Big Data Analysis and Clinical Decision Making in Diabetes

Dennis P. Scanlon, PhD: Robert, at Joslin, and with your clinicians, you are sitting on a lot of data. You’re 1 organization. Are you partnering with others, or sort of analytically looking at your own data? I mean, one worries, obviously, of 1 institution or 1 organization and basing any conclusions on just an N of 1. How, in practice, are clinicians looking at this world of big data?

Robert Gabbay, MD, PhD, FACP: It’s sort of interesting. I think there is a long-standing tradition in medicine of providers going on their personal experience. The evidence on that is not very good, because you get 1 outlying case that something bad happens and that really changes practice much more effectively than many good cases. And so, it’s not really a balanced analysis.

The next step up from that is looking across an institution, and I think that data is helpful. We do look at that and we also have worked with a number of pharmaceutical companies and others to be able to mine that kind of data. But I think the real opportunity is in data sets that are far larger, whether those are health plan or merging health plan data with health system data that has laboratory data as well. That’s where I think we have the opportunity to do much more than we have done. That’s starting to happen, but there’s a huge opportunity.

Dennis P. Scanlon, PhD: So, if you look at a company like Aetna, or other payers, you’ve got variation across the number of providers, location, and patients. But you may not have that lab data or detailed clinical data to merge with that. That would be important.

Kenneth Snow, MD, MBA: That’s a key aspect. One of the hopes for the future is that we’ll be able to integrate that type of data, more effectively, into the data that we have through various relationships we have with the providers. Where we’re able to share that information and be able to bring the power of the information that’s collected on the individual patient level—lab data, physical findings, etcetera—but also bring it to a level where we’re talking about not necessarily hundreds or thousands but, now, talking millions of folks that we’re looking at in the analysis.

Robert Gabbay, MD, PhD, FACP: The American College of Cardiology has a large diabetes registry, now. I think it’s grown to about a million patients with diabetes with all of that data there. I think you’re going to see more and more of that, and there are companies that are data aggregating from various sources from claims data, pharmacy data, etcetera. I think those really big data sets are where you’ll be able to answer these kinds of questions more effectively, because you can do much better propensity matching.