The price of poker

Sentara Health and Anthem Blue Cross Blue Shield of Virginia are duking it out in public. One side usually has the weaker hand, the structure says which, and the settlement lands in your claims either way.

Contract fights between a health system and a carrier have been argued in public for years, so that part is not a call anybody needed me to make. What I have said for the past two years is that we would see more of them, and with forty-two of them checked further down, and the published count of these disputes now running about three times what it was in 2022, that is the part that is arriving. This piece lays out the anatomy of what most of them look like, then works through one in our own back yard that opened at the end of July.

Disputes tug heartstrings. The resolution tugs purse strings. Whichever side you land on, the useful posture is neither party’s. And none of this is a gotcha. It is a deep dive on the public record, which is the one thing these disputes rarely get, before anybody blinks.

When a health system and a carrier take a contract dispute public, the market is asked to decide who is right. The useful question is which of the two can afford to walk.

Here is the basic anatomy:

• A hospital that leaves a network loses a share of one payer’s commercial volume for a bounded period. It keeps most of its emergency volume, and it gets paid more for what it keeps.
• A network that loses a hospital loses an important part of what it is selling: in-network coverage.

At face value, this is an asymmetric argument.

What happens when a hospital goes out of network?

Matthew Fiedler and Bich Ly identified actual network-status changes in commercial claims data and measured what happens next: the hospital retains roughly 12 percent of its non-emergency in-network volume, at prices more than twice as high, and roughly 84 percent of its emergency volume, because those patients are not choosing [1]. The patients in need of elective care walk. The emergency book stays and reprices, and the value of that option is large enough that bargaining models ignoring it overstate hospital marginal costs by about 20 percent [2].

For the carrier, the exposure is the product. A network is a list of buildings, and what it is worth to an employer depends on which buildings are on it. Lose the one important one your members already use and there is nothing left to explain.

What is happening between Anthem and Sentara in Virginia?

On July 31, 2026, Sentara Health served notice on Anthem Blue Cross Blue Shield of Virginia that it would let certain commercial, Medicare and Medicaid agreements expire. Both parties put up campaign pages, and both quote real numbers that check out. Anthem’s is at anthem.com/update/sentara and Sentara’s at sentara.com/aboutus/news/keep-sentara. Both are captured as of August 22, 2026, with archived copies linked from the sources and method page published with this piece, because pages like these come down when a dispute settles.

Here is a wrinkle the coverage leaves out. Sentara is an insurance company too. Sentara Health Plans is a licensed Virginia carrier and the second largest in the individual market at 17 percent, behind Elevance at 38 [3]. It is one of the five Cardinal Care managed-care organizations, competing directly against Anthem’s Medicaid entity, HealthKeepers, Inc., inside the same state program [4]. And it appears by name as a payer in the price files below, including at hospitals it does not own.

The margin looks terrible and is not what it looks like

The $593 million premium deficiency reserve was booked in 2024 against losses expected in 2025, and about $296 million of it was released back in the first half of 2025 [9]. Strip that and the Medicare Advantage line too and the consolidated loss is roughly two points. Sentara lost $168 million on $645 million of Medicare Advantage premium, and largely exited the line effective January 1, 2026, dropping its HMO, MA-PD and chronic-condition products while keeping its dual-eligible plan [25]. Underneath, the hospital business made money in each of the three most recent filed years [7]: 12.3 percent in 2024, 12.4 in 2023, 9.8 in 2022. The consolidated loss is effectively an insurance loss.

What does Anthem already pay Sentara hospitals?

CMS has required hospitals to publish negotiated prices under 45 CFR 180.50 since 2021, and since January 1, 2026 its schema has carried a median allowed amount actually paid, drawn from remittance data over a 12 to 15 month lookback, with a claim count beside it [24]. A negotiated rate is a term; an allowed amount is a result.

Which network, and why nobody can tell you

Neither party names one. KeyCare, HealthKeepers, Pathway, High Performance Network, BlueCard, PPO, EPO, HMO and POS appear zero times in connection with the termination, across both campaign sites, both FAQs, the July 31 release and every outlet that has covered it. Anthem’s scope stops at “(subject to exclusions)” and never defines them [11]. Sentara’s stops at “certain.”

