Berwick on Analytics: Technology Is Ready, but Doctors Need Help

Scott Mace, for HealthLeaders Media , May 15, 2012

If Marcus Welby, MD, were practicing on TV today, would he be letting data drive his decision-making? I’m on a journey to find the answer to this and related questions. Last week this journey took me to Atlanta for a HealthLeaders Media Roundtable on business intelligence and predictive analytics, and then onward to North Carolina for a conference dedicated to healthcare analytics.

While in North Carolina, I got to sit down with Don Berwick, MD, former administrator at the Centers for Medicare & Medicaid Services, and prior to that, founding CEO of the Institute for Healthcare Improvement. We talked about data analytics, but our discussion ranged far and wide around healthcare IT. Here is a portion of our conversation.

HealthLeaders Media: How far along is healthcare with its adoption of analytics?
Berwick: I certainly see the potential. At CMS we did do some early trials with Oak Ridge National Laboratory, which has tremendous data capacity. [CMS] gave them access to privacy-protected Medicare information. They have tremendous analytic capacity, and it was stunning what they did. I remember visiting Oak Ridge, and they had modeled some uses of Medicare and Medicaid data, and they were coming up with insights right away, of geographic patterns of variation that I don’t think Health Services researchers knew about.

HealthLeaders Media: Why was Oak Ridge doing this? I don’t think of them usually in this space.

Berwick: Oak Ridge is not just a Department of Energy supplier. They work with other government agencies that want to contract with them to do essentially analytics and data mining. The one place I saw analytics working was in our early work on predictive analytics for fraud. The Affordable Care Act suite of efforts to reduce fraud involves the traditional what they call pay-and-chase, which is enforcement. You find something wrong and you prosecute. We were working with the Department of Justice and the FBI and local law enforcement to catch criminals. That’s traditional and effective. You need to do it. But it’s, after all, after the horse has left the barn.

So upstream from that, there’s prevention. Make sure that the people that want to offer home healthcare or durable medical equipment, that they’re qualified to do so, they don’t have a history that makes you suspicious, and since there’s a very high concentration of fraud in certain parts of Medicare payment, one’s able to target prequalification as an area. But I thought the most promising was predictive analytics, which was take the data and turn loose the ability to go through it looking for weird patterns. The technology was ready.

Along about this time, I took a vacation with my wife in Turkey. I got online to buy a ticket for an internal flight in Turkey from Istanbul to an interior village, and I’d say a minute or two later, my cell phone rang, and it was American Express saying, “Just checking—a purchase was made in Turkey. Is this you?” Well it’s the same thing, where we can get not just retrospective but almost real-time signals. I remember the first run of predictive analytics, the volume of insights and ideas and hot spots that were spotted, it was really something.

HealthLeaders Media: I was listening to a podcast with Dr. Lynn Vogel, CIO of MD Anderson Cancer Center, who told an interviewer that the number of facts going into physicians’ decisions on treatment is growing exponentially. What’s the healthcare system going to do about that?

Berwick: It is doing [things] about it. I don’t know what Dr. Vogel meant by that comment particularly, but there are at least two meanings. One is that the science base is expanding vastly. A few years ago, I did a Google search to see how many randomized clinical trials in the world are underway right now, and the number that came up in the search was 40,000. At any one time there are 40,000 randomized trials underway. The concept that an individual human mind could possibly search through that and find out what’s relevant to your needs when I see you in my office, that obviously is folly. There has to be some intermediation of the science so that someone, some technologically supported, trusted agent is digesting that and making it available. We’re quite a way along there. We’ve had wonderful work, professionally led by the American College of Physicians and the American College of Surgeons and the cardiologists to do exactly that: intermediate between the scientific wealth and direct application. So I think that’s going okay. I’m sure it could be better. There’s been some ambivalence in public policy around this. This is the big debate about clinical effectiveness research, which is this kind of confusing question as to should we use science in decision-making? We actually are asking that question. Hopefully we’ll get over that at some point.

On the [care] delivery side, the data stream is overwhelming now. Every beep on a monitor is a data point. There are some successes. For example, Intermountain Healthcare has for years had in the ICU real-time data collection, multiple facts about a patient that can help guide action in real time. You’ll see more and more of that. I don’t think that’s conventional. I think most docs are still bathing in that kind of data without much help.

HealthLeaders Media: In all the debate about Meaningful Use, it seems to me that the incentives don’t encourage knowledge transfer from the successful innovators, like Intermountain, to those who are following.

Berwick: We’re just barely in Phase 1. The concept of meaningful use is going to grow. What doctor in the end would not want access, if it were technically available, to the answer to the question, Who does the best at this, or What does Mayo think, or Where’s the best science? We’re just in an adolescent moment in terms of evolution of that kind of knowledge transfer. Right now, my daughter is a second-year resident in medicine at the Brigham hospital in Boston. When I was in training, the message was, put it all in your head. Get it in your head. Read it, memorize it, and then spout it out at Rounds. They carry their iPhones on Rounds. So a question comes up, and they’re using their iPhones and iPads right there to get the information. Why would they store it in their head if it’s in the world? That’s a basic human factors design idea. Knowledge in the world is more useful than knowledge in the head. Young doctors and nurses are coming from a totally different mindset about access to knowledge.