Universal Health Services, Inc. ($UHS)

Good morning. My name is Sarah Conrad, and I'm part of the health care services team here at GS. Today, I'm joined by UHS with CFO, Steve Bilton; and Darren Lehrich, who's VP of IR. Thank you guys so much for joining us.

So I guess I want to start off on the volume and demand environment, which I think has been super topical after a peer report last week. So on your recent 1Q 25th earnings call, you reiterated a 2% to 3% 2026 volume growth framework, and then at recent conferences, you've articulated that 1Q volume was roughly in line with the low end of the range after adjusting for Fluent weather. Can you help us frame any emerging industry dynamics in the second quarter that we should continue that we should consider that could influence how acute volumes are trending in the second quarter and into the second half of 2016 relative to the full year guidance.

So there, I think sort of consistent with our policy in the last several quarters, we really haven't been commenting on intra-quarter volumes, and I think we'll continue to sort of stick to that policy.

And just are there any inputs that you can share that you have the highest conviction to drive increased acute volume growth for UHS from here, including like added capacity, market growth, physician alignment, any other factors?

Yes. I mean we did talk in our first quarter call and at conferences subsequent to that. about the fact that we've added a significant amount of acute capacity, I think, 178 beds, in 3 distinct projects or discrete projects in the second quarter, a tower in our Liquid Ranch facility on the West Coast of Florida. -- new floor in our Henderson or facility in Las Vegas and then a replacement facility in versside County, California, that add some incremental beds into replacement facility at aset. So those are projects that because they're at existing hospitals should ramp up more quickly than de novo project. We also talked in our first quarter call about the continued ramp of the Cedar Hill Hospital in Washington, D.C., which we acknowledge, got off to a bit of a slow start, but we think whose both volumes and earnings are sort of more weighted to the second half of the year.

That's super helpful. And then I guess, as we think about the previous acute margin path, you had previously outlined a path back towards 16% to 16.5% acute margins. But we've seen some shifting industry dynamics. Is it reasonable to assume that your thinking may be evolving on the appropriate long-term margin target? And can any of these headwinds they offset by potential AI or productivity improvements?

So that margin commentary was made as the industry and we were emerging from COVID when margins had been diminished fairly significantly. And I think if you look at our 2025 margins on a same-store basis, taking out the impact of the novo facilities, I think we were at 15.8%. So pretty close to that target that we had set. And I think we effectively had recovered from the COVID pressures with moderating labor costs and lower acuity patients, eliminating those cover patients who are a significant profitability drag. Obviously, now as we look forward into the next several years, there are some new headwinds, all of which I think have been discussed at great length. This year, there were the the ending of the exchange subsidies. And next year, we have the Medicaid work requirements and then 2028, we've got the beginning of supplemental payment reductions. -- we certainly have every intention of trying to counter the impact of those. And to your point, I think technology plays a significant part in that. We have talked a lot about our recent technology investments, AI and otherwise in the revenue cycle. And I think we suggested that those improvements broadly added maybe 50 basis points to revenue per unit in 2025 for the acute division. We're undertaking a very similar third-party consulting review of our behavioral revenue cycle this year and into next year, also beginning to implement some clinical applications AI applications that I think should serve as an aid and reducing length of stay and increasing productivity, et cetera. So yes, there's any number of headwinds as well into the next several years, but I think there's also a significant number of opportunities to offset those.

I think at our recent headquarters visit a few weeks ago, we talked a lot about the AI productivity, which was driving revenue per adjusted admission. Can you give us just a little bit of framing there as we should think about the benefit in 2021 and then into 2026.

Yes. So again, I think in our 2025 quarterly releases, we've, I think, estimated that we were enjoying maybe a 50 basis point increase in revenue per adjusted admission as a result of some of the improvements made in the revenue cycle. And these, I think, included more accurate coating, both ER and invasion coating, more effective denials management and denial appeals and that sort of thing. It's a little bit hard to parse it out exactly because it's kind of a fluid environment, and we know that the payers are also increasing their investment in technology and they're being more aggressive. Clearly, they've seen improvements in their medical loss ratios in the last several quarters. So I think we feel at a minimum, we're keeping pace with what's happening on the payer side. but it's sometimes hard to parse out exactly well, what's due to the revenue cycle improvements, what's due to other changes we might be making.

