Stick a fork in me, I am done with this week. It was a busy week and then you add in a chief of staff (Eddie, the dog) that would whine every 2 hours at night to go outside to deal with a stomach issue and I’m toast. On one hand, #grateful, but ready for this to pass. Literally.
Having One’s Cake and Hiding It Too. On April 20, STAT’s Bob Herman published a breakdown of the 500+ comment letters filed on the Department of Labor’s (DOL’s) proposed pharmacy benefit manager (PBM) transparency rule. The January proposal would require PBMs to disclose rebates, affiliated group purchasing organization (GPO) fees, and other revenue streams to employers under ERISA, with an audit mechanism built in.
The PBM industry (PCMA, CVS, Cigna, UnitedHealthcare, AHIP, BCBSA) spent half their comment letters arguing the rule is illegal. PCMA’s Tim Dube invoked the 2024 Supreme Court ruling that weakened agency rulemaking authority, claiming DOL can’t use ERISA to regulate “aspects of the pharmaceutical marketplace that exist outside the ERISA plan relationship.” Translation: we’ll take the employer’s money, but you can’t make us tell them how we spend it.
The timing problem is real. Congress passed a PBM reform law in February (signed by Trump) that requires similar disclosures but doesn’t take effect until August 2028. DOL’s rule would go live this July. The industry argument is basically: you’re creating a parallel track before the law you already passed kicks in, so withdraw the rule and wait two years. Which preserves opacity for another 28 months.
PhRMA, AbbVie and AstraZeneca all asked DOL to keep net drug prices confidential and used solely for fiduciary oversight purposes. Bristol Myers Squibb’s went further asking to make it aggregate-level only, not per-drug.
The Purchaser Business Group on Health reached out to 47 insurers and third-party administrators on behalf of five large employers and most said they wouldn’t disclose compensation or conflicts until legally required.
DOL’s final rule is expected this summer. Litigation will follow.
Four Caps and a Deselection. On Tuesday, the National Pharmaceutical Council published a fact sheet analyzing how four state prescription drug affordability boards (PDABs) with authority to set upper payment limits might affect patient access. The boards — Colorado, Maryland, Minnesota, and Washington — can cap what insurers reimburse for selected drugs.
The findings aren’t encouraging. Compared to non-PDAB states, patients in these four jurisdictions often face more restrictive formulary coverage, less favorable cost-sharing tiers, and higher prior authorization (PA) rates for the same medications. Colorado and Maryland members have more favorable cost-sharing for some drugs than those in non-PDAB states, but less generous formulary coverage and more common PA. Minnesota patients face better formulary coverage but less favorable tiering, more PA, and higher cost-sharing. Washington’s story: implementation “may prompt payers to use any of the identified levers” to offset losses, meaning formulary exclusion, adverse tiering, or PA to recoup upper-payment-limit-related revenue hits.
Translation: when states cap reimbursement, payers won’t absorb the difference quietly. They will shift costs to patients or restrict access through utilization management.
License to Refill. In January, Utah launched a 12-month pilot allowing an artificial intelligence (AI) system to autonomously renew certain prescriptions for chronic conditions. After a physician reviews the first 250 renewals, the algorithm operates without human oversight.
The regulatory innovation: Utah’s 2024 AI Policy Act created a “learning laboratory” that temporarily waives state-law requirements for selected participants. The AI handles 192 commonly prescribed drugs — lisinopril, levothyroxine, sertraline, metformin.
But there’s a federal law problem.
The Food, Drug, and Cosmetic Act requires prescription dispensing to occur “upon a written prescription of a practitioner licensed by law to administer such drug.” The Doctronic system reportedly hasn’t been reviewed by the Food and Drug Administration (FDA), raising two questions: Does “practitioner” apply to algorithms? And if Utah “licensed” the AI under state law, does that satisfy federal requirements?
Most legal experts read “practitioner” as meaning a person, not software. Even if pending federal legislation (the Healthy Technology Act of 2025) expands the definition to include AI, Doctronic would still need FDA clearance as a medical device, which apparently hasn’t happened.
