Here & Now

Noon Wednesday with guest Justin Sydnor

Marisa Wojcik:

Welcome to Noon Wednesday, I’m Marisa Wojcik, a multimedia journalist with Here & Now on Wisconsin Public Television. This week we were inspired by a recent ProPublica article that talks about big data. And all of the personal information that health insurers might have and how that could potentially be used to do different things like set prices. So, joining me to talk about this is Justin Sydnor. He’s a UW Madison Professor and expert on health insurance and risk management. And Justin, thanks so much for being here.

Justin Sydnor:

Thanks for having me.

Marisa Wojcik:

I want to start by asking you what does the phrase social determinants of health mean? It kind of sounds like a very jargony term. But it comes up quite a bit, especially in this article.

Justin Sydnor:

It’s really a broad term that people use to refer to all the ways that your social environment and your economic environment end up shaping your health outcomes and your health spending. Maybe a good example would be take something like diabetes. If you’re going to get diabetes and how you’ll respond to treatment will be based in part on genetic and biological factors about you. But it will also be based on things that are related to sort of where you live and the culture of how you’ve grown up. Do you live for instance, in a place where the norm is to eat fast food most of the time? And where all of your friends and your family do that. That may make it more likely you’ll get diabetes but also harder to manage once you’re diagnosed. And that’s really what people are talking about when they talk about social determinants.

Marisa Wojcik:

Also in this article you can request from LexisNexis, which is an insurance broker, to see what kind of information they have about you. I did that for myself personally. There’s a lot of papers here even though, I’m not that old but none of it has to do with my medical history. It has a lot of where I’ve lived. Some information about some of my assets. But, largely, nothing to do with my health. So, what kind of things are they using this information for? Or do we even know?

Justin Sydnor:

To some extent, we don’t fully know. So you know, what the insurers and what these data brokers like LexisNexis and Optum are doing in this space is a little bit cloudy right now. And they don’t have to publicly release all the information they have or exactly what they’re doing. But what we know is that health insurers have a bunch of data about your health. They have access to your medical records and to your claims. There are lots of regulations around that data about you.

So it’s hard actually, for a health insurer to for instance, sell other people information about your medical spending. It’s kind of comforting actually that LexisNexis didn’t spit out for you a record of all your medical claims. There are good federal regulations to protect your health data in that way. But, what these companies can do is gather a whole bunch of other data about you. So there’s information that we might be familiar with from our credit profile.

So they can see things about where you’ve lived. What borrowing you’ve done. Whether you have an auto loan. Whether you have anything you haven’t paid. But increasingly they’re also collecting a lot of data from social media sites and from the internet about other things about you. Sort of who your friends are, maybe your political inclinations, and a range of other sources of data about you. And increasingly some of these data providers are starting to try to merge that to medical records.

Either by selling this data file with all this other social and personal information about you to health insurers or for a couple of these companies, they’re linked to the really large national health insurers. And they themselves are merging these out sized data sources with the health claims.

Marisa Wojcik:

Do we know if they’re using that information to set prices and you said with the definition of social determinants of health? Based on knowing that I used to live in this geographical area they could make assumptions about my health patterns or my eating patterns or anything like that?

Justin Sydnor:

In most markets where you might get your health insurance, they currently can’t and don’t use this type of data to set the rates. So in particular, if you get your health insurance through an employer like many of us, like I do for instance. There’s laws that prevent them from pricing people differently for your insurance there. If you get your health insurance through the Affordable Care Act exchanges, like healthcare.gov, also they can’t price you differently there.

There are some places where they can set different individual rates. In particular for instance, if you’re looking on the private market for a short term health plan, which has been in the news a lot lately. There they could potentially use these types of data to set different prices for you. Now, we don’t actually have much evidence about the extent to which health insurers have actually found these data useful for doing that. And whether any of them really are doing it yet.

Marisa Wojcik:

So we shouldn’t be afraid quite yet?

Justin Sydnor:

Yes, so it’s more the potential than the reality at this point. I think that’s actually true for the insurers. For the most part they probably don’t yet know themselves the extent to which gathering this type of information about you could be useful. You can imagine that it might. But it’s not obvious that it will be.

Marisa Wojcik:

Since it’s not regulated, is it in our best interest to just trust that they’re going to use it in the right way and in the ProPublica article they talk about oh, we’re trying to tailor our services to you. We’re trying to get a better product to you. Similar to Facebook using our personal information to personalize ads or Google, or any other source that’s kind of collecting all of this data on us?

Justin Sydnor:

Yeah, so I think whether this is a positive or a negative thing, you know, is something I think we should be worried about. Even though they aren’t necessarily using it right now. And I think there’s an argument where it could be beneficial. And that’s the argument that the insurers are making. That potentially they could use these data to better tailor health services. So I think a good example of that would be, let’s go back to my diabetes example.

So let’s say you get diagnosed with diabetes. They may have found out that getting people to manage their diabetes correctly, exercising more, changing weight loss, regularly taking their medicines, can benefit from something like a nurse calling you or maybe even making a house visit. But sort of thing’s really expensive. So it would really benefit them to be able to target those resources mostly to the people who are going to be affected by it most.

So in theory if they had a model that could predict you’re the sort of person who will struggle with these things and could benefit from a nurse. They could devote resources better. Now, I think in practice, that’s still not a reality that I’ve heard of really in any clinical setting. We don’t for instance, have published health studies showing examples where health systems are using data in that innovative of a way. That’s still a little bit far out. But that’s sort of a dream scenario.