And then I guess 1 more acute follow-up from our visit. You had talked about the growing emphasis on growing your acute adjacent outpatient assets, including ASCs, -- can you frame where you are in the process, right? -- now? And just any color you can give us around this initiative.

Yes. I mean, like everybody else, we certainly have acknowledged that certainly over the last few years and maybe even for longer than that, there's been a continued shift to outpatient. That is certainly a payer preference they continue to view outpatient, whether it's ASCs or freestanding imaging or whatever as lower cost settings of care and are encouraging patients or maybe more than in patients or incenting patients to use more outpatient facilities. I think patients prefer generally to be treated in an outpatient facility if it's clinically appropriate. And we've certainly tried to participate in that. Probably the most success we've had on the acute outpatient side is in freestanding emergency departments. We've got, I think, at last count, 35 freestanding EDs. We'll have another several more by the end of this year. and these serve a great many purposes. They're just another very convenient access point for patients. They've allowed our ED volumes to grow patients like the convenience, they tend to be sort of a more pleasant experience for the less acute less emergent patients. We also have about a dozen ASCs in our various markets. I think we have at least 1 ASC in every 1 of our larger markets. But plan over the next several years to double or even triple the amount of ASCs that we have. And I think in both segments, we view the likelihood that outpatient over the next several years will grow faster than inpatient, and we want to make sure that we participate in that growth, having the appropriate number of facilities and geographically, properly geographically dispersed in [indiscernible].

Okay. I want to pivot over to the behavioral segment. So you recently highlighted that the key drivers to move more deeply into that 2% to 3% growth range. are the increased labor investment, long baring shifts to more aggressive outpatient strategy. As you think about those 2 variables, what do you think is more important of accelerating behavioral health volume growth and what would be the operational milestones that would give you confidence that the BH segment is moving to a sustainably stronger volume growth tempo?

Yes. I mean I'd repeat the comment that I just made about the acute segment, and maybe this is, I think, a slightly newer development on the behavioral side that is the shift to outpatient, which I think has been underway for probably a decade or more on the acute side, I think it's a more recent development on the behavioral side. But clearly, we've seen, I think payers are reporting increased demand for behavioral services on the outpatient side. And in fairness, we have historically been an inpatient-centric company and I think have only participated and enjoyed this growth in outpatient demand to a limited degree. And I think we feel like we've done a great number of things in the last several years to really increase the focus on outpatient growth in behavioral. One is, I think we've reorganized to a significant degree. We've created dedicated outpatient personnel, meaning these are folks who are focused exclusively on growing outpatient, they're incented, they're held accountable exclusively on growing outpatient focusing on those patients who we create, if you will, these are patients who are discharged from our facilities as in patients, but who need further care, either what we would describe as the most acute level of care, which we would describe as partial hospitalization or something a little bit less intense and we call intensive outpatient, somewhere between I would say, high single digits, low double-digit percentage of our patients require that sort of care, but we only capture a small portion of those patients. And so to your point, what the hurdle is there, the obstacles are often geography patients may live far away from the hospital. And while they were willing to make that drive or the trip for -- to be an inpatient, they're not willing to do it every day or 3 days a week to be an outpatient. So geography makes a difference. And sometimes, we just don't have the appropriate number of therapists to do that. And that's why we were so enthusiastic about the tax base acquisition because what the Talk Space acquisition, when it's closed, hopefully, in the third quarter, will allow us to do is to offer much more of a virtual alternative so that if a patient wants to continue in our system, and have that continuity. They now have a virtual option that, quite frankly, they probably didn't have before. We had some limited amount of virtualizes but not many. And then the other piece is talk space had this panel or has this panel of 6,000 therapists. Again, this has been an issue for us. They believe strongly that those 6,000 therapists have a significant amount of incremental capacity can treat more patients. That's very helpful. But they also have this infrastructure to recruit more therapists in a way that probably has economies and effectiveness beyond what we have. So yes, I mean, I think focus on outpatient -- we are building through our 1,000 branches, branded freestanding facilities, more access points for patients that are not necessarily associated with our hospitals, either geographically or even sort of on a branded basis. And then I think the opportunity with talk space to really build this continuum that goes from sort of the lowest acuity virtual offerings to the highest acuity inpatient that we for many years. And then everything in between, which I think is really that in-between space is really perhaps the most significant opportunity.

