Case Study 19.2 — Contamination as the Historical Pattern

The argument of this case study, stated up front so you can disagree with it as you read: the harms that end up mattering most in medicine are usually not the ones anybody was testing for. They are the ones nobody had thought to look for, discovered late, and identified only because somebody had been keeping records for an unrelated reason.

If that is right, it has an uncomfortable corollary for this chapter. A market with no records does not merely fail to catch the harms we know about. It is structurally incapable of discovering the category of harm that history says will matter.

The anchor case is Chapter 3's cadaver growth hormone. Three other episodes are brought in briefly, not to pile on, but because the pattern only becomes visible when you see it repeat.

The anchor: cadaver-derived growth hormone

Before recombinant DNA technology made it possible to manufacture human growth hormone in cell culture, there was exactly one source: human pituitary glands, collected at autopsy. National programs in several countries collected glands, extracted and purified the hormone, and distributed it to children with growth hormone deficiency — children who, without it, would not grow.

This was good medicine, conducted by serious people, for a real indication, with real benefit. Hold onto that. It is not a story about carelessness.

The preparations were tested. They were tested for the things that were known to require testing: identity of the hormone, potency, sterility, the standard microbiological concerns of the era. Those tests were passed.

They were not tested for a transmissible agent that causes a spongiform encephalopathy, because at the time the programs were running, the relevant biology was not understood well enough for anyone to formulate the test. Some batches carried the agent that causes Creutzfeldt-Jakob disease.

The incubation period ran to years and, in many cases, decades. Recipients developed a fatal, untreatable neurodegenerative disease long after their treatment had ended — after they had left pediatric care, after the programs had been superseded by recombinant hormone, after the connection to a childhood course of injections had become the least obvious hypothesis available.

It was identified anyway, and the reason it was identified is the whole point of this case study. The programs were documented. There were records of who had received hormone, from which batch, and when. When cases of an unusual disease began to appear in an unusual population, there was a common thread available to pull, and clinicians pulled it. Even so it took years, in a supervised, institutional, nationally organized program with named physicians and follow-up registries.

Now perform the counterfactual that §19.8 asks for. Suppose the same contamination had been present in material moving through a market with no distribution records, no lot identification, no prescriber of record, and no mechanism for anyone to be contacted afterward. Rare neurological disease, twenty and thirty years later, scattered across geography, presenting to unrelated clinicians, in people who in many cases would not have mentioned an exposure they had stopped thinking about. There would be no cluster, because there would be no denominator, no registry, and no question anyone thought to ask. The deaths would still have occurred. The cause would simply never have been found.

The pattern, seen three more times

Elixir Sulfanilamide, 1937. Sulfanilamide was a genuinely useful antibacterial. To make a liquid formulation, a manufacturer dissolved it in diethylene glycol, which is sweet, dissolved the drug nicely, and is poisonous. The active ingredient was fine. The problem was in the part nobody was required to test, because no law then required toxicity testing of a finished formulation. The resulting deaths produced the 1938 Federal Food, Drug, and Cosmetic Act. The harm was not in the drug; it was in the vehicle.

Thalidomide, late 1950s and early 1960s. A sedative that worked, marketed widely, and produced severe birth defects. Testing at the time did not systematically examine effects on the developing fetus — not because anyone concealed the results, but because the question was not part of the standard battery. The harm was in the endpoint nobody was measuring.

The 2012 fungal meningitis outbreak. A compounding pharmacy produced injections contaminated with fungus. Dozens of people died. Everything about the substance being compounded was fine; the failure was environmental and procedural, in the sterility of the process rather than the identity of the drug. It is the reason the 503B outsourcing facility category exists at all. The harm was in the manufacturing environment, not the molecule.

Four episodes, four different locations for the failure: a transmissible agent nobody could yet test for; a solvent; an unexamined endpoint; a contaminated facility. In none of them was the active ingredient the problem. In all of them the testing that was performed passed.

What this pattern implies for an unregulated market

First, the six questions in §19.3 are a floor, not a ceiling. They enumerate the failures we currently know how to look for. History suggests that the failure which ends up mattering will be a seventh thing, and the honest position is that we cannot name it in advance. That is an argument for systems — documentation, traceability, surveillance — rather than for any particular test, because systems are what let you discover a problem you did not anticipate.

Second, "nothing has gone wrong so far" is the weakest possible reassurance in exactly this domain. The cadaver hormone programs ran for decades before the first case appeared. A market operating for five years with no visible catastrophe has demonstrated approximately nothing about long-latency risk, and the shorter the market's history, the less its silence means.

Third — and this is where the case study cuts against a comfortable reading — none of this is an argument against peptides. The cadaver hormone was natural, human-derived, and bioidentical in the most literal sense the word permits. Being natural did not protect anyone. Sulfanilamide was a good drug in a bad solvent. Thalidomide was effective at what it was sold for. What failed in every case was the checking, not the chemistry — and checking is a property of a system, not of a molecule.

Fourth, detection capability is itself a safety feature. We do not usually think of it that way. We think of safety as a property of the substance. But the difference between the cadaver hormone tragedy and a hypothetical undocumented version of it is not a difference in how many people were harmed. It is a difference in whether anyone ever found out — which determines whether the harm stops.

Discussion questions

1. State this case study's central argument in one sentence, then make the strongest objection to it you can. (A good objection notes that four selected episodes are not a random sample of medical history. Take that objection seriously and say what would count as better evidence for the claim.)

2. The cadaver growth hormone preparations passed the tests that existed. Explain why that fact makes the case more rather than less relevant to a chapter about unregulated products, and what it implies about how much reassurance a full panel of current tests can provide.

3. Work the counterfactual carefully. List, step by step, the specific pieces of information that allowed the connection between the hormone programs and the disease to be made. For each one, state whether an unregulated market would have it.

4. The case study claims that detection capability is a safety feature. Argue for and against that framing. Does it hold for an individual person deciding what to do, or only for a population?

5. All four episodes involve failures outside the active ingredient — a solvent, a facility, an unmeasured endpoint, an unknown agent. Using §19.3, map each episode onto the closest of the six quality questions, and identify which episode maps onto none of them. What does that residue suggest?

6. §19.10 insists that the risk word is unregulated, not peptide, and this case study is built on a substance that was entirely natural. Write a paragraph explaining to someone who believes natural substances are inherently safer why this history is the strongest available counterexample — without implying that synthetic substances are therefore safer either.