Maybe they could do this. And having this social data about you might be useful for that. The other reason they may want it though, is that in the markets where they can charge different prices, or where they can decide whether or not to accept you as a patient or as a customer. There they might want to have a model that takes as much information as they can and makes a prediction or an estimate of how much your health cost is going to be.

That sort of thing that I think most of us as consumers are rightfully should be much more wary about. Because the main goal there is on their standpoint to avoid people who are likely to need a lot of healthcare. And that sort of cherry picking is problematic both for the individual. But also broadly for the health market. Those people end up needing to get care somewhere.

Marisa Wojcik:

And also is it fair to maybe assume that I have a certain lifestyle just based on this information but they actually don’t know for sure?

Justin Sydnor:

You’re raising actually a really fascinating point. That’s broadly relevant for this whole movement towards big data and using predictive models. So from the insurance company’s standpoint, all that really matters is whether this information they have about you predicts on average that you’re likely to be higher cost. If they use it then for a whole bunch of people, they’ll be identifying on average a bunch of people who are highest cost.

And if the model’s good, that will help them better predict. But for any one individual, the prediction could be really bad. And the insurer that’s not so important. Because they’re based on averages. They’re trying to actually do the best they can, given the information they have. But for you as an individual it can be deeply unfair.

And it can create a lot of unexpected cost or for instance being rejected for health insurance coverage in certain markets based on things that both feel like a privacy invasion but also may not actually for you be relevant to your health. So there can be a lot of errors in the data and to have it still be something that’s useful to the company.

Marisa Wojcik:

Is there any sort of movement towards regulating what kind of information can be used about us, assumed about us?

Justin Sydnor:

I think there’s a fair amount of interest and movement in this direction. I don’t think a lot of it is very far down the road. In other countries, so in Europe for instance, there are typically very strong national privacy laws that make it much harder to do any of this type of thing. Here in the US we don’t have very strong laws in that way except for narrow pockets like your specific medical information.

There has been some legislation that I think is, I think best way to think about it is being floated and considered but hasn’t yet moved significantly forward at the national level to put some more regulation on this. And, you can see some parallels with things like credit reporting. Credit reporting bureaus over initially, just gathered lots of information about your spending and your debts and things and were pretty unrestricted.

And over time there have at least been some laws passed that force them to share with you all the information they know. And we can see for instance from your example, that’s not yet true in this space. So Lexis has to give you some of the information they have but there’s not a clear regulation about them needing to share everything. And there aren’t clear limits to what they can do with it. Or for instance what they’re allowed to merge it with.

Marisa Wojcik:

So without strict regulation yet, do we know if there are things that, if I wanted to at least limit the footprint of information that’s out there about me. Is there anything that I can do to do that?

Justin Sydnor:

After we arranged this interview, I looked into that a little bit myself. I’m not a privacy expert. Main conclusion I got is that it’s pretty difficult. A lot of the information I think, for most of us who are active on the internet, and other things is really already out there. And there are so many different sources of places where we leave digital breadcrumbs. That knowing how to turn off that tap entirely I think is going to be hard.

So when you look into these things you’ll see there are certain web browsers that you can use that try to limit the amount of data that’s shared about you. There’s privacy settings you can turn on on your phone and other things. But if you think about all the sources, I think realistically it’s going to be hard for most of us to sort of turn off that tap.

And so I think for those of us who are concerned about these issues, it’s probably more supporting the regulations of how these data are going to be used and supporting both regulations about what data people can track about us and use for commercial purposes and others. But also potentially thinking in the insurance space about what the regulations are that create incentives for firms to want to use these data.

So what we can think about is the main problem here, is we’re worried that insurers or health providers might have an incentive to avoid the sickest people. That depends in part how strong that incentive is for them. Depends in part on the sort of incentives they have. And how their payments are affected by who is in that risk pool. And whether they can avoid those people.

So in certain markets where you have to sell insurance to everyone, making a prediction that I’m less healthy than you isn’t that valuable to the insurer and they’re a little bit less likely to do it. In other markets where we give insurers a lot of discretion to charge people different prices or to decide they’re not willing to offer insurance to other people. In those markets they have a lot of incentive to try to figure out who the healthy person is and who the less healthy person is.

Marisa Wojcik:

Looking at Europe though, we can see that it is possible to put these regulations in place. And they are enforceable?

Justin Sydnor:

Yup, so I think that’s clearly true. That there’s a lot that can be done. You know with any regulation there’s going to be there’s going to be some down sides. The sort of thing you see here is that a lot of the way that I think these data get generated come from things we do online and services we use for free. That we all value a lot. And part of why they’re free is that they sell the data about you. So there will be tradeoffs.

But I think most people would prefer to sort of face that trade off head on and recognize that if I’m only getting this thing for free because you’re selling data about me that I didn’t realize to people like health insurers for uses that are a little bit dubious at times, I think many people would be okay giving up that free service. But time will tell on that.

Marisa Wojcik:

So some choices may need to be made in the future. All right, thank you so much for joining us and telling us more about this.

Justin Sydnor:

Thank you very much.

Marisa Wojcik:

If you liked this interview, please share it and we do include the ProPublica article link in the description so be sure to check that out and read more about it. For more from Here & Now and Wisconsin Public Television, visit wpt.org and thanks for joining us on Noon Wednesday.