Thank you André ( @AndreZimmerman) for tunnelling under the Lancet Paywall with a copy of ISIS-2.

This allows me to check whether @mshunshin, who frequently waxes lyrical about this paper, and has so far been my only source of knowledge of it, is bullshitting.

Let's see! #FOAMED

Obviously don't download this copy: you should pay big $ to get it from the paywall instead. https://drive.google.com/open?id=1y171uSpfC02uHgXZnFiY3AGcnJ7CZj_i

You know how most papers begin,

"Despite the decades of progress(reference a few random trials), the morbidity and mortality of X disease is still unacceptably high (ref a few random observational studies)..."

"We at the Gilgamesh Foundation are strong believers in co-operation, leveraging the expertise of the Aardvark Institute, the Innovation Hub of the Kardashian Foundation, and the Patient-Centred, Future-Directed, Personalised, Genome-Wide, .... blah blah"

Here is how to do it, the masters of in-your-face, cut-the-crap, get-to-the-point'ism.

Bam!

I am looking at that sentence.

1. Every word is needed.

2. People may not have realised it.

3. And it is funny.

Brilliant!

Even the quotation marks around "only" are needed.

It is a masterpiece of economy.

What about the word "strict"? Is that (strictly!) needed?

Well, all those decades ago, they were *already* thinking of the nincompoops who would sterilise a bag of saline, and then spit in it.

The same people who would randomize a trial and then de-randomize for analysis.

One word.

In one word, they spit upon the idiocy of reshuffling the arms to make a more positive result.

Thoooof.

We are only one sentence into this paper, might I add.

There is an excellently concise and clear intro, with Headings (subsequently banned by the Lancet, sadly) explaining *why* Aspirin, Strepto, and Factorial design.

The methods are clear and amusinginly honest.

Ooops:

Shame there was no such things as online appendices at the time, or I sense they would have shown the poor programmer person wearing a dunce cap and with the eyes inadequately covered up by a too-small black rectangle.

Kaboom!

About 20% reduction in ~1month vascular deaths, with EITHER streptokinase OR aspirin.

And about 40% when you give both. i.e. no interaction, just pure additiveness.

Awesome!

Reviewers: "But what about subgroup analysis? Men? Women? Tall people? Short people? People with a nervous tic or a stammer? Did they respond differently?"

Authors: "Sod off"

Reviewers: "Errrrr... See these 'ere buttons, ACCEPT and REJECT...?"

Authors: "OK, how about this..."

Reviewers: "Thank you for being reasonable, i.e. doing what we tell you to do."

However, having left the reviewers grinning with self-congratulation, the ISIS-2 authors continued as follows.

And included this profound but little recognised truth:

While giving the reviewers an extra stabbing, at the cost of only two words.

And for maximum humour, look how they *began* their subgroup analysis plot.

What is the first categorisation?

And indeed, here it is.

People with Lemini or Gibra, don't go to hospital with your STEMI! Aspirin kills you, slightly.

Instead, find a friend, born at another time of year, and send them in your place.

These were the people, and that was the era that was.

Papers that are enjoyable to read.

Here's hoping it will return!