From snout to tail

Early in my career in employee benefits, we watched specialty drugs arrive all at once. For the sake of this conversation, we will stay with one of them.

Humira caught a lot of attention. We saw it on television, and we saw it in claims. It became a routine sight, which is part of what made it remarkable, because it was a four figure per fill medication showing up on ordinary groups.

This was around the same time a lot of plans were adopting HSA qualified coverage for the first time. In open enrollment meetings we would commonly heed the warning that a single fill of Humira would satisfy the entire HSA minimum deductible, and that got people's attention in a way that no benefit summary ever did. The math was not subtle. In 2013, Humira's list price was $1,153 per syringe, and the minimum deductible for an HSA qualified plan was $1,250 for self only coverage. [1][2] One monthly fill ran roughly $2,306. It cleared the deductible nearly twice over.

Fast forward. AbbVie has raised Humira's list price 27 times since launch. By 2021 it was $2,984 per syringe, roughly $77,600 a year. [3] By 2023 it was $6,922 per monthly carton, close to $90,000 a year. [4] Then the biosimilars arrived, nine of them in 2023, and the lowest priced one came to market at $995 per carton, an 85% discount. [4] Worth sitting with that number for a second, because even at an 85% discount, the cheap option in 2023 cost roughly double what Humira itself cost at launch in 2003. [1]

The biosimilar that actually took the share is worth knowing by name. For a full year after launch, adalimumab biosimilars sat at 2% to 3% of the market despite discounts as deep as 85%, because the three largest PBMs kept the reference product on formulary. [5] That changed in April 2024, when one of them removed Humira from its major national commercial formularies in favor of private label biosimilars, including Hyrimoz, co-labeled through a subsidiary the PBM had launched the year before. New weekly prescriptions for Hyrimoz went from roughly 640 to 8,300 inside of a week. [6] By that August the company reported converting 97% of its commercial Humira volume to a biosimilar. The other two major PBMs have since adopted versions of the same private label approach. [5]

Price did not move that market, nor did employers gather en masse to force the decision. A formulary decision did.

Today Humira sits in a crowded list of biologics and biosimilars. It is less likely to make an underwriter nervous, since the cost is often dwarfed.

Here is the part that matters. In 2013, one patient's annual list price for Humira ran about five times what an employer paid for a full year of single coverage. By 2023 it was closer to eleven times. Across all of that, nothing in a copay column moved because of any of it.

The front of the store

Let's break this down.

Plan sponsors spend a significant share of their time analyzing copays, deductibles, coinsurance, and out of pocket maximums. Consultants build the grids. Committees compare the columns. Benchmarking reports rank them against peers.

All of that is the front of the store. It is what you see when you are window shopping. It is not where the sausage gets made.

What those columns actually decide is how much of the plan's cost the employee finances, and in what form. Call it plan design arbitrage. Moving a deductible from $1,000 to $2,000 does not make a plan cost less. It moves a slice of a fixed liability from one balance sheet to another, and the sponsor books the difference as savings while the exposure sits with the employee, generally unmeasured.

This is not a criticism of anyone doing it. It is a rational response to a budget, and in a lot of years it is the only lever a sponsor can pull inside a renewal cycle. But it is worth being clear about what the lever does. It reallocates. It does not reduce.

Why everyone looks at the window display

I understood this from the other side of the table before I ever worked in benefits.

I was a high school teacher. I was hired mid year to fill a vacancy, and I had moved to a new state I had visited exactly once, for the interview, over 300 miles from the place I had called home for 22 years. None of the onboarding was memorable. What I do remember is the open enrollment meeting the following school year. We sat in an auditorium with a pile of paper in front of us and several plans to choose from. I compared them and made what felt like an easy choice, an HMO with a $0 deductible, $5 copays, that cost me less than $5 per paycheck. Call it the $0/$5/$5 plan.

I was right to pick it, given what I could see. Everything I could see was the front of the store. It was the easiest benefits decision I've ever had to make, after all, I was on a teacher's salary, and it was the most affordable plan.

