan Udall Foundation for the FDA. That's a foundation set up by Congress. Right now its major project is research on how to use big databases to study the safety of medicines on the market.

TROY MCCALL: This is a new science and much work needs to be done to develop and continue to improve the methods behind this.

GREENFIELDBOYCE: Troy McCall is managing the project for the foundation. This foundation does get some money from the FDA but supports its research with other funds. And so far all the money for this particular project has come from pharmaceutical companies like Merck and Pfizer. McCall says he doesn't think it's seen as a, quote, "industry project" because the work is so important.

MCCALL: Clearly at the end of the day this is all to benefit patients and ensure that drugs that are on the market are safe and that no drugs are taken from the market that are safe as well.

GREENFIELDBOYCE: Drug companies do have a big interest in what stays on or off the market. I asked Woodcock at the FDA if she had any concerns that the industry is funding the development of the scientific methods that will then be used to help make regulatory decisions.

WOODCOCK: It would be very difficult to develop a method that was going to favor your drug, when you're developing a general method I think. But what I would say is, OK, who is going to develop these methods? We need them developed.

GREENFIELDBOYCE: Foundation officials say, it operates transparently and has different stakeholders represented on its governing boards. But one board member, who represents consumers, thinks the lack of independent funding is troubling.

DIANA ZUCKERMAN: I think that creates an appearance of a conflict of interest and potentially a real conflict of interest.

GREENFIELDBOYCE: Diana Zuckerman is president of a nonprofit called the National Center for Health Research.

ZUCKERMAN: If all the money is coming from the pharmaceutical companies, whose livelihoods are going to be affected by what the project finds, I just think that's an untenable situation.

GREENFIELDBOYCE: She worries that industry funding might influence the methods that eventually get used for finding drug safety signals in big data. And says that could potentially limit what this new approach will reveal about the medicines Americans take every day. Nell Greenfieldboyce, NPR News.

Copyright © 2014 NPR. All rights reserved. Visit our website terms of use and permissions pages at for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.