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.