Ethics in medical trials

[steuard]

There's a thought-provoking article on medical ethics in today's New York Times: New Drugs Stir Debate on Basic Rules of Clinical Trials. The crux of the article is this: controlled clinical trials are the gold standard of medical evidence because they give the clearest evidence about whether one treatment is better than the alternatives. If we have no clear expectation for which treatment is more effective, then there's no ethical dilemma: the control group might well be better off. But the more compelling the evidence that one treatment will turn out to be superior, the greater the moral difficulty in refusing that treatment to the study's controls.

As the article points out, there have been times in the past when we've short-circuited the trial/approval process for treatments that seemed clearly better, only to learn later that they actually hurt survival rates. Careful testing really is important! And if you release a promising new treatment before formal testing, you may never be able to convince people to participate in a trial that could deny it to them. But none of that makes it any easier to look someone in the eye and tell them, "You have to die so we can make sure this new drug works as well as we think it does."

So there's the question: Is there any scientifically acceptable alternative to the usual 50/50 controlled trial that would reduce the ethical dilemma in cases like this?

Or to frame that in a concrete (if extreme) scenario: Imagine that every participant in a small Phase 1 trial took a single pill, and their advanced cancer immediately vanished and showed no sign of recurring after six months. What sort of full-scale trial would you demand as hard evidence that the pill was effective? How far from this extreme would the results have to be before you could accept a standard trial?

Comments

[feuervogel]

The control group for new treatments for HIV and cancer doesn't get placebo. They get current best practices/standard of care. A lot of studies for diseases where the ethics of giving no treatment are less clear-cut, like hypertension, are designed as non-inferiority, as well (ie, "is our new ACE inhibitor as good as this other one?" which results in way too many me-too drugs that don't have any clear advantages over the other ones.)

[ukelele.livejournal.com]

I finally got around to reading Nurture Shock, and one of the chapters concerns a preschool program so effective that attempts to do science on it keep getting short-circuited; either the sites switch the control groups wholly over to the new program (because it's so effective they feel it unethical to continue the control), or they lose their funding (because it was for helping at-risk kids, which their students no longer are).

[beth_leonard]

That's the exact thought that crossed my mind too. Why is it ok for pre-schools, but not for dying people?

I know the FDA is designed to protect us, but man, when it comes to terminal cancer and anyone given less than 1 year to live (I include the extreme elderly who want to experiment on themselves here too) I'm inclined to just let the companies and the sales reps have at it.

It will still take some promising results to convince people to try something other than "best standard of care," but you'll always find somebody willing to give it a shot. I'm thinking of the holistic crazies who will forgo chemo "best standard of care" treatments in favor of herbs they can buy from Whole PaycheckFoods. Enroll those people in the first round. They'll feel better about it too.

When you amass enough anecdote, then more and more mainstream people will pile on. The clinical trial isn't vs. people enrolled in a scientific study and denied promising treatments they want, it's vs. published survival rates for best standard of care.

Something like this shouldn't be done for OTC drugs, but when you look at the medical costs of treating cancer, and discover a large portion is the drugs, and a large portion of those drug costs are the clinical trials for people who are already at death's door, it just screams out for a better system.

We've got to have a better system, because people at death's door will pay ALL THEY HAVE for just a few more months of life, and that is not a good recipe for low costs of those treatments.

--Beth

[zathrus.livejournal.com]

OK, now I want to read this book. Now. I wonder if my library has it.

There are entirely too many dimensions along which parenting and teaching can be optimized. I can't even find time to read all the books about all of them, let alone time to implement them all....

Newt

[feuervogel]

This blog post may be relevant to your interests: http://scienceblogs.com/insolence/2010/09/balancing_scientific_rigor_versus_patien.php

[jon-leonard.livejournal.com]

The core question is really this: How do you know that the drug is worthwhile? That's a scientific question, and a statistics-heavy one at that. You need to set up your experiment so as to gain useful information, ideally with as low a cost in resources and human suffering as possible.

In this case, the article makes it sound as if the drug is already established as useful -- but then why hasn't the FDA approved it already? I'll happily admit that the FDA is overzealous in keeping drugs off the market, but as described, we really don't know yet that the drug's useful. And there are certainly drugs that look promising, but on further study don't turn out to be as useful as hoped, Thalidomide being probably the obvious example. But there are also things like a Parkinson's drug that does wonders in preventing plaque formation, at least for a few months; then the disease returns with a vengeance, no better off than untreated.

A non 50/50 trial might make sense: We have a pretty good idea how cancers behave with current treatments (at least, on average), so limiting the size of the control group might make sense. To be honest, this seems like a fairly straightforward problem in information theory: You can get a certain amount of information from each experimental subject (limited because of uncontrollable noisy factors), and presumably there's some threshold amount of information where you'd approve a drug. (But maybe information theory isn't that straightforward, and it doesn't seem to be in the MD curriculum.)