In the more well researched followup by the same person, the author of Astral Codex, we see that the FDA did not block or slow the approval. It was that the manufacturer had no interest in even submitting a application. In fact, the FDA regularly granted usage exemptions, granted a broad exemption for research, and even directly funded one of the first trials even though they are not primarily a funding body.
The actual moral of the story maybe is that we should let third partys submit applications? It seems like there would be little risk of bad outcomes as they still need to do the approval process. It aligns incentives better as the people ultimately at risk can deploy resources comparable to their QoL improvement instead of the expected increased sales volume at current prices. There is a obvious problem that the production process is inextricably linked to the final output so a non-manufacturer has limited control and insight into a core component that introduces the possibility of regression, but it seems relatively doable to develop protocols for handling that if they do not already exist.
edit: To be fair, once the application was submitted it still took 6 years for approval. There are likely large procedural costs in doing the exemptions and testing that make inaction preferable. So, in some sense exemptions and lacking interest in approval may be problems of their own creation, but on net the overall outcome seems fairly reasonable. Actual improvements in procedural matters are likely too nuanced for a random layperson like myself to comment on.
> The actual moral of the story maybe is that we should let third partys submit applications?
The application isn't a single page form, it's hundreds to thousands of pages detailing everything about the clinical trials ranging from the success criteria for each phase, the doctors and hospitals involved in the trials, and even tiny details like how much blood is going to be drawn, when it will be drawn, and how it's going to be tested.
They're impossible to write without the participation of almost everyone involved in the process since a significant part of premarketing approval is quality control of the final drug manufacturing process.
Yes? I meant they should be allowed to run independent clinical trials and then submit for approval.
I also did point out that the integral nature of the manufacturing process is a obvious potential problem. It, however, does not seem like a insurmountable problem. The specifications on sourced components and the quality control processes applied to them is already integral to the drug manufacturing process. This would just move it one level further up where the final drug produced by some existing production process is a “component”. They also already approve drugs for different uses and evaluate existing drugs for new uses, so it does not seem that qualitatively different from some existing approval procedures. I am not sure of the exact details and am not proposing a specific solution, but it certainly does not seem to be a deal breaker.
Designing clinical trials from scratch is a very difficult and time consuming process so almost all of them are modeled after previous successful trials targeting the same class of medicine - as specific to the new drug as possible.
There's a lot of copy pasting followed by customization - often by "contract research organization" consultants who specialize in different types of clinical trials - but there are no "frameworks" as we would understand it in software engineering.
In the more well researched followup by the same person, the author of Astral Codex, we see that the FDA did not block or slow the approval. It was that the manufacturer had no interest in even submitting a application. In fact, the FDA regularly granted usage exemptions, granted a broad exemption for research, and even directly funded one of the first trials even though they are not primarily a funding body.
The actual moral of the story maybe is that we should let third partys submit applications? It seems like there would be little risk of bad outcomes as they still need to do the approval process. It aligns incentives better as the people ultimately at risk can deploy resources comparable to their QoL improvement instead of the expected increased sales volume at current prices. There is a obvious problem that the production process is inextricably linked to the final output so a non-manufacturer has limited control and insight into a core component that introduces the possibility of regression, but it seems relatively doable to develop protocols for handling that if they do not already exist.
edit: To be fair, once the application was submitted it still took 6 years for approval. There are likely large procedural costs in doing the exemptions and testing that make inaction preferable. So, in some sense exemptions and lacking interest in approval may be problems of their own creation, but on net the overall outcome seems fairly reasonable. Actual improvements in procedural matters are likely too nuanced for a random layperson like myself to comment on.