Your first point about lack of funding for non-pharmaceutical interventions seems to me a pervasive problem in our capitalist society -- prevention doesn't make tons of money like drugs/devices do. Moreover, sociologically, it seems it is hard to "get credit" for *preventing* illness as opposed to *curing* illness (this actually extends to events like financial crises as well, I think). This I think is also a tension between the "public health" community, which thinks more in terms of populations and prevention, and the "medical" community, which tends to think in terms of individuals and cures. Any thoughts on how to bridge this gap, both at the professional level as well as more broadly in society?
You're the best writer on these issues by a whole order of magnitude, because you can understand and communicate so well across academic disciplines. Happy to subscribe.
The gold standards of N95/P100 masks are used to protect medical personnel. Why not protect everyone? A useful study would investigate the utility of covering the exhalation ports of the industrial versions of N95/P100 masks with 'add-on' filters of varying filtration parameters. There must be something waiting at the end of the lock down 2.0 tunnel that will protect the user in the general population. Semi source-control with face coverings has and will fail. Although somewhat counter intuitive, personnel protective equipment is probably the best 'source-control'. If I don't get it, I can't give it to anyone.
Excellent post, thank you. In my upcoming Phil of Science class I'll certainly be using (more) Covid-19 related examples when arguing that we shouldn't always demand RTCs. Since I am from Denmark I'd like to add a few details on the much debated Danish RTC on masks. It is of serious concern that at least one of the authors of the study claims that it gets rejected from top-journals because the results are controversial - while hinting that the study does not provide evidence for masks working. However, while the sample size is 6000 (sounds large, eh?!) it was always going to be too small. They seem to have expected 2% of the population to become infected in the 30 days they were tracking participants. Even with such an incidence rate they would still only have been able to show an effect size of 50% (based on a simple power analysis). So, even in this hypothetical situation they planned for, they could only have found a positive effect if masks protected the wearer really well. A well-designed study had realized that mask experts wouldn't have expected such a big number.
Now, in fact only 0.003% of the population got a positive test in Denmark in May (when the study ran) - even if we assume lots of undetected infections and multiply this with 16, and thus assume 0.05% got infected, they would no one (out of 3000!) in the mask condition to be infected, to show an significant result.
So, the research design is just flawed, and we should keep this in mind, whenever they do get published somewhere. I could add additional details, e.g. the fact that they haven't tested if masks work, but only tested if it works to send a bunch of masks to 3000 people, send them a weekly reminder, and then hope they actually wear the mask and rely on their survey based answers to track that (I could go on).
Briefly put: Running RTCs on masks doesn't work for many reasons, but even when you manage to run a study that sounds big (6000 people), you still face challenges that make your results far less impactful than intended.
Ouch that sounds like terrible power (in terms of analysis power for those not statistically-inclined). I almost wish they would publish it somewhere so we could put an end to the speculation that it showed something but that's why it's being suppressed.
2.1% infected in non-mask group, 1.8% in mask-group. These infection numbers are really high, and certainly higher than in the general population. So, the sample was particuar.
On RCTs, I just mention the COVID-NMA website that maintains a database of trials about Covid-19 treatments and vaccines. You can see there a dashboard of treatment trials and several other charts :
There are two things that have hurt the development of RTCs, at least in Europe :
1. The communication and disinformation over HCQ as a proven cure, which made many patients wanting to be prescribed HCQ instead of being a part of a trial.
2. Once the first wave was over in Europe, there wasn't enough patients to be enrolled in trials anymore.
Unfortunately the second point didn't apply to the US, and doesn't hold anymore in Europe...
It is a catastrophic failure, in my opinion, that we did not start randomization on day one. It should have been part of the pandemic playbook, and should have been done through hospital coordination even if the CDC wasn't on board, and Europe had even less excuses except not being ready.
Your first point about lack of funding for non-pharmaceutical interventions seems to me a pervasive problem in our capitalist society -- prevention doesn't make tons of money like drugs/devices do. Moreover, sociologically, it seems it is hard to "get credit" for *preventing* illness as opposed to *curing* illness (this actually extends to events like financial crises as well, I think). This I think is also a tension between the "public health" community, which thinks more in terms of populations and prevention, and the "medical" community, which tends to think in terms of individuals and cures. Any thoughts on how to bridge this gap, both at the professional level as well as more broadly in society?
