The GLP-1 boom didn’t just stress pharmacy budgets — it exposed a deeper failure in the machinery of drug coverage. Behind every approval or denial sits a maze of lifetime limits, indication-specific exclusions, carve-in exceptions, network rules, and dynamic lookbacks that often aren’t visible in the PA document the prescriber thinks they’re complying with. Legacy systems can handle complexity, but they were built for linear drug therapy. GLP-1s are anything but linear.

They now span metabolic pathways from diabetes to MASH to obstructive sleep apnea, with manufacturers chasing new greenfields and employers quietly recalibrating benefits behind closed doors. The result isn’t just inconsistency — it’s a collapse of fairness and provenance in favor of prioritizing speed and cost containment.

The Rule System Was Never Built for This

Consider the lifetime limit. It exists in two flavors: financial caps and quantity caps. Either way, it places an arbitrary ceiling on how long a beneficiary can access a drug within a benefit design — sometimes measured in dollars, sometimes in units dispensed. For a medication class designed to manage chronic, lifelong metabolic conditions, this creates an invisible expiration date on care that patients rarely see coming.

Traditional coverage logic assumes a patient starts a drug, stabilizes, and either continues indefinitely or stops. GLP-1s don’t follow that script. They require titration schedules. They span multiple indications with different coverage rules. They interact with comorbidities that shift which pathway a patient falls into. And increasingly, they’re subject to employer-driven benefit redesigns that happen without patient awareness.

The coverage machinery wasn’t designed for drugs that behave like this. It was built for statins and ACE inhibitors — drugs with clear indications, stable dosing, and predictable utilization patterns.

When the Rules Don’t Match the Criteria

Here’s the part that frustrates clinicians and pharmacists alike: what happens at the point of sale isn’t always aligned with the prior authorization criteria documents.

In practice, coverage isn’t “does this drug require PA.” It’s a chain of conditions: who wrote it, where it’s filled, how it’s billed, how recently it was filled, whether the diagnosis matches the benefit’s allowed pathways, and whether the plan considers this a continuation versus a restart. How robust is the logic? Prior authorization is only one gate; adjudication is the maze. Coverage is a conditional graph, not a rule sheet — and most of that graph is invisible at the point of prescribing.

The PA form might list specific clinical requirements. The adjudication engine running the claim has its own logic — network restrictions, dispense limits, lookback periods, retail-before-mail sequencing (or sometimes no retail at all!) — that operates independently.

Here’s what the patient experiences: The clinician submits the PA and gets approval. The patient fills once at retail. The second fill rejects because the employer benefit requires two retail fills before mail. The third attempt rejects because the plan’s refill-too-soon window is tighter for GLP-1s. Nobody is “wrong” — the rule chain is just invisible until the end users (patients) engage with the healthcare endpoints.

These aren’t rare corner cases — they’re common failure modes. And there are too many variations to count, let alone document in a way that prescribers can use.

Submit the wrong code on first review and you’re starting over. That’s not a clinical failure. It’s a systems failure.

Consistency Fails First

When a coverage system comes under stress, something must give. Speed gets prioritized because denials create backlogs. Accuracy gets attempted because errors generate appeals. Legitimacy gets protected because plans need defensible decisions.

Consistency is what collapses quietly.

Carve-ins and carve-outs reshape access in ways beneficiaries can’t see. One employer decides GLP-1s are a cost containment target and adjusts their benefit design. Another keeps coverage, but two quarters later, realizes the population health “benefits” do not outweigh the cost uptick. A third carves the entire class out to a specialty vendor with different network rules. The downstream impact gets assessed via data and reporting by SMEs so that compliant notification can take place — but the patient experience is fragmented, inconsistent, and often inexplicable.

This is closed-door bureaucracy meeting an open-air cultural phenomenon. The media has covered the sensationalized positives and negatives of GLP-1s endlessly. What they haven’t covered is that the lion’s share of employers that pay their own claims have had this drug class in their sniper scope for a long while. The utilization surge and cost burden have made GLP-1 benefits a line item that gets scrutinized in ways most therapeutic classes never face.

Patients don’t see any of this. They see a prescription. They see a denial. They see a number to call. And when they call, they often find that no one can explain the full chain of logic that led to their outcome.

The Choice We’re Making by Default

When legitimate access collapses under its own complexity, shadow markets bloom. The compounded semaglutide phenomenon isn’t mysterious — it’s a classic smuggle scenario, the kind economist Jagdish Bhagwati described decades ago. When the benefit of defecting far exceeds the risk of being caught or penalized, defection becomes rational.

Patients aren’t choosing compounded products because they distrust the pharmaceutical supply chain. They’re choosing them because the legitimate pathway has become so opaque, so inconsistent, and so difficult to navigate that the alternative starts to look reasonable.

This is the real cost of coverage opacity. Not just denied claims or administrative burden, but the slow erosion of trust in a system that can’t explain itself. No one loses faith in drug coverage over a single GLP-1 denial. They lose faith when the logic behind the denial doesn’t exist anywhere they can see.

GLP-1s are the stress test. The question is whether anyone builds the infrastructure to make coverage logic legible — or whether we keep pretending the machinery still works while patients quietly find their own way around it.