Case Study 2 — Rationing

What happens when a peptide you cannot stop taking becomes a pricing question

Type: Real, public, ongoing · Tier 1 structural facts, Tier 2 magnitudes · Relevance: §11.9 · Direct precedent for Chapter 12

Background: a drug you cannot pause

Most drugs can be interrupted. Missing a statin for a week has consequences measured in fractions of a percent of lifetime cardiovascular risk. Missing an antihypertensive for a few days is not usually an emergency.

Insulin in type 1 diabetes is not like this. Without it, glucose rises, the body switches to metabolizing fat, ketones accumulate, and diabetic ketoacidosis develops — a medical emergency that can be fatal within days.

So the therapy has a property that almost nothing else in this book shares: it cannot be rationed safely, and the person taking it knows that.

Which makes cost a different kind of problem. For most drugs, an unaffordable price produces non-adherence and worse average outcomes over years. For insulin, it produces a specific, acute, foreseeable failure mode.

What rationing actually looks like

The behavior is documented and it is not a single thing. Reported patterns include:

  • Stretching — using less than the prescribed dose to make a vial last longer
  • Skipping — omitting doses, particularly mealtime doses
  • Delaying refills until money is available
  • Substituting to older, cheaper formulations without clinical supervision
  • Choosing between insulin and other necessities

Surveys of people with diabetes have consistently reported that a meaningful minority have engaged in cost-related underuse of insulin. Reported proportions vary substantially by population, insurance status, and survey method, and this book will not quote a figure — but the finding is consistent enough across studies that its existence is not in question.

And there are documented deaths. Cases of young adults dying of diabetic ketoacidosis after rationing insulin — frequently after aging out of a parent's insurance coverage — have been reported and publicly investigated. These are individually documented, not statistical inferences.

Why this specific drug

Several features combine, and it is worth separating them because Chapter 12 will ask which ones transfer.

WHY INSULIN IS THE WORST CASE

  ① CANNOT BE STOPPED
     Type 1 diabetes without insulin is fatal in days to weeks.
     There is no "wait until next month."

  ② NO THERAPEUTIC ALTERNATIVE
     For type 1, there is no other drug class. This is not a market
     with substitutes.

  ③ DEMAND IS PERFECTLY INELASTIC
     A person who needs insulin will pay whatever they can. Economically,
     this is the definition of a market where price is not disciplined
     by demand.

  ④ THE PATIENT IS THE RESIDUAL PAYER
     List price is not what most insured payers pay. It is very close to
     what an UNINSURED person pays. So the rebate system that lowers net
     prices for insurers RAISES the effective price for the uninsured.

  ⑤ COVERAGE IS DISCONTINUOUS
     Insurance status changes at predictable life transitions — aging off a
     parent's plan, losing a job, changing jobs. The documented deaths
     cluster at exactly these transitions.

  ────────────────────────────────────────────────────────────────────────
  ①–③ are properties of the DISEASE. ④–⑤ are properties of a HEALTH
  SYSTEM. Only the second set is fixable by policy, and only the second
  set varies between countries.

That last line is the analytically important one. Type 1 diabetes is equally unforgiving everywhere. The rationing deaths are not. Which locates the cause.

⚠️ Hype Check — the two bad versions of this story

Version one: "greedy pharmaceutical companies are killing diabetics." Emotionally satisfying and analytically weak. It cannot explain why the same companies' products cost far less in other countries, which is the fact that most needs explaining. Manufacturers set list prices, and the intermediary structure that converts a list price into what a patient pays is not theirs alone. A story that stops at "greed" cannot generate a fix, because it identifies no mechanism.

Version two: "insulin is complex to manufacture and prices reflect innovation." Also weak. Manufacturing a modern analog is genuinely non-trivial and it does not explain a severalfold international price difference for identical products made in the same facilities. And "innovation" is a claim about R&D recovery that would have to be argued rather than asserted.

The version that supports a fix identifies the mechanism: list prices set high because rebates flow to intermediaries; a patient population whose demand cannot fall; a subgroup paying near list price; slow biosimilar competition for regulatory reasons; and coverage discontinuities at predictable life transitions.

Each of those is separately addressable, and several have been addressed — out-of-pocket caps, direct-purchase programs, and biosimilar entry have all reduced costs for many people in recent years. The situation is better than it was and is not resolved. Any specific figure here would be outdated by the time you read it, which is itself the point: this is a policy variable rather than a physical constant.

Why this belongs in a peptide book

Because Chapter 12 is about to ask the same questions about a newer class, and this is the century-long version of the experiment.

The GLP-1 receptor agonists share properties ③ and ④ with insulin: demand that is not price-disciplined, and a list-versus-net structure that concentrates cost on the least insured. They do not share property ① — stopping a GLP-1 agonist is not an emergency; it produces weight regain over months.

That difference matters and it cuts in a direction people find counterintuitive. Because discontinuation is not acutely dangerous, cost-driven discontinuation of a GLP-1 agonist produces no deaths and generates no headlines. It produces a slow, invisible reversal of benefit in a large population — which is harder to see, harder to attribute, and therefore easier to ignore.

Insulin's rationing deaths are terrible and they are legible. Chapter 12's version of this problem may be larger and will be almost entirely illegible.

What this case teaches

A drug's price interacts with the disease's biology. The same price produces different harms for a therapy you cannot stop than for one you can. Analyzing cost without the pharmacology gets this wrong.

Locate the cause by asking what varies. Type 1 diabetes is identical across countries; the deaths are not. That single observation does more analytical work than any argument about motives.

And "chronic therapy" is a cost claim as much as a clinical one. Chapter 8 §8.9 established that GLP-1 drugs for obesity are chronic therapy — benefit persists only while taken. The clinical statement and the economic statement are the same statement, and insulin is the demonstration of where that leads over a century.

Discussion questions

  1. Separate the five features in the diagram into properties of the disease and properties of the health system. Which set explains international variation, and what follows?

  2. Version one and version two of the story are both described as analytically weak. Construct the strongest version of each, then say what a fix-generating account requires that neither provides.

  3. Insulin cannot be stopped; GLP-1 agonists can. Explain why this makes the GLP-1 cost problem potentially larger and definitely less visible.

  4. The documented deaths cluster at insurance transitions — aging off a parent's plan, job loss. What does clustering at predictable transitions suggest about tractability?

  5. This book declines to quote a figure for the proportion of people who ration insulin. Is that appropriate caution or evasion? What would you need to quote one responsibly?

  6. Forward to Chapter 12. Which of insulin's five features do the GLP-1 drugs share, and which do they not? Write the prediction that follows, and check it against Chapter 12 when you get there.