[zaroth]

Comparing old vs new is how it's done in cancer research. It's inhumane to prescribe placebo to someone in serious pain. You give them the old or the new, and they report how they felt.

In this case, since there is no concept of old or new, it's an entirely new class of treatment. So you give the current best-practice treatment without anti-fungal, and you give the same with anti-fungal.

Anyone in the study is agreeing at the onset that they have a 50/50 chance they will either get the current best practice, or the exact same thing but also the anti-fungal. These are volunteers who are helping move the anti-fungal research forward for everyone, in exchange for a chance of getting early access. They can also leave the study at any time after starting, but if they are in the anti-fungal group they would lose access to it if they did.

If that's not enough, there are cases (e.g. in cancer research) of de-blinding the study when there's a clear/strong positive effect and providing early access to all enrolled if it becomes obvious the new treatment is widely beneficial.

So I think we have the theory behind the ethics fairly well wrapped on how to roll out these drugs. I don't think in practice it works so well, the studies are often poorly run and poorly administered making them extremely more costly than they should be.

[protomyth]

The place where I worked did a pretty good job at administering the studies. The cost was in all the sampling and shipping. When your study is doing 1,000 sample kits with 5 samples every week day for multiple years, its going to cost.

I am still a bit clueless with all the research we have done with Alzheimer's why we need another control group?

[toast0]

The control group, with double blinds, provides evidence that the change in treatment causes the change in outcome. If you test without a control, the change in outcome could be a result of any other changes between the test protocol and previous control groups.

You might be able to do a small study without a control to test feasibility of the hypothesis, but I don't know if that would pass ethics review; would probably depend on the side effects of the anti-fungal. Another less invasive way would be to test for fungal presence in groups with and without the disease, likely through autopsies, since sampling a live brain seems intrusive.

[protomyth]

Here's my problem, we know the outcome of Alzheimer's, surly we have the data to go ahead without a control group? We've had them before.

[toast0]

Without the control group, you don't know if the change you think you made is the change that made a difference.

[zaroth]

I think it really comes down to the strength of the effect. If the anti-fungal has only a marginal effect on outcome, a large n control group helps tease out statistical significance. If the anti-fungal has a strong readily observable effect, the whole study gets un-blinded and everyone starts getting it pretty quickly anyway.

The reality is the control group and the test group are not the exact same people and often it's not even completely true everyone is dealing with the "exact same" disease (different genes, different mutations, different past treatment regimens, etc.). I definitely believe there are cases where a control is setup because "that's how it must be done" and not because it's going to actually provide any useful data at the end of the day to tell you anything about how the test group actually performed.

Another way I've seen this done which is a little less wasteful, and when you have high confidence in your new treatment, is to tweak the standard treatment marginally in some way you think might provide some small improvement in both groups, but in a way that could never justify its own full study. Like you modify the dosing schedule, add some vitimin, diet, or exercise change in some way you think will help both groups. But again you can only afford to do that if you are expecting a very strong effect from your new additive treatment which would blow away your tweak statistically, else you risk hiding a small benefit of the new treatment in your tweak.

In other words, you give every participant the best damn chance and the best care you can legally give them, and 50% of them know they are on something new which could be a blockbuster, and if it is, the whole group will get early access to it anyway.

[protomyth]

Ok, but don't we have a body of knowledge on Alzheimer's that having another control group is redundant? Why do we keep needing new control groups for things that are going to kill people and we know are going to kill people? I cannot get by the ethical problem of knowing we're kill 50% of the people in the study to provide data we already have.

[toast0]

The problem is, if something related to how you run your test of the anti-fungal has an effect on the disease, you're going to assume its the anti-fungal; but maybe it's not. If you have a control group, and both groups do better than expected (compared to the body of evidence), then you've learned something you wouldn't have without a control group (and you'll have to work hard to figure out what changed in this study vs other studies). If anti-fungals are a clear winner, it's likely that the study protocol would be changed to give everyone the drug after early results.

[DanBC]

The control is either "no treatment" or "treatment as usual".

Those have problems of blinding, and treatment vs no treatment is difficult to get through ethics committees.

[zaroth]

Is that 1,000 patients enrolled each being tested daily? That is a heck of a lot of testing!

Obviously depends on the patient population (in/out patient) but IMO a huge component of the cost is pushed off on patients having to spend a day getting to the study facility to give blood and urine versus dropping it at a local lab.

Then again, if Theranos tells us anything about variability of lab results, you certainly have to take every precaution to protect the integrity of your data.