Hacker News Reader

Comment by lutusp | Hacker News Reader

> If the drugs had no effect, then why are people (ab)using drugs like Adderall to improve cognition, etc?

You're making an assumption here -- that people's trust in, and use of, drugs argues that it's not a mere placebo effect. But there are any number of accounts of people persuading themselves that a drug is reliable, for a drug that is in fact a placebo. In fact, that's how the placebo effect is defined.

The only way to sort out these issues is with a double-blind prospective trial. Unfortunately, such trials are nearly nonexistent for this class of drugs, and those trials that have been conducted, are often run badly or published in very selective ways, creating the impression that the drugs are more effective than they really are.

Reference (PDF warning): http://alphachoices.com/repository/assets/pdf/EmperorsNewDru...

View on HN

Whoa, wait a second here. Your reference article is arguing that there isn't any difference between these drugs and placebos, so by definition it was a "placebo controlled trial". To say that "double-blind prospective trials are nearly non-existent" is very wrong. Unless you are referring to very old psychiatric drugs, most of them have gone through such trials.

There was a lot of disagreement about the paper you referenced. CNS drug are notoriously hard to get positive clinical trial data for. The reasons are: (1) placebo effect is notoriously difficult to control for in psychiatric trials, (2) the clinical trial endpoints are very blunt measures of efficacy.

But you know what? That's the best we have right now. There certainly is a strong requirement for efficacy in psychiatric drugs as evidenced by how many have failed to get approved by the FDA.

The sad thing is that most big drug companies have completely pulled out of doing R&D for new CNS drugs. They are just too hard to get approved. The end result is that patients with psychiatric illnesses will end up with fewer treatments options unless this changes.

> To say that "double-blind prospective trials are nearly non-existent" is very wrong.

You need to read each word and think about it. In a prospective trial, subjects are chosen in advance, they aren't drawn from an existing population of people who have already decided to seek psychological help.

A "double-blind, prospective trial" would have to sign people up from a cross-section of the population who have no interest in, or association with, psychological treatments. Then they would be divided into two groups. One group would receive the drug under test, the other would receive a placebo. And the trial would have to be designed so that the two groups wold not be able to tell to which group they belonged. Unfortunately, about the drugs under test, most of them have side effects, which means the subjects aren't blinded (the placebo group have no problem realizing who they are).

Even will all its defects and shortcomings, what I have just described is unheard of in psychological testing.

> Unless you are referring to very old psychiatric drugs, most of them have gone through such trials.

Absolutely false. The kind of design I am describing is simply unheard of. Most trials are of self-selected psychological treatment clients, and it's not uncommon for the "control group" to be what are euphemistically called "no-treatment controls". Guess what that means -- it means the experimental group is compared to people who are told to go home with no treatment and no sham treatment.

> There was a lot of disagreement about the paper you referenced.

Yes, of course -- and that's the way it should be. But there have been enough similar studies and meta-analyses that the outcome is not in question.

> But you know what? That's the best we have right now.

Different topic.

> There certainly is a strong requirement for efficacy in psychiatric drugs as evidenced by how many have failed to get approved by the FDA.

Most of the drugs that just were proven ineffective were not ever tested for efficacy as psychological drugs, they were prescribed off-label -- they were originally approved for some other purpose.

> The end result is that patients with psychiatric illnesses will end up with fewer treatments options unless this changes.

If the treatments were effective, that would be a shame. But they aren't, so it isn't.

Prospective study was probably not the correct term, since that's an epidemiological study. "Placebo controlled study" is probably a better term. And a placebo controlled study is the best evidence for a medical treatment - doing epidemiological studies to determine medication effectiveness doesn't seem right.

Consider an SSRI study for depression. I want to know if someone who is depressed and takes it does better thank someone who is depressed and doesn't take it, or takes a different drug. I don't care what it does in a healthy population, since they probably won't be taking it. I mean, the effects of SSRIs in healthy individuals is an interesting topic, but not relevant if I'm investigating depression treatments.

> And a placebo controlled study is the best evidence for a medical treatment ...

Yes, and with respect to the drugs we're talking about, subsequent meta-analyses show that the original studies were all biased in favor of their sponsors and all the drugs produce the same results, which argues for a massive placebo effect.

For years, studies were published that appeared to support various drug approaches to mental illness, but those studies have been invalidated by more careful analysis that included (as just one well-known example) unpublished studies that came to conclusions not favorable to the drug companies' outlook.

> Consider an SSRI study for depression.

Yes, let's do that. As things stand right now, studies of drugs that increase serotonin levels show the same effects as those drugs that decrease serotonin levels. This causes open-minded scientists to doubt the efficacy of this entire approach.

