Thread by @ProfDFrancis

Suppose you motor-bike to work every day. @rallamee nags you, "Why don't you wear a crash helmet, it is safer, blah blah blah."

If you wear a crash helmet today (ONLY), how will it improve your survival curve?

(Hazard = death risk on a particular day)

That baffled people, sorry. This isn't supposed to be the hard bit.

Wearing a crash helmet TODAY makes my motorcycle riding safer for me TODAY.

I go back to no-helmet from tomorrow onwards. Does the fact that I had worn a helmet today, make tomorrows ride safer?

Being a non-smoker today helps me not die today.

Does whether I smoke today have any influence on the probability of me surviving through the whole of 8 Dec 2030, GIVEN THAT I survive up to the end of 7 Dec 2030?

i.e. can smoking today cause _future_ death?

OK, we have now established that there are two ways a daily-use protective thing can work.

Way 1. Only reduces your instantaneous risk of dying ON THE DAYS YOU DO IT. e.g. crash helmets, airbags, looking carefully when you cross the road, avoiding injecting random illegal drugs.

Way 2. Only reduces your LONG TERM FUTURE DAILY RISK of dying, by a tiny little bit. No special advantage today, but the advantages over many days builds up.

e.g. Exercising like @LCC_DrMalik rather than slumping in front of computer like @ProfDFrancis

An intervention might benefit by both routes.

25 years ago, we assumed statins worked entirely by Way 2. They slowed down the furring up of the tubes, so they made it take longer to get into a bad state, and end up in the hands of the cardiologists.

More recently, people have argued that they work by Way 1, namely anti-inflammatory effects, which can be very local in time.

This allows, for example, the almost-instant reduction in reinfarction that we see in some statin trials post-MI.

If Way 1 was the only Way, participating in an RCT of statins would lower your risk ONLY during the trial. After the trial, when both arms move to an approximately equal rate of taking statins, within a few days, both arms would move on to have the same daily risk of CV events.

The RCT data show a prolonged lower daily hazard (for years) in the statin arm versus the placebo arm after exiting the trial. This suggests Way 2 is important.

I personally think that Way 2 is what I care about.

I have near zero expectation of dying TODAY from CV dis.

But in the long term, CV dis is the major _preventable_ cause of death for me, since I don't do hangliding, skydiving, smoking, injecting myself with random needles, etc

The reason *I* take a statin is to get the benefit of Way 2.

I don't take aspirin because I think it benefits mainly through Way 1, i.e. stopping a bad plaque from turning into an MI. I am guessing it doesn't stop plaque developing though.

I would rather have the prevention of onset of bad plaque (plus any short lasting anti-inflammatory anti-rupture effect) with a statin,

than no prevention of onset (plus short lasting anti-thrombosis effect, plus tummy ulcer risk etc)

So I don't care about that paper. I actually just take a statin and have no idea where I lie on the risk score.

I don't care. I also don't care what my risk of dying today in a car crash is: I just wear a seatbelt as it is less effort to do it than waste time calculating.

But some people get all frothy about it.

They argue that "Oh only 1 out of N statin treated people actually BENEFIT from a statin, blah blah, all the other people are just being poisoned by the mafia of the Medical Establishment, Big Pharma, Illuminati, Chulthu, etc."

In those days I cared about saving silly people like that. So with my fellows, we wrote this paper to explain it for them.

If you believe that statins slow down atheroma progression, then it is irrelevant to you how many people's MI's were prevented DURING THE DURATION A TRIAL.

Because you are not doing it for benefit during the trial. You are doing it to keep your pipes cleaner for *decades* to come. That is enough reason to do it, just like you try to be active, and eat healthily. You don't turn it into a big song-and-dance and argument.

For the people who don't believe in slowing of atheroma progression (I pity them, but they exist), then the only way that statins can prevent events is through Way 1.

An instantaneous effect during the time you are taking them.

For pure-Way-1 believers, who whinge that "Oh, only 1/X'th of people die of CV disease, and only 1/Y of them have their lives saved, so the NNT is 1/(X*Y)," that paper is there to shut them up.

If there is an average ~1 year life span gain benefit, that is good enough for me.

If they want to argue that "only 1/10 of people actually get benefit", that is fine, but then the benefit for those people is (obviously) an average of 10 years.

We didn't actually need to write the paper to prove it, it's kind of obvious.

Most of the time, we spent arguing over "Which is better, Matlab or R?" and in the end presented the software in both languages.

The strange thing was that people found it non-obvious.

If I give payments to 100 people so that their wallets increase by an average of $1 each, this COULD be by giving $1 to each person.

But if someone comes and whinges that "90% of people didn't get anything", then it AUTOMATICALLY follows that the remaining 10% got $10 each on average.

There is no alternative. No calculation necessary.

So the paper was for the "Only Way 1" believers.

I am a Way 2 believer and I think Stephen Senn is, too.

Obviously both ways contribute, but I think the contribution of Way 2 is so powerful that it doesn't matter whether Way 1 happens or not.