Does An Insurance Company Really Want To Develop New Therapies?

by: Derek Lowe

Via David Shaywitz on Twitter, I’ve been seeing comments from the insurance company startup Clover Health that raised my eyebrows a bit.

The misalignment between existing health insurers and their customers is that insurers view customers in terms of annual income instead of customer lifetime value. This is where (Vivek) Garipalli believes Clover’s edge will be — capturing the data around customer lifetime value

“At Clover, we’ve built everything around this centralized view of our customer and try to capture as much data as we can about each customer … The traditional insurer is not set up to drive workflow management with their customers.”

OK, I have no objections to that view, except the usually slightly creeped-out ones, common to this era, about how so many companies are devoting so much time, effort, and money to capturing every single possible bit of data about me that they can. I mean, I use Google, and I use Amazon, and I leave as much of digital footprint as anyone else, so I know that a lot of people know a lot more about me that I might think. But still. Anyway, it’s not that part that I’m wondering about, it’s this one:

Those data sets could eventually help Clover expand beyond insurance into therapy and care itself, Garipalli told Drew Armstrong of Bloomberg in a fireside chat.

“We’re building one of the most longitudinal datasets in healthcare, and that might mean we start building our own therapeutics.”

It’s hard to say what he’s driving at there. It might be the (somewhat annoying phrase alert) “digital therapeutics” market, which to me is Silicon Valley’s way of saying “See, we’re making a a big difference in human health, even though we’re still writing code and making small electronics just like we always have”. I’m open to the idea that some of these things could be valuable, but I’d like to see data. Verily’s “Project Baseline” features such devices, and the first patients have started in their study (to some fanfare), and we’ll see how good a data set they can build.

There’s also the AI/machine learning/deep learning/pick your buzzphrase area, which people very much want to apply to huge health data sets. (Those terms are not necessarily equivalent, I need to add). I see no reason that that sort of thing can’t work, but it’s suffered from a lot of overly raised expectations so far. The opinion that IBM has been “choking on its own hype” about its Watson system’s applications to medicine seems accurate to me. Perhaps this is what Clover has in mind, though?

If, on the other hand, they are thinking about making new medicines based on therapeutic hypotheses that come out of their customer data, that’s an even bigger jump, and it’s one that I would imagine that an insurance company is ill-prepared to make, to put it as gently as I can. I sometimes roll my eyes about (for example) academic research departments trying to discover new drugs and take them all the way to the clinic, but I have a lot more faith in them than I do in an insurance company. Hopping directly into the drug research business from a completely non-physical-sciences industry and background sounds like a recipe to lose all your money, unless you’re just turning yourself into a source of funding and not much else.

But in that way, that’s where I can see the insurance folks really making an impact: in funding clinical trials, particularly comparative ones between existing therapies. People have talked about this for some time, and recent efforts are explicitly trying to get these sorts of things going. In these cases, though, the insurance companies will mostly be providing the money for clinical research organizations to do the work. If Clover or others can come up with worthwhile ideas based on their own data analysis, so much the better. We’ll see eventually what they really have in mind.

Disclosure: None.