The clinical case is equally uncertain. Automated refills can improve adherence, which matters for chronic disease control. But some medications on Utah’s list require careful dose adjustments and monitoring for toxicity or disease progression.
If the FDA doesn’t intervene, the pilot runs through January 2027. Then we’ll know if autonomous prescription renewal worked.
Side Effects May Include Litigation. The FDA asked Congress this month for new authority to crack down on prescription drug ads that lack “fair balance” between benefits and risks. In its fiscal year 2027 budget request, the agency said direct-to-consumer (DTC) pharmaceutical advertising is “frequently misleading and confusing.”
The U.S. and New Zealand are the only two countries that allow DTC drug ads. The industry spent more than $10 billion on them last year.
The FDA is already moving without waiting for Congress. The agency has sent thousands of warnings to manufacturers over misleading ads and closed a loophole allowing certain side effects to be listed on linked websites instead of in the ad itself. Late last month, Sens. Roger Marshall (R-Kan.) and Dick Durbin (D-Ill.) urged the agency to require pre-clearance reviews for certain drug ads before they air.
The crackdown has unusual crossover appeal. Health Secretary Robert F. Kennedy Jr.’s “Make America Healthy Again” base and progressives both believe drugmakers are getting away with deceptive marketing. Bernie Sanders (I-Vt.) has proposed banning drug ads altogether. Sens. Josh Hawley (R-Mo.) and Jeanne Shaheen (D-N.H.) introduced a bill preventing manufacturers from claiming business deductions on DTC advertising.
But there’s a First Amendment problem. Some legal experts say last month’s Supreme Court decision protecting therapist-patient conversations as constitutionally protected speech could be a signal on what the courts might rule. Tighter FDA regulation increases the likelihood of a constitutional challenge.
No One Price Shops a Cure. Last week, the American Journal of Managed Care published a provocative argument: patient cost-sharing for curative cell and gene therapies is policy theater that creates access barriers without serving any legitimate function.
The piece, co-authored by former Humana exec William Shrank and University of Michigan’s Mark Fendrick, makes a straightforward case. These therapies now routinely cost more than $3 million. They’re delivered at a handful of specialized centers to patients meeting strict clinical criteria. There’s no risk of overuse or discretionary consumption. Yet commercially insured and Medicare beneficiaries can still face thousands in out-of-pocket costs through deductibles and coinsurance, often hitting annual maximums while managing chronic illness, disability, and lost income.
LOVED this piece. A few years ago, Express Scripts was quoted in one of the industry journals saying this exact thing … essentially, why bother charging patients for these? The traditional justification for copays collapses here. Cost-sharing is meant to discourage low-value utilization and encourage price sensitivity.
In a category where eligibility is narrow, clinical oversight is intense, and inappropriate use is basically impossible, financial barriers undermine access for the populations who need these therapies most.
The authors propose a better accountability mechanism: require long-term patient participation in outcomes registries as a condition of coverage with zero cost-sharing.
Two important caveats. First, therapies with limited or uncertain effectiveness may warrant different treatment. Coverage with evidence development has precedent when clinical benefit is modest or still evolving. Second, conditions with existing effective chronic therapies (like type 1 diabetes) present a different calculus when curative options arrive at multimillion-dollar price tags.
But for therapies offering durable remission or cure with no alternative, the conclusion is hard to argue with: access should be guided by probability of clinical benefit, not ability to pay. As the pipeline grows, establishing that norm now matters more than the dollar amount patients contribute today.
Resources to Love. Vertical integration chart. Thank you Drug Channels.
Reimbursement Fundamentals – What Metrics Miss
Pharmaceutical policy runs on measurements that look one layer too shallow. We count manufacturers and declare supply chains diversified. We average program results and declare models successful or failed. We track approval rates and declare utilization management working. And then we’re surprised when the system breaks in ways the data said it shouldn’t.
