The road to hell is paved with biologic plausibility
I’ve never gone viral or had an even remotely interesting social media experience.
I certainly don’t have a particularly large Twitter presence. I don’t tweet all that much and I don’t have many followers.
But last week, I posted a fairly unremarkable tweet about the stupid hydroxychloroquine “controversy” that was retweeted by a political operative named Rick Wilson (who has a million Twitter followers), and suddenly my phone wouldn’t stop buzzing.
Look at the Tweet along with its numbers of retweets and likes - it’s not even a really big number of impressions, but it was by far the most social media engagement I’ve ever had (and caused me to turn off Twitter notifications).
To see the nature of the responses, comments, and messages I got left me feeling disappointed about the future of humanity.
“Didn’t you look at the “field studies” that showed an impressive benefit?”
“My neighbor, a nurse, said all her nursing Facebook board members are using it and it works”
“I understand that hydroxychloroquine and zinc work early in Covid”
“How do you explain the Henry Ford Hospital system that announced positive results a few weeks ago?”
It’s really sad to be living in a world where interpretation of the data on hydroxychloroquine has turned everyone into Reviewer 2.
The first of my points in the tweet - that the experience of clinicians on the ground with hydroxychloroquine wasn’t impressive - is so much less important than my second point:
If hydroxychloroquine really had a benefit in Covid-19, we would have seen it in the randomized controlled trials.
Why do you keep talking about randomized controlled trials? I’m about to unsubscribe from your newsletter
I’ve written over and over again in this newsletter about the importance of doing randomized clinical trials to determine if medications and treatments work.
I know that I’m a broken record.
But headlines in the popular press - and the doctors who overpromote the findings of research or their own experience - make it *really hard* to follow the news without being misled about what works and what doesn’t.
Think about this through the simple framework of pneumonia.
About 10% of patients with pneumonia who are admitted to the hospital die
About 1% of patients with pneumonia who are treated at home die
Does this mean that being admitted to the hospital causes you to die of pneumonia?
Of course not! Being admitted to the hospital is an indicator of how sick you are. Confounding variables can make our observations untrustworthy, which is where randomization comes into play.
There are a million different factors that will play a role in a patient’s overall outcome and may influence a doctor’s decision to prescribe a specific treatment. You can’t just look at what happened during the course of normal treatment and then do fancy statistics to sort out what caused benefit and what caused harm.
The only way to tell if a treatment actually causes your outcome to be different is to take two groups of similar patients and randomly assign one group to treatment A and one group to treatment B.
Anyone who works in the tech industry is familiar with A/B testing. It’s the same concept.
Treatments need to be tested before they become the standard of care.
Actemra - tocilizumab - is a non-politicized example in Covid-19
In the beginning of the pandemic, we were giving tocilizumab (a arthritis drug that blocks activation of a substance in the immune system called interleukin-6) to a select group of patients with severe Covid-19 infection.
We had an expert-created set of guidelines about who to administer the drug to and under which clinical circumstances someone should receive it.
Our anecdotal experience was encouraging - some of these patients seemed to get better - and we were cautiously bullish on the prospects of this (very expensive!) potential Covid treatment. The non-randomized data were encouraging too.
But when studied in a randomized way, the drug failed to improve the outcome in Covid infection.
Again and again (and again), we see that being fooled by randomness is a real thing
Establishing a treatment without evidence to back it up is the medical equivalent of crossing the street without looking.
Just because it “makes sense” that a treatment will work based on our current understanding of the biology doesn’t mean that it will work in reality.
Armchair experts - or even real experts! - need to be cautious about extrapolating a theoretical mechanism of benefit to assume there is actual benefit without proper testing.
When it comes to Covid-19, hydroxychloroquine simply doesn’t work, no matter how many reports you hear that it does.
The same goes for Actemra/tocilizumab and Kaletra/lopinavir-ritonavir.
The road to hell is paved with biologic plausibility (and travels right alongside Twitter commentary).
PS: If you’re enjoying my newsletter, please subscribe and send to your friends and family!