The numbers each side quotes are no more checkable. Anthem’s page says Sentara is seeking “as much as 30% over the next three years” and cites nothing for it [11]. Sentara’s FAQ names no figure at all, describing “a modest, single-digit overall rate adjustment”; the 6.2 percent widely reported, and the roughly one percent reduction attributed to Anthem’s counter, both came from the same Sentara executive at the same July 31 briefing [23]. Anthem has never confirmed the counter it is said to have made, and its own FAQ concedes the two figures are not measuring the same thing. Anthem in turn says Sentara wants to remove “industry-standard patient protections, like claims reviews” [11], which is a claim about terms in an unsigned contract that no filing anywhere records. Even the dates disagree: Anthem’s page puts the Medicaid expiry at January 27, 2027 and Sentara’s July 31 release at January 28. Both are still posted.

The figures below show each of Anthem’s commercial products separately rather than blended, and you can find yours. The PPO leads because it is the employer book and the larger one: wherever both products clear the CMS reporting threshold it carries between 64 and 83 percent of Anthem’s claims [10]. The same reading runs on any hospital in the country, and the chart, the files and the prompt that rebuild it out of your own facilities are already published.

These are allowed amounts, not billed charges

The obvious objection to a $59,701 sepsis admission is that it looks like a chargemaster number. It is not. Every figure here is the median allowed amount from remittance data, and at every hospital it sits at or just below that hospital’s own filed rate: $59,701 against a filed $63,446 at Norfolk General, $37,829 against $39,846 at Inova Fairfax. These are amounts paid rather than amounts asked.

The more useful thing that check turns up is that the gap is not a chargemaster artifact. Norfolk General bills Medicare 2.4 times what Martha Jefferson bills for the same DRG, $87,872 against $35,875, and Anthem’s allowed amount is only 1.6 times as much. The chargemaster gap is the wider of the two. What is left sits in the negotiated base rate, which is the thing actually being fought over.

Price tracks leverage, inside one system, under one insurer

For a sepsis admission the comparison is genuinely like for like. All four Sentara hospitals and Inova price MS-DRG 871 the same way, off a negotiated DRG base rate with an outlier provision on top, the contract’s own stop loss rather than an employer’s, and the algorithm field in each file says so [10]. Anthem pays $59,701 at Sentara Norfolk General and $26,577 at Sentara Northern Virginia. Heart failure runs the same order.

That gap is worth setting against a floor. Two Inova hospitals that file under one tax identification number and publish the same rows in the same order land within 1.01 times of each other at the median on what was actually paid, and within 1.02 on all three admissions in Figure 3. Two buildings on effectively one contract come in within two percent. These two, inside one system, are 2.25 times apart.

The buildings are not interchangeable and neither are their markets. Norfolk General sits in Hampton Roads, where a competitor-commissioned study put Sentara at just over 72 percent of inpatient share five years ago [13]. It is the region’s only civilian Level I adult trauma center [14], and Sentara calls it the only Level I adult trauma center and burn trauma unit there [26]. Martha Jefferson is across town from UVA. Sentara Northern Virginia is in Inova’s back yard, and it is the only Sentara hospital that also contracts with CareFirst, because part of Prince William County is CareFirst’s Blues territory rather than Anthem’s [12].

Sentara Virginia Beach General is where the obvious explanation fails. It is in Hampton Roads, it is on the January list, and Anthem pays $38,462 there, in line with Inova and UVA rather than with Norfolk General. The obvious explanation would be an alternative down the road. On the map there is not one. Once every Sentara hospital is out of network, the nearest acute-care hospital that is not Sentara’s is Chesapeake Regional, 20.8 road miles away, and there is nothing closer. Norfolk General, the most expensive building in this set, has three non-Sentara hospitals nearer than Virginia Beach General’s single closest, and the first of them is 4.3 miles away. Whatever Anthem is paying for at Norfolk General, it is not distance.

What is being priced is whether the service has a substitute, not whether the building does. Bon Secours Maryview is four road miles from Norfolk General. It is not a Level I adult trauma center, and Norfolk General is the only civilian one in Hampton Roads [14]. Run the same test the other way and the cheapest hospital Sentara files is the one sitting inside somebody else’s territory: Sentara Northern Virginia has five non-Sentara hospitals inside twenty-one road miles, three of them Inova’s, the nearest 15.5 miles away and Inova’s flagship at twenty.

The same comparison on the HMO book

Run the same comparison on Anthem’s HMO book, which Sentara’s own participation list calls Anthem BCBS HMO - Healthkeepers [12], and the ordering does not move. Norfolk General is still the most expensive hospital in the set, Sentara Northern Virginia still the least, on every admission. One more column falls out of the same files.