This week’s release of the U.S. Pharmacopeia’s 2025 Vulnerable Medicine List demonstrates the problem in concrete terms. Of the 100 essential drugs identified as structurally at risk, 48 have at least one key starting material (KSM) manufactured in only one country. Many of these drugs weren’t on last year’s list. They weren’t flagged as vulnerable because, at the finished-dosage level, they looked fine: multiple manufacturers, broad geographic distribution, low shortage risk scores. The redundancy was real at the tier we were measuring. It just wasn’t real where it mattered.
A drug might have five API manufacturers across three continents and a dozen finished-dosage facilities worldwide. That looks like resilience. But if every one of those manufacturers sources the same KSM from a single facility in a single country, the perceived redundancy is an illusion. The chokepoint was always there.
This isn’t unique to supply chains. The same structural flaw shows up across pharmaceutical policy when we measure at the wrong level of aggregation.
Take the Oncology Care Model evaluation published in JAMA last month. The headline: net loss to Medicare, $639 million more in program payments than spending reductions over six years. That framing shaped policy. The Enhancing Oncology Model, OCM’s successor, launched in 2023 with slashed payments, mandatory two-sided risk from day one, and narrower eligibility. It has 41 participating practices compared to OCM’s 202.
But the aggregate result masked the underlying story. Eighty practices exited OCM before completion, many leaving precisely when CMS began requiring two-sided risk for practices that hadn’t earned performance-based payments. The reported savings averaged data from practices that collected enhanced payments for years without meaningful transformation alongside 24 practices in two-sided risk that generated substantial, sustained spending reductions. A separate evaluation found most OCM savings came from just those 24 practices, representing 34% of episode volume.
The question the evaluation couldn’t answer: did OCM’s model design fail, or did aggregate results get diluted? More importantly, the data that would have clarified that question arrived in 2024, after EOM had already launched based on incomplete interim results. Policy moved faster than evidence because we measured at the wrong tier and didn’t wait for visibility at the right one.
Prior authorization follows the same pattern. Last week’s JAMA Health Forum study tracked branded prescriptions initially rejected by PA. Fifty-four percent were eventually approved. That sounds like the system working: more approvals than denials. But the approval rate is the wrong measurement.
Dig one layer deeper and the picture changes. Thirty-five percent were processed on the same-day, but 44% were approved after a median six-day delay, and 46% were ultimately denied after the full PA review process. Patients with multiple chronic conditions faced 5% lower approval rates. Medicaid enrollees faced 8% lower approval rates than Medicare or commercial. Prescriptions requiring multiple rounds of PA review had 37% lower probability of same-day processing.
The binary outcome of approved or denied obscures what’s actually happening: a multi-day administrative process that delays treatment for two-thirds of initially rejected prescriptions, eventually denies nearly half, and produces systematically worse outcomes for patients with complex conditions and fewer resources. Measuring approval rates makes PA look functional. Measuring time to treatment and patient-level disparities reveals it’s functioning more as friction than clinical gatekeeping.
The pattern repeats because we default to measuring what’s easiest to see. Those metrics are available, standardized, and comparable across time. But they’re also surface-level.
The policy cost is tangible. Enhancing Oncology Model now struggles with participation one-fifth the size of its predecessor. Supply chain investments get allocated based on finished-dosage vulnerabilities while upstream KSM concentration creates invisible single points of failure. Prior authorization gets defended as clinically appropriate because approval rates look reasonable, while patients with complex conditions wait six days and face higher denial rates.
The fix isn’t more data. The fix is looking at the right tier before making policy decisions. That means practice-level results before designing successor payment models. Upstream KSM sourcing before declaring supply chains resilient. Patient-level outcomes, not just claim-level approval rates, before defending utilization management as working.
And critically, it means waiting for visibility at the right level before locking in policy responses. OCM’s final evaluation showed savings quadrupling by year six, the signal EOM needed but didn’t have when it launched. The vulnerable medicines list now includes KSM data that reveals risks invisible in 2024. The prior authorization study quantifies delays and disparities that binary approval rates conceal.
You can’t build supply chain resilience if you don’t know where the chokepoint is. You can’t design a successor model if you don’t know which practices succeeded. You can’t fix prior authorization if you’re measuring the wrong outcome.
Redundancy is only real if it exists at the tier that matters. The rest is illusion.