Employees ask for richer cost shares because those are the numbers printed on the page in front of them, sponsors optimize cost shares because that is what employees ask for, and benchmarking rewards sponsors for landing near their peers on exactly those columns and rows.

Given that, the whole market converges on the window display. Benchmarking tells you whether yours looks like everyone else's. It tells you very little about what is on the loading dock, or better yet, on the butcher's table.

How the sausage is made

When an employer chooses an ecosystem, they are buying into a financing model. Plan design moves what the plan costs an employee, though the expense and the forward risk get financed regardless, by someone, whatever the columns say.

Which makes it worth looking at what is actually going into the sausage, because that is where the seismic shift has happened.

Overall health care spending does not distribute anything close to evenly. Working from the 2023 MEPS data, roughly 14% of the population had no health expenses at all, the lowest spending half of the population accounted for 3% of all spending and averaged $433 apiece, and the highest spending 5% accounted for nearly half of all spending, averaging $72,918. [7] The top 1% averaged $150,467.

Set a cost share column against that. A plan with a $1,000 deductible has no discernible difference from a $0 deductible for a $100,000 claimant. For reference, the 2025 average single deductible was $1,886 among covered workers who have one, and $1,670 at firms with 200 or more employees. [8] Against a six figure claim, that is a rounding error. Your plan, depending on size, likely has several such claimants.

The top of that distribution is also where the growth is. Tokio Marine HCC's 2026 Annual Market Report, drawing on six years of its own claims data, found that stop loss claims exceeding the specific deductible by at least $2 million rose 213% since policy year 2020, and claims exceeding it by at least $500,000 rose 114% over the same period. [9] Cancer accounts for just over 35% of total paid stop loss claims, cardiovascular near 13%.

Infants under one average $1.37 million in severity on claims above $500,000, and the largest single reported claim in 2025 was a perinatal case at $8.93 million. [9] There is a leveraging effect here as well. Once a claim breaches the specific deductible, all of the medical and pharmacy trend above that point runs into the stop loss layer, which is a good part of why renewals in that market are expected to stay tight through at least 2027. [9]

No deductible, copay, or out of pocket maximum touches any of that.

Money in versus money out

Underneath the tail, trend compounds. PwC put group medical cost trend at 8.5% for 2026 when it first published, then restated 2026 up to 9.0% and projected 9.0% again for 2027. [10] Aon landed at 9.5%, over $17,000 per employee. [11] Mercer's 6.5% is measured after employers make plan design changes, with underlying trend near 9%. [12] Pharmacy runs about 2.5 points above medical. [10]

Take a plan routinely spending $5 million a year. At 6% trend it is spending $6.69 million in year five. Add a single point for utilization and it is $7.01 million, which is $322,000 more in that year alone and roughly $890,000 more across the five years. Run it at 9%, which is roughly where published trend actually sits, and year five is $7.69 million, about $2.74 million more than the 6% path across the five years combined.

Nobody signs up for that. It arrives anyway, one renewal at a time, in increments small enough that each individual year looks survivable.

The question is already in court

There is a version of this argument being made in court, and it lands on the same seam.

The Consolidated Appropriations Act, at Section 202, requires brokers and consultants on group health plans to disclose their compensation to the plan sponsor in writing, which puts a document in the hands of the people making these decisions. [13] Separately, a set of class actions filed since 2024 has tested whether the prudence standard that governs 401(k) fee decisions reaches how a health plan buys pharmacy benefits. Lewandowski v. Johnson & Johnson, Navarro v. Wells Fargo, and Stern v. JPMorgan each name the employer and its benefits committee rather than a vendor. [14]

Worth being clear about where those stand, because the commentary has generally run ahead of the docket. None has been decided on the merits. Lewandowski was dismissed a second time in November 2025 for lack of standing and is on appeal to the Third Circuit, and Navarro went the same way.

But look at the open question underneath them. It is whether formulary and PBM selection is a fiduciary function or a plan design function. That is the front of the store versus the back of it, being argued as a matter of law.

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.