The Case of the Vanished p-Value

Just as I was wondering how people manage to turn “nothing” into “WOW!”, there came a knock. Three times. Hesitant. Like a reviewer who knows they’re about to make someone cry. A messenger brought me an envelope. On it, only this: “Primary endpoint missed. Still a triumph of the clinical trial. Please examine the study manuscript.” I smiled. “Ah,” I murmured. “A classic. The culprit’s name is: spin.”

Crime Scene: The Abstract: Science’s Pedestrian Zone

I opened the attached manuscript. Even the title sounded like someone had thrown confetti into the statistics machine. And sure enough, the abstract read like this:

• “clinically relevant”
  
• “promising”
  
• “could be a game changer” (I visibly flinched)
  
• “favorable safety profile”
  
• and my personal favorite: “trend towards improvement”
  

I turned the page. And there it was. The sentence that gives everything away like cigar ash on freshly mopped parquet: The primary endpoint was not met.
I set the paper down as if it had just lied to my face. “So,” I said to the room, “we have a conviction… and yet the lawyer delivers a victory speech. That isn’t just creative writing. That’s an offense.”

The Suspects: Three Elegant Con Artists

I know this gang. Infamous. Always wearing the same perfumes: hope, urgency, and “please cite us.”
Suspect No. 1: Sir Secondary Endpoint: He stands there, perfectly groomed, and says: “Sure, the main prize was empty… but look! I’ve got three consolation prizes, all statistically significant!”
Suspect No. 2: Lady Subgroup: She whispers: “But in left-handed women with green socks born in the first half of the year, it worked brilliantly.” And everyone nods, because nobody wants to admit how bizarre that just sounded.
Suspect No. 3: Captain Post-hoc: He only shows up after the party and goes: “I had another idea…” And suddenly, one analysis becomes a full-blown improv theatre show.

The Motive: Your Brain Loves Stories

Now it gets neuro-detectively delicious because the problem isn’t only the paper. The problem is your head. And mine, too.
Our brains have some charming quirks:

• They love coherence: a good story feels true.
  
• They love reward: “maybe it works after all!” Dopamine applauds.
  
• They hate disappointment: “it didn’t work” sounds like rain on vacation.
  

And that’s exactly where spin steps in: it takes a sober result and throws a warm blanket over it. Not crude. Not obvious. Elegant with so many little tricks that even smart people suddenly say: “Well then! That sounds so much better!” I call it: cognitive hypnosis in paper format.

The Chase: I Enter the Subgroup Salon

I went deeper into the manuscript where the lights are dim: the discussion. There I found a paragraph that read like a salesman in a suit that costs way too much: “Although the primary endpoint was not met, the consistent signals across multiple secondary parameters…”
I raised an eyebrow. “Consistent?” I asked softly. “Or merely… randomly pretty when arranged just right?”
Out came my notebook, my most reliable tool: the prefrontal flashlight. With it, I shine light where the story likes to cast shadows:

• Was the primary endpoint pre-specified?
  
• Were lots of endpoints tried until something “worked”?
  
• How big is the effect really and how certain is it?
  
• Are the “positive” findings clinically meaningful, or just statistically decorative?
  

The authors had scattered little bright spots everywhere, like candles in a dark church. But I wasn’t looking for scientific atmosphere. I was looking for evidence. Unshakable facts.
And then I found the trail: one secondary endpoint was “significant,” but so many had been tested that I could practically hear chance giggling. “Aha,” I said. “The culprit wasn’t a lie. The culprit was… selection romance.”

The Big Twist: Spin Doesn’t Just Steal Truth — It Steals Decisions

Here comes the part where even my brother might fall silent (briefly, mind you). Because when spin succeeds, this happens:

• A doctor skims the abstract between appointments.
• A patient hopefully asks about “the new breakthrough.”
  
• A clinic debates off-label use.
  
• And suddenly, medicine moves not because of evidence, but because of storytelling power.

Spin is not a literary problem. Spin is a clinical risk. It’s the glitter on the water that makes you forget how deep it is.

The Resolution: I Interrogate the Primary Endpoint

I went back to the beginning. Always back to the beginning. Because good cases aren’t solved in the discussion. They’re solved with one question: What was promised and was it delivered?
The primary endpoint sat there like an unremarkable witness nobody wants to interview. I sat down across from it, smiled politely, and asked: “Did you win?”
It shook its head. “And why is everyone cheering?” It shrugged. That’s witnesses for you. They tell the truth, but they don’t decorate it. I stood up. Case closed. “The culprit,” I said, “isn’t necessarily evil. Often it’s just… ambitious. And its accomplice is our brain which prefers to drink hope rather than uncertainty.”

My Personal Spin Detector for You

Next time you see a “nearly brilliant” paper, do this:

1. Ask first about the primary endpoint. Not the prettiest sentence.
   
2. Don’t let secondary endpoints seduce you. They’re clues, not verdicts.
   
3. Distrust subgroups like you’d distrust a count with a monocle. Pretty, rarely reliable.
   
4. Look at effect size and uncertainty. Not only “Is it significant?” but “Is it sensible?”
   
5. If the conclusion is more euphoric than the results, it smells like perfume. And perfume is rarely evidence.
   

Epilogue: A quiet success

Later I sat again in my London room, the candle small, the tea irresponsibly strong.
And I thought about today’s culprit, so quiet that many don’t even recognize it as a culprit.
Spin isn’t a blunt fraud with fake numbers. Spin is subtler: it’s the art of styling a “non-result” so it feels like a “near-success.” You take a primary endpoint that isn’t significant, place it politely on a sofa (with cushions), and instead show the audience the beautifully lit display case of secondary endpoints, subgroups, or “trends.”
Not lied about, but dramaturgically redecorated. Like a magic trick where the coin doesn’t vanish… it just ends up very skillfully in a pocket.
And this isn’t just my London-by-the-fireplace philosophy: I recently read a paper that examined exactly this pattern in MS trials specifically in studies where the primary endpoint wasn’t met. Result: spin appeared in more than half of the publications.
In other words: even if the main key doesn’t fit the lock, sometimes people jiggle the doorknob so charmingly that it sounds like they’re already inside. Why does it work? Because our brains love stories. “Maybe it works after all!” is cinema with popcorn for your gray matter. “It didn’t work” is a textbook chapter without pictures.
I almost blew out the candle, paused  and grinned. “Spin,” I murmured, “is the makeup layered over evidence of non-effect. And my job isn’t to ban it… it’s to spot it in the half-light.”