How could regulations slow consolidation of physician practices into hospitals?

© sveta - stock.adobe.com

Consolidation in U.S. health care has gotten no small amount of attention in recent years.

The added analysis has included some warnings about potential bad effects for physicians and patients. Congress and the administrations of President Donald J. Trump and his predecessor, President Joe Biden, all have offered proposals on ways to protect patients by halting the growth of hospitals, health systems and insurers. But regulation has not followed quickly enough for critics who argue health care needs robust competition, not just a few big payers or health systems controlling local or regional markets.

“Are Hospital Acquisitions of Physician Practices Anticompetitive?” is a new study that quantified how hospital systems have been adding physician practices from 2008 to 2016. Although the data may reach back years, the authors found “little reason to believe this trend has or will reverse” now, unless there are changes to American health care policy.

Zack Cooper, PhD
_{© Yale University}

The study is based on sources such as the American Hospital Association or surveys of SK&A, now Iqvia OneKey. The researchers used those sources combined with Medicare Data on Provider Practice and Specialty with taxpayer identification numbers, physician national provider identifier numbers, and claims from insurer UnitedHealth. It was a massive dataset, so researchers trained artificial intelligence to analyze it.

Co-author Zack Cooper, PhD, spoke with Medical Economics about the findings and the research methods they used to assemble a picture of business integration in recent years. He is associate professor of public health (health policy) and economics in Yale University’s Institution for Social and Policy Studies.

Co-authors are Stuart Craig, Aristotle Epanomeritakis, Matthew Grennan, Joseph Martinez, Fiona Scott Morton and Ashley Swanson, coordinated with NBER and the Yale University Tobin Center for Economic Policy. They published an accompanying policy brief, “Hospitals are gobbling up physician practices — and health care prices are rising as a result.”

This transcript has been edited for length and clarity.

Medical Economics: At the state and federal level, what policies would you like to see lawmakers or regulatory boards consider that might slow down consolidation and integration of physicians’ practices into hospital systems?

Zack Cooper, PhD: It's a pretty hard issue to address. The first thing is thinking about what's leading some of these hospitals to buy up practices and physicians to sell. And one of the big drivers of integration is the way Medicare pays physicians, and they'll end up paying more to physicians that are in practices owned by hospitals than they will to practices that are independent. And so that creates a really strong incentive for physicians to sell and hospitals to buy. We got to get rid of that. I know that's something the Trump administration is looking into now, and there have been a bunch of piecemeal bites at that over the last couple years. So that's a real thing that we've got to change, introducing so-called site-neutral billing reform. The second is really thinking about antitrust enforcement in this sector, and that's really tricky, because a lot of these deals are actually pretty small. But I think you can do a couple of things. The first is, we offer a bit of a roadmap for which deal is the most problematic, and the deals that are the most problematic tend to be ones where one of the parties already has a lot of negotiating power, so maybe it’s like a monopoly, a hospital. The other are going to be deals that are going to shift a whole lot of patient referrals. And so it could be a deal where one physician is already sending a lot of our patients to a bigger hospital, she gets bought by a different hospital, and it's going to steer all her patients to that new owner. That's the kind of deal that's going to raise prices pretty substantially and we want to really take a look at that. And then I think the third is because there are a lot of little deals, we might want to offer some barriers that sort of slow down the merger process. It could be things like requiring hospitals to say how these mergers are going to benefit the community, how they're going to impact quality, and have them state affirmatively the good that these transactions are going to do, and why that's going to outweigh the harm that we just showed pretty convincingly occurs when these types of transactions get consummated.

Medical Economics: You and your colleagues used machine learning algorithms to identify those integration events. Were you happy with that process? And do you think that kind of machine learning and AI are going to have the potential to improve accuracy of scientific and economic study in the future?

Zack Cooper, PhD: It's really a tool that makes tasks that would take too much time for a person to do, more accessible. There are lots of physicians in the U.S., on the order of about a million. And what we wanted to know was, how many of those physicians are in a practice owned by a hospital? One of the things that we could do is manually code those one by one, but it's really hard to code a million observations. What we did is got all the data that you need to figure that out, publicly accessible, and data on press reports about which hospitals buying which physicians, tax reporting from the hospitals themselves, data from the Medicare program, basically anything we can get our hands on. And for about 5% of physicians in the U.S., we coded them up manually. We had a team of people working week after week, month after month, year after year, to say, who's owned by who. And then what we did is, we taught a machine learning algorithm how to basically do that for itself. You basically say, look, here's the data, here's what we figured out, and here's how we did it, here's how you, machine, can do it. And then we see how accurate we can get it. And once we have that algorithm built up, we then set the data, set the algorithm loose on another 5% of the data, and then we audited that data ourselves. So we sort of had truth, and we trained the machine on truth, then we set the thing loose, and then we audited a portion of that, and we saw that we actually predict integration really, really accurately. And then we set the thing loose on the other 90% of data. The real advantage there is not that it's doing anything differently than we could do by hand, it's that for us to do it by hand, it would have taken 20 years. And so instead of taking it 20 years, it took a year or two. I think it just sort of opens up the range of discovery that we can push forward. And I think the more that we can be at the frontier, the more interesting topics we can look at, hopefully in the long run, the better the public policy we get from the research we produce.

Medical Economics: Our main audience is primary care physicians. What would you like to say to them, or what would you like them to know?

Zack Cooper, PhD: It's just the key thrust of the paper: There's a big change happening in the physician sector, and the question, obviously, is how it will affect them, day to day, and then how it affects their patients. One way that it can affect patients is by making care more expensive. And for a lot of folks in this country, a 15% increase in the price of a physician visit represents a really, really big barrier to accessing care.