At both Inova hospitals Anthem’s PPO and its HealthKeepers HMO carry the identical rate, to the cent, on every admission tested. At the Sentara hospitals the two diverge, and the size of the divergence ranks in exactly the order the leverage does: 1.047 at Sentara Northern Virginia in Inova’s back yard, about 1.15 at Martha Jefferson and Virginia Beach General, and 1.310 at Norfolk General. Where the hospital has a substitute, the two products price closer together. Where it does not, the hospital appears to hold the upper hand, if price alone is taken to be the fact of the matter. Note what that ordering does not follow. Norfolk General has a competitor 4.3 road miles away and the widest spread of the four. Sentara Northern Virginia’s nearest is 15.5 miles away and its spread is the narrowest. The two in between are effectively tied. Proximity is not doing the work. HealthKeepers is the cheaper of the two at every Sentara hospital, by four percent at Sentara Northern Virginia and twenty-four at Norfolk General.

The last column is the one I have not seen run for this dispute. Sentara owns a health plan, and it files its own negotiated rates in these same files, so the file shows what Sentara’s own insurer has contracted to pay at a Sentara hospital for the same admission in the same year. At every hospital where both file, Sentara’s own insurer has agreed to pay more than Anthem: 1.4 percent more at Virginia Beach General, 4.4 at Sentara Northern Virginia, 8.5 at Norfolk General, 2.0 at Inova Fair Oaks and 11.9 percent more at Inova Fairfax, a hospital Sentara does not own. At Martha Jefferson it files no rate at all.

I am not going to tell you what that means. An intercompany rate is a transfer price rather than a market one, and the Inova figures are an arm’s-length purchase by a smaller plan with less to trade. What it is not is an assumption: these are filed dollars, published by the hospitals, nothing suppressed.

The delivery number does not compare, and that is the finding

Vaginal delivery does not compare cleanly, and the reason it does not is the point. Read the contract terms out of the files and Anthem holds a carved-out normal delivery case rate of $9,563 at Martha Jefferson and $9,556 at Sentara Northern Virginia. At Sentara Norfolk General and Sentara Virginia Beach General there is no delivery carve-out at all; a delivery falls to the general DRG base rate [10]. So the $18,428 Anthem paid at Norfolk General against $7,246 at Martha Jefferson is not one price against another. It is a negotiated carve-out against the absence of one, and the absence costs about two and a half times as much.

That is the leverage argument stated properly. Anthem got a delivery case rate in the two markets where Sentara competes and did not get one in the market where it does not. The same pattern shows in the outlier terms: the stop loss triggers at $117,347 of charges at Martha Jefferson and pays 89.2 percent of the excess, against $224,154 and 54.8 percent at Norfolk General [10].

How do prices compare to Medicare?

A multiple of Medicare reads out negotiating position wherever the Medicare payments themselves are close, because Medicare has already adjusted for the things that make hospitals different. The multiple runs from 3.99 at Sentara Norfolk General down to 1.79 at UVA Medical Center, and the top of that range is not the hospital with the fewest neighbors.

Norfolk General and Sentara Northern Virginia are the pair where that reading is cleanest. Medicare’s payments at those two hospitals sit seven percent apart, so the gap between 3.99 and 1.90 lives almost entirely in what Anthem agreed to pay. Where the Medicare payments are far apart the multiple stops being a clean read, and the two hospitals in Charlottesville are the case that shows it.

Do the higher paid hospitals score better on quality?

The instinct when two parties argue about price is to ask which hospital is better. Sort the same seven hospitals three ways and watch what happens.

The orderings do not line up. The most expensive hospital in the set is not the highest rated; it holds four CMS stars and the lowest Leapfrog grade of the seven. The two five-star hospitals sit in the middle of the price list. Whatever Anthem is paying for at Norfolk General, the public quality measures do not describe it, and paying more is not the same as buying the rating.

Widen the frame and the sources stop agreeing with each other as well.

Norfolk General is top-ranked in its metro on the 2026-27 U.S. News list and third in the Commonwealth [20], and carries a Leapfrog C in both cycles shown [18]. Four public systems rate the same buildings and reach different answers.

