The Epic bill most health systems plan for arrives with go-live. Buffalo, N.Y.-based Kaleida Health’s May 30 launch came with $24.5 million in one-time operating expenses through June, tied to equipment, staff and provider training, and a temporary reduction in some outpatient services. New York City-based Memorial Sloan Kettering Cancer Center tied $177 million of its 2025 operating expense growth to Epic go-live support and training.
Those numbers are steep, but they show up in a business case before the project starts. The costs now catching health system technology leaders off guard come later, and they don’t behave like a contract price. They rise with how patients, clinicians and AI models use the system, and in some cases with choices the health system never made.
At Charlotte, N.C.-based Advocate Health, the number of patients who linked their MyChart accounts to ChatGPT rose more than 7,000% from July to August. Each linked account, and each request a patient enters into ChatGPT, generates API calls against Advocate’s Epic EHR. Bobbie Byrne, MD, executive vice president and CIO, told Becker’s in October that Advocate did nothing to enable the connections.
“There is some cost to us based on the volume of patients and the number of requests each patient enters into Chat. There is compute power required for this, so that will be a cost to someone,” Dr. Byrne said. “Right now we can absorb the amounts, but I would be concerned if the growth rate escalates.”
Epic has acknowledged the load. Seth Howard, the company’s executive vice president of research and development, addressed the issue at Epic’s Open@Epic developer conference Oct. 21, pointing to support for FHIR notifications that would let connected apps stop repeatedly polling health systems’ records for new information.
“This is only increasing. Our customers are definitely noticing the extra traffic,” Mr. Howard said.
Under Epic’s model, the infrastructure carrying that traffic belongs to the health system.
“Epic ships code to our customers, but they own the system and they own the data,” Mr. Howard said. “So customers are in control.”
The AI features Epic sells directly carry a similar problem. At Epic’s Users Group Meeting in August, several CIOs described a hybrid pricing model, with some features billed at a flat rate and others by consumption, that makes the suite hard to forecast. Sha Edathumparampil, chief digital and information officer of Coral Gables, Fla.-based Baptist Health South Florida, compared it with competing products.
“The pricing model is super interesting. If you look at some of the competing products, [ambient listening tools are] one license fee,” he said. “But Epic, because of rising token costs and the economics around it, [has] a fixed fee and then a variable … [charge], which depends on which frontier model you use. It’s only increasing — some increase faster than others, so that’s a concern.”
For Amy Trainor, BSN, RN, vice president and CIO of New Orleans-based Ochsner Health, that variability changes the budgeting calendar.
“Planning a budget a year in advance is a little more difficult than it’s been in the past,” she said.
Nor does she expect a cushion if a health system misjudges demand.
“I don’t imagine there’s … a refund policy. Like, ‘Hey, you guys, sorry you missed it.’ … I just don’t envision that,” Ms. Trainor said.
The pricing is also generating costs that never appear on an Epic invoice. Ms. Trainor said Ochsner is adding roles that didn’t exist a year earlier: “more QA, more engineering, financial FinOps for each team that has a consumption piece.”
Aaron Miri, DHA, chief digital and information officer of Jacksonville, Fla.-based Baptist Health, laid out another trade-off: running AI models on the system’s own servers to avoid per-token fees.
“We’ve also got to have some on-premise [capacity] to be able to run some of those models locally, so we do not incur those costs,” Dr. Miri said. “Now there’s GPU costs on the other side, and we’ve got to consider [that]. It’s a calculus … we just haven’t gotten there yet as an industry.”
Some of the new spending isn’t optional. Eric Podradchik, vice president of digital clinical systems at Somerville, Mass.-based Mass General Brigham, noted in August that Epic’s cloud data platform, Cogito Cloud, has effectively become mandatory.
“You’re a customer, and you’re being forced to implement something to maintain your current platform. At the same time, it’s providing the capabilities that you want, so it’ll be interesting to see how they develop it and how that translates to impact from capabilities as well as system sustainability and cost management,” he said.
Epic has said it is working on the visibility gap. The company reported in August it is building individualized cost-tracking and forecasting tools similar to what Microsoft offers its enterprise customers. Ms. Trainor said the current view stops short of that.
“Epic doesn’t have quite [that] all the way down to the user,” she said. “Right now it’s by function … and by time.”
None of that means the investment doesn’t pay off. In October, health system leaders pointed to gains in revenue capture, operational efficiency and reduced downtime as they weighed whether Epic had recouped its cost. Kaleida expects lower go-live costs, a return on the investment and other margin initiatives to deliver a positive operating margin by year’s end.
The newer costs are harder to plan around because the health system doesn’t fully control the meter. Dr. Byrne said she supports patients using ChatGPT, even as the bill lands on Advocate.
“We certainly want patients to feel empowered in their care. If using ChatGPT helps patients further understand their health, I want to support it,” she said. “Like many things in our healthcare system, there is a disconnect between who benefits and who pays.”
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.