What are sources of real world data?

What are sources of real world data?

Inherently, as I alluded to, real world data are complex. They're messy because, again, life is messy.

And so I think the best type of real world, the basically the more representative real world data are, this data source is for the patient sitting in front of me, the more useful it is for them.

So the data sets that I like the most for real world data are the ones that come from anonymized patient data from across America, not just one center, not just a coalition of academic centers, but everywhere, including community sites, including, you know, Veterans Affairs, VA hospitals, including, you know, other hospitals where patients, you know, get care that might not be represented in these kind of big academic center databases.

And so that I think would be the best type of real world data, that comes again from looking at patients regardless of the treatment setting and using it. Other uses, the real world data, which are real world, I would argue not quite as real world if you can use real is not a binary yes or no, but somewhere in between would be.

For example, I have conducted data with, you know, patient data, at my center at the Fred Hutchinson cancer center in Seattle where we get IRB approval. So it is approved, a use of the patient data, where we use anonymized data to look at, for example, does dexamethasone caused weight gain or, diabetes or blood sugar issues or other complications, or looking at other questions therein looking at a single center is worth of data.

That's another example where I think that is a little bit more robust in some ways, because we're able to get more information about each patient because they’re our own patients, however, they’re the only patients being treated again at an academic center in a big city. How useful that is to some patients, I don't know. One source of real world data we didn't talk about would be claims data as well.

So many, researchers have also used, for example, Medicare data, something called SEER, which is a database that Medicare maintains where you look at claims. So looking at, hey, how often was Medicare billed for this medication called bortezomib in patients with myeloma? And for example, with that, did we notice that patients who got bortezomib twice per week versus once per week, if there was any difference in how long they stayed on that insurance plan, for how long before we saw the treatment switch to a different myeloma treatment and so forth.

That I think has some pros because, you know, some with insurance company may follow them across different centers. They may be able to capture, for example, that the patient went to urgent care at a different hospital, and wasn't at the main center or something along those lines?

However, claims data are also messy because as people listening to this will know you know, for example, lenalidomide in myeloma is not always covered by insurance.

People sometimes have grants. Sometimes you request, you know, a, what's a good example, venetoclax. We prescribe 100 milligram capsules. But what the patient actually takes might be 200mg or 400mg, or they take 2 or 4 of those pills. And a claims database cannot tell you how much of that pill or how often they're actually taking it.

That's another example of real world data, there are, you know, large, companies that are devoted to mining these data in a de-identified manner from different centers. And so I think there's many different ways of looking at it. I think the one unifying theme across everything I mentioned is that real world data, you know, it is they are real data points that are being used for research.

They do require IRB approval. IRB is an institutional review board, meaning that before I could even use this data, even if the patient's name is scrubbed, patient information is scrubbed. I need to have a research plan, and I need to have my either my hospital or someone else's hospital approve the research because this is a useful use of patient data and roll with it.