Reference: http://www.newyorker.com/online/blogs/elements/2013/09/psych...

Quote: "While S.S.R.I.s surely alter serotonin metabolism, those changes do not explain why the drugs work, nor do they explain why they have proven to be no more effective than placebos in clinical trials."

> I mean, the effects of SSRIs in healthy individuals is an interesting topic, but not relevant if I'm investigating depression treatments.

Such studies -- with the diagnosed depressed -- show no clinically significant result that can distinguish the outcome from the placebo effect. Those are the facts.

If SSRIs are no more effective than placebo, then:

* Why do some SSRIs work for some people while others don't?

* Why does increasing the dose of a specific SSRI have an increased effect?

Is it not possible that the real reason that SSRIs might be no more effective than placebo in clinical trials is that we don't quite understand the reasons that they work yet? (i.e. a specific SSRI might only have a real effect on 15% of the non-control group)

> If SSRIs are no more effective than placebo, then:

> * Why do some SSRIs work for some people while others don't?

The placebo effect. Surely you don't think everyone has the same placebo effect? Also, you're not describing the same study with the same participants and different drugs, you're describing different studies, different subjects, and different drugs -- all of which can change the outcome.

> * Why does increasing the dose of a specific SSRI have an increased effect?

The placebo effect. If people believe that a placebo is working, then more placebo produces more illusion.

Also, your use of the term "increased effect" is misleading. The largest doses show an effect, smaller doses show no effect. One theory about that is that the large-dose drug's side effects alert the subjects that they're not in the control group. A proposal has been made to create "active" placebos, placebos that mimic the side effects of the drug under test but not the property under test. This idea hasn't been tried yet.

> Is it not possible that the real reason that SSRIs might be no more effective than placebo in clinical trials is that we don't quite understand the reasons that they work yet?

You're overlooking the fact that placebos produce the same effects as the drugs under test. Our understanding or not understanding isn't an issue, because they've been demonstrated to be indistinguishable from a sugar pill.

And our not having a mechanism is definitely a criticism of this kind of treatment -- it would be like administering a mystery drug that proves essential to health, then deciding to increase the dosage, without understanding that the mystery drug is fatal in large doses (like vitamin A).

Wait a second, "prospective" trial does not mean that you draw patients from a pool who have no interest in treatment. Prospective simply means you have decided what the endpoints will be before the trial begins.

The inverse of a prospective trial is a retrospective trial where you have a set of data and you go looking for signals.

> Wait a second, "prospective" trial does not mean that you draw patients from a pool who have no interest in treatment.

Then it's a good thing that I never said or implied that. The subjects must be drawn from a uniform population, not from a population that have already come forward for treatment -- otherwise the result is biased by the experimental population at the outset.

> Prospective simply means you have decided what the endpoints will be before the trial begins.

No, it means the experimental design doesn't recruit from a pre-existing population of subjects with a history or established preferences, instead you choose random subjects from the population at large.

> The inverse of a prospective trial is a retrospective trial where you have a set of data and you go looking for signals.

There's nothing stopping you from actually learning what these terms mean. A prospective study recruits subjects from the population at large, people with no particular interest in, or knowledge of, the topic being studied. A retrospective study relies on self-selected subjects, people who are already engaged with what you want to study.

For example, let's say I want to study the link between daily vitamin use and intelligence -- do people who take a daily vitamin become more intelligent as a result?

I begin to design a study. Shall I enlist people who already take a daily vitamin (retrospective), or do I enlist people with no interest in the issue, divide them into two groups, make one of the two groups take a daily vitamin, and make the other group take a placebo (prospective)?

See the difference? In the retrospective, any conclusion I make may result from a confusion of cause and effect -- for example, if the vitamin-takers turn out to be more intelligent, that might result from a factor like intelligent people taking a vitamin because of their increased awareness of health issues, in other words, it might be an effect, not a cause.

The prospective experimental and control groups, one of which got a vitamin, and the other which got a placebo, is much more likely to produce a meaningful result.

Well you did a much better job describing how you define retrospective vs. prospective trial design. I still think the way you are defining it is not accurate, since retrospective trial designs have nothing to do with self-selection. That is a separate matter. All retrospective means is "looks back in time" at data that has already been collected.

However, you argument stills doesn't hold much water. How can a psychiatric drug be used retrospectively unless it has already been approved for use? If you're complaining about trial data on off-label use, then I won't argue, but the FDA does not accept retrospective trials to support drug approval.

Reply on news.ycombinator.com