The case for having insurance companies pay physicians for prior authorizations

© Nuttapong punna - stock.adobe.com

An automated clearinghouse that standardizes and transfers health care information across the nation could be a way to streamline the prior authorization process that has become a drag on patient care.

But it’s not the only way.

Economist David M. Cutler, PhD, proposed creation of a pricing system for insurers and other payers to reimburse physicians for the work they do on prior authorizations. And while current technology would support a health care automated clearinghouse, there are more opportunities for automation as physicians and payers integrate artificial intelligence (AI) into the computerized exchange of information.

Cutler outlined a plan for a health care automated clearinghouse in “Reducing Administrative Costs in U.S. Health Care,” a 2020 policy proposal that considers methods to lower that spending that accounts for up to one-third of health care expenses in the United States. He is the Otto Eckstein Professor of Applied Economics at Harvard University, and the author of “Your Money Or Your Life: Strong Medicine for America’s Health Care System.”

Cutler spoke to Medical Economics about his concepts and how they could affect the prior authorization process. This transcript has been edited for length and clarity.

Medical Economics: In your paper, you suggested another way to improve the process could be to attach a price to prior authorization. Can you explain how that would work?

David M. Cutler, PhD
_{© Harvard University}

David M. Cutler, PhD: One of the tenets of economics is that if you want someone to do something for you, you typically pay them to do it. So, think about two transactions. One is when the provider wants to get paid for an image and the insurer says fine, I’ll pay you. The second is, the provider says I want to get paid for an image and the insurer says I'll pay you if you meet requirements A, B, and C, and by the way, Xerox those and send them over to me. In the second of those, the payer is asking the provider to do more stuff. In addition to looking at the patient, reviewing the medical records, deciding what's appropriate, they're also saying, and I want you to provide all this information to me. So, just like you don't ask someone to fix your house for free, don't ask someone to give you a ton of information for free. I think that if the payers had to pay the providers more when they required more information, they would then think more carefully about what information they want. Currently, it's easy to impose more requirements – hey look, you want to provide the anti-hypertensive for another month, fine, send over this form. There's no reason to have that form, but there's no reason to get rid of it if you're the insurer because there's no savings from getting rid of it. On your end, it's very minor savings. Having a price there would make the payer think about, gee, do I really want to do prior authorization in this? Is it worth it to me? Or is this just a waste of time and it's not even worth that money? And so that's what I want the price they're primarily for, is to say, I want you to think about whether it's worthwhile for you to gather this information. If it is, that's fine, pay $10, gather the information, or $20 or whatever it is, but if it's not, then just get rid of it.

Medical Economics: Coming up in May, the American Medical Association's CPT Editorial Panel will consider an application that would create new CPT billing codes for physician and staff time spent on prior authorizations. Are you familiar with that proposal?

"Reducing Administrative Costs in U.S. Health Care," by economist David M. Cutler, PhD, proposes the nation create a new automated clearinghouse for health care information. © The Hamilton Project - Brookings

David M. Cutler: I am familiar with the proposal. As we were talking about, I in general think that payers should pay when they require prior authorization or equivalently get a discount when they don't require prior authorization. In general, I'm not a super huge fan of creating like tons of CPT codes, especially at different levels so that if you have the nurse do it instead of the clinical person, coding person, then you get paid more. I wouldn't want to make it too complicated, but I do think there ought to be some price attached to it. And I think that would really cause the payers to think a lot about which one they want to do, which ones they want to do. And conversely, the way that I think about it now is if they're paying on average $50 and half the time they're caught having $10 worth of prior auth and half the time they're not, I would rather than say OK, now I want you to pay $60, I would say I want you to pay $45 if you don't require it and $55 if you do, so the average is the same but at least you're paying more if you require more from them. I don't want to see this as a way to – not that I'm against giving more money to primary care docs, I'm not against it, but just this is not the way to do that. If we do that, we should do that because we want to do that, not because we've snuck in some backdoor way to get them more money.

Medical Economics: How can artificial intelligence (AI) affect the future processes for prior authorization, quality measurement and data sharing in the U.S. health care system?

David M. Cutler: I do think AI is going to have a huge impact. When people hear about AI in health care, the first thing they think about is, OK, so the computer is going to diagnose me. And that's hard, that's very hard, because there are things that a doctor can observe that the computer cannot and information that the doctor takes in that the computer does not. Not that computers are irrelevant, but just that they won’t replace doctors, they'll supplement them. Where I think AI can have the biggest impact is in the administrative costs. Lots of things that we require people to do, we can instead use AI to do. I'll just give you an example, it's not even AI, but just a very simple example. If you go to Walmart and you buy a roll of paper towels, Walmart needs to tell the paper towel supplier to send a new roll of paper towels – one has disappeared, they have to get a new one there and so on and so forth. How many people are involved in that? The correct answer is zero. The computer at Walmart talks to the computer at the paper towel company and so no human being has to be involved. So that's an example where if you just programmed the thing simply you can get rid of all the people. Here you'll probably need AI because it's a more complicated transaction for things like prior authorization. It's not just I want to do an image, fine do an image. It's, I want to do an image, OK, I need to make sure that it conforms with whatever guidelines there are that are appropriate. But for the same thing, we now use people to do that and we can use AI instead. And the AI really, truly can substitute for people in that. So, with prior authorization, with billing, with all of that stuff, we can use AI, not exclusively AI, but all sorts of AI-like things and automation things to remove the people component which is really expensive, takes a lot of time and the mistakes people just do, they make mistakes. We can do a whole lot better with that.