Thread by @ProfDFrancis

Summer Sunday quiz
-----------------------

from Sunny Plymouth, Pearl of Europe

#DreadMed #DooMed

Look at this table of baseline characteristics.

Control arm consists of 22+2 = 24 people.
Therapy arm consists of 20+1=21 people.

How many people is that, altogether?

Nothing in life is that easy.

Think again.

NOW, how many people were randomized?

Don't rush to judgement.

Try this.

Third time lucky:

How many people were randomized?

Glad I've made you skeptical.

Now the arithmetic puzzle of the day. No need for 13 times table this time , as I can't bear being humiliated by my own PhD students (have a demotion @mshunshin).

Calculate the mean increase in EF in the active arm.

What is the mean increase in EF in the cell therapy arm? (Ignore the other arm for the purpose of this question)

Suppose 10 staff of ORBITA HQ drink different amounts of orange juice at Rasha's (and PCI's) 40th birthday party.

5 glasses
6.5 glasses
8 glasses

Etc.

Suppose the MEAN amount drunk is 7.0 glasses.

If i listed the individual number of glasses drunk, what would the mean be?

Look at the coloured bars on the right (ignore the black ones on the left).

Imagine they are the number of glasses of orange juice drunk by ORBITA HQ revellers.

The company providing the bar give Rasha a bill saying "the mean number of glasses drunk was 7, so hand over the £"

What will she say?

Well, now let me tell you that those were in fact the individual patient EF changes in the cell therapy arm.

Is there any (any, ANY, A N Y) possibility that their mean is 7.0?

Review of pic without red line:

If the graph is correct, is there any chance that the statement in the abstract, of a 7.0 unit mean increase in EF in the cell arm is true?

But actually it has to be true because this information was used to bring in investors.

The prospectus to the investors says a 25% improvement in ejection fraction.

I guess they are arguing a 7 percentage point increase from about 27 points is about 25%.

Cheeky but defensible.

But not if even the 7 is a lie.

Sergio good question.

It's not the fault of the journal or the peer reviewers.

It's the fault of the authors for submitting incorrect stuff.

And the fault of the journal for not resolving the situation once pointed out.

When the research was redone properly, the treatment was of course found totally ineffective.

The company kept the investors money though, and changed its name.

Source

onlinejacc.org/content/accj/6…