Not linear is not the same as unrelated, and it is worth saying where the relationship does hold, because it holds in one of these three markets and not in the other two. Start with what the seven have in common. Five of them price within 3.4 percent of each other, from $37,213 to $38,462, across three markets and three operators. Only two sit outside that band, and both are Sentara: against the average of those five, $37,828, Norfolk General is 58 percent above and Sentara Northern Virginia 30 percent below. That is the shape to keep in mind while reading the quality columns.

In northern Virginia, you get what you pay for. The two Inova hospitals hold five CMS stars and Leapfrog A grades in both cycles, and they price inside the band with everyone else. Sentara Northern Virginia holds three stars, a Leapfrog grade that fell from A to B, and the highest heart-failure readmission ratio of the seven, and Anthem pays $26,577 there against $37,829 at Inova Fairfax. In that market the cheaper building is also the worse-rated one, and the ordering is exactly what an employer would expect.

In Charlottesville, price and quality both come out level. Sentara Martha Jefferson and UVA Medical Center are 3.9 road miles apart, both hold four CMS stars and a Leapfrog B, and Anthem pays within one percent of the same amount at each: $37,213 and $37,517. Martha Jefferson carries the set’s only hospital-acquired-condition penalty, which is the one thing separating them. Note what the multiple of Medicare does to that pair, because it is the clearest warning in this piece about the measure. As multiples they look nothing alike, 3.42 against 1.79, and the whole of that gap is in the denominator: Medicare pays UVA $21,015 for the admission and Martha Jefferson $10,867, because Medicare adds teaching and disproportionate-share money to one of these buildings and not the other. On the number an employer actually pays, the two hospitals cost the same.

In Hampton Roads it inverts. Norfolk General and Virginia Beach General hold the same four CMS stars and the same three-star patient experience, and neither is penalized on hospital-acquired conditions. Norfolk General is better on heart-failure readmissions. It also carries a Leapfrog C against Virginia Beach General’s A, and Anthem pays $21,239 more for the same admission there, a 55 percent premium. Four public rating systems look at those two buildings and none of them finds the difference the price is charging for. The thing that separates them is not on any scorecard: one of them is the region’s Level I adult trauma center and the other is not.

How many members does the Anthem and Sentara dispute affect?

None of this is a reason to shrug at the outcome. Roughly 380,000 Anthem members in Virginia sit somewhere in this dispute [11], about 215,000 of them in Hampton Roads [22], and the January dates are narrower still: about 43,000 commercial and Medicare members on January 1, then about 62,000 Medicaid members at the end of the month [23]. Those two do not add to 380,000, and the difference is the reason the scope question above matters: Sentara’s release counts everyone whose access lapses at some point [11], while the product-level split counts only the January cohort. The rest fall in the December 2027 tier, or nowhere the published record locates them.

The Medicaid cohort is the one with no practical alternative: a commercially insured family in Virginia Beach drives the 20.8 miles to Chesapeake Regional, while a Medicaid enrollee changing managed-care organization mid-year to follow a hospital is a different exercise. Sentara’s release names seniors, people with disabilities and lower-income families as the ones hurt most, and the order in which the agreements expire is consistent with that.

Emergency and scheduled care do not sit in the same place under the federal surprise-billing rules, and which of the two your exposed admissions fall into is the first thing I would put to your plan’s counsel. What governs is the language in your own plan document.

And hold on to who goes first. The government programs are all in the opening wave: every Medicare Advantage agreement in the notice expires with 2026, and the Medicaid line follows before January is out, while the commercial book empties in stages that run to the end of 2027. The people at the front of that line are the ones with the least room to move. Medicare Advantage and Medicaid members are, as groups, older, lower-income, or managing a disability, and a longer drive to a substitute facility is a different proposition there than it is for a commuter with a PPO card. That should not get lost in the product-line arithmetic.

The other side of that ledger deserves the same honesty. The shorthand is that Medicaid underpays and commercial makes up the difference. Sentara Hospitals’ own 2024 Schedule H does not say that: Medicaid revenue of $970.2 million against $788.8 million of cost, and Medicare revenue of $669.3 million against $758.3 million [7]. On the system’s own filing Medicaid revenue exceeded Medicaid cost by about $181 million, while Medicare fell $89 million short. That Medicaid line includes supplemental and directed payments, and Virginia funds part of its program through assessments levied on the hospitals themselves, so the true net is narrower than the gap. It is still not the direction the shorthand predicts.

Both parties are describing real pressures accurately. Neither one describes the pressures on the other side. The employer between them pays for the resolution and never gets to read it.

