Using patient experience data for drug approvals--SciAm article

https://blogs.scientificamerican.com/observations/a-streamlined-pathway-to-drug-approval/

Author advocates for using patient experience data and open trials in lieu of RCT.

 

Here is a link to the paper they published.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6478354/

 

Of course the example of rituximab comes to mind.

Imagine if we just went from the anecdotal / open-label data and made rituximab the standard treatment for ME/CFS. We decide we're so convinced it's effective that virtually every patient gets it. We may have never had any means of coming to the knowledge that it wasn't effective - of unfooling ourselves.

I understand that there are rapidly terminal or otherwise devastating diseases without approved treatment for which it may well be ethical to administer experimental treatments. And that there are diseases so rare that it's difficult to set up trials.

But if a treatment really is profoundly beneficial for some condition, it seems to me that it will prove out in well-designed trials, even if the trials don't look exactly like a typical textbook example. In my opinion, better to spend a few years, maybe even a decade or two, doing due diligence and get reliable results for posterity. It beats out hoping and guessing.

 

Both together definitely makes sense. This would solve the growing problem of EBM randomized (sometimes controlled) trials where the cherry-picked results completely contradict patient experience. The contradictions are simply buried and dismissed because it is allowed. This would put an end to this practice. So it would be good for patients, terrible for mediocre researchers. Unfortunately it's pretty obvious which priority is more important.

Although it's only a matter of time before machine learning and natural language processing allow freewheeling patient testimony to be analyzed, thematized and made into formal summaries, at low cost, without the human distortion of biased researchers doing selective reporting. I think it will naturally happen and it will be interesting to have access to anonymized comments from participants in psychosocial research.

 

I agree. There is plenty of flexibility within the format of obtaining reliable data from adequately controlled studies and if a drug really works it is not hard to show it quickly. You can do a watertight controlled study on a single patient with a chronic condition.

I think there is a mistake in the way the author conceives the alternatives. She talks of a
'path to drug approval that involves real-world, patient experience data' but most formal trials do that. The key issue is controlling for bias. That does not need a huge trial of 640 patients (where systematic bias will be even more troublesome). But it does need adequately controlled observation.