Thread by @ProfDFrancis

Yes that sentence made me laugh.

Of course it can't be "stenosis per se" because there is no single index of "stenosis per se" for a patient.

https://twitter.com/ajaykirtane/status/1336277018175008772

If you had a high resolution map of everything what would you use?

How about tightest percent stenosis anywhere in coro?

If you thought it was great then you are going to rate a person with an isolated 75% stenosis of the distal RCA worse than a person with five 70%ish stenoses of each of the LAD, CX, RCA, Om1 and D1.

Then you would definitely lose when the plaque burden people rated the multivessel guy as higher risk.

So then you reach out to Francis Industries.

"Ah," we say.

"We are so clever. We are gonna make a continuous variable, one for each vessel.

Then we take the average of all 3 vessels.

Global Stenosis Index

Only $9995 per vessel."

NOW who rates as higher stenosis ?

Right but then Kirtanics Inc decides to break the stranglehold of the Francis industries patent.

"Francis are stupid. How can you only count each vessel as just one thing?

Use the Ajaytronic index!

Calculate stenosis in each of 19 Coro segments and total them."

What's the matter with you people?

Which is greater

Average of 70 70 70

Or

Average of 75 0 0

Yes it's bloody obvious.

How come this?

I can't believe I let you guys put metal things in my heart.

So global Stenosis Index is going to be better than "worst vessel"

But Ajaytronic index will be even better because it will correctly rate a guy with 70% lesions in all 19 (or whatever) coronary segments as worse than someone with a single distal 75% lesion in each of the 3 big arteries.

But before Ajay can buy his mansion in Monte Carlo along comes Roxanna

With her Merantic index

"19 segments is too few, of course!
We are Mehrancorp use 256 slice CT, at least until we can get the 8192 slice CT in"

"So we divide the coros into 65536 segments blah blah blah"

And of course she will win.

The only person who can trump this sequence of one upmanship is the person who realises the mathematical consequence of smallening the segments is that they converge to a limit of 0ish.

The numbered distinct segments become a continuum.

And Frank "don't dichotomise anything, ever" Harrell will be that person.

With his method,

Frank's Ultimate Continuous Kinetic Stenosis Analysis of Kernel Envelope

... he will definitely win.

And what would he have done?

Integrated the stenosis over the total coronary tree.

Very clever.

Until he gets sued.

By the people who do the total plaque burden.

"That's our thing!"

THAT is why total burden will keep winning.

That's why whenever you want your own godforsaken index to win, the number 1 rule in your clinical trials is, what?

Because if you DO measure total burden, total burden will win.

So don't.

Just measure your thing.

Most people are so stupid that they won't realise what you are doing.

They will just be amazed,

"oh Darrels Flow Reserve is so amazing, isn't it? So prognostic!"

Go look at all the indices of risk on coronary angiogram.

How many compared the fancy expensive thing with plaque burden measured in any way (even if stupid)?

Looking forward to seeing any counterexamples below!

>