How do hospital and insurer contract standoffs usually end?

FTI Consulting has been counting these since 2022, searching news databases each quarter for publicly reported contract negotiations. Its annual totals run 51 disputes in 2022, 86 in 2023 and 133 in 2024, and the four quarters of 2025 as it published them add to about 170, which is my arithmetic on its figures rather than a total it states. That is somebody else’s count, which is most of why it is worth having. Two caveats belong with it. FTI revised its fourth-quarter 2025 figure from 76 to 83 between two of its own posts, so this is a tally of media coverage rather than a registry. And its first quarter of 2026 came in at 22, which FTI reads as the seasonal lull at the start of a plan year [21].

Underneath that count I checked forty-two disputes myself, on a rule anybody can re-run: a multi-hospital system or a flagship academic medical center, a contract covering employer-sponsored coverage, live at some point between September 2025 and August 2026, and a page about it on at least one of the two parties’ own websites. All forty-two are listed on the sources and method page with what happened to each.

Three things fall out of them. Fifteen never went out of network at all, settling at or before the deadline. Eleven went out and came back, most inside a few weeks and the fastest in ten days, and several came back retroactive to the day they left, so members ended up with no gap at all. And Medicare Advantage is the line that does not come back. Commercial returns, and that book gets carved out or abandoned, which is what happened at Memorial Hermann and at Health First.

The long end is real, though. Ten are still out. The most notable is Johns Hopkins Medicine and UnitedHealthcare, out since August 25, 2025, which is now more than twelve months, with both sides saying they concluded negotiations without reaching an agreement [21]. Broward Health has been out of Florida Blue since July of last year. The remaining six have their deadline ahead of them rather than behind, and Anthem and Sentara is one of the six.

Now put the anatomy from the top of this piece beside the prices. A hospital that goes out of network loses a share of one payer’s commercial volume for a bounded period, keeps its emergency volume, and gets paid more for what it keeps. A network that loses a hospital loses an important part of what it sells. The hospital here is the only Level I adult trauma center in its region [14], inside a system that held just over 72 percent of Hampton Roads inpatient share five years ago [13]. And the price already on file says which way that asymmetry has been running: 3.99 times Medicare at Norfolk General against 1.90 at the Sentara hospital sitting inside another system’s territory.

What the price files do not show is staying power. How long either side can hold turns on days cash and on how much of Sentara’s commercial book Anthem represents, and this piece does not attempt to read either. So none of this is a prediction. It is the arithmetic of the position each side is bargaining from, and most of it was settled long before either of them built a website.

I could not find disclosed terms in any of the forty-two. What does get published is each side’s opening position, usually with a number attached: more than 20 percent in one year, a 30 percent cut, 60 percent over three years [21]. Those numbers never appear again once a deal is signed. There is no published study measuring the rate a hospital wins in the contract signed after a public standoff, and there cannot be one, because the parties announce a deal and publish nothing about it. Bad news travels fast. Good news travels without the facts.

What should a plan sponsor do during a network dispute?

This is the hardest section to write, because the honest version is short. The opening weeks of a public standoff are each party’s bid for your members’ attention. The pages built for it are campaign material, aimed at your members and written to move them. Both sides are prioritizing their own interests, which is exactly what you should expect, because this is the business of healthcare. Three moves cover what a sponsor can actually do.

Consider the facts. The record behind the campaign pages is public, and most of it was used above. Establish which of your facilities are named, which product lines the notice touches, and which dates apply to your population, and put those dates on a calendar. Where the notice does not say, ask your carrier in writing and keep the answer.

Run the gut check. Take the substitution question from the top of this piece and answer it for your own census. Figure 4 is that answer for this dispute, and the same map can be drawn for any market from the same public files. If nothing your members rely on lacks an alternative, the standoff is noise. If one building does, that building is what the negotiation is about.

Build the backup plan before you need it. Draft the member communication now, name the alternative facilities while nobody is anxious, and put the continuity-of-care question to the carrier before the deadline rather than after. Most of these settle. The plan you never send costs a morning; the one you improvise in January costs more.

And hold all of it loosely. As this piece keeps finding, there is always more to the story.

Fringe Theory is independent and unaffiliated. Views expressed are my own and do not represent those of my employer. Nothing here is legal, tax, medical, or investment advice. Figures described as modeled or derived are my own arithmetic from the public sources listed, not disclosed data.