"the failings of our expert communities and how they relate to society" - this is happening again with rapid antigen tests: so many experts saying how it's a "bad idea", and media reporting "missed cases". Exactly like the mask situation where the community benefits is played down or even ignored. Here's a recent headline: https://www.theguardian.com/world/2020/nov/05/operation-moonshot-rapid-covid-test-missed-over-50-of-cases-in-pilot
I know! The argument that we cannot trust people or explain things properly is so pervasive in public health that it's really terrible.
You're the best writer on these issues by a whole order of magnitude, because you can understand and communicate so well across academic disciplines. Happy to subscribe.
The gold standards of N95/P100 masks are used to protect medical personnel. Why not protect everyone? A useful study would investigate the utility of covering the exhalation ports of the industrial versions of N95/P100 masks with 'add-on' filters of varying filtration parameters. There must be something waiting at the end of the lock down 2.0 tunnel that will protect the user in the general population. Semi source-control with face coverings has and will fail. Although somewhat counter intuitive, personnel protective equipment is probably the best 'source-control'. If I don't get it, I can't give it to anyone.
Excellent post, thank you. In my upcoming Phil of Science class I'll certainly be using (more) Covid-19 related examples when arguing that we shouldn't always demand RTCs. Since I am from Denmark I'd like to add a few details on the much debated Danish RTC on masks. It is of serious concern that at least one of the authors of the study claims that it gets rejected from top-journals because the results are controversial - while hinting that the study does not provide evidence for masks working. However, while the sample size is 6000 (sounds large, eh?!) it was always going to be too small. They seem to have expected 2% of the population to become infected in the 30 days they were tracking participants. Even with such an incidence rate they would still only have been able to show an effect size of 50% (based on a simple power analysis). So, even in this hypothetical situation they planned for, they could only have found a positive effect if masks protected the wearer really well. A well-designed study had realized that mask experts wouldn't have expected such a big number.
Now, in fact only 0.003% of the population got a positive test in Denmark in May (when the study ran) - even if we assume lots of undetected infections and multiply this with 16, and thus assume 0.05% got infected, they would no one (out of 3000!) in the mask condition to be infected, to show an significant result.
So, the research design is just flawed, and we should keep this in mind, whenever they do get published somewhere. I could add additional details, e.g. the fact that they haven't tested if masks work, but only tested if it works to send a bunch of masks to 3000 people, send them a weekly reminder, and then hope they actually wear the mask and rely on their survey based answers to track that (I could go on).
Briefly put: Running RTCs on masks doesn't work for many reasons, but even when you manage to run a study that sounds big (6000 people), you still face challenges that make your results far less impactful than intended.
Ouch that sounds like terrible power (in terms of analysis power for those not statistically-inclined). I almost wish they would publish it somewhere so we could put an end to the speculation that it showed something but that's why it's being suppressed.
Results from the Danish facemask study will be released at 2 pm UK time today, 18th of November.
2.1% infected in non-mask group, 1.8% in mask-group. These infection numbers are really high, and certainly higher than in the general population. So, the sample was particuar.
On RCTs, I just mention the COVID-NMA website that maintains a database of trials about Covid-19 treatments and vaccines. You can see there a dashboard of treatment trials and several other charts :
- Global dashboard : https://covid-nma.com/dataviz/
- Delay of trials : https://covid-nma.com/research_delay/
- Tested treatment : https://covid-nma.com/treatments_tested/
There are two things that have hurt the development of RTCs, at least in Europe :
1. The communication and disinformation over HCQ as a proven cure, which made many patients wanting to be prescribed HCQ instead of being a part of a trial.
2. Once the first wave was over in Europe, there wasn't enough patients to be enrolled in trials anymore.
Unfortunately the second point didn't apply to the US, and doesn't hold anymore in Europe...
It is a catastrophic failure, in my opinion, that we did not start randomization on day one. It should have been part of the pandemic playbook, and should have been done through hospital coordination even if the CDC wasn't on board, and Europe had even less excuses except not being ready.