NEW Paper 2022! @vkprasadlab
Once again we analyze the use (& MISUSE) of the parachute analogy in biomedicine
This time we trace citations to the original BMJ 2004 paper & the Yeh remix!
Let me walk you through the paper..
[THREAD]
First, you need to know that a parachute has a huge effect size. Without one, falling from plane, u are sure but dead. With one, deaths are ~1 per 10 million jumps.
Effect size, ARR = 99.999....
AKA amazing. Such a huge effect size, you don't need any trial to know it works!
Second, know that many in medicine love to believe they have something just as good.
My chemotherapy drug -- it's a parachute
A cloth mask to prevent covid -- a parachute
The Impella or Aortic balloon pump - two parachutes
renal artery stenting - parachute
Of course, when people in medicine claim to have a parachute, they often are telling Tall tales
Renal artery stenting has multiple, negative RCTs. hardly seems like a parachute!
Cloth masking didn't work in the Bangladesh RCT, so.. not sure it is a parachute!
Doctors, public health experts or policy makers mostly cry "parachute" to avoid subjecting their favorite intervention to randomized testing.
Enter our study, we trace cites to 2 very famous papers which pushed the parachute analogy-- both half joking
We classified the context with which biomedical authors references these papers, and whether they named a SPECIFIC practice as a parachute.
First, these were the attitudes among biomedical researchers citing these papers. A few conceded RCTs gold standard, but others claimed that observational studies or common sense could supplant RCTs, as with parachutes themselves!
Second, authors analogized 21 specific practices to parachutes, but we found 2/3rds had been tested in RCTs already!
Moreover, in the 5 cases of success, this was the Absolute Risk Reduction
Recall, Parachutes were 99.999...%
7.2 - 44%. That is decent, but it ain't no parachute, get outta here!
We conclude simply, "The parachute analogy appears to be misused and misunderstood in biomedicine."
A final note...
In the decades to come post COVID, the greatest failure will be a failure to generate better data.
It is forgivable to be ignorant in March 2020, but we failed:
Zero randomized trials of school closure
Zero randomized trials business closures
Zero randomized trials on masking in schools
Zero randomized trials on masking kids
Zero randomized trials on masking in USA
Zero randomized trials of 3 vs 6 ft
Zero randomized trials of cohorting
Zero randomized trials of installing HEPA filters
Zero randomized trials of masking, post vaccine
Zero randomized trials testing alternative dosing strategies of vax (post phase 3)
Zero randomized trials of whatever foolishness we are doing on college campuses
Zero randomized trials of asymptomatic testing
Zero randomized trials of quarantine duration
Zero randomized trials of testing for preschool/ school
Zero randomized trials of plexiglass
Zero randomized trials of wearing your mask from the door to table, and table to bathroom, but at no other points during your restaurant experience
Yet many of these things were wrongfully called parachutes by at least some people who should know better
Leaving a pandemic as ignorant as we entered is awful
It is also why all of these issues have become political and not scientific
Science should be ashamed of itself
In 100 years we will look as primitive as the folks who survived the plagues of the middle ages
And...
Abusing the parachute analogy played a small part in the foolishness that we find ourselves in
This is an intriguing Q. A few hypotheses 1. Vax'd young healthy college kids have succumb to fear-mongering & have irrational worry about own breakthrough 2. This cognitive bias more common on Left, where most youth are... 1/5
3. Any student who protests is called ableist, eugenicist or white supremacist. Hyperbolic use of these terms is chilling for speech and demeans what these terms actually mean.
4. The culture of wanting to take retribution against someone for one little thing they said has reached an apogee, students want to take no risks.
Dazzling, daring, stunning piece by JH Russell and @DmpPatterson in @tabletmag tellin the true story of masking during COVID 👇
"The overselling of policies began in the early days of the pandemic with the problematic #masks4all movement...."
must read
🧵 tabletmag.com/sections/scien…
First, they are spot on. Many don't see it, but Trump is so influential, He sets the sides!
When he was for schools, Liberals lost their mind and worked to keep them closed; AAP did a bizarre flip flop & sided with unions to keep them closed.
And the same for masks 👇
Absolutely right again 👇
Inexperienced & naive twitter pundits kept saying the pandemic would vanish if we all worse a mask or n95 for 14, 21 days, a claim....
Wow! 👇
secondary endpoint
potential for infinite looks at data
yet still, scant events (50?) of dubious clinical value (some asx?)
Plus other issues, including durability discussed in video
But postponing/ flip flopping is going to poison vax confidence
White house should not have pressured FDA on boosters
Gruber and Kraus should still be there
Politics cannot dictate vaccine approval
This is a total fiasco
This administration is doing serious, serious damage to vaccine confidence
The backlash will hurt a lot of people
States will get rid of good programs to encourage routine childhood immunization
Vaccines become even more political
This administration does not have a plan here
They should have awknowledged natural immunity
and tried for just 1 dose in people with no prior NI
Lots of talk of wearing masks.
The policy question is: what happens when you advise people to:
Wear cloth masks
Wear surgical
Wear n95s
You have to combine mask properties with real world compliance
Ans:
No benefit
11% benefit in setting w no immunity & high bias rct
No trial
The 11% benefit should have an* because there is a serious risk of concealment being revealed in the study due to imbalance and study population by arm
If you pair this imbalance with the idea that people will be slightly less likely to report symptoms if they were enrolled on the margin. You have a big problem