Case Study 14.1 — Twelve Men: Reading Rudman 1990 as a Paper

Why this case

Almost nobody who cites the Rudman study has read it. That is not a criticism of anyone in particular; it is how citation works in every field. A result enters circulation, gets summarized, gets summarized again, and after a few hops the summary is what people are actually arguing about.

Growth hormone is the extreme case. A single six-month study of twelve men, published in 1990, is still — thirty-five years later — the load-bearing citation under a global industry. That makes it the perfect object for the skill this book is really teaching: reading a paper for its scope rather than for its conclusion.

We are going to do that here. Not to debunk the paper — the paper is fine — but to see precisely where the gap opened between what was found and what has been claimed.

What was done

In July 1990, the New England Journal of Medicine published a report by Daniel Rudman and colleagues on the effects of human growth hormone in men over 60 years old. Recombinant growth hormone had been available for only a few years; before that, supply had been limited to what could be extracted from cadaver pituitaries, and there had never been enough of it to ask a question like this one.

The design, in the terms a reviewer would use:

The result, in this book's canonical wording:

Twelve men over 60, six months of growth hormone versus untreated controls: increased lean mass and decreased fat mass. It did not measure strength, function, or long-term safety.

That last sentence is not editorial. It is a description of the methods section.

The figure

🔬 Read the Study

```text FIGURE CS14.1 — "The scope of a paper, and the scope of a claim" [real published trial]

THE STUDY Rudman D, et al. Effects of human growth hormone in men over 60 years old. New England Journal of Medicine, 1990. Twelve men over 60, selected in part for low IGF-1, received growth hormone for six months. A larger comparison group of similar men received no treatment. Body composition was the measured outcome.

THE QUESTION In older men with low IGF-1, does six months of growth hormone administration change body composition?

WHAT IT SHOWS Yes. Lean body mass rose and adipose mass fell in the treated men relative to untreated controls. The direction matches the physiology exactly: growth hormone drives lipolysis and nitrogen retention (§14.2). The finding has been reproduced many times since, in larger and better-controlled trials. Treat it as real.

WHAT IT Whether the men were STRONGER. Whether they could do MORE. Whether DOESN'T they fell less, broke fewer bones, stayed independent longer, felt better on a validated instrument, or lived longer. Whether any of it persisted past six months. Whether the effect generalizes to older men with normal IGF-1, to women, or to anyone under 60. Whether harms occurring in fewer than roughly one participant in twelve exist — a twelve-person study cannot see them. And because controls were untreated rather than given placebo, it cannot cleanly separate the drug from the knowledge of receiving it.

THE VERDICT A competent small study that answered a narrow question about a surrogate endpoint, and answered it correctly. Its scope ends at six months, twelve men, and body composition. Every popular use of it exceeds all three boundaries at once.

THE LESSON Do not ask "is this study good?" Ask "is the claim inside this study's scope?" A study's scope is fixed by its design — who, how long, against what, measuring what — and no amount of subsequent enthusiasm extends it. The Rudman paper has been asked for thirty-five years to answer questions it was never built to address, and it has, understandably, failed to. ```

What the study could support

Be precise about this, because being unfair to the paper is its own kind of error.

It supports a claim about physiology. Administering growth hormone to older men changes body composition in the direction the mechanism predicts. That is a genuine contribution. In 1990, recombinant growth hormone was new, and the question of whether the aging body still responded to it was a real open question with a real answer. Rudman answered it.

It supports the generation of a hypothesis. If body composition responds, it is entirely reasonable to ask next whether function responds. That is exactly what a small study is for: it is a hypothesis-generating instrument, and generating this hypothesis was legitimate.

It supports the design of the next study. Which, to the field's credit, is largely what happened in the research literature — larger, randomized, placebo-controlled trials followed through the 1990s and 2000s, and were eventually pooled in systematic review. The scientific process worked. It was the commercial process that skipped a step.

What the study could not support

Any claim about function. Not measured. Not inferable. A person can gain lean mass and get no stronger; lean body mass as measured by these methods includes water, and fluid retention is a known growth hormone effect (§14.7). That is not a hypothetical objection — it is a specific alternative explanation for part of the observed change that the study could not exclude.

Any claim about aging. "Aging" is not an endpoint. Nothing in the study measured a rate of aging, because no such measurement was available then and its availability is still contested now.

Any claim about safety. Twelve people for six months. Whatever this study shows, it does not show that a thing is safe, and it could not have.

Any claim about people not in it. Older men with low IGF-1. Not older men generally, not women, not the 45-year-old reading a clinic's website.

How twelve men became an industry

The mechanism of transmission is worth tracing, because it is generic and you will see it again.

The finding was real. This matters enormously. Claims built on fabricated or fragile findings tend to collapse under scrutiny. Claims built on a real finding that has been stretched are extremely durable, because every challenge can be answered by pointing back at the real finding — "there's a New England Journal study" — which is true.

The final sentence was dropped. Citing the body composition result without the scope restriction is technically accurate and functionally misleading. This is the single most common form of citation abuse in the whole peptide space.

The mechanism filled the gap. Where the data stopped, the story continued: growth hormone builds tissue, tissue building is what aging loses, therefore. Chapter 5's third frozen rule exists to block exactly this move.

The supply already existed. Recombinant production had removed the constraint that had defined the field. There was, for the first time, more growth hormone available than there were patients with deficiency. A finding that pointed the surplus somewhere arrived at a moment of maximum readiness.

The effect was visible. A user who loses fat and gains lean mass over a few months experiences exactly what the study found — which feels like confirmation of claims the study never tested. This is why the hype is so resistant to correction: the user is not imagining anything.

The journal itself would eventually note publicly that the article was being misrepresented in advertising for anti-aging products. That is an unusual step, and it tells you how far the gap had opened.

What happened next in the literature

The research community did what it should have. Randomized, placebo-controlled trials in healthy older adults accumulated over the following fifteen years and were eventually synthesized. The picture that emerged, most prominently in a mid-2000s systematic review in a major internal medicine journal:

• Body composition changed — Rudman replicated.
• Strength and functional capacity did not convincingly improve — the surrogate moved and the outcome did not follow.
• Adverse effects were substantially more frequent in treated participants: soft tissue edema, arthralgia, carpal tunnel syndrome, gynecomastia in men, impaired glucose metabolism.

Note the shape of that result. It is not "the original finding was wrong." It is "the original finding was right, and the inference drawn from it was not." Those are very different failures, and the second one is much harder to see coming.

Discussion questions

1. The Rudman study used untreated controls rather than placebo controls. For a body composition endpoint measured by instrument, how much does that matter? Construct the strongest argument that it matters very little, then the strongest argument that it matters more than it appears.

2. Lean body mass as measured in this study includes body water, and growth hormone causes fluid retention. Explain why this is not a quibble, and describe what measurement would distinguish the two interpretations.

3. The chapter insists the sentence "It did not measure strength, function, or long-term safety" must accompany every mention of this study. Is that an unreasonable standard to apply to citation generally? Where else would you apply it, and what would happen to scientific writing if it were universal?

4. The finding was real and the inference was wrong. Which is more dangerous for public understanding: a fabricated finding, or a real finding with an unsupported inference attached? Defend your answer with at least one example from outside this chapter.

5. Later randomized trials found the body composition change but not the functional improvement. Explain why this is not the same as proving growth hormone has no functional effect in healthy older adults, and then explain why the ❌ rating is nonetheless correct.

6. Suppose you had been a reviewer for this paper in 1990. Knowing nothing of what would follow, what would you have asked the authors to add, remove, or qualify? Would any realistic revision have prevented what happened?