Yes. And then as we talk a little bit more about the talk space acquisition in the 1,000 branches program, I think 1 of the more interesting points from our headquarters visit was that only a low single-digit percent of discharged patients transition into UHS' own step-down programs today. So what are the biggest barriers to improving conversion? And where do you think we could go?

Yes. So I think this is what I touched on before. I think -- so when a couple of things are important. So when a patient is being discharged, number one, we've really got to be focused on that discharge planning. And I think historically, we may not have been as focused as we should have been. We were so focused on the inpatient part of their stay and other stays. We didn't always necessarily focus on their, I'll call it, after care. Where I think way more focused on it today that's part of it, just the focus part. Secondly, we've got to make sure we have the therapists we continue to invest in hiring, but also, again, one of the, I think, exciting things about the Talkspace acquisition is significant incremental availability of therapists that we'll be able to access once the deal closes. And then just that virtual option that some patients prefer. There are some patients who actually prefer inpatient or face-to-face care. There are some patients who prefer virtual care. Having Talkspace sort of, if you will, under the UHS tent will allow us to really offer patients and offer payers to provide care in the setting that makes the most sense, both clinically and financially.

Okay. So I want to pivot a little bit to the policy side. So this year, we have the expiration of the ACA enhanced subsidies. And you framed this ACA issue as more of like a payer mix collectability issue versus volume. I guess now that we are into June, can you give us any update on the volume progression that you've seen and just as you look toward the markets, can you frame like how important is the absolute like total volume of exchange patients versus that metal tier versus utilization of these patients? And how we just should be thinking about this transition?

Yes. So the reason we made the assumption that the lapse of the exchange subsidies was likely to be -- the impact was likely to be felt on payer mix, and I'll call it bad debt or uncompensated care rather than on volume is when we looked at that exchange population, they tended to behave in a way that we would say was very similar to our Medicaid population that is they were very ER centric, meaning most of the care that they sought was through the ER. That's sort of where they began their treatment process. They didn't necessarily have their own private physicians, et cetera. And so our notion or our assumption was that even if they lost their coverage, they would continue to come to the ER in the same patterns in the same utilization as they had before. The issue was now they would be coming without coverage, and that would create an uncompensated or bad debt burden for us. And what we said in our assumptions was we thought that 25% to 30% of our exchange population would lose their subsidies, will lose their coverage as a result of the subsidies lapsing and probably 80% to 90% of them would not be able to get their coverage. Like other providers, I think we commented in Q1 that we weren't seeing the loss of exchange patients at that rate, at that full rate that we expected in -- for the full year. And I think that was always our expectation in part because we have this view that we're going to continue to learn more holistically, I think, how many patients have really lost their coverage because we still have patients who come to the hospital with an exchange coverage card or whatever. But we'll find out later when we go to bill and we go to verify whatever that the patients have not paid their premiums -- and as a consequence, they really don't have coverage. And so I think we -- and I think our other peer providers said in Q1 was we still needed another quarter or 2 to really gauge whether we -- whether our assumptions were correct or not. I think at this point, we continue to believe our assumptions seem reasonable and that impact will sort of grow as the year goes on. But I think we continue to feel like we've estimated it reasonably accurately and shouldn't be materially short in any event.

Okay. That's super helpful framing. I think the other thing we've been very focused on is regarding the Medicaid state directed payment reforms. There was a recently announced proposed rule where CMS expanded SDP reformed to cover nearly all Medicaid services, including BH and they sized total savings north of $700 billion, what's your initial interpretation of the proposal and how it impacts your business and the industry overall?

Yes. So we didn't really -- as we read the proposal view it as containing anything terribly new or significant. When we estimated and we've been very transparent both in estimating what our benefit is from the DBP payments and how our benefit is likely to be reduced based on the provisions of OB3 beginning in 2028. And those assumptions have always included our behavioral hospitals and the DBPs that our behavioral hospitals are receiving. So yes, I think in the context of how does this make us think about our overall DBP payments and the estimates that we've made about the reductions that begin in 2020, not significant. I think one of the more significant, what I'll describe as administrative or mechanical parts of the proposal are CMS is suggesting that the states and the payers really have to adjudicate these DPP claims on an individual claim basis, which they don't really do now. that's, I think, an administrative burden that I think we'll be commented on, I'm sure, by payers and providers and other in the states pretty extensively in the in the comment process. But as far as sort of the ultimate impact, I don't think we felt like the rule at a different impact than what we had already been estimating.

And then, I guess, just staying on that SDP, you recently got visibility into the 2025 retroactive Florida SVP approval. You disclosed an expected approximately $100 million retroactive benefit in the second quarter. We're still waiting for that 2026 program to be approved. How should we balance thinking about this and the benefit to 2026?

Yes. So I think one of the challenges we've always had is predicting with any sort of accuracy or specificity when these programs will be approved CMS kind of moves at their own pace, and it's difficult to predict. I think it's worth noting. And I think generally, whenever we express a point of view about a program and it's likely approval, we're really just echoing what that specific state has told its hospitals. So in this case, Florida says, as we all know that the 2025 program has been approved, we believe the 2026 or they believe the 2026 program is -- will also be approved. We don't have enough knowledge of the specific elements of the plan to know exactly what that impact would be on us or we certainly don't know when the timing would be but have an expectation that at some point, we'll also be recognizing the benefit from a 2026 Florida program.

And then I want to talk a little bit about your AI initiatives. You've provided a lot of visibility into some of your AI investments and key initiatives. You cited roughly 2 dozen active projects and a further pipeline under evaluation. Can you talk a little bit about what you're most excited about in development or being deployed today? And then also relative to how the current market is thinking about your acute and BH businesses, where do you see the most upside opportunity from AI?

Yes. So I think we've talked about as you said, a couple of dozen projects where I think we've been specific about several. We've used an AI application from a vendor for coding of emergency room patients, as you might imagine, is or more streamlined, less complicated exercise encoding for inpatients. There are hundreds, close to 1,000 different DRGs for inpatients. There's a handful of essentially acuity codes for ER patients. We've found that AI can code more accurately. We've gone through a significant sort of, I'll call it, parallel measurement period where we'll code manually and we'll go to the AI application and we'll compare the 2 and find that AI application is more consistent, more accurate. So that's been a benefit to us. We've talked about doing a number of things elsewhere in the revenue cycle. We know that payers have been using AI to generate denials and denial letters, et cetera, for a number of years. we've started recently to use AI to write those denial appeal letters, and we find that now a nurse or another clinician who used to spend an hour writing an appeal letter now spends 5 minutes reviewing denial appeal AI-generated ladder. So driving efficiency in that regard. We have found on the behavioral side that AI is helping us in the intake process. So a lot of times, when we are referred to patient by an acute care emergency room or a community mental health center, et cetera, they will send us over an extensive medical record sometimes it could be 10, 50, 60 pages long. And historically, the nurse or a psychologist if somebody is having to review that. Well, now we've got an application that is summarizing [indiscernible] surprise that in I'll call it sort of in an executive summary, and now the clinician is reading page or 1.5 page , which is really accelerating the intake process and reducing the amount of time we have to spend on the intake process. So again, just another example, we've been using AI to make follow-up calls when a patient is discharged from an acute hospital, they'll generally receive a post-discharge call 24, 48 hours after their discharge. They're going to be asked how they're doing, whether they've made their follow-up doctors' appointments, whether they've filled their prescriptions, what their pain level is and that has generally been an effective tool in reducing readmissions and increasing patient satisfaction, et cetera. We found that now 40% of those calls are made by AI agents. Patients are notified upfront, but that's an AI agent on the phone. About 40% of them continue the call. And interestingly, those that do, for the most part, tend to be very satisfied. The AI agents have more patients, I think, sometimes than our own clinicians. They're willing to stand the phone longer with patients, the patients like that. And I think again, just another example of how we're driving productivity through just offer one more because you talked about sort of productivity or we talked earlier, productivity improvements. One of the productivity improvements we see on the acute side in the next several years is continued reduction in length of stay. Length of stay in the acute business rose dramatically during the pandemic. It's come down significantly since then. But I think we believe there's still opportunity to improve it even more. And interestingly, historically, we sort of measure our effectiveness on length of stay post discharge, meaning we would look at what a patient's length of stay was compared to what was [indiscernible] sort of it's called the Medicare geometric mean length of stay. But today, we've got an AI application that at the beginning of a patient's stay based on their diagnosis, based on their condition, et cetera, projects what their length of stay should be and we can sort of measure our effective in real time, how we're doing against that, et cetera. So there are those -- now again, all these things are sort of tied to other issues. One of the challenges we've had with length of stay over the last several years is we've got patients who are being -- or need to be discharged to some other setting of care, skilled nursing, nursing home, rehab, et cetera. And often patients are sort of held in the hospital because there's not an available bed in -- or capacity in one of those places. That's not something we can necessarily control. So that's a challenge. But also, we've got in many of our emergency rooms, we've got to back up of patients who are holding waiting for a bed. And so to the degree that we can add capacity where it's appropriate. We talked about some of those examples earlier. We can help that throughput, and I think that helps length of stay as well because patients who are being held in the emergency are probably not getting the same level of diagnostics, et cetera, that they'll be getting once they get to the floor.

And then you've been very careful that frame AI is a multiyear margin opportunity. You've noted a couple of areas we should be watching, including length of stay, and I assume revenue per adjusted admission as well. Are there any other important milestones that we should be watching to see these AI investments like trickle through the P&L?

Yes. I mean I think the other sort of obvious one we touched on is labor productivity. So in a couple of the examples that I gave, the denials management, the preparation of denials appeals in the post-discharge calls, we're eliminating or at least reducing dramatically the amount of human time that has been devoted to that. So I think over the next several years, as we look to drive productivity, to drive efficiency, a lot of that will be AI or technology generator technology initiated.

Then I want to talk a little bit about capital deployment. So the tax space acquisition is expected to close in the third quarter of 2026. You've talked a little bit in this panel about continued investments in de novo and outpatient assets. Just as we think about capital deployment, potential ongoing buybacks, how are you thinking about the relative returns from deploying capital into organic acute assets versus behavioral outpatient expansion repurchases at today's valuation. Just any framing would be super helpful.

Sure. So once the Talkspace deal is completed, again, we hope in early Q3, our leverage level should be in the low 2s. It's still at the low end of a range that we've talked about for a long time in that 2 to 3 range. So we certainly have the ability continue to deploy capital in all the ways that you suggested, including continued CapEx investment, which I think will be skewed towards outpatient in both segments. Selective M&A where it makes sense and continued share repurchase, probably goes without thing, but we view our current share price is fairly compelling at this point. We've been a very active acquirer of our own shares for many years now. I think over the last 10 to 12 years, we've repurchased about 40% of the company shares. I think we would anticipate continuing to be an active repurchaser. And I don't think our leverage levels while I would say I don't think we will prevent us from doing that, but I think honestly, they'll provide us an incentive to continue to be active.

And then I want to talk a little bit about the capacity ramp with your new Florida hospital that's now opening, additional beds are coming online. Can you talk about how we should think about the ramp of these assets toward your total company margins? What are some of the key variables that we should be considering?

Yes. Most new hospitals take somewhere between 12 and 18 months to ramp up to kind of divisional averages of occupancy and margins, et cetera. And I would think our new Florida hospital would be the same. I think the only time we've seen hospitals ramp up faster than that tends to be in the Las Vegas market. we found West Henderson was literally profitable in its first quarter of operation, which was really extraordinary. But yes, I mean, we're very excited about the new hospital, which is about 2 hours north of here in North Palm Beach County in very desirable demographic area, growing area, very well situated. Darren and I both were at the opening of the hospital just a little over a month ago. So we're very enthusiastic about that. But we certainly know. And I think embedded in our guidance for the year is the notion that this year, it will be a bit of a drag. But by -- we would think by the end of next year should be sort of closing in on more like divisional performance.

Okay. And we've got just about 1 minute left. So I just want to ask on 2Q volume modeling. In the first quarter, we had a few one-timers. We have flu. We had some seasonal dynamics, spring break timing. Are there any lapping dynamics or onetime items that we should be considering?

Yes. I mean so we aren't providing an intra-quarter commentary. As Steve mentioned, I think from a seasonality perspective, The only thing that I think we'd want to just call out is what Steve was just talking about as it relates to the opening of our hospital in Palm Beach Gardens, Florida. So that did open in May as you would expect a new hospital de novo to have some drag on overall profitability as it ramps. So that will be in the Q2. But as it relates to your volume question, not commenting.

Yes. Had to try. And with that, we're out of time. Steve, Darren, thank you so much for joining